pmid stringlengths 8 8 | pmcid stringlengths 8 11 ⌀ | source stringclasses 2
values | rank int64 1 9.78k | sections unknown | tokens int64 3 46.7k |
|---|---|---|---|---|---|
32372006 | PMC7200721 | pmc | 7,658 | {
"abstract": "Plant growth promoting rhizobacteria can improve plant health by providing enhanced nutrition, disease suppression and abiotic stress resistance, and have potential to contribute to sustainable agriculture. We have developed a sphagnum peat-based compost platform for investigating plant-microbe interactions. The chemical, physical and biological status of the system can be manipulated to understand the relative importance of these factors for plant health, demonstrated using three case studies: 1. Nutrient depleted compost retained its structure, but plants grown in this medium were severely stunted in growth due to removal of essential soluble nutrients - particularly, nitrogen, phosphorus and potassium. Compost nutrient status was replenished with the addition of selected soluble nutrients, validated by plant biomass; 2. When comparing milled and unmilled compost, we found nutrient status to be more important than matrix structure for plant growth; 3. In compost deficient in soluble P, supplemented with an insoluble inorganic form of P (Ca 3 (PO 4 ) 2 ), application of a phosphate solubilising Pseudomonas strain to plant roots provides a significant growth boost when compared with a Pseudomonas strain incapable of solubilising Ca 3 (PO 4 ) 2 . Our findings show that the compost system can be manipulated to impose biotic and abiotic stresses for testing how microbial inoculants influence plant growth.",
"conclusion": "Conclusion Washing compost significantly reduces soluble nutrient pools, including N, P and K, without affecting its structure or pH. We show that nutrient levels can be manipulated to simulate different nutrient deficiencies and study the potential for microbial inoculants to overcome these deficits. The compost provides a standardised testing system defined in some of its biological, chemical and physical attributes. This defined system creates a platform for detailed and complex plant-microbe interactions to be studied and the simple manipulation of nutritional status provides a reliable screening platform for PGPR. The system is closer to conditions found in the field than the commonly used inert mineral substrates or hydroponics, so we believe that results are more relevant for sustainable agriculture. It has great potential as a platform to provide rigorous and reproducible pre-field testing of commercial microbial inoculants to replace or complement mineral fertilizers.",
"introduction": "Introduction Novel and optimised alternatives to artificial fertilisers, which can be deleterious to the environment, are required to achieve sustainable intensification of agriculture 1 . Plant growth promoting microbes found at the root-soil interface – part of the root microbiome – have an underexploited arsenal of capabilities that can improve plant health, growth and nutrient status 2 . The network of interactions between microbial partners and the host plant is complex and not well understood. This knowledge is essential for the production of plant growth promoting bioinoculants, which are predicted to be an important part of the solution to feeding the world’s growing population 3 – 5 . Plant growth promoting rhizobacteria (PGPR) have been shown to increase acquisition of many nutrients and improve plant health and productivity where nutrients may be limiting. These nutrients include iron 6 , phosphorus 7 and potassium 8 . The ability of PGPR to mitigate abiotic stresses such as drought 9 , salt 10 or alkaline soil 11 has been demonstrated in many crop plants. Potato plants exposed to all the aforementioned stresses in combination survived when inoculated with two Bacillus species 12 . The mechanism of tolerance, induced by the bacteria, was shown to be an increase in reactive oxygen species scavenging enzymes and an improvement in photosynthetic performance 12 . Pseudomonas species are important PGPR, overcoming drought stress in maize 13 and often possessing many of the characteristic traits of plant growth promotion: production of siderophores, indole-acetic acid and ammonia 14 . Microbiome research is also focusing on the potential of beneficial microbes to improve agricultural resilience and soil health, enhancing practices such as phytoremediation 15 , the creation of disease suppressive soils 16 , and reducing dependency on artificial fertilisers 17 . Bacterial inocula have been shown to reduce incidence of disease across a broad range of both fungal and bacterial diseases, enhanced further when mixed inocula are used 18 . In order to assess the potential benefits of PGPR there is a need for the development of a defined testing platform. Previous work has used simplified inert substrates, such as sand, glass beads, vermiculite or perlite, but these are less conducive to root growth and are a much less intricate and realistic matrix than soil 7 , 19 . Conversely, the use of real soil is often unsuitable due to inherent difficulties in altering soil nutrients, without also affecting soil physical and biological status, making the importance of each individual component for plant health difficult to quantify. We argue that sphagnum peat moss compost, hereafter referred to as compost, provides a suitable substrate to test factors that influence plant growth. Other systems are either too simple to replicate in vivo conditions well enough to test inoculants, such as vermiculite, or too complex to manipulate nutrient levels, such as soil. Compost is an irregular, organic matter matrix with absorption and adsorption properties that provides chemical complexity and an excellent structure for plant growth whilst being easy to manipulate and implement stresses. Tomato plants grown in substrates containing peat showed improved growth compared with using other media such as sand and perlite 20 , and growth of daisies was also improved when vermiculite was supplemented with compost and sphagnum moss peat 21 . Compost provides the optimal structure for plant growth, allowing the effect of implemented stresses or microbial inoculants to be measured. Many experiments use in vitro functional screening assays to predict the ability of bacteria to promote plant growth via biotic and abiotic stress resistance. While this is the accepted method, it is important to be able to test that these abilities can be replicated in an in planta system 22 . For example, genomic screening of bacteria for plant growth promoting genes is a useful indicator of potential PGPR function. However, plant growth promoting genes are often not expressed in planta , perhaps due to an incompatible plant-microbe interaction. It is therefore necessary to test potential inoculants in vivo to confirm their plant growth promoting efficacy 23 . In order to develop a system to test microbial interactions in planta , it is essential to be able to define the physical matrix and soluble nutrient status of the system, and to understand how changing these factors affect plant growth. In this work we describe a defined compost system and demonstrate the use of this medium with three case studies: 1) Compost nutrient status – removal and reconstitution; 2) Compost physical structure status – the effect of pore size on plant growth; 3) The deployment of microbial inoculants, with the example of a P solubilising bacterial isolate, to mitigate soluble-P deficiency.",
"discussion": "Discussion Nutrient limited compost system As expected, washing the compost reduces soluble nutrient availability and plants subsequently grown in this medium have a significantly reduced biomass. Neither the bulk density or pH of the compost changes significantly after washing so the reduction in above ground growth was attributed to the removal of many of the soluble nutrients, primarily N, P and K (Supplementary table S2 ). In order to reconstitute the nutrient levels of the washed compost system, two different nutrient solutions were trialled to find the most efficient and successful recovery of plant growth. It is essential that nutrient levels can be easily manipulated to simulate different nutrient deficiencies for use in experiments with plant growth promoting microbes. Hoagland’s solution provided highly effective and reliable recovery of plant growth when used in the phosphate solubilisation bioassay (Supplementary table S1 ). This method involved daily watering with a prescribed dose of Hoagland’s solution, allowing nutrients to be replaced gradually. Although Hoagland’s solution can easily omit P, the recipe adds many different sources of N, in combination with other micronutrients. As such, this method is not suitable for mineral N depletion experiments. Removing all sources of N would mean many other essential micronutrients that constitute Hoagland’s solution would be eliminated. The second nutrient solution tested was a modified version of Letcombe’s solution for wheat (Supplementary table S1 ). One major advantage of this approach is that the nutrients are given as one single dose at the beginning of the experiment and thereafter plants are watered with tap water, significantly reducing the workload. Reconstituting the compost with the exact amount of N and P calculated to have been removed by washing was insufficient to fully recover plant growth. In order to support plant growth to the same biomass level as that of the unwashed compost, it was necessary to add 5X the original amount of nutrients. It is likely that this can be attributed to the loss of fine particles and their associated nutrients during the washing process. These particles have a high surface area:volume ratio and the mineralisation of organic nutrients to inorganic soluble forms will be reduced in washed compost. Two further concentrations were also tested (10X and 20X), but these had no further effect on plant growth and biomass indicating that the system was replete with nutrients. The nutrient status of the washed compost is more important than pore size for plant biomass (Fig. 2 ). Washing compost removes nutrients and results in a pronounced reduction in plant biomass. However, when comparing wheat grown in unwashed compost, milling resulted in a reduced wheat biomass, whereas for washed compost milling had no effect on wheat biomass. It may be the case that under nutrient replete conditions that the reduced biomass seen with milling could be attributed to reduced oxygenation of the soil, water logging and compaction affecting root structure 33 , but this is secondary to the effect of nutrient depletion. The compost system is a more intricate, ecologically complex matrix than for example, sand or hydroponics. This will mean simulations can be closer to those found in the field – the translation of experiments from compost to field may be more realistic than, for instance, in vitro experiments or those using an inert mineral substrate such as vermiculite. Additionally, real soil is not suitable for experiments such as these as it is difficult to manipulate their biotic, abiotic and physical states. The development of soils depleted in particular nutrients would take many years to achieve, whereas the compost system presented here can be readily manipulated for the control of these factors. The main application of this system is in testing microbial interactions and their impact on plant health. The ability of microbes, either as singular inoculants or as synthetic communities, to enhance plant growth can be screened in a standardised, easily manipulated system that can simulate many environmental conditions: nutrient deficiencies, altered physical structure, a reduced microbial pool and abiotic stresses such as drought or flooding. Influence of phosphate-solubilising pseudomonads on the aerial biomass of wheat in P-deficient compost The P solubilising bacterial strain was chosen as it had the highest PSI score. The non-solubilising bacterial strain was chosen as it was not able to solubilise P but was the same species, came from the same host and was morphologically similar to the P solubilising strain. When wheat was inoculated with phosphate-solubilising bacteria ( Pseudomonas sp.), under depleted soluble P condition but supplemented with recalcitrant P, it had increased above ground biomass relative to wheat inoculated with a non-solubilising Pseudomonas species, or a no bacteria control. This suggests that the P solubilising Pseudomonas sp. was able to liberate P from inorganic, insoluble Ca 3 (PO 4 ) 2 , promoting plant growth. The inclusion of a no bacteria control treatment addresses the potential issue of pH difference between agar and compost systems influencing Ca 3 (PO 4 ) 2 solubilisation. The in vitro screen used for P solubilisation is at pH 7, whereas the compost system has a pH ~6. As the compost substrate is slightly more acidic than PVK agar, the recalcitrant source of P could be more soluble in the compost system, making PO 4 in the Ca 3 (PO 4 ) 2 more readily available to the plant. However, our data shows that plants inoculated with bacteria unable to solubilise P and plants grown with no bacterial inoculant have a similar biomass ( p = 0.8), and the addition of the P solubilising bacterial inoculant increases wheat biomass relative to the other treatments ( p = 0.02). This demonstrates that it is the addition of Psol that benefits the growth of wheat, either directly or indirectly, in the compost system. Isolates screened in vitro for their ability to solubilise phosphate may not replicate this in planta . The use of PVK media and Ca 3 (PO 4 ) 2 as the recalcitrant source of phosphate is well documented 34 , but soils contain many kinds of metal-phosphate sources and other organic P sources, so screening for other P solubilisation mechanisms in the future should also be considered, to replicate conditions found naturally in soil 35 . Rhizosphere competence, the chemical, biological and physical properties of the growth medium, and microbial competition are all factors that may influence the potential of an inoculant to perform the same functions in planta that are observed in vitro . Isolates screened using this technique, including Enterobacter and Burkholderia sp. - both displaying a high phosphate solubilising ability in vitro - yielded contrasting results in planta , either increasing or having no effect on the biomass of sorghum respectively 36 . Our experiment used compost supplemented with Ca 3 (PO 4 ) 2 , thus in vitro screening using the same phosphate source was appropriate. For use in natural soil systems, the use of mixed communities with a variety of PGPR functions and phosphate solubilisation abilities is likely to give more success in field situations. The use of plant growth promoting bioinoculants is an exciting avenue for exploration of alternatives to mineral fertilisers in agriculture. Previously published work focusing on phosphate solubilising microbes has conflicting results. For example, pea plants had increased biomass when inoculated with Pseudomonas sp., using Ca 3 (PO 4 ) 2 as the recalcitrant P source 7 , and the growth of walnuts was improved by separate inoculation with Pseudomonas chlororaphis and Pseudomonas fluorescens , previously screened for phosphate solubilisation using PVK media 37 . The ability to manipulate the washed compost system means high-throughput screenings of plant growth promoters under specific conditions, such as nutrient depletion, pathogen challenge or drought, are possible in a consistent and defined matrix."
} | 3,877 |
37110944 | PMC10145448 | pmc | 7,659 | {
"abstract": "Electrospinning is a process to produce versatile nanoscale fibers. In this process, synthetic and natural polymers can be combined to produce novel, blended materials with a range of physical, chemical, and biological properties. We electrospun biocompatible, blended fibrinogen:polycaprolactone (PCL) nanofibers with diameters ranging from 40 nm to 600 nm, at 25:75 and 75:25 blend ratios and determined their mechanical properties using a combined atomic force/optical microscopy technique. Fiber extensibility (breaking strain), elastic limit, and stress relaxation times depended on blend ratios but not fiber diameter. As the fibrinogen:PCL ratio increased from 25:75 to 75:25, extensibility decreased from 120% to 63% and elastic limit decreased from a range between 18% and 40% to a range between 12% and 27%. Stiffness-related properties, including the Young’s modulus, rupture stress, and the total and relaxed, elastic moduli (Kelvin model), strongly depended on fiber diameter. For diameters less than 150 nm, these stiffness-related quantities varied approximately as D −2 ; above 300 nm the diameter dependence leveled off. 50 nm fibers were five–ten times stiffer than 300 nm fibers. These findings indicate that fiber diameter, in addition to fiber material, critically affects nanofiber properties. Drawing on previously published data, a summary of the mechanical properties for fibrinogen:PCL nanofibers with ratios of 100:0, 75:25, 50:50, 25:75 and 0:100 is provided.",
"conclusion": "5. Conclusions The goal of the present study was to determine the mechanical properties of electrospun fibrinogen:PCL nanofibers as a function of fiber diameter and fibrinogen:PCL ratios. A combined atomic force microscopy/optical microscopy technique was used to determine the mechanical properties of the electrospun hybrid fiber. Extensibility, elasticity, and fast and slow relaxation times depended on the fibrinogen:PCL ratio, as their values increased when the PCL ratio increased from 25% to 75%. Stiffness-related properties, including breaking stress, Young’s, total, and relaxed, elastic moduli strongly depend on the fiber diameter, in addition to the fibrinogen:PCL ratio. Below a fiber diameter of 150 nm, the value of these properties strongly increased, and with decreasing diameter, above 150 nm, the diameter dependence leveled off. These data complement previous data on electrospun, blended fibrinogen:PCL nanofibers, thus completing a library of mechanical properties for fibrinogen:PCL nanofibers with ratios ranging from 100:0 to 0:100. The diameter dependence of stiffness-related properties is intriguing and future studies should focus on investigating the molecular mechanisms of this observation. Additionally, the blended fibers should be tested in biological and other applications.",
"introduction": "1. Introduction Electrospun nanofibers have gained prominence in recent years due to their versatility and unique properties. Large surface area to volume ratios, nanoscale size, and a wide range of physical and biochemical properties make electrospun fibers an attractive material for various fields such as tissue engineering [ 1 , 2 , 3 ], medication delivery [ 4 ], textile manufacture [ 5 , 6 ], filtration [ 7 , 8 ], and clean energy (batteries, solar panels, fuel cells) [ 9 , 10 ]. Although several techniques exist to generate ultra-thin fibers, electrospinning is one of the most economical and straightforward processes. Electrospinning offers several advantages, including ease of use, scalability, and adjustability [ 11 ]. This technique utilizes a high electric field to produce fibers on the nanoscale using polymer solutions of synthetic or natural polymers [ 12 , 13 ]. It allows control of fiber diameter, mesh pore size, and surface morphology [ 14 , 15 ], and, if desired, the fibers may be infused with additional small molecules. Furthermore, electrospinning enables the creation of diverse structures, including hollow [ 16 ], core-shell [ 17 ], multilayer [ 18 ], and nanowires [ 19 ], providing great versatility in the nanofiber design for various demands in the applications [ 20 ]. Several factors affect electrospinning, including solution composition, processing parameters (flow rate, electric field strength), and ambient conditions (temperature, humidity) [ 21 , 22 , 23 , 24 , 25 , 26 ]. Understanding and adjusting these parameters allows the production of nanofibers that meet the requirements of specific applications. Over the past years, natural polymers such as collagen, fibrinogen, and elastin were successfully electrospun to nanofibers for potential uses such as tissue engineering scaffolds, wound dressings, and various other biomedical applications [ 27 , 28 , 29 , 30 , 31 ]. A growing interest exists in creating new materials by blending natural polymers with synthetic ones [ 32 , 33 ]. Blending polymers can produce new materials with distinctive structural, mechanical and biochemical properties. Studies showed that blending natural and synthetic polymers can improve mechanical stability, as natural materials are often weaker than synthetic ones [ 34 , 35 , 36 ]. Blending synthetic and natural polymers may result in the formation of either covalent or physical bonds between the polymer chains [ 37 , 38 ]. These interlinks may create a strong and stable network structure for some polymers that can significantly enhance the mechanical properties of the resulting material. However, there may also be compatibility problems for other polymers resulting in weaker materials. Further, blending synthetic materials with bioactive proteins may endow nanofibers with biological functionality [ 39 , 40 , 41 ]. Fibrinogen is a soluble protein with a molecular weight of 340 kDa and is primarily found in the blood plasma. Fibrinogen’s principal function is forming fibrin fibers, which provide a mechanical and structural scaffold for blood clots at the site of an injury to a blood vessel [ 42 ]. Electrospun fibrinogen nanofiber-based scaffolds have been successfully fabricated for potential tissue engineering applications [ 28 , 43 ]. Even though pure fibrinogen fibers have excellent properties regarding interactions with cells, such as supporting cell adhesion, proliferation, and differentiation, some applications may require altered mechanical properties [ 44 ]. Although fibrinogen possesses appealing characteristics such as high biocompatibility, biodegradability, and the capacity to mimic the natural extracellular matrix, its use in biomedical applications has received less attention compared to other natural polymers like collagen, elastin, and chitosan. Moreover, only a few studies have investigated the potential for blending fibrinogen with synthetic polymers. Hence, further research is needed to fully understand the potential of fibrinogen as a biomaterial and optimize its properties for specific applications. Polycaprolactone (PCL) is a synthetic polyester polymer with a partially crystalline structure and a low melting point (~60 °C). PCL has been widely used in biomedical applications due to its biodegradability, biocompatibility, mechanical stability, and relative softness [ 45 , 46 ]. Moreover, electrospun PCL networks can mimic the structure of the native extracellular matrices (ECM) [ 47 , 48 ]. Hence, blending fibrinogen with PCL may produce a new biomaterial that meets mechanical and biochemical scaffold design requirements. Polymeric materials must fulfill several requirements to be suitable as a scaffold for biomedical applications. Among these requirements, mechanical properties are essential when fabricating scaffolds. The ideal scaffold should mimic the physical and chemical structure of the microenvironment of the ECM. Moreover, it should possess enough mechanical strength and integrity to withstand various forces when handled or implanted. For example, electrospun scaffolds for tissue engineering vascular grafts may serve as a conduit for blood flow. Thus, it must sustain the forces put on it without rupturing or being permanently deformed as in an aneurysm [ 49 ]. Until recently, little research was done in determining the mechanical properties of single electrospun nanofibers. Previous research mostly investigated the mechanical properties of the entire scaffold or mat. The mechanical properties of scaffolds strongly depend on several factors such as network porosity, single fiber properties, fiber junctions, fiber distribution, and the alignment of the fibers [ 50 , 51 , 52 ]. Obtaining precise information on how various factors affect scaffold mechanical strength is challenging. Prior knowledge of the modulus and strength of the single fibers forming the mat is critical to build a mechanical model of scaffolds. The atomic force microscope (AFM) is one of the most suitable tools for the mechanical analysis and topographic characterization of soft matter, such as polymers, biological materials (proteins, DNA, cells), colloids, nanoparticles, and soft materials, such as gels, elastomers, and hydrogels [ 53 ]. A microindentation test can be conducted using AFM, which involves applying a controlled load to a small probe tip and measuring the resulting indentation depth on the sample’s surface. The force–indentation depth curve obtained from microindentation testing can provide information about the material’s mechanical properties, such as the elastic modulus, hardness, and toughness [ 54 ]. In addition, the atomic force microscope (AFM) can perform tensile tests on soft materials by connecting the AFM tip to the sample and then stretching it until it breaks while simultaneously measuring the force and displacement. The resulting force–displacement curve obtained from tensile testing can provide valuable data about the material’s tensile strength, strain at break, and elasticity under tension. We used a combined AFM and optical microscope to investigate the mechanical properties of electrospun, blended 75:25 and 25:75 fibrinogen:PCL nanofibers. The current work expands the results of previous studies and in total, we have now investigated the mechanical properties of electrospun fibrinogen:PCL nanofibers with ratios of 0:100, 25:75, 50:50, 75:25 and 100:0 [ 55 , 56 , 57 ]. Taken together, this work provides a library of mechanical properties for dry, electrospun fibrinogen:PCL nanofibers.",
"discussion": "4. Discussion We employed the electrospinning technique to fabricate blended fibrinogen:PCL fibers with two different ratios (75:25) and (25:75). These hybrid fibers had diameters ranging from about 40 nm to 600 nm. Several mechanical properties, including maximum strain and stress, moduli, energy loss and elasticity of the electrospun fibers, were determined using a nanofiber pulling technique based on a combined AFM/inverted optical microscope. A summary of the total findings is given in Table 1 and Table 2 . As a part of our experimental procedure, we glued the nanofibers to the ridges in the substrate with epoxy to prevent them from slipping when pulled by the AFM probe. Due to the tendency of fibers with high PCL content to slip, as reported before [ 59 ], it was necessary to anchor these fibers so that the mechanical properties could be measured accurately. In some cases, the epoxy leaked into the grooves or along the fibers; these fibers were not used for measurements. We allowed the glue to cure at ambient temperature for at least 24 h before manipulation to ensure they were adequately fixed to the ridges. Drawing on data from previous work [ 55 , 56 , 57 ], Table 3 summarizes the key mechanical properties of electrospun fibrinogen:PCL fibers with ratios of 100:0, 75:25, 50:50, 25:75, 0:100. This provides a fuller picture of how the different ratios affect nanofiber properties. Pure PCL fibers display the largest extensibility at 133%, which drops slightly and linearly with increasing fibrinogen concentrations to 120% (25:75) and then 110% (50:50). As the fibrinogen:PCL ratio increases to 75:25, the extensibility strongly decreases to 63%, less than half the extensibility of electrospun pure PCL fibers, as seen in Figure 6 . This strong decrease is somewhat surprising as the extensibility of pure fibrinogen increases again to 110%. One explanation for this trend in the extensibility could be that PCL dominates blended nanofiber properties for ratios of 50:50 and higher. At these higher PCL concentrations, the PCL polymer chains may still be able to interact with each other and are entangled with each other. For the 75:25 fibrinogen:PCL fibers, the PCL polymers may lose their ability to interact with each other and their entanglement is diminished. The increased extensibility and the high modulus of the pure fibrinogen fibers, indicates that the interactions between fibrinogen polymers in the pure fibrinogen fibers are strong, and likely mediated through long flexible chains. These results also suggest that the fibrinogen polymers and PCL polymers may not interact as well with each other as PCL–PCL interactions and fibrinogen–fibrinogen interactions. The effect of the PCL ratio on fiber extensibility we observed agrees with the trend seen in previous reports. Miele et al. found that doubling the weight ratio of PCL in a mixture of PCL and collagen increased strain at the failure from 12.8 ± 0.8% to 63 ± 9% [ 62 ]. Mobarakeh showed that the elongation of PCL/gelatin nanofibers significantly increased from ~35% to ~170% when the weight ratio of PCL increased from 50% to 70% [ 63 ]. The latter two studies evaluated the mechanical properties of aligned fiber scaffolds rather than single fibers. Like the extensibility, the elasticity of the blended fibers also increased when the PCL fraction in fibrinogen:PCL fibers was increased. The 25:75 fibrinogen:PCL fibers were permanently deformed between 18 ± 6% and 40 ± 14% strain. In contrast, the deformation of the 75:25 fibrinogen:PCL fibers occurred between 12 ± 4% and 27 ± 10% strain. Sharpe et al. reported a much lower elastic limit for 50:50 fibrinogen/PCL fibers [ 56 ]. They found that the 50:50 fibers could be stretched to 5 ± 5% strain before permanent damage was seen. It is not clear why the elastic limit of the 50:50 fibers was much smaller compared to the other ratios, although we followed the same fiber preparations and the same experimental method. However, comparing our findings for blended fibers to pure fibrinogen and pure PCL fibers, the elastic limit of the blended fibers with two different ratios is within the elastic limit range of pure PCL (24%) and pure fibrinogen (15%), as seen in Table 3 . The stiffness (Young’s modulus) of the fibrinogen:PCL fibers strongly depends on the fibrinogen:PCL ratio. At first, we will compare large diameter fibers, ignoring the strong diameter dependence of the small diameter nanofibers. Electrospun, pure fibrinogen fibers are stiffest with a modulus of 4200 MPa, and electrospun pure PCL fibers are softest with a modulus of 380 MPa. The modulus of blended fibers generally decreases with deceasing fibrinogen. The modulus is dominated by the PCL properties for the 0:100 and 25:75 fibrinogen:PCL fibers with both having a modulus around 300 MPa, which is similar to the modulus of bulk PCL [ 64 ]. The modulus increases by a factor of about 5 for the 50:50 fibers, and by another factor of about 3 for the pure fibrinogen fibers (100:0). In the 75:25 fibrinogen:PCL fibers there may be a compatibility discrepancy between fibrinogen and PCL, which is why the trend of increasing modulus is broken for these fibers. Another key finding is that the Young’s modulus, the rupture stress, and the total and relaxed, elastic moduli (from incremental stress–strain curves) strongly depend on fiber diameter. Fibers with smaller diameters are significantly stiffer than fibers with larger diameters and stiffer than bulk (for PCL), as seen in Table 2 . A power law fit to the data yields an exponent of approximately −2 for fibers with a diameter less than about 150 nm; above this diameter, the moduli approach a constant value. This diameter dependence is unusual because the modulus of a homogeneous material is a material constant that should be independent of the dimensions of the material. Intriguingly, this diameter dependence may be a general phenomenon for electrospun and some natural nanofibers. It was recently reported for pure electrospun PCL fibers [ 57 ], for which the Young’s modulus strongly increased as the diameter decreased below 100 nm; however, fibers with a diameter greater than 100 nm showed a weak dependence on diameter. It was also seen in natural fibrin fibers [ 65 ]. For the electrospun, pure fibrinogen fibers and for the 50:50 fibrinogen:PCL fibers, the diameter dependence was not investigated. The diameter dependence may be explained by (at least) two different models. In one model, the nanofiber does not have a uniform cross-sectional density, but instead it has a higher density in the center (core) of the fiber, which decreases toward the periphery. Such a model was invoked by Li et al. to explain the diameter dependence of the modulus for fibrin fibers [ 65 ], which form the mechanical and structural scaffold of a blood clot. These authors proposed a fibrin fiber model in which the protofibril density of the fiber decreases with increasing diameter, D , as Y ∝ D − 1.6 . Similarly, this model was also used by Alharbi et al. to explain the diameter dependence of electrospun PCL fibers [ 57 ]. In a second model, the increases in the modulus are due to the higher surface-to-volume ratio of smaller fibers compared to larger fibers, and it assumes that the polymer chains in the surface region of the fiber are much more oriented and aligned than the fibers in the core [ 66 ]. Our data cannot conclusively distinguish between these two models, and more studies are needed to validate molecular models. For instance, experiments that could determine the density and alignment of polymers within nanofibers as a function of diameter might be able to shed light on the molecular origins of the observed diameter dependence of the modulus. In cyclic stress–strain curves, all blended fibers with different ratios showed a monotonic increase in energy loss with increasing strains. The strong dependence of dissipated energy on strain has been previously reported for several natural and electrospun fibers, such as fibrin fibers, collagen, and electrospun PCL [ 57 , 67 , 68 ]. For electrospun PCL fiber, the energy loss increased from 38% at small strain (10%) to 66% at high strain (70%). We can estimate the loss energy of electrospun fibrinogen fibers from the elastic limit reported in [ 55 ] to be 14% at 3% strain to roughly 65% at 13% strain. Significant energy loss at small strain has also been observed for the electrospun collagen, as the dissipated energy at 12% strain was 80%. Our findings showed that the energy loss increases from 28% at a small strain of <10% to 76–85% at larger than 70% strain for (75:25) and (25:75) fibrinogen:PCL fiber, respectively. The energy loss seems to reach a plateau at a strain of 30–40% with an energy loss of 72–68%. The large energy loss indicates a large viscous component of the deformation. Besides the fact that these data contribute to building a library that principally focuses on the mechanical properties of single electrospun fiber, the data also provide insight into the mechanical behavior of hybrid fibers, which will help to optimize and design materials suitable for use as scaffolds in many biomedical applications. Future work should focus on understanding how the mechanical properties of individual fibers contribute to the mechanical behavior of meshes made up of these fibers. Additionally, the diameter dependence of the modulus is an intriguing novel property of nanofibers and warrants further exploration."
} | 4,969 |
37738354 | PMC10516493 | pmc | 7,661 | {
"abstract": "Agriculture is driving biodiversity loss, and future bioenergy cropping systems have the potential to ameliorate or exacerbate these effects. Using a long-term experimental array of 10 bioenergy cropping systems, we quantified diversity of plants, invertebrates, vertebrates, and microbes in each crop. For many taxonomic groups, alternative annual cropping systems provided no biodiversity benefits when compared to corn (the business-as-usual bioenergy crop in the United States), and simple perennial grass–based systems provided only modest gains. In contrast, for most animal groups, richness in plant-diverse perennial systems was much higher than in annual crops or simple perennial systems. Microbial richness patterns were more eclectic, although some groups responded positively to plant diversity. Future agricultural landscapes incorporating plant-diverse perennial bioenergy cropping systems could be of high conservation value. However, increased use of annual crops will continue to have negative effects, and simple perennial grass systems may provide little improvement over annual crops.",
"introduction": "INTRODUCTION Most pathways to addressing climate change require expanded use of bioenergy ( 1 ). However, widespread adoption of bioenergy crop feedstocks can transform landscapes, strongly affecting ecosystems and biodiversity ( 2 – 5 ). The direction and magnitude of these changes will depend on which cropping systems are adopted, which types of ecosystems they replace on the landscape, and how these crops are managed. There are currently competing visions for which crops could be grown and how they should be incorporated into managed landscapes. In the United States, corn is a dominant crop and is the business-as-usual option for producing ethanol from grain, with ~40% of harvest currently allocated to biofuel production ( 6 ). However, expanding the footprint of corn or other intensive cropping systems dismantles natural ecosystems, reduces biodiversity, and undermines key ecosystem services that agriculture ultimately depends on ( 4 , 7 – 9 ). Corn ethanol may also be more carbon intensive than gasoline after emissions from land use change are taken into account ( 5 ). Given these sustainability concerns, low-input perennial crops are a compelling alternative to annual systems as a bioenergy source. Biodiversity in perennial biofuel crops is usually higher than in arable crops ( 8 , 10 , 11 ). In addition, in general, adding perennial features to agricultural landscapes can promote valuable ecosystem services locally and at landscape scales by increasing structural complexity ( 8 , 12 – 16 ). Still, perennial biofuel systems tend to be less biodiverse than natural and seminatural reference systems ( 2 , 11 , 17 ), so their net effects on ecosystems and biodiversity (in addition to their net carbon balance) will depend both on characteristics of the crops in question and on which ecosystems they replace. In some contexts, biofuel crops could replace natural or seminatural systems, whereas in others, they may replace arable crops, particularly on land where traditional crops are unprofitable ( 18 – 21 ). We address two important knowledge gaps concerning biodiversity in bioenergy crops. First, there are many potential cropping systems that could become widely adopted, but there have been virtually no empirical experiments carried out to compare candidate cropping systems simultaneously and measure how they affect local biodiversity. Instead, most of our ability to forecast comes from literature reviews and meta-analyses composed of narrower studies ( 2 , 10 , 17 ), databases ( 11 ), or simulations ( 22 ). These syntheses have shown us that perennial and/or second-generation bioenergy crops tend to be more biodiverse than annuals and that both types of systems are less biodiverse than natural systems. However, they are not well suited to make detailed comparisons between perennial crop types. Second, the taxonomic scope of past work on this topic has been limited, often focusing on one to two animal groups per study ( 23 – 28 ), but available evidence suggests that different lineages of organisms can respond inconsistently to cropping systems ( 2 , 10 , 11 , 17 ). We also have only a nascent understanding of soil microbial communities across different bioenergy crops despite their roles governing key ecological processes ( 29 – 36 ). As a result, our ability to make broad statements about the impacts of candidate bioenergy crops on biodiversity and ecosystem functioning, particularly within the wide range of low-input perennial systems that could be adopted, is limited. To address these knowledge gaps, we conducted broad biodiversity censuses in 10 bioenergy cropping systems grown in a well-established long-term experimental array (28 m–by–40 m plots, five replicates for each crop). We surveyed a spectrum of crop and grassland types that could be widely adopted or expanded in North America, including three intensive annual systems (corn, sorghum, and sorghum with winter cover crop), four simple grass-based perennial systems ( Miscanthus , mature and newly establishing switchgrass stands, and a native prairie grass mix), and three complex perennial polycultures (reconstructed prairie, successional vegetation, and short-rotation poplar). Within the annual systems, corn is currently the dominant North American bioenergy feedstock; energy sorghum has been developed as an alternative, having yields that approach corn but with greater resource use efficiency ( 37 , 38 ). Among the simple perennial systems that we censused, Miscanthus × giganteus is a perennial and fast-growing sterile hybrid grass that produces dense, high-yielding bamboo-like thickets ( 37 , 38 ), and switchgrass ( Panicum virgatum ) is a perennial prairie grass native to North America from which several high-yielding varieties have been developed for bioenergy ( 39 , 40 ). We censused communities in two switchgrass systems, including mature stands originally seeded in 2008 and newly establishing stands that contained dead cover crop residue ( Trifolium pratense ) in the spring and increasing switchgrass cover as the season progressed. We also censused a native grass mix containing switchgrass plus four native perennial grasses that are characteristic of prairies in the U.S. Midwest. Among the complex perennial polycultures that we sampled, reconstructed prairie plots contained a seeded mix of 18 native prairie species (6 grasses, 9 forbs, and 3 legumes) plus limited volunteers. Successional vegetation plots contained unmanaged volunteer vegetation, and short-rotation coppice poplar plots contained Populus NM-6 ( Populus nigra × Populus maximowiczii ), with a diverse and mostly unmanaged understory. Poplars were in their third year of growth during data collection. Within each of the 10 cropping systems, we profiled macro- and microorganismal groups including plants, bees, butterflies, ground beetles, ants, birds, small mammals, prokaryotes, fungi, and microeukaryotes (i.e., eukaryotes that are not plants, animals, or fungi; fungi were censused separately). We used a mix of in field identification, trapping, and metagenomics based on environmental DNA. We had three objectives. First, we compared taxonomic richness of all organismal groups, as well as abundance of each animal group, across the 10 cropping systems. We expected these community attributes to vary across systems, and beyond the general pattern of higher diversity in perennial crops than in annuals, we expected to observe strong variation across the different perennial systems. Second, we measured community dissimilarity for each taxonomic group, partitioned into nestedness and turnover. These are independent sources of dissimilarity and are relevant to biodiversity conservation because they indicate whether communities differ because they contain unique species assemblages or because one is a species-poor subset of the other. Last, we examined the gradient of plant diversity and composition that these cropping systems comprise and demonstrate its role in shaping diversity of other groups of organisms. We expected this relationship to take the form of positive correlations between plant richness and that of other groups of organisms.",
"discussion": "DISCUSSION Our findings illustrate the widely contrasting effects that bioenergy crop expansion could have for biodiversity. For many animal groups, the strongest differences in species richness were not between annual and perennial crops but between the different types of perennial crops. Specifically, biodiversity gains in simple perennial systems relative to corn were often minimal and were dwarfed by those in complex perennial systems with more plant species ( Fig. 1 ). Therefore, dichotomizing between annual and perennial bioenergy systems can overlook the glaring biodiversity differences that occur between different types of perennial bioenergy habitats. We also found that for several groups, particularly bees, butterflies, and ants, communities were highly nested ( Fig. 4 ). That is, species-poor communities in low-quality habitats were merely subsets of communities found in more diverse crops rather than containing unique species. We interpret this as meaning species-poor communities in simpler cropping systems contain extreme habitat generalists and/or species that arrive through local and landscape spillover but would be unlikely to persist using resources in those habitats alone ( 41 ). We conclude that while simple perennial cropping systems offer many important benefits over annual systems, such as reduced input requirements and increased carbon storage, nutrient, and soil retention, the best outcomes for conservation of macroorganisms lie in complex perennial bioenergy habitats with higher plant diversity. Patterns of richness in microbial communities differed from those of plants and animals, indicating that different factors shape the dispersal and persistence of microorganisms. In most (but not all) instances, richness was higher in perennial systems than in corn, but unlike for plants and animals, it did not increase sharply in complex perennial systems compared to simple ones. Prokaryote responses in the soil matrix and phyllosphere were opposite most other groups and decreased relative to corn in many of the treatments, suggesting that drivers of bacterial diversity differ from those of other groups. Sources of community dissimilarity also contrasted between micro- and macroorganisms, as microbial community differences were much more strongly driven by turnover than nestedness in most cases ( Fig. 5 ). This study used an experimental field array instead of field-scale cropping systems. This allowed for intensive, standardized, simultaneous side-by-side comparisons that eliminated variation in field size, landscape species pool, and physical geography. It also allowed us to evaluate crops that have not yet been widely deployed on the landscape and to include realistic harvest practices. Because plots in the array were at a smaller spatial scale than what is agronomically realistic, the differences that we measured between crops are likely to be conservative. We base this on three lines of reasoning. First, communities in small patches and field edges are partly made up of organisms donated from nearby habitats ( 41 ). If we had censused larger fields, then the influence of cross-habitat spillover would likely be reduced and differences in richness and community composition would be more pronounced. Second, large fields in aggregate produce coarse-grained landscapes, which offer limited resource complementarity for organisms that forage across multiple habitat types ( 42 ). In contrast to this, in our study, we observed (for example) bees nesting in open soil or crop residue in corn and sorghum but nectaring on flowers that occurred only in other habitats. At realistic scales, resource complementarity between habitats would be reduced. Third, when widely deployed in a landscape, crops can strongly shape the species pool of that landscape ( 43 ), and in simple landscapes dominated by species-poor crops, many species would be absent altogether. This type of effect was outside the scope of our experiment, which tested whether various crops were being used or colonized by different subsets of a single landscape-scale species pool. Last, we note that for some taxonomic groups with limited dispersal ability, the crop types in this experiment were relatively spatially independent from one another, whereas for others (e.g., bees, butterflies, and birds), the array functioned as a choice experiment in which all crop types were available, but they were more likely to visit and be detected in cropping systems that contain resources that they use. Within a crop type, management intensity and intracrop variation can strongly shape biodiversity outcomes. Richness of many groups in our experiment was highest in poplar; short-rotation coppicing systems can be quite biodiverse, but in practice, this depends on management, understory vegetation, and growth stage of the planting ( 44 – 48 ). In our experiment, poplar plots contained a diverse understory of volunteer plants, which may or may not occur depending on management and landscape context. We expect that coppicing systems with fewer plant species in the understory would not support the diverse animal communities observed in our study. Communities in coppicing systems also change strongly over time, particularly as the canopy closes and after harvest, and at landscape scales, diversity can be optimized by staggering stand age and timing of coppicing ( 47 ). Along the same lines, the switchgrass stands in our study were treated with broadleaf-specific herbicides and contained almost no forbs. If management had allowed more weeds to persist, then richness of some groups probably would have been higher. Past studies of switchgrass found that animal diversity levels were closer to prairie than what we measured here, but often, a large amount of plant biomass in those switchgrass treatments was made up of other species ( 49 , 50 ). Last, in annual systems, cover crops can provide a range of ecosystem services and generally enhance biodiversity ( 51 ). However, we found no evidence that cover crops enhanced biodiversity in the sorghum treatments in this study, and for a few groups (e.g. ants), abundance decreased where cover crops were used. We suspect that the increased disturbance associated with planting and terminating the crop outweighed any potential benefits. In general, much more research is needed to design grass- and woody-based management systems that optimize both bioenergy yield and biodiversity gains. Microbial richness responses were quite eclectic, differing strongly depending on the kingdoms and habitat compartments that we sampled. Recent meta-analyses show that richness and abundance of soil fungi and prokaryotes are generally positively correlated with plant species richness ( 52 , 53 ). In our study, fungal richness conformed to this trend in both soils and roots, but prokaryotes did not. We are unsure of why this difference occurred, but there is considerable variability in the plant-microbial diversity correlation that can occur because of differences in habitat type and sampling scale ( 53 ), and some evidence suggests that fungi may respond more strongly to plant diversity than prokaryotes do ( 52 ). The lack of effect on prokaryote richness could also have to do with the relatively small spatial scale of our experimental array or because turnover in species and functional groups occurred without strong changes in total richness. Breaking each microbial group down into the many functional guilds that they include will yield important additional information about how microbial diversity and function vary across these plant communities. Pathways to biodiverse bioenergy landscapes Bioenergy adoption can cause adverse land use change in which croplands are expanded, natural habitats are destroyed, and landscapes are simplified ( 54 – 56 ). Most evidence indicates that converting any seminatural system to bioenergy crops will incur biodiversity costs even if the crops are perennial ( 2 , 17 ). While we did not census natural reference habitats such as hardwood forests in this study, the costs of converting this type of ecosystem to bioenergy crops are likely to be high both in terms of biodiversity impact and carbon debt. Our results show that the biodiversity cost of converting seminatural habitats to biofuels will depend strongly on crop type. For example, converting successional grasslands on abandoned farmland to miscanthus would result in strong biodiversity loss. However, converting them to higher-yielding prairie or woody coppicing systems could have neutral or positive effects for biodiversity. In some contexts, demand for bioenergy could enhance biodiversity and ecosystem services in agricultural landscapes if diverse perennial systems are adopted. Farmers have increasing access to precision tools like spatially explicit yield monitoring and profitability mapping, allowing them to identify subfield areas that consistently lose money, contribute disproportionately to soil loss and nutrient pollution, and would be better suited to perennial crops ( 18 , 19 ). Converting unprofitable subfield areas minimizes indirect land use change by removing crops from the least productive areas; precision techniques could also compensate for yield losses incurred if they increase yield in other parts of fields due to enhanced management (i.e., variable rate fertilization). Adding perennial bioenergy habitat in this manner would increase local biodiversity and enhance the broad suite of ecosystem services that occur when perennial elements are incorporated into arable fields ( 15 , 16 ). It would also add both compositional and configurational complexity to simplified landscapes, bringing about improvements to biodiversity and ecosystem services at larger spatial scales ( 8 , 13 , 14 , 43 , 57 , 58 )."
} | 4,545 |
36271396 | PMC9587672 | pmc | 7,662 | {
"abstract": "Background Process and function that underlie the assembly of a rhizosphere microbial community may be strongly linked to the maintenance of plant health. However, their assembly processes and functional changes in the deterioration of soilborne disease remain unclear. Here, we investigated features of rhizosphere microbiomes related to Fusarium wilt disease and assessed their assembly by comparison pair of diseased/healthy sequencing data. The untargeted metabolomics was employed to explore potential community assembly drivers, and shotgun metagenome sequencing was used to reveal the mechanisms of metabolite-mediated process after soil conditioning. Results Results showed the deterministic assembly process associated with diseased rhizosphere microbiomes, and this process was significantly correlated to five metabolites (tocopherol acetate, citrulline, galactitol, octadecylglycerol, and behenic acid). Application of the metabolites resulted in a deterministic assembly of microbiome with the high morbidity of watermelon. Furthermore, metabolite conditioning was found to weaken the function of autotoxin degradation undertaken by specific bacterial group ( Bradyrhizobium , Streptomyces , Variovorax , Pseudomonas , and Sphingomonas ) while promoting the metabolism of small-molecule sugars and acids initiated from another bacterial group ( Anaeromyxobacter , Bdellovibrio , Conexibacter , Flavobacterium , and Gemmatimonas ). \n Video Abstract Conclusion These findings strongly suggest that shifts in a metabolite-mediated microbial community assembly process underpin the deterministic establishment of soilborne Fusarium wilt disease and reveal avenues for future research focusing on ameliorating crop loss due to this pathogen. Supplementary Information The online version contains supplementary material available at 10.1186/s40168-022-01375-z.",
"conclusion": "Conclusion In this study, metabolites that were more abundant in diseased soils shaped the mechanisms by which microbial communities assemble and correspondingly the community compositions. The characteristics of the soils treated with these exudates were similar to natural diseased soils surveyed from many locations and cropping systems. This was consistent both in terms of the relative abundances of the distinguishing taxa and with regard to deterministic processes driving community assembly. Soils with these characteristics exhibited a higher disease incidence of fusarium wilt in watermelon. Together, our study revealed inherent differences in the composition of diseased and “healthy” rhizosphere microbiomes and identified dominant rhizosphere metabolites that drove the assembly of metabolite-responsive microbial groups contributing significantly to the characteristics of a diseased rhizosphere microbiome. This study provides a theoretical framework for the underlying causes in the establishment of a “diseased-state” rhizosphere microbial community that informs future control of fusarium wilt disease.",
"discussion": "Discussion In this study, we combined global bacterial high-throughput sequencing data of fusarium wilt rhizosphere-associated soil samples from multiple independent studies and crops for the identification of microbial community characteristics associated with disease. Lower bacterial community diversity was associated with disease, in concert with earlier findings [ 20 ]. Based on network analyses, a low number of connections were associated with the diseased network, reflecting less robust microbe-microbe interactions within the community. Previous studies also associated disease with lower connectivity in microbial networks. For instance, the number of network edges decreased in a fusarium wilt diseased microbial network in banana [ 21 ], and more connections were present in a network associated with healthy rhizosphere soils rather than diseased samples [ 22 ]. The presence of Kaistobacter , Mesorhizobium , Bacillus , Anaeromyxobacter , Bdellovibrio , Conexibacter , and Flavobacterium in the diseased samples was determined as the microbial feature that distinguished the diseased rhizosphere microbiomes. However, the majority of the top 50 most abundant microbial taxa identified through cross-validation were also more abundant in diseased rhizosphere soils than in healthy soils. This indicates that a diseased microbiome may have more uniform characteristics than that of a healthy and diverse microbiome. Diseased communities also exhibited lower variation in community composition among samples, compared to larger variations exhibited by the healthy samples. This infers a homogenization effect associated with biotic stress from fusarium wilt disease that is similar to the impact of abiotic stresses such as drought and salinity [ 23 ]. This homogenization effect served as a basis to examine the rhizosphere microbiome assembly processes under fusarium wilt disease pressure. Four basic processes (diversification, dispersal, selection, and drift) can contribute to microbial community assembly [ 24 ] and subsequently can be used to describe the microbial assembly process under different environmental scenarios [ 25 – 27 ]. In this study, we explored the assembly process of the rhizosphere microbial community under fusarium wilt disease versus that in “healthy” soils. We found that variable selection process dominated in diseased rhizosphere bacterial communities, while stochastic processes dominated the assembly process within healthy sample microbiomes. This suggests the presence of a strong microbial selection pressure within the diseased plant rhizosphere. Recent advances in metabolomics have greatly advanced our understanding of plant-microbe interactions. Within the rhizosphere soil, plants exude organic metabolites to support microbial activity and, in turn, receive beneficial services from soil microbes [ 28 ]. A multistep model for root microbiome assembly from bulk soil has been proposed and supported with rice [ 29 ] and grapevines [ 30 ]. Dynamic root exudate profiles were associated with microbial community assembly patterns in a reference plant: wild oat ( Avena barbata ) [ 21 ]. These interactions appear to be two-way, as microbiomes were shown to condition soils by reprogramming root exudation profiles [ 31 ]. Specific root exudates have been associated with F. oxysporum disease spread in Lisianthus [ 20 ]. In this study, rhizosphere metabolites differed between diseased and healthy samples across multiple sites. Five metabolites (tocopherol acetate, citrulline, galactitol, octadecylglycerol, and behenic acid), enriched in the diseased rhizosphere soil, were considered to be “key” components that drove microbial community assembly. Several lines of evidence in literature indicate that these metabolites are associated with biotic stresses. Among these, citrulline has been found to be enriched in plants when exposed to multiple stressors [ 32 ] and also in the rhizosphere of fusarium wilt-diseased watermelon [ 33 ]. Tocopherol acetate is a member of the vitamin E family and is increased in host plants under multiple stresses [ 34 ]; meanwhile, behenic acid was enriched in sesame upon salinity stress [ 35 ]. We found higher abundances of tocopherol acetate and behenic acid in diseased rhizosphere soils, which appeared to be vital to the deterministic process of assembly within the diseased microbial community in our validation experiment. Thus, we suggest that the enrichment of some exudate constituents may be a common response of host plants to biotic and abiotic stresses. Nonetheless, not all rhizosphere metabolites alter microbial community assembly processes. For example, in a previous study [ 36 ], four organic acid exudates from cucumber (citric acid, pyruvate acid, succinic acid, and fumarate) were shown not to affect the microbial community assembly process (Supplementary Fig. 14 ). Collectively, the enrichment of five selected metabolites here within diseased rhizosphere soils may play an important role in the process of rhizosphere microbial community assembly as well as plant susceptibility to disease. We found the ability of autotoxin degradation was decreased in diseased rhizosphere soil, which would be one of the mechanisms of disease happening as previous studies have shown that autotoxin accumulation would cause continuous cropping obstacles by nutrient imbalance and microbial dysfunction [ 7 , 37 , 38 ]. We further found that these functional abilities decline due to the decrease of relative abundance within Bradyrhizobium , Streptomyces , Variovorax , Pseudomonas , and Sphingomonas . These bacterial groups have been reported to play multiple beneficial roles, such as antibiotics production, root colonization, and ISR activation, to maintain plant health [ 3 , 39 – 42 ]. Conversely, the metabolic of “readily available carbon” (small-molecule sugars and organic acids) was significantly enriched in diseased soil. These could promote the readily available carbon metabolism and, thus, increase the emergence and abundance of pathogens [ 43 , 44 ]. Small-molecule organic acids could help plants defend against Fusarium wilt in several ways, such as pathogen growth inhibition [ 45 ], resistance improvement [ 45 ], and beneficial microorganisms recruitment [ 46 ]. However, the enhancement of organic acids metabolism leads to the weakness of the above potential beneficial function. Five feature microbes, Anaeromyxobacter , Bdellovibrio , Conexibacter , Gemmatimonas , and Flavobacterium , were the main contributors to small-molecule organic acids metabolism. Previously, all of the five bacterial groups have been uncovered in soil and/or rhizosphere environments, with one, Conexibacter , even being recognized as a pathogen [ 47 ]. Hence, both FM1 and FM2 may have important ecological roles in maintaining the health status of plants. However, further research is needed to verify the roles of these “potentially important” species in maintaining plant health or in the formation of the fusarium wilt-diseased microbial community."
} | 2,523 |
34868130 | PMC8632824 | pmc | 7,663 | {
"abstract": "Renewable fuels are needed to replace fossil fuels in the immediate future. Lignocellulosic bioenergy crops provide a renewable alternative that sequesters atmospheric carbon. To prevent displacement of food crops, it would be advantageous to grow biofuel crops on marginal lands. These lands will likely face more frequent and extreme drought conditions than conventional agricultural land, so it is crucial to see how proposed bioenergy crops fare under these conditions and how that may affect lignocellulosic biomass composition and saccharification properties. We found that while drought impacts the plant cell wall of Sorghum bicolor differently according to tissue and timing of drought induction, drought-induced cell wall compositional modifications are relatively minor and produce no negative effect on biomass conversion. This contrasts with the cell wall-related transcriptome, which had a varied range of highly variable genes (HVGs) within four cell wall-related GO categories, depending on the tissues surveyed and time of drought induction. Further, many HVGs had expression changes in which putative impacts were not seen in the physical cell wall or which were in opposition to their putative impacts. Interestingly, most pre-flowering drought-induced cell wall changes occurred in the leaf, with matrix and lignin compositional changes that did not persist after recovery from drought. Most measurable physical post-flowering cell wall changes occurred in the root, affecting mainly polysaccharide composition and cross-linking. This study couples transcriptomics to cell wall chemical analyses of a C4 grass experiencing progressive and differing drought stresses in the field. As such, we can analyze the cell wall-specific response to agriculturally relevant drought stresses on the transcriptomic level and see whether those changes translate to compositional or biomass conversion differences. Our results bolster the conclusion that drought stress does not substantially affect the cell wall composition of specific aerial and subterranean biomass nor impede enzymatic hydrolysis of leaf biomass, a positive result for biorefinery processes. Coupled with previously reported results on the root microbiome and rhizosphere and whole transcriptome analyses of this study, we can formulate and test hypotheses on individual gene candidates’ function in mediating drought stress in the grass cell wall, as demonstrated in sorghum.",
"introduction": "Introduction As fossil fuel resources decline and usage continues to have adverse effects on global climate, it is crucial to develop alternative fuel sources to supply energy demand. Lignocellulosic bioenergy crops have the advantage of being renewable energy resources with reduced carbon emissions compared to petroleum-based fuel. Recent studies have demonstrated that the use of lignocellulosic biofuels has the potential to sequester carbon from the atmosphere by increasing soil organic carbon, which may help mitigate increased carbon release to the atmosphere ( Vanholme et al., 2013 ; Taptich et al., 2018 ; Wu et al., 2018 ). Despite these advantages, lignocellulosic biofuels have several requirements to meet before they can be considered competitive with petroleum-based fuels. The foremost requirement is cost competitiveness, which is primarily dependent on the recalcitrance of the cell wall and the ability to valorize waste products from biofuel production. Cell wall recalcitrance can often be simplified to several key factors: the amount and complexity of lignin, the amount and crystallinity of cellulose, the presence of biomass conversion inhibitors such as acetate, and the ratio of C5:C6 sugars, with a low ratio tending to favor saccharification efficiency and conversion into fuels ( Somerville et al., 2010 ; Bosch and Hazen, 2013 ; van der Weijde et al., 2013 ; Kumar and Sharma, 2017 ; Kim, 2018 ). In addition to biomass conversion efficiency, lignocellulosic feedstocks should ideally be grown on land that is not suitable for food crops to avoid displacing food crops. A major factor that makes cropland marginal is low rainfall and/or increased drought. With irrigation water being a severely limited, expensive, and energetically intensive resource, lignocellulosic feedstocks must be robust enough to generate high biomass yields even under adverse conditions like drought. In addition to high biomass yields, the lignocellulosic properties of the feedstocks would ideally not be altered by drought ( Somerville et al., 2010 ; van der Weijde et al., 2013 ; Vanholme et al., 2013 ). The impact of drought on the plant cell wall-related genes has been investigated in transcriptomic and proteomic studies ( Wu et al., 2001 ; Vincent et al., 2005 ; Zhu et al., 2007 ; Sasidharan et al., 2011 ; Le Gall et al., 2015 ; Nakano et al., 2015 ; Tenhaken, 2015 ; Lenk et al., 2019 ). Over the last several years, researchers have also performed basic chemical analyses of plant cell walls produced under drought conditions. However, these studies are limited in several ways. First, most transcriptomic studies lacked cell wall analysis to verify that transcriptomic changes translated into chemical changes in the wall. Secondly, much of the chemical cell wall analyses that were performed were either on specialized tissue types, from multiple plant lineages, or were under conditions unlikely to be encountered outside of a laboratory ( Iraki et al., 1989 ; Wakabayashi et al., 1997 ; Piro et al., 2003 ; Konno et al., 2008 ; Leucci et al., 2008 ; Rakszegi et al., 2013 ). Thirdly, more recent studies analyzing compositional and sugar yield differences use techniques that do not distinguish between different polysaccharides nor are they coupled to transcriptomic analyses ( Emerson et al., 2014 ; Ottaiano et al., 2017 ; van der Weijde et al., 2017 ; Hoover et al., 2018 ). These variables can make it incredibly difficult to draw anything other than very broad conclusions for a given plant on how the cell wall will respond to drought. Several grasses have been proposed as lignocellulosic feedstocks including: Sorghum bicolor, Panicum virgatum , and Miscanthus spp. ( van der Weijde et al., 2013 ; Mullet et al., 2014 ). These grasses have all been selected for their high vegetative biomass yields and for their increased tolerance to prolonged drought and nutrient scarcity ( van der Weijde et al., 2013 ; Mullet et al., 2014 ). S. bicolor has several advantages as a crop and as an experimental system including: high yield under water and nitrogen-deficient conditions, a sequenced genome, genetic tractability, and demonstrated genetic modification techniques ( Howe et al., 2006 ; Liu and Godwin, 2012 ; van der Weijde et al., 2013 ; Mullet et al., 2014 ; Wu et al., 2014 ; Abdel-Ghany et al., 2016 ; McCormick et al., 2018 ). Furthermore, S. bicolor has the advantage of having different either pre-flowering drought-tolerant or post-flowering drought-tolerant cultivars ( Xu et al., 2000 ; Borrell et al., 2014 ). For this study, we used RTx430, a pre-flowering drought-tolerant line, and BTx642, a post-flowering drought-tolerant line ( Xu et al., 2000 ; Borrell et al., 2014 ), to explore how some different drought-tolerance strategies affect cell wall composition under both pre-flowering and post-flowering drought stress.",
"discussion": "Discussion Transcriptome to Cell Wall One of the biggest questions answered by this study is whether the transcriptome of droughted plants, when compared to well-watered plants, can accurately predict chemical changes within the plant cell wall. Concerning the cell wall-related HVGs, the transcriptome does not necessarily translate into measurable chemical wall changes. While some transcriptional changes resulted in expected cell wall changes, many were either correlated with no compositional changes or even the opposite of the expected changes. While many reported effects of drought stress on the plant cell wall are derived from transcriptomic studies that are rarely experimentally validated ( Buchanan et al., 2005 ; Dugas et al., 2011 ; Le Gall et al., 2015 ; Tenhaken, 2015 ), this study demonstrates the need to validate the chemical and physical changes in the cell wall, especially as the tissues and conditions with the highest number of HVGs had the fewest changes in the wall, while the leaves experiencing pre-flowering drought induction had the fewest number of HVGs but the greatest compositional changes in the cell wall. These results support a growing body of literature reporting contrasting transcriptome results concerning drought stress, with Brachypodium distachyon demonstrating a wall-related transcriptome that did not always align with the metabolome in response to drought stress ( Lenk et al., 2019 ). Despite significant differences in expression of genes with cell wall-related GO terms, we detected few changes in the composition and saccharification efficiency of the cell walls of both S. bicolor cultivars in response to drought stress. This disconnect between transcriptome and wall composition can be explained by several different scenarios. The first scenario is that the HVGs encode enzymes that are minor contributors to wall biosynthesis and modification, with more active members of the gene families remaining constant under drought stress. Alternatively, there could be a large differential expression in these genes that is specific to a cell or tissue type, but the resulting change in composition is masked by that of other cell types. Importantly, such ‘minor contributor’ genes could still have an important role in drought response even if their role is not observable by bulk cell wall analysis of whole tissues or organs. Future studies, e.g., using reverse genetics with HVGs would help to clarify that. Another scenario could be that the enzymatic action is again occurring, but compositional changes are masked by the flux of carbon into and out of the cell wall. For example, it is possible that a great deal of new wall material is being deposited and/or modified, but either this new material or old material is being exuded from the wall for signaling and/or symbiotic processes with the rhizosphere ( Xiao and Anderson, 2013 ; Anderson, 2016 ; Xu et al., 2018 ). Assuming that transcription does result in increased active enzyme levels, it is possible that the activity levels of the enzymes encoded by these HVGs are very low, or that the actual enzymatic activity is different from its putative function, as most of these encoded enzymes have not had any demonstrated activity in sorghum. A last scenario is one in which differential expression of cell wall-related genes, particularly those involved in biosynthesis and modification, could simply be the result of cells and their walls either ceasing to expand and grow, a known and visually obvious phenotype of progressive drought in leaves, or continuing to maintain extensibility in the roots despite reduced soil moisture availability. The ability to pinpoint the most likely set of scenarios will rely on further experiments exploring the composition and architecture of specific polysaccharides (xylan and pectin branching), exploring root sugar exudates, and exploring composition in a tissue and/or cell-specific context. Further useful experiments would also explore the cell wall-related proteome of these plants to understand whether differential expression also leads to differential enzyme accumulation. Of the compositional changes that were observed, the most prominent changes happened to be in the matrix – namely, an increase in pectic rhamnose and galacturonic acid, arabinose, and glucose in the young leaves of RTx430 plants experiencing a pre-flowering drought stress and an increase in arabinose and galactose in the roots of BTx642 plants experiencing post-flowering drought stress. The changes in young leaves likely indicate an increase in pectin, arabinosylated xylan, and mixed-linkage glucan, although changes in abundance of arabinogalactan proteins could also be possible. Because these changes occurred in the pre-flowering drought-tolerant RTx430, but not BTx642, it is possible that these pectic and hemicellulosic differences contribute to pre-flowering drought tolerance. Root changes likely indicate an increase in arabinosylated xylan and either pectic galactan or arabinogalactan proteins. Despite not seeing an increase in any GALS1, PAGR, or GT31 homolog expression ( Liwanag et al., 2012 ; Geshi et al., 2013 ; Stonebloom et al., 2016 ), we still observed an increase in galactose in the roots of BTx642. This could be due to an entirely different and unidentified galactosyltransferase, or alternatively it could be that known galactosyltransferase activity is regulated on the enzymatic level and thus is not observed in the transcriptome. In the same vein, we also did not observe any differential expression of GAUT or RRT1 homologs ( Orfila et al., 2005 ; Atmodjo et al., 2011 ; Takenaka et al., 2018 ) during pre-flowering drought stress in RTx430 plants to account for changes in rhamnose and galacturonic acid. Pectin has been implicated in the plant drought response in the past ( Micheli, 2001 ; An et al., 2008 ; Peaucelle et al., 2012 ; Liu et al., 2014 ), and pectic galactan and arabinan from rhamnogalacturonan I have been proposed to behave as “plasticizers” in drought and desiccation conditions for several different plants ( Moore et al., 2008 ; Dinakar and Bartels, 2013 ). However, this focus on pectin’s role in the plant drought response has primarily been relegated to dicotyledons, in which pectins comprise a much greater percentage of the plant primary cell wall. These results coupled with that in the literature indicate that even in plant lineages where pectin is a very small portion of the plant primary and secondary cell wall, this polysaccharide may still play an outsize role in responding to and/or mitigating plant stress. Moreover, the over-representation of genes encoding pectin modifying enzymes amongst the HVGs in this study is striking, given that there are relatively few in sorghum, with only 16 pectic lyases and 35 carbohydrate esterases (including pectin methylesterases) ( Rai et al., 2016 ). More work looking into pectic side chain composition, pectin esterification, and cleavage of pectic oligosaccharides in response to drought stress can help clarify the role of pectin in these processes. An increase in matrix glucose was observed which often, but not always, correlated with a similar transcriptomic response in the CSLF gene family, which is known to encode mixed-linkage glucan synthases ( Buckeridge et al., 2004 ). It is possible that this increase in glucose is in part derived from another cell wall polysaccharide, most likely amorphous cellulose that is released during hydrolysis with TFA. As with the pectin study, it is possible that either an unidentified glucosyltransferase is responsible for this increase in glucose or that the known glucosyltransferases are regulated tightly at the enzymatic level and the transcriptomic regulation does not reflect that. However, this is in contrast to previous literature in which mixed-linkage glucan decreased in the aerial biomass of other grasses, with a mixed response in the roots ( Wakabayashi et al., 1997 ; Piro et al., 2003 ; Konno et al., 2008 ; Leucci et al., 2008 ; Rakszegi et al., 2013 ). MLG is deposited in the expanding cell wall and is correlated with increased wall extensibility ( Buckeridge et al., 2004 ; Vega-Sanchez et al., 2012 ). Secondary Cell Wall Deposition Transcriptomic changes in other putative cell wall biosynthetic genes such as CESAs, the GT61 family responsible for glycosylating the xylan backbone ( Anders et al., 2012 ), and lignin biosynthetic enzymes ( Schilmiller et al., 2009 ) were not reflected in cell wall compositional analyses. These upregulated genes could simply be a proxy for continued growth, including cell expansion and secondary cell wall deposition. Secondary cell wall CESAs were upregulated in older leaves of both cultivars in response to both post-flowering drought stress and pre-flowering drought stress recovery. This suggests an increase in secondary cell wall deposition in these older leaves as they either recover from drought stress or experience drought stress post-anthesis, likely from continued cell expansion and/or increased maturation. However, it does not suggest that the total secondary cell walls are more enriched in cellulose or lignin, which is borne out in the compositional wall analyses. Cell Wall Modification Increased extensibility of root cell walls in response to low water potential linked to expansin abundance and activity has been explored and verified in the roots of maize seedlings exposed to drought or osmotic stress ( Wu et al., 1996 , 2001 ). Our own transcriptomic results align well with transcriptomic surveys of other grasses exposed to drought stress in relation to wall expanding genes such as expansins and XTHs ( Buchanan et al., 2005 ; Dugas et al., 2011 ; Le Gall et al., 2015 ; Lenk et al., 2019 ). Expansin activity during drought stress results in more highly expansible cell walls, as has been shown in the apical domain of the maize root ( Wu et al., 1996 ), and overexpression of a rose expansin in Arabidopsis thaliana has previously conferred drought tolerance ( Lu et al., 2013 ). Interestingly, expansin expression decreased with distance from the root apex, implying that expansin activity during drought was focused to provide increased extensibility to the root apex, the portion of the root still expanding for deeper water reserves ( Wu et al., 2001 ). Our data shows that expansin activity in the basal root is reduced during drought conditions, suggesting that in both genotypes, the basal root has decreased extensibility in response to drought stress. Interestingly, re-watered roots of the RTx430 pre-flowering drought tolerant genotype had increased expansin and XTH expression levels during the pre-flowering drought recovery period. As indicated by Xu et al. (2018) , RTx430 appeared to have a faster recovery from pre-flowering drought stress, both in biomass recovery, but also in root microbiome recovery. This may be another indicator of pre-flowering drought tolerance, in which RTx430 is primed to resume cell division and expansion more rapidly after re-watering, which in turn gives a performance boost to the plant and its associated microbial communities via increased carbon flux. Biomass Conversion Efficiency On a more practical level, this study indicates that drought stress does not impact saccharification efficiency of cell wall sugars in either cultivar and in fact can contribute to saccharification efficiency through starch accumulation in the RTx430 line (an increase of up to 66% in some assayed timepoints). As bioenergy crops like sorghum will likely be exposed to greater frequencies and durations of drought stress, this is welcome news, as saccharification penalties combined with potential biomass yield penalties would make bioenergy crops even less competitive in the energy market. From this study, combined with previously published data using this field trial ( Xu et al., 2018 ; Varoquaux et al., 2019 ), we can conclude that the sorghum cultivars RTx430 and BTx642 experience no saccharification penalties under both pre-flowering and post-flowering drought stress. This study suggests that biorefineries do not need to be concerned with significant changes in biomass properties depending on drought patterns. One caveat to this conclusion is that the compositional and saccharification analyses were done on leaves, while stalks were not analyzed in this study. Since stalks are a major part of the aboveground biomass, it will be important to determine if the stalk cell walls and their saccharification are also largely unaffected in similar drought experiments. Summary and Significance While there have been several studies detailing the transcriptomic differences of drought-stressed grasses ( Buchanan et al., 2005 ; Dugas et al., 2011 ; Lenk et al., 2019 ), and several studies detailing cell wall changes in grasses affected by drought stress ( Vincent et al., 2005 ; Emerson et al., 2014 ; Ottaiano et al., 2017 ; van der Weijde et al., 2017 ; El Hage et al., 2018 ; Hoover et al., 2018 ), this is the first study to detail the cell wall-related transcriptomic differences and the cell wall differences in a drought-tolerant C4 grass affected by several different drought stresses in the field. As with previous transcriptomic studies ( Buchanan et al., 2005 ; Spollen et al., 2008 ; Dugas et al., 2011 ; Le Gall et al., 2015 ; Tenhaken, 2015 ; Rasheed et al., 2016 ; Lenk et al., 2019 ; Varoquaux et al., 2019 ), we see that GO terms and categories relating to the cell wall are significantly affected by drought stress, although the transcriptomic response to drought stress is large in sorghum, with more than 40% of expressed genes experiencing an effect on expression patterns relative to well-watered conditions ( Varoquaux et al., 2019 ). In our own study, we noted a wide range of 15–117 cell wall related HVGs depending on the genotype and drought treatment, out of a total of 520 cell wall biosynthesis and modification genes already described in sorghum ( Rai et al., 2016 ). Across these studies, genes thought to be involved in cell wall modification, particularly involving the extensibility of the wall, are particularly affected by drought. Conversely, recent studies on field-grown bioenergy grasses, including switchgrass, Miscanthus , and corn stover, demonstrate contrasting effects on cell wall composition and saccharification ( Emerson et al., 2014 ; Ottaiano et al., 2017 ; Hoover et al., 2018 ). Despite the decrease in structural sugars found in these studies, sugar conversion was often unaffected or improved in plants exposed to drought stress, possibly due to a less recalcitrant cell wall. Our results, coupled with findings from these studies, indicate that drought stress does not result in biomass conversion penalties when using enzymatic hydrolysis, indicating that the cell wall is as or more accessible to enzymatic hydrolysis. This does not rule out the presence of microbial inhibitors. Our findings suggest that large compositional changes in the cell wall do not occur during drought stress, but this does not rule out structural changes in the cell wall and amongst its components. Importantly, drought, while predicted to have an increase in rigidification of tissues used for biomass processing, does not seem to affect the recalcitrance of the wall in the drought-tolerant S. bicolor RTx430 and BTx642 genotypes."
} | 5,709 |
34760409 | PMC8564097 | pmc | 7,665 | {
"abstract": "Abstract Plant root symbionts, namely mycorrhizal fungi, can be characterized using a variety of methods, but most of these rely on DNA. While Sanger sequencing still fulfills particular research objectives, next‐generation sequencing currently dominates the field, thus understanding how the two methods differ is important for identifying both opportunities and limitations to characterizing fungal communities. In addition to testing sequencing methods, we also examined how roots and soils may yield different fungal communities and how disturbance may affect those differences. We sequenced DNA from ectomycorrhizal fungi colonizing roots of Pinus banksiana and found that operational taxonomic unit richness was higher, and compositional variance lower, for Illumina MiSeq–sequenced communities compared to Sanger‐sequenced communities. We also found that fungal communities associated with roots were distinct in composition compared to those associated with soils and, moreover, that soil‐associated fungi were more clustered in composition than those of roots. Finally, we found community dissimilarity between roots and soils was insensitive to disturbance; however, rarefying read counts had a sizeable influence on trends in fungal richness. Although interest in mycorrhizal communities is typically focused on the abiotic and biotic filters sorting fungal species, our study shows that the choice of methods to sample, sequence, and analyze DNA can also influence the estimation of community composition.",
"conclusion": "CONCLUSIONS We reviewed outcomes of choices on sampling, sequencing, and analyzing DNA from plant root symbionts (i.e., mycorrhizal fungi) on fungal community composition and OTU richness. As roots are habitat for fungi belonging to other trophic guilds (e.g., Unuk et al., 2019 ), we also considered effects on endophytic, saprotrophic, pathogenic, and unidentified fungi. We first investigated the influence of sequencing method on fungal community composition by comparing root‐associated fungal communities identified by Sanger and Illumina (high‐throughput) sequencing. We found that by selecting EM roots and Sanger sequencing the fungi colonizing those roots, we were able to target EM fungi and avoid those belonging to other trophic guilds. Illumina sequencing captured a wide range of trophic groups; however, many sequences were uninformative because the sequences could not be matched to an identified species of fungi. When accounting for differences in sequencing depth, the two sequencing methods were comparable. To detect shifts in common fungi comprising communities, more samples would be required for Sanger versus Illumina sequencing. Next, we investigated how sampling roots versus soils might impact community composition and found that the two habitats capture relatively distinct fungal communities. Higher inter‐sample variability in community composition among root versus soil samples indicates that more sampling would be necessary to capture community shifts of root‐associated fungi. Dissimilarity between root‐ and soil‐associated fungal communities was insensitive to forest disturbance (harvesting), suggesting the two communities change in composition to the same extent with recovery. In other words, root‐associated fungal communities do not seem more sensitive than soil‐associated fungal communities to forest disturbance, despite the death of trees (Pec and Cahill, 2019 ). Finally, we found that samples highly variable in sequencing depth are particularly sensitive to rarefaction with respect to estimating richness. Researchers may want to carefully weigh rarefying (Haegeman et al., 2013 ), excluding samples, or incorporating other normalization and transformation methods (McMurdie and Holmes, 2013 ; Love et al., 2014 ) when working with samples that are extremely unequal in sequencing depth.",
"discussion": "DISCUSSION We first investigated the influence of sequencing method on fungal community composition by comparing root‐associated fungal communities identified by Sanger and Illumina (high‐throughput) sequencing. Regardless of sequencing method, EM fungi were the dominant functional guild, and similar taxa made up the common OTUs. OTU richness was higher for Illumina sequencing; this was likely due to: (1) Illumina sequencing is not inhibited by species co‐occurrence (i.e., mixed DNA template), (2) sequencing depth is much higher than that of Sanger (Sanger: 432 sequences across 30 samples; Illumina: 1,018,520 sequences across the same 30 samples), and (3) DNA from fine roots, including EM root tips, was extracted and likely increased the pool of root‐associated fungi. Extrapolating the species accumulation curve of Sanger sequences to the number of samples submitted (i.e., 709 EM root tips), estimates of OTU richness are similar between the two methods (Sanger: mean = 97, SE = 0.3; Illumina: mean = 102, SE = 10.3) (Appendix S3 ). Thus, if the success rate of Sanger sequencing were improved through, for example, quality control procedures at critical stages in construction, sequence assembly, and annotation (Ma and Fedorova, 2010 ), this method is likely comparable to Illumina in capturing OTU richness. Furthermore, Sanger's shallow sequencing depth caused differences in fungal community composition, and inter‐sample variability was higher in Sanger‐sequenced versus Illumina‐sequenced communities. The distinct pattern of inter‐sample variability is the law of large numbers at work: increasing the sample size lowers the sample variance. Because ecological communities have inverse J‐shaped species abundance distributions, intensely sampling the common species (i.e., what is effectively done through Illumina sequencing) will result in community convergence. If detecting shifts in common EM fungal species colonizing roots is the study goal, Sanger sequencing may be adequate. However, given the inter‐sample variability, a higher number of Sanger‐sequenced samples would be required to detect these shifts compared to those generated by Illumina. If capturing fungal diversity is the study goal, especially across functional guilds, Illumina is the method of choice. However, most of the diversity in OTUs generated by Illumina had “unidentified” or “unresolved” assignments. Without taxonomic assignment, the function or biology of species comprising fungal communities cannot be inferred or interpreted. Importantly, richness and composition of Sanger‐ and Illumina‐sequenced communities should not be compared without adjusting for sampling intensity. The next choice in methods we explored was how sampling roots or soils may affect fungal community composition. We found that roots and soils yielded different fungal communities, and this remained consistent regardless of forest disturbance, rarefaction of sequences, and fungal community (total versus EM fungi). Furthermore, we found that fungal communities in soils were more clustered in composition than those associated with roots, suggesting that soil conditions may act as stronger environmental filters than roots. Fungi occupying roots may be buffered from soil heterogeneity (Beck et al., 2015 ; Goldmann et al., 2016 ). The deterministic processes underlying community assembly emerging from environmental variation in soils may be weakened by the presence of living roots, for which colonization is mostly governed by stochastic priority effects (Kennedy et al., 2009 ). Alternatively, when regional species pools are much greater in richness than local communities, the influence of stochasticity can be greater in local communities (Chase, 2003 ; Chase and Myers, 2011 ). However, given the inconsistent patterns in fungal richness between roots and soils in our study, it is difficult to assess where the higher diversity truly lies. As part of investigating the differences in fungal community composition between roots and soils, we also characterized the response of root‐based and soil‐based fungi to forest clearing, to test whether this disturbance decouples the two community types. We found that for both total and EM fungi, community composition did not increasingly diverge with disturbance, rather the distinctness of fungal communities between roots and soils persisted and represented two “views” of belowground fungi. The final choice in methods we explored was the effect of rarefaction on fungal communities. In our case, the sample on which rarefaction was based (4209 reads) had 13% of the reads than the next smallest sample (33,401 reads). Although rarefying had the desirable effect of normalizing read counts across samples, it had a sizeable influence on trends in OTU richness to the extent that patterns may be distorted. Rarefying sequence reads that have large inter‐sample variability may affect the among/between‐group patterns observed (McMurdie and Holmes, 2014 ). For total and EM fungal communities, rarefaction resulted in changes to between‐group patterns in OTU community richness. In our study, where sequencing depth varied substantially among soil samples, rarefaction showed a completely different response of fungal communities to disturbance. Without rarefaction, soil fungal communities increased in richness with disturbance, whereas with rarefaction, richness remained stable. We suggest that researchers should carefully evaluate inter‐sample variability before deciding whether to rarefy data. For data sets with large inter‐sample variability such as ours, researchers may want to consider analyzing both raw and rarefied data, then presenting both results if the data sets differ or showing only the rarefied results if no differences are observed. While rarefied and non‐rarefied data sets generated different patterns for community richness, patterns in the composition of fungal communities and functional guilds between intact and disturbed forests were similar. Although the majority of taxa in fungal communities are “rare,” the observed patterns are primarily driven by the most dominant taxa. This suggests that when choosing between numbers of samples and sequencing depth, investing in the former will provide more power to detect changes in community composition or turnover."
} | 2,551 |
36035726 | PMC9404334 | pmc | 7,666 | {
"abstract": "Co-inoculation of arbuscular mycorrhizal fungi (AMF) and bacteria can synergically and potentially increase nitrogen use efficiency (NUE) in plants, thus, reducing nitrogen (N) fertilizers use and their environmental impact. However, limited research is available on AMF-bacteria interaction, and the definition of synergisms or antagonistic effects is unexplored. In this study, we adopted a response surface methodology (RSM) to assess the optimal combination of AMF ( Rhizoglomus irregulare and Funneliformis mosseae ) and Bacillus megaterium (a PGPR—plant growth promoting rhizobacteria) formulations to maximize agronomical and chemical parameters linked to N utilization in maize ( Zea mays L.). The fitted mathematical models, and also 3D response surface and contour plots, allowed us to determine the optimal AMF and bacterial doses, which are approximately accorded to 2.1 kg ha –1 of both formulations. These levels provided the maximum values of SPAD, aspartate, and glutamate. On the contrary, agronomic parameters were not affected, except for the nitrogen harvest index (NHI), which was slightly affected ( p -value of < 0.10) and indicated a higher N accumulation in grain following inoculation with 4.1 and 0.1 kg ha –1 of AMF and B. megaterium , respectively. Nonetheless, the identification of the saddle points for asparagine and the tendency to differently allocate N when AMF or PGPR were used alone, pointed out the complexity of microorganism interaction and suggests the need for further investigations aimed at unraveling the mechanisms underlying this symbiosis.",
"introduction": "Introduction Nitrogen (N) represents a major nutrient for plants, being an essential component of proteins, nucleotides, chlorophyll, and a broad range of secondary metabolites. Cereals grains, providing 60% of the food necessary to feed the world’s population, require significant inputs of N to achieve optimum yields. Nevertheless, N availability can represent a limiting condition since nitrate (NO 3 – ) and ammonium (NH 4 + ), representing the readily available N pool, account for only 2% of total soil N content ( Moreau et al., 2019 ). At the same time, the increasing use of synthetic N fertilizers over the last decades has posed concerns about the contamination of surface and groundwater bodies by nitrate, the impairment of biodiversity, and the emission of greenhouse gases ( Ahmed et al., 2017 ). In line with the EU “Farm to fork” strategy, which aims at increasing the agricultural land managed under organic farming by 25%, scientific efforts are being done to maintain (or even increase) crop productivity by efficiently using organic fertilizers, which have been recognized to increase N losses very often, especially if applied inappropriately ( Maris et al., 2021 ). Such an increase in N use efficiency can significantly minimize N losses and the consequent adverse impacts on ecosystems while decreasing costs for organic fertilizers ( Galloway et al., 2014 ). All the pathways of N cycling in soils mainly depend on the edaphic conditions, agronomic management, climate, crop genetics, and finally determines N availability, transfer, transformation, and losses ( Congreves et al., 2021 ). Many scientific contributions have reported best agronomic practices and breeding solutions to enhance NUE, commonly defined as the plant biomass accumulation per unit of soil N available ( Peng et al., 2006 ; van Bueren and Struik, 2017 ; Anas et al., 2020 ). N assimilation by plants involves the GS–GOGAT pathway, for which glutamine, asparagine, glutamate, and aspartate are upstream key intermediates; once incorporated into organic compounds, N is then distributed to a broad range of different N-containing compounds. Over the last two decades, biotechnological engineering of the amino acid metabolism has led to promising results for the improvement of NUE, especially the content of glutamine, asparagine, glutamate, and aspartate has been adopted as an interesting target to carry out these studies, being involved in plant N utilization and storage after assimilation ( Dellero, 2020 ; The et al., 2021 ). Recently, especially in a framework of sustainable crop production, the inoculation with beneficial microorganisms has gained importance in such plant NUE increase ( Di Benedetto et al., 2017 ; Verzeaux et al., 2017 ; Dalla Costa et al., 2021 ). Arbuscular mycorrhizal fungi (AMF) have been reported to increase NUE by developing the symbiotic association with most terrestrial plants, favoring access to N uptake in a larger soil volume ( Verzeaux et al., 2017 ). Interestingly, the specific up-regulation of NO 3 – and NH 4 + transporters has been observed in AMF colonized roots compared with the non-colonized plants ( Courty et al., 2015 ; Garcia et al., 2016 ). In cereals such as sorghum, maize, and rice, AMT3.1 plant NH 4 + transporter transcripts were specifically up-regulated following the mycorrhizal colonization ( Koegel et al., 2017 ). Similarly, higher NUE has been observed following beneficial associations with N2-fixing bacteria such as Rhizobia, Frankia sp., and Cyanobacteria, and some other diazotrophs like Azospirillum spp., Herbaspirillum spp., and Paenibacillus spp. Interestingly, a community of mycorrhizosphere bacteria living strictly associated with AMF has been suggested to encompass improved crop performances, acting synergistically with AMF ( Oldroyd et al., 2011 ; Santi et al., 2013 ; Udvardi and Poole, 2013 ; Agnolucci et al., 2015 ; Kollah et al., 2016 ; Mus et al., 2016 ; Rosenblueth et al., 2018 ). In this regard, a high degree of specificity between bacteria and AMF has been proposed, and this tripartite association can potentially increase NUE in plants ( Tajini et al., 2012 ; Giovannini et al., 2020 ; Paul et al., 2020 ). The dynamic assembly of the rhizosphere microbial community depends on a large set of factors and is driven by an intricate set of belowground chemical communications ( van Dam and Bouwmeester, 2016 ; Qu et al., 2020 ). However, very little information is available in the literature regarding optimizing AMF-bacteria co-inoculation to manipulate rhizomicrobiome and improve plant performance. The present study aimed at optimizing the co-inoculation between mycorrhiza and Bacillus megaterium using the response surface methodology (RSM), with reference to enhanced N utilization in maize ( Zea mays L.). The RSM approach has been chosen to account for the interaction(s) between the fungal and bacterial inoculum in the framework of the complex rhizosphere community. Similarly, different plant-based indices of NUE (i.e., NHI and NUtE; López-Bellido and López-Bellido, 2001 ) and yield of maize have been considered to account for the different assimilation, metabolization, mobilization processes, as well as the translocation to reproductive portions, to overcome the temporal and spatial edges of NUE indices ( Congreves et al., 2021 ).",
"discussion": "Discussion The choice of the best AMF-bacteria combination to improve NUE and thus ensure efficient and sustainable food production has become a burning question in the perspective of developing new strategies for the ecological transition of agriculture ( Agnolucci et al., 2019 ). B. megaterium and AMF are both supposed to be potential bio-fertilizer agents able to play a key role in plant growth promotion. Their single and dual inoculation has revealed beneficial emerging properties in plant symbiosis, mostly translated into enhanced plant growth, yield, and abiotic stress tolerance ( Ortiz et al., 2015 ; Khalid et al., 2017 ). RSM allows testing of multiple factors with complex interactions, modeling the system mathematically and using a limited number of experimental trials. However, despite the presence of studies based on the application of RSM models to improve plant growth, yield, and NUE, the statistical optimization of the best performing inoculum which can boost plant N acquisition is still limited in literature ( Peng et al., 2014 ; Gundi et al., 2018 ; Naili et al., 2018 ; Mazumdar et al., 2021 ). Several studies have shed light on the ability of AMF or PGPR to increase plant NUE, reporting higher grain and biomass values, as well as higher levels of NUtE, in different crops. In maize, clear evidence of improved NUE have been observed following inoculation with bacterial strains of P. fluorescens S3X and C. necator 1C2 ( Pereira et al., 2020 ), Bacillus megaterium ( Ganugi et al., 2022 ), and other. Concerning AMF, positive results were reported by single inoculation with R. irregulare BEG72 and F. mosseae BEG234, and coupled application of Rhizophagus irregularis and Bacillus spp. ( Adesemoye et al., 2008 ). Nevertheless, understanding the specific mycorrhizal and bacterial synergic action requires considerable study since PGPR–AMF affinity and colonization efficiency seem to be strongly related to the fungus species and origin. Marulanda-Aguirre et al. (2008) found different results for B. megaterium when inoculated on lettuce ( Lactuca sativa L.) coupled with Glomus constrictum autochthonous, G.constrictum from a collection or commercial G.intraradices . In addition, this study highlights how an optimized mycorrhizal-to-bacterial inoculation rate on maize seeds at planting might be a success factor for the establishment and effectiveness of an AMF–PGPR symbiosis. Interestingly, our optimization aimed at enhancing maize NUE revealed similar RSM values for maximum SPAD, aspartate, and glutamate concentrations, which correspond to the intermediate doses of AMF and B. megaterium . Together with indicating a treatment effect on these NUE indices, this point confirmed the tripartite synergy between plants, mycorrhizae, and microbial communities for plant nutrition. Such synergy can be translated into a significant increase in N acquisition compared with the single inoculation of AMF or B. megaterium ( Hestrin et al., 2019 ). However, the identification of saddle points for asparagine and glutamine pointed to the complexity of microorganisms’ interactions. The presence of such intricate microbial interactions indicates the need for further deep analyses, focused on experimental runs at points along this path, using the observed response values for guidance on where to locate the next factorial experiment ( Lenth, 2009 ; Jan et al., 2021 ). Despite the lack of significant treatment effect on most of the agronomic parameters, our data revealed interesting trends which is worth exploring in the future. On the one hand, it seems that maize plant biomass and biomass N-uptake could be increased by the concomitant action of AMF and B. megaterium at > 2.1. On the other hand, grain yield, grain N-uptake, and NHI showed a tendency to be increased preferentially by the single action of either AMF or B. megaterium , while lower values were registered under co-inoculation conditions. This latter statement was further corroborated by the fact that NHI (which was close to being significantly affected by the treatments) had the highest values with 4.1 kg ha –1 of AMF or B. megaterium , alternatively. These results suggest that (i) combining AMF and B. megaterium at a significant dose (> 2 kg ha –1 ) may have the potential to enhance plant tissues growth toward vegetative biomass, and (ii) if AMF and B. megaterium are inoculated alone, it is likely that grain-related agronomic parameters could benefit more than biomass-related ones. In fact, distinct effects provided by AMF, PGPR, and their co-inoculation in terms of drought resistance and essential oil yield have been previously reported in myrtle ( Azizi et al., 2021 ). Similarly, it was reported that symbiotic efficiency may be hindered by interaction(s) with other biofertilizers together with being impacted by agro-practices ( Kuila and Ghosh, 2022 ). In another work, Pacheco et al., evaluated the effect of AMF and Pseudomonas putida (PSB) single and co-inoculation on P uptake, productivity, and P concentration in maize ( Pacheco et al., 2021 ). Interestingly, these authors reported that the microbial inoculants enhanced plant P uptake, that the presence of PSB increased biomass per unit of P taken up, and that the microbial inoculants altered P allocation within the plant, reducing grain P concentration ( Pacheco et al., 2021 ). This distinct effect of the microbial inoculants on biomass production and nutrients allocation agrees with the results from Lozano Olivério Salvador and co-workers, who observed specific effects in terms of dry weight, symbiotic efficiency, chlorophyll content, and nitrogen accumulation when AMF or rhizobacteria were applied with compost to soybean ( Salvador et al., 2022 ). Despite the fact that mechanisms underlying the effect of single and combined inoculum are still unknown, amino acid metabolism represents an interesting target for crop NUE improvement, being actively involved in plant NUtE ( Dellero, 2020 ). In particular, the glutamine synthetase/glutamate:2-oxoglutarate aminotransferase (GS/GOGAT) cycle represents the major route for plant ammonia assimilation, followed by asparagine synthetase (ASN) and glutamate dehydrogenase (GDH) ( Harrison et al., 2000 ). Accordingly, experimental studies on maize have pointed out the increased concentration of aspartate and glutamate following R. irregularis and Gigaspora margarita inoculation ( Matsumura et al., 2013 ; Hu and Chen, 2020 ), while the enhanced activity of ammonium assimilating enzymes (GS and GDH) have been detected with Azospirillum bacterial treatment ( Ribaudo et al., 2001 ). Here, for the first time, we reported how different rates of mycorrhiza-to- Bacillus co-inoculation may be considered—to some extent—as a driver of differential N partitioning between grain and vegetative biomass in the maize plant. This should be considered to address future studies to shed light on how such biotic interactions may steer plant physiology toward grain rather than vegetative tissues N accumulation. Nevertheless, it should be stated some important aspects of the experiment. At first, it must be pointed out that the Aegis Sym irriga ® product is predominantly but not totally made by AMF, being associated—as the most commercial inocula—to other bacteria. For this reason, the effect of this inoculum on plant-based indices of NUE, as well as that obtained following its interaction with the PGPR product, is not necessarily ascribable to the exclusive presence of mycorrhiza. Second, to further complicate matters, plant responses should be studied under a wide range of soil-climate conditions, including different soil types and soil fertility, and temperate to dry climates. Particularly, physico-chemical soil properties, such as nutrient availability, pH, structure, organic matter content, and texture, can affect the release of plant root exudates and, consequently, the availability and the interactions with soil microbial communities ( Neumann et al., 2014 ). It is likely that plant-microbe symbiosis is profoundly influenced by external environmental conditions, namely, temperature and moisture ( Cheng et al., 2019 ). As consequence, it should be clarified that our results are strictly related to a specific pedoclimatic experimental context and, for this reason, maximum and saddle points for each NUE index cannot be certainly considered as absolute values exactly reproducible under different plant growth conditions. Equally, the lack of significant effects in some agronomic indices may be reasonable because of the peculiar temperate and fertile conditions underlying the study, which is why it would be unreasonable to draw a conclusion about a lack of general effect of microbial inocula on the plant NUE. Notwithstanding, the application of prediction models may considerably contribute to advancements in the agricultural sector, particularly, under challenging factors interaction conditions."
} | 4,002 |
22746823 | null | s2 | 7,667 | {
"abstract": "The genetic rules that dictate legume-rhizobium compatibility have been investigated for decades, but the causes of incompatibility occurring at late stages of the nodulation process are not well understood. An evaluation of naturally diverse legume (genus Medicago) and rhizobium (genus Sinorhizobium) isolates has revealed numerous instances in which Sinorhizobium strains induce and occupy nodules that are only minimally beneficial to certain Medicago hosts. Using these ineffective strain-host pairs, we identified gain-of-compatibility (GOC) rhizobial variants. We show that GOC variants arise by loss of specific large accessory plasmids, which we call HR plasmids due to their effect on symbiotic host range. Transfer of HR plasmids to a symbiotically effective rhizobium strain can convert it to incompatibility, indicating that HR plasmids can act autonomously in diverse strain backgrounds. We provide evidence that HR plasmids may encode machinery for their horizontal transfer. On hosts in which HR plasmids impair N fixation, the plasmids also enhance competitiveness for nodule occupancy, showing that naturally occurring, transferrable accessory genes can convert beneficial rhizobia to a more exploitative lifestyle. This observation raises important questions about agricultural management, the ecological stability of mutualisms, and the genetic factors that distinguish beneficial symbionts from parasites."
} | 356 |
30556129 | PMC6375383 | pmc | 7,668 | {
"abstract": "Abstract An enduring challenge for ecology is identifying the drivers of ecosystem and population stability. In a spatially explicit context, key features to consider are landscape spatial structure, local interactions, and dispersal. Substantial work has been done on each of these features as a driver of stability, but little is known on the interplay between them. Missing has been a more integrative approach, able to map and identify different dynamical regimes, predicting a system's response to perturbations. Here we first consider a simple scenario, i.e., the recovery of a homogeneous metapopulation from a single localized pulse disturbance. The analysis of this scenario reveals three fundamental recovery regimes: Isolated Regime when dispersal is not significant, Rescue Regime when dispersal mediates recovery, and Mixing Regime when perturbations spread throughout the system. Despite its simplicity, our approach leads to remarkably general predictions. These include the qualitatively different outcomes of various scenarios of habitat fragmentation, the surprising benefits of local extinctions on population persistence at the transition between regimes, and the productivity shifts of metacommunities in a changing environment. This study thus provides context to known results and insight into future directions of research.",
"introduction": "Introduction How can dispersal, ecosystem size, and local dynamics interact to determine recovery from a disturbance? This question is fundamental to ecology, not only due to its relevance for conservation and management, but because it connects key concepts of ecology, such as stability, landscape, metapopulations, and disturbance. Dispersal plays a fundamental role in all aspects of ecology, affecting the stability of populations (Abbott 2011 ), biodiversity patterns (Haegeman and Loreau 2014 ), trophic interactions (McCann et al. 2005 ) and evolutionary dynamics (Baskett et al. 2007 ). Dispersal is often studied because of two main effects it has on ecosystems: sustaining diversity (Kerr et al. 2002 ) and generating population synchrony (Lande et al. 1999 , Abbott 2011 ). When dispersal is weak, it can promote diversity, allowing populations to benefit from spatial insurance effects, whereby good patches prevent local extinctions in less favored locations (Loreau et al. 2003 ). This effect is fundamental in the context of biodiversity loss caused by human‐induced landscape fragmentation, which impedes dispersal (Burkey 1989 , Fischer and Lindenmayer 2007 ). Dispersal, however, is not always beneficial. Strong dispersal may synchronize population dynamics and cause global extinctions. It can inhibit spatial insurance effects, causing generalist species to competitively exclude specialists (Abbott 2011 ). The opposite scenarios described, i.e., extinctions caused by dispersal limitations, or global synchrony due to strong dispersal, are extreme cases where there is a clear separation of timescales between the local dynamics and the time it takes to disperse across the system. In between is an intermediate regime without a clear separation of timescales, which has not been investigated much or even well defined. Not all relevant spatial aspects of ecosystems are centered on dispersal and interactions across space. Sheer size is also important as spatial processes are effectively mediated by the system size (Galiana et al. 2018 ). Larger regions can allow for substantial spatial heterogeneity, from asynchrony due to nonlinear local dynamics or disturbance regimes (Bjørnstad et al. 1999 , Gouhier and Guichard 2007 ), to an imposed structure due to topography or climatic gradients (Qian et al. 2009 ). Spatial averaging over such heterogeneities has led to many well‐known concepts in ecology, such as the species–area relationship (Connor and McCoy 1979 ) and landscape equilibrium (Turner 1989 ). The ecological concepts discussed above, such as synchrony and spatial averaging, are non‐trivial due to local dynamics that act in conjunction with spatial effects. Ecology has long focused on local non‐spatial behavior, with central issues such as the diversity–stability debate (MacArthur 1955 , May 1973 , McCann 2000 , Loreau and de Mazancourt 2013 ) largely addressed by focusing on local interactions between species. Assumptions on local dynamics vary greatly from linear behavior around an equilibrium to highly nonlinear dynamics far from it. This is evident in stability research where noisy time series are typically assumed to be close to equilibrium (Ives 1999 ), while catastrophic regime shifts are inherently nonlinear (Scheffer and Carpenter 2003 ). When considering spatial aspects, however, most studies implicitly or explicitly assume linear behavior, while research into nonlinear behavior is mostly focused on specific scenarios such as emergent stationary spatial patterns in drylands (von Hardenberg et al. 2010 ) or chaotic behavior of algae blooms (Franks 1997 ). Stability is a central notion in ecology, and it can be defined in various ways, which are typically context dependent (Grimm and Wissel 1997 ). Nevertheless, the concept of stability comes down fundamentally to the ability of the system to recover from a perturbation (Arnoldi et al. 2018 ), which may be affected by the timing of the perturbation (e.g., constant or single event), the dynamical aspect considered (e.g., rate of convergence or disturbance strength withstood), or the central measure recovered (e.g., biodiversity or overall biomass). Recent work has investigated the stability of populations and ecosystems in a spatial context, by explicitly considering the issues of dispersal, system size and local dynamics (Yaari et al. 2012 , Dai et al. 2013 , Plitzko and Drossel 2015 , Fox et al. 2017 , Gilarranz et al. 2017 , Wang et al. 2017 ). The scenarios considered, however, are often quite specific, and it is difficult to draw general conclusions from them. Moreover, since there is no clear framework in which to understand the phenomena described, results are hard to compare, limiting the potential for synthesis. Here we propose that answering the preliminary question of how dispersal, ecosystem size, and local dynamics interact to determine recovery from a disturbance provides a unifying framework to understand and compare the dynamical behavior of spatially extended ecosystems. We first address this preliminary question in a simplified setting, i.e., a metapopulation subject to a disturbance (a sudden change in abundance) in a uniform one‐dimensional landscape. We monitor the time needed for the system to return to its pre‐disturbed state. This notion is easily measured and understood, while having clear relations to other notions of stability (Arnoldi et al. 2016 ). This allows us to draw an exhaustive map of dynamical behavior, predicting the transitions between three qualitatively different recovery regimes: Isolated Regime (IR), Rescue Regime (RR), and Mixing Regime (MR). In IR, dispersal is not essential for recovery. In RR, propagation of biomass is key for recovery. Finally, in MR, recovery occurs after the disturbance's effect has spread throughout the system. We then translate this approach to more complex ecological settings: early warning signals of catastrophic transitions, the interplay between local extinctions and metapopulation persistence, and productivity of metacommunities in a changing environment. Our approach thus defines a powerful methodology for the analysis of spatial ecosystems and proposes new directions of research for fundamental and applied ecology.",
"discussion": "Discussion By analyzing the response to a disturbance of a simple yet spatially explicit model, we could make powerful predictions about the stability properties of various complex spatial ecosystems, ranging from metacommunities in a changing environment to populations in a fragmented landscape. We highlighted three regimes of recovery, Isolated (IR), Rescue (RR), and Mixing (MR), and showed how these regimes depend on both the properties of the system, and of the disturbance that is imposed on it, thus mapping these regimes onto the space of system and disturbance parameters. The recovery processes involved in each of these regimes are qualitatively different. In MR, the system first homogenizes before local processes drive the system back to equilibrium, whereas in RR the recovery is driven by the propagation of biomass from undisturbed regions into the disturbed ones. The relationship between time to recovery and disturbance property (extent, intensity, and strength) thus differs substantially between regimes. More generally, any prediction about the effect of a given parameter on the system's stability will strongly depend on the regime the system is in. To determine the three recovery regimes, there are two main constants to consider. The first constant is determined by three‐dimensional parameters of the system { d , r , L } , which combine to form a nondimensional constant L r / d that we call the effective system size. When this constant is very small, then the system is well mixed, and therefore in MR. When this is not the case, the limiting non‐spatial timescale τ 0 (scaled by r ), needs to be considered, as it determines the effective reach of rescue dynamics. When this effective reach is much smaller than the effective system size, then the system acts as multiple isolated sites, and hence is in IR. Otherwise, if the local timescales allow for a large effective reach (compared to effective size), then the system is in RR, in which undisturbed domains of the system are the main instigator of recovery and thus stability. From a mapping of these regimes we could predict the effects of fragmentation and global change on basic features of ecosystem stability. What matters in this context is not only where the system is on the map, but where it is going. When global change moves the system closer to collapse, local dynamics slow down, and therefore the system is pushed towards MR. This is the premise of various early warning signals of catastrophic transitions (e.g., recovery length of Dai et al. 2013 ), made explicit within our framework. These indicators essentially measure how close the system is from MR (although being in this regime does not imply a collapse). Our mapping is especially useful when considering the combined effect of fragmentation and global change, since fragmentation may push the system towards IR, in contrast to the effect of slowing down due to global change. This means that the recovery regime may not change at all, so that early warning signals that measure the distance to MR will not detect the impending collapse. A much clearer picture is gained from knowledge of the system's trajectory on the map of recovery regimes. Following from our definition of recovery regimes, it follows that only in RR can there be an interplay of spatial scales, between the characteristic scale of the system and any scale that is imposed on it. This leads us to ask what phenomena are specific to this regime, especially since it is relatively less explored than IR and MR. In particular, the transition regions between RR and the other two recovery regimes, which may take place over a large range of parameter space, can be expected to be especially interesting and display exotic behavior, since here there is no clear timescale separation. Indeed, the surprising results shown by Fox et al. ( 2017 ), where local extinctions save the predator population from total extinction, well demonstrate this notion that the transition regions can show particular and unintuitive behavior. In the presentation of our results we focused on the recovery of the system from a single disturbance. In this restrictive context, the definition of a non‐spatial timescale τ 0 is a result of the interaction between local dynamics and disturbance intensity. More generally, the relevant definition of τ 0 depends on the dynamical scenario in question, as shown in the example of biomass productivity (Fig. 7 ). In this case, the period over which conditions change defines the relevant non‐spatial timescale τ 0 . Other definitions of τ 0 can be made, depending on the perturbations the system undergoes, as is detailed in Appendix S2 . For instance, Yaari et al. ( 2012 ) investigated the spatial scaling of metapopulation persistence subject to demographic stochasticity, and universally found three distinct regimes across a gradient of dispersal. If one wanted to compare recovery regimes to the scaling regimes of Yaari et al., one would define τ 0 as the local time to extinction, which must be compared to the time needed for rescue dynamics to take place. This work demonstrates that despite inherent complexities in the dynamics of populations and ecosystems, strong qualitative predictions can be made from the analysis of a simple and generic model. Indeed, we studied the dynamics of a single species at equilibrium in a uniform one‐dimensional landscape, and considered its response to a single disturbance. However, our example of fragmentation scenarios shows that we can apply our methodology to more complex spatial structure. Moreover, the predictions shown for the predator–prey and the metacommunity systems clearly show that our framework can be applied to multiple species systems. These two examples also show that considering a system disturbed from equilibrium did not limit our predictive ability, as the predator–prey system was one where multiple disturbances occurred and the dynamics exhibited large oscillations, while the metacommunity system had no explicit disturbance at all, but rather a continuously changing environment. Overall these examples suggest that there are universal properties of ecosystem dynamics in a spatial settings, that can be unraveled from dimensional considerations. Heterogeneities in space, time or species properties will impact the recovery regimes, however, if they are sufficiently strong. As noted in Appendix S5 , both demographic and environmental noise (temporal heterogeneities) can have an effect on recovery regimes. In particular, they tend to make the transition between IR and RR more gradual, while having a minimal effect on the RR‐MR transition. Spatial heterogeneity can be expected to have a similar effect, having a strong impact on local dynamics when dispersal is weak, but less so when dispersal is strong, due to the spatial‐averaging effect of dispersal (Loreau et al. 2003 ). The case of species‐rich communities (heterogeneity of the community) is the most intriguing. In the most complex case of strong collective nonlinearities (e.g., complex succession dynamics, multiple equilibria, and so on) and heterogeneous dispersal abilities, more work is needed to say how to correctly apply and generalize our framework. Nonetheless, in addition to the specific examples given in Examples , there are at least two general cases for which our approach can be expected to directly apply. The first and more straightforward case is when a dominant species governs local dynamics (a simple case of strong heterogeneity). The second case, corresponding to moderate heterogeneity, would be a community comprised of many species with weak interactions and similar dispersal abilities. In this case, local dynamics will be a collective outcome of species interactions, but we may nonetheless define and predict transitions between recovery regimes with respect to a collective rate of local recovery. Determining the relevant nondimensional constants that govern the recovery regimes is not always straightforward, but it sets a clear direction toward a deeper understanding of the spatial processes in a system of interest. For instance, our analysis of the fragmentation scenarios showed that the effective system size is essentially the average shortest path between two sites, rather than the total number of sites in the network. The issue of estimating the relevant parameters should be particularly interesting in the context of complex species interactions, due to collective emergent behavior. For instance, the characteristic timescale of a community, such as its local recovery time, may be an emergent property of the assembly process. Moreover, in the case of strong heterogeneities between species (e.g., substantially different dispersal abilities between trophic levels) it may be that more nondimensional constants are needed to faithfully describe the system's dynamics. For instance, a higher trophic level may spread much faster than its resource (McCann et al. 2005 ), leading to a combination of IR and MR. Our study is essentially based on the presumption that we can learn a great deal on an ecological system by performing a simple calculation using our knowledge of its dimensional properties. The examples we have presented show that this claim has merit, and our methodology can indeed provide new insights into dynamics of spatial systems. This study takes us one step further towards a quantitative understanding of the response to disturbances of spatially extended ecological systems, drawing for the first time a clear link between the opposite cases of a well‐mixed system and a set of isolated sites. Considering the proliferation of theoretical and empirical studies, and the growing sets of observational and experimental data, being able to qualitatively compare different systems is of vital importance. Our study suggests a simple way of performing such a mapping, using limited information about a given system. Such an approach is relevant for both theoretical models and empirical data, paving the way towards a novel synthetic view on ecosystem dynamics in space."
} | 4,451 |
39194466 | PMC11351569 | pmc | 7,669 | {
"abstract": "The mimesis of biological mechanisms by artificial devices constitutes the modern, rapidly expanding, multidisciplinary biomimetics sector. In the broader bioinspiration perspective, however, bioarchitectures may perform independent functions without necessarily mimicking their biological generators. In this paper, we explore such Bioarchitectonic notions and demonstrate three-dimensional photonics by the exact replication of insect organs using ultra-porous silica aerogels. The subsequent conformal systolic transformation yields their miniaturized affine ‘clones’ having higher mass density and refractive index. Focusing on the paradigms of ommatidia , the compound eye of the hornet Vespa crabro flavofasciata and the microtrichia of the scarab Protaetia cuprea phoebe , we fabricate their aerogel replicas and derivative clones and investigate their photonic functionalities. Ultralight aerogel microlens arrays are proven to be functional photonic devices having a focal length f ~ 1000 μm and f-number f/30 in the visible spectrum. Stepwise systolic transformation yields denser and affine functional elements, ultimately fused silica clones, exhibiting strong focusing properties due to their very short focal length of f ~ 35 μm and f/3.5. The fabricated transparent aerogel and xerogel replicas of microtrichia demonstrate a remarkable optical waveguiding performance, delivering light to their sub-100 nm nanotips. Dense fused silica conical clones deliver light through sub-50 nm nanotips, enabling nanoscale light–matter interactions. Super-resolution bioarchitectonics offers new and alternative tools and promises novel developments and applications in nanophotonics and other nanotechnology sectors.",
"conclusion": "4. Conclusions Biological development offers an unlimited variety of forms, exhibiting both unsurpassable complexity and mathematical order. This work draws inspiration from architectural concepts that have been utilized for millennia and explores the application of bioarchitectural forms and their derivatives to create 3D functional devices for photonics. These devices do not necessarily replicate the mechanisms or processes of the parent organisms, instead they offer fundamentally different functionalities that may not be achievable through other means. In this work, we investigated a novel bioarchitectonic 3D nanofabrication approach based on the direct replication of structural elements of insects using ultra-porous silica aerogels. Aiming at photonics devices and applications, two paradigms have been investigated using natural specimens of the European hornet Vespa crabro flavofasciata and the ‘golden beetle’ scarabeo Protaetia cuprea phoebe . Aerogel and xerogel replicas of hexagonal ommatidia cornea of the compound eye and microtrichia , microneedles, were successfully fabricated, and their functionalities were demonstrated and analyzed. First results confirmed that silica aerogels having extremely low refractive index can be used to produce ultralightweight refractive micro- and nano-optics. In the following stages, these freeform bioarchitectonic elements underwent systolic transformation, which resulted in their densification and dimensional miniaturization. This process preserved their conformality to the original master structures provided by the parent biological organ. Three-dimensional fused silica clones of the natural compound eye having hexagonal microlenses with approximately 10 μm diagonal, focus collimated visible light at focal lengths f ~ 35 μm with a f-number, f/3.5. Transparent fused silica replicas of microtrichia waveguide and deliver the light through their sub-50 nm nanotips. Such bioarchitectonic concepts and methods promise to exploit the wealth of complex natural forms, to realize novel freeform devices that surpass the art of mainstream biomimetics, and impact critically on crossdisciplinarity nanotechnologies.",
"introduction": "1. Introduction Nanotechnology in three-dimensional (3D) space opens many avenues for future spatiotemporal light–matter interactions in the molecular and atomistic domain. Even though advanced planar lithographic processing technologies enable fabrication in the sub-10 nm scale, building 3D functional structures in space remains a challenge. Diffraction of light and particle waves imposes fundamental physical limitations on the resolution of lithographic processing and alternative new tools are needed to reach miniaturization limits and explore the inherent and far-reaching potential of 3D nanodevices. Current technology employs two-dimensional (2D) planar processing methods to build 3D devices on planar substrates [ 1 ]. However, the fabrication of high-resolution monolithic macroscopic 3D structures of arbitrary ‘freeform’ stereometry necessitates the deployment of alternative concepts. In recent decades, state-of-the-art maskless techniques [ 2 ] have emerged, including stereolithographic methods, additive manufacturing, formative casting and molding. Direct laser materials growth and processing [ 3 ] now incorporates laser-induced material transfer [ 4 ], even of sensitive biological materials [ 5 ], and artificial intelligence optimization methods [ 6 ]. Additive manufacturing by laser sintering methods is now an industrial reality, achieving, however, a relatively low spatial resolution on the scale of tens of micrometers. Multiphoton laser polymerization [ 7 ] achieves 3D structures with submicron precision. Experimental results have shown the capacity for a sub-100 nm resolution by using special polymer and hybrid resins [ 8 ]. Even though the latter method is capable of high-resolution fabrication of predesigned freeform 3D objects, it exhibits significant disadvantages of limited physical processing volume, low processing speed, high complexity and high cost. Self-assembly methods have been developed for the formation of 2D and 3D lattices using polymers [ 9 ], liquid crystals, nanocrystals and polystyrene microspheres [ 10 ]. These lattices have been used in turn as templates for lithographic processing, assuming the role of exposure masks in 2D and 3D device fabrication, which are both of interest in nanophotonics [ 11 ]. This method enables the formation of regular lattices, but it is not capable of freeform 3D object fabrication. The complexity, the reliability and fabrication speed usually associated with multistep processing are persistent disadvantages. Nature offers an unlimited variety of 3D structures, organs, optimized via biological evolution to perform extremely complex physical, chemical and biological operations [ 12 ]. Organs serve the specific needs of living creatures and plants, while exhibiting unparalleled curves and forms [ 13 ]. The interaction of man with the natural world has been the inspirational playground for the conceptualization and fabrication of artificial functional objects since the ancient times [ 14 ]. This art evolved to the modern, broad and rapidly expanding technological domain of biomimicry and biomimetics [ 15 ]. Advanced concepts and devices mimicking the forms, functions and processes of plants and animal life cover several diverse fields and respective technologies, from biological chemistry [ 16 ] to mechanics [ 17 ], photonics [ 18 ] and machine vision [ 19 ], to information technology and neuromorphic processors [ 20 ], fluid dynamics [ 21 ] and others that have recently been reviewed [ 22 , 23 ]. Focusing on photonics, considerable research has been directed toward the intriguing phenomena of structural coloration and the compound vision of insects. Nanostructures of butterfly wings [ 24 ] and scarab bodies [ 25 ] have been thoroughly studied, being most relevant to diffractive and photonic band gap (PBG) structures [ 26 ]. The fabrication of 3D PBGs in the visible spectrum is still presenting several technological challenges [ 27 ]. Natural structures, such as those seen in the Morpho rhetenor butterfly [ 28 ], cicada wings [ 29 ] and golden beetles [ 30 ], could inspire novel solutions [ 31 ]. More relevant to the paradigms addressed in this work is the compound eye of insects. Following the classical studies of Hooke [ 32 ], this vision organ has attracted huge interest and inspired the technology of microlens arrays. Extensive studies have been devoted to its anatomy and biological operation [ 33 ], as well as its potential in biomimetic applications [ 34 ], from imaging [ 35 ] and wide-angle machine vision [ 36 ], to sensing [ 37 ] and micro/nanofabrication. A variety of methods have been deployed for the fabrication of artificial biomimetic lens arrays [ 38 ]. Depending on the materials used and the dimensional scale, combined subtractive and additive methods have been employed [ 39 ], including lithographic processing [ 40 ], femtosecond laser microfabrication [ 41 , 42 , 43 ], self-assembly and template replication [ 44 , 45 ] and additive manufacturing [ 46 ]. Of relevance is the formation of microlens arrays distributed over a hemispherical surface [ 47 , 48 ], closely mimicking the natural compound eye. Ultrawide hybrid optics for machine vision [ 49 ] have also been demonstrated, enabling superposition [ 50 ] or apposition [ 51 ] imaging for enhanced image quality [ 52 ] and wide-field angular coverage. Natural microneedles found on the bodies and wings of insects represent another example of interest in this work. Biomimetic microneedle array technology [ 53 ] is attracting considerable attention, finding applications in transdermal drug delivery [ 54 ], the administration of immunobiological substances, cosmetic treatments, cell sensing and disease diagnosis [ 55 ]. The fabrication of such structures has been achieved by a variety of lithographic microfabrication methods [ 56 ], including multiphoton polymerization [ 57 ], and etching [ 58 ]. Photonic nanotips have been developed in silicon for antireflection coatings (black silicon) [ 59 ], fiber optic nanotips for scanning near-field optical microscopy (SNOM) [ 60 ], fiber sensors [ 61 ], photoimprinting [ 62 ], metamaterial nanotip [ 63 ], silicon carbide photonic arrays for light delivery [ 64 ], and spectral control [ 65 ]. On the other hand, and beyond bioinspiration, the realistic biomimetic nanotip arrays such as, for example, the semiconductor nanotips inspired by cicada wings [ 66 ] have been overlooked and they merit further attention. Direct replication of natural elements by casting has been developed as a supporting step in scanning electron microscopy (SEM) of plant specimens [ 67 , 68 , 69 ] following procedures like those applied in dentistry. More recently, tin-oxide calcination of Morpho butterfly wings [ 70 ] resulted in biomimetic photonic structures and demonstrated variable structural coloration. Direct biomimetic replication of butterfly wings has also been performed by soft lithography to investigate optofluidic synergetic properties [ 71 ]. In our recent research [ 72 ], we have replicated artificial diffractive elements in aerogels and demonstrated stepwise systolic processing, yielding physically downsized micro- and nanoscale structures. The most recent replication of compound eye of dragon fly [ 73 ] has demonstrated gradient ommatidium arrays for multi-focus imaging and enhanced vision acuity. In addition, cicada wings have been used as templates for fabricating flexible polydimethylsiloxane (PDMS) nanoporous substrates, which have been loaded with uniform Ag nanoparticles and employed in portable surface-enhanced Raman scattering (SERS) systems [ 74 ]. Relevant to our work is the thermal transformation of silica aerogel into dense fused-silica glass [ 75 , 76 ]. Vitrification of sol–gel casts has been used to form microlens arrays [ 77 ]. In our previous studies [ 78 ], we laser patterned aerogel solid objects to produce surface relief and in-volume-void embedded microstructures. Subsequent thermal processing of these decorated monoliths produced miniature fused silica replicas of the original master objects. Both the bulk monolith and its embodied micropatterns were conformally, simultaneously, and analogously downsized. This systolic transformation enables the manufacture of 3D freeform solids that are geometrically affine to aerogel master (parent) objects but have dimensions which are submultiple to their master. As a result, the smallest feature crafted by any pattering method is transformed to its affine clone feature, size-reduced by a systolic factor (SF) determined by the materials and methods used. Consequently, this method facilitates super-resolution fabrication of functional 3D devices of arbitrary, freeform stereometry, that is not achievable by any other means to date. In the present work, we replicate Bioarchitectural forms in aerogel and transform them into functional elements. These elements utilize the organized and unified structure of the specific bioarchitectural design, without necessarily imitating or reproducing the biological processes of the master (parent) organ. Essentially, they are a kind of Bioarchitectonic elements. Despite being produced through this relevant approach, they differ from the established modern concept of Biomimetics . Focusing on optics and photonics, we replicate bioarchitectures in ultra-porous aerogels [ 79 ]. We form, transform, demonstrate, and characterize the functional operation of these 3D artificial photonic elements for the first time to our knowledge. Although silica aerogels are ultra-porous materials exhibiting extremely low mass density (<0.25 gcm −3 ) and refractive index (n < 1.1), our results prove that they can be used to form ultralightweight refractive optical elements. In our approach, we have chosen the well-known paradigm of compound eye and investigated three affine versions of the complete head of European hornet Vespa crabro flavofasciata embodying ommatidia microlens arrays. In another paradigm, the ‘golden beetle’ scarab Protaetia cuprea phoebe provided microtrichia arrays found on its wings. They have been transformed into optically transparent microneedles that waveguide light and deliver it through their nanotips. The photonic performance of microtrichia replicas is completely unrelated to the biological function of natural microtrichia, thus highlighting the distinct bioarchitectonic approach we address in this work.",
"discussion": "3. Results and Discussion 3.1. Bioarchitectonics in Biomimetics Biomimetics , a concept and technology developed over thousands of years, has evolved into a broad, rapidly expanding, and highly interdisciplinary field spanning many disciplines. In its modern definition, biomimetics involves the study of biological forms, actions, mechanisms, and processes, targeting to synthesize artificial products and mechanisms that mimic the natural ones, from the macroscopic to the molecular scale. Current developments embrace micro/nanotechnology methods, yielding novel chemical/biochemical processes and cutting-edge functional devices, some of which are already reaching the marketplace. In a broader context, Biomimetics would be conceived as a subset notion of Bioinspiration . This encompasses all concepts, forms, topologies, mechanisms, methods, and products that have a reference to biological development. However, one may note that the majority of biomimetics research and technology concerns mimicking biological forms, mechanisms, and processes. Building on ideas of architecture and art, we explore in the present work the independent use of forms that do not necessarily mimic the functions of their biological generators. This architectonic approach represents a new and emerging category within the broader biomimetics sector, distinct from current mainstream biomimetics. By a fruitful selection of structural forms, materials and methods, we create Bioarchitectonic devices that function differently than their biological counterparts. In this context, we examined two well-known paradigms, the compound eye ‘ommaditia ’ and the microneedles ‘microtrichia’ . We fabricated aerogel and xerogel devices of compound eyes of hornet, of the species Vespa Crabro Flavofasciata and microtrichia ensembles of scarabs, of the species Protaetia Cuprea Phoebe, as functional photonic devices. In spite of their extremely low mass density and refractive index, the optical performance of these devices demonstrates their potential for manufacturing ultra-lightweight optical elements. In addition, systolic miniaturization of these replicas resulted in transparent fused silica glass photonic devices, which are miniaturized clones that perform differently from their natural parents. 3.2. The Paradigm of Biorchitectonic Compound Eye Silica aerogel replicas of the compound eye of the hornet Vespa crabro flavofasciata have been fabricated using high-temperature supercritical drying and multistep ambient-pressure drying of wet gels, as described in previous sections. Figure 3 presents an example of an aerogel replica monolith fabricated via high-temperature, high-pressure supercritical drying. Figure 3 a presents a far view of the hornet’s head replica, while Figure 3 b provides a close-up view of the hexagonal microlens array structure selected from the indicated spatial location. Although the 3D aerogel object consists of a very brittle nanoporous air-filled skeleton, these replicas exhibit remarkable structural uniformity and reproduction fidelity. The average diagonal dimension of each lenslet ranges from D ~ 30–35 μm, indicating a ~ 10% deviation from the dimensions of the natural cornea. We reiterate that the natural variance and the plurality of microlenses prevent exact one-to-one comparison between the fabricated elements and their natural counterparts. The optical function of the fabricated compound eye optic was investigated by using the optical microscope under white-light, and color-filtered, plane-wave transillumination, directed bottom-up through the microscope column. This arrangement allowed for the observation of the refractive surface and the ensemble of focused beamlets, the latter serving as the experimental functional test case. A 20× objective having numerical aperture NA = 0.2 was used for imaging, in conjunction with the CMOS camera. In addition, the imaging beam profiler operating at low light levels and below saturation allowed for accurate intensity quantification, as well as facile visualization by employing pseudo-coloration. Figure 3 c shows the image of the surface captured by the beam profiler. The surface was set at the working distance of the microscope objective, as indicated in the drawing of Figure 3 d. The microlens array foci, as observed in Figure 3 e, are recorded by lowering the microscope sample-stage as illustrated in the drawing of Figure 3 f. Experimentally, quasi-Gaussian focal spot profiles were observed at a distance h ~ 195 ± 5 μm above the aerogel surface, as measured by the micrometric translation stage of the microscope. The measured FWHM can be related to the ideal Gaussian spot size by: (3) F W H M = 1.18 × w o = 2 × 1.18 × λ π F # \nwhere w o is the ideal Gaussian spot size, λ is the wavelength of light and F/# = f/D is the f-number of the focusing optic. The elevation h ~ 195 μm and wavelength λ = 500 nm, imply f-number of F/6, which yields an estimated FWHM ~ 2.2 μm. This result is in contrast to the measured FWHM ~ 15 ± 1 μm value shown in Figure 3 e. Therefore, we may reasonably conclude that the measured apex-to-focal point distance, h, corresponds to the ‘back focal length’, f b ~ 195 μm of the thick optical element. We further investigated by SEM microscopy to directly measure the radius of curvature of both the natural cornea specimen and its aerogel replica, as shown in Figure 4 . Owing to the fragility of aerogel, a precise dissection of the aerogel replica was not possible and thus we employed a high inclination profile to verify the result. Figure 4 a presents the cross section of the natural compound eye of a hornet specimen, indicating the cornea surface (A) and the underlying multilayer structure (B). This is followed by the conical taper segments and rhabdoms (C), the latter being damaged due the dissection. Notably, fine damages and blemishes on the natural eye are replicated in the artificial elements. Figure 4 b shows a close-up view in which we measured the linear dimensions of the chord, c ~ 33.75 ± 0.01 μm, and the sag, s ~ 2.85 ± 0.01 μm, which were used to estimate the radius of curvature. Assuming a spherical surface of the cornea, we applied the power-of-point theorem for the major circle of the sphere: (4) c 2 2 = s 2 R − s \nand estimated the radius of curvature as R ~ 51.38 ± 0.02 μm. Bearing in mind the dimensional tolerances and the natural variance of the ommatidia geometry, this value agrees well with the value estimated using projection geometry applied on the high inclination profile of the aerogel replica shown in Figure 4 c. We used nitrogen porosimetry and volume-fraction estimation (SiO 2 -Air) to determine the effective refractive index n aero of the high purity and optical quality aerogel. These samples exhibit ~90% porosity, yielding mass density of ρ aero ~ 0.22 gcm −3 , and n aero ~ 1.046 at λ = 500 nm. Considering that only the convex surface of the aerogel replica, having radius R ~ 47 μm, is involved in the focusing action, we use Equation (1) and estimate the paraxial focal length of the aerogel microlens as f aero ~ 1030 μm, and the f-number, as F / 32. Using this value in Equation (3) we estimate the FWHM ~ 12 μm, which is in reasonable agreement with the experimentally measured FWHM ~ 15 ± 1 μm shown in Figure 3 e. The free-space propagation of a real, non-diffraction-limited, optical beam can be described in terms of the beam propagation factor M 2 [ 89 ] defined as: (5) Μ 2 = W ο Θ ο w o θ ο = π λ W ο Θ ο = π λ D 2 f b F W H M 1.18 \nwhere, λ, is the wavelength of light, W o , the real focal spot-size and, Θ ο , the measured divergence. These values differ from those of an ideal diffraction-limited Gaussian beam, having respective spot-size, w o , and divergence, θ ο . In the present case, we derived the divergence, Θ ο , by considering the filled microlens aperture, D, the focal plane position along the propagation axis measured as the back-focal-length, f b , and the focal spot W o , determined by the measured FWHM at the focus. Applying Equation (5) with these experimental parameters we estimate M 2 ~ 7, implying a considerable deviation from an ideal Gaussian. This discussion highlights the potential of ultra-porous aerogel materials to produce refractive optical elements, provided that high optical quality is attained. Unlike microlens arrays produced by other methods, these elements are extremely lightweight, with an estimated mass of the individual refractive lenslet to be of the order of ~1 ng. The near unity refractive index naturally produces a relatively long focal length, even though it can be tuned by adjusting the materials density and the micro-optic curvature. Furthermore, the extremely low thermal conductivity of aerogel makes it attractive in various applications. Similar results of optical function have been obtained using quasi monochromatic illumination provided by the blue (B) and red (R) dichroic filters of the RGB standard, respectively centered at 470 nm and 630 nm of the spectrum. Figure 5 b shows selected areas of the compound optic surface relayed using the blue filter (B) and Figure 5 c presents the corresponding microlens foci. The curvature of the compound eye surface prevents simultaneous sharp image focus across the entire image field. The focusing conditions are presented in the schematic of Figure 5 a with exaggerated dimensions for clarity. The paraxial ray propagates through the thick bulk and focuses at the back focal points, F B , and F R for the blue and red respectively. These focal points are positioned respectively at the back focal lengths f b (B) and f b (R) distances, as they are measured from the microlens apex. Considering the focal points F B and F R, and the second principal point H 2 defined on axis by the principal plane H 2 , the focal lengths of the optic are f B = |F B H 2 | and f R = |F R H 2 |, for blue and red light respectively. Experimentally, the elevation distance between the two image settings, i.e., focus on surface and focus on focal plane, records the back-focal length f b (B) ~ 185 ± 5 μm of the individual lenslet for blue light as shown in Figure 5 b,c, and the back focal length f b (R) ~ 200 ± 5 μm for red light as depicted in Figure 5 d,e. Assuming a common achromatic second principal plane H 2 we can make an account of the aerogel dispersion, v, as: (6) v = 1 − n a e r o , R 1 − n a e r o , B = f B f R ~ 0.98 We note that this value is not the Abbe V-Number but it is only a relevant figure-of-merit. In the second stage of our investigation, xerogel structures were fabricated by casting and drying alcogels. Particular attention was paid to ensure uniform solvent extraction and smooth demolding to prevent dimensional discrepancies. Natural drying is the primary systolic step, delivering a downsized monolithic xerogel replica of the hornet head as shown in Figure 6 a. Figure 6 b presents a close-up view of the hexagonal microlenses. By averaging the measured hexagon diagonals as D xero ~ 16 ± 2 μm we estimate a systolic factor SF × ~2.1. Figure 6 c,d, present the two focusing positions at the surface and the microlens focal plane, respectively. The focus was positioned approximately at 90 ± 5 μm above the object surface. A Gaussian focal spot having FWHM ~ 4 μm is observed in Figure 6 d. After the primary systolic process, the resulting microlens radius of curvature is estimated as R xero ~ 22 μm. In addition, nitrogen porosimetry of xerogel gives a porosity value of ~50 %, implying a xerogel refractive index n xero ~ 1.23. These values yield an estimate of the focal length f xero ~ 98 μm, at F/6. By using Equation (3) we estimate the diffraction-limited Gaussian focal spot FWHM ~ 2.2μm that is comparable to the observed FWHM ~ 4 ± 1 μm in Figure 6 d. Using Equation (5), we estimate the factor M 2 ~ 1.8 verifying the nearly diffraction-limited performance of the microlens. Figure 7 illustrates the imaging function of the xerogel microlens array under quasi monochromatic conditions provided by the blue (B) and red (R) filters of the standard RBG filter set. We reiterate that the high curvature does not allow sharp imaging over the field. A considerable difference between the foci positions measured experimentally is observed. The difference in focus setting in Figure 7 a,b measures the back focal length f b (B) ~ 60 ± 5 μm for the blue range. Similarly, using Figure 7 c,d, we measure the back focal length f b (R) ~ 90 ± 5 μm for the red spectral region. Applying similar arguments, we infer by Equation (6) a measure for the xerogel dispersion figure-of-merit v ~ 0.73. Viscous flow vitrification of both aerogel and xerogel replicas was performed through thermal processing at approximately 1100 °C. Figure 8 a shows a miniaturized hornet head replica composed of dense fused silica, fabricated by vitrification of a monolithic aerogel replica. Submicron original defects and droplet-type defects, probably formed upon solidification of xerogel microparticles. By comparing the diagonal of the original aerogel master object to the measured diagonal of the individual microlens elements in Figure 8 b, D ~ 10 μm, we deduce the systolic transformation factor SF × ~3.4. The optical image of the surface under white light illumination is depicted in Figure 8 c. The focusing action of the microlenses in Figure 8 d allows for the direct measurement of the microlens back focal length, which is approximately f b,f-s ~ 30 μm. Considering the systolic factor SF × 3.4, we can estimate the radius of curvature of the individual microlens element to be R f-s ~ 14 μm. Such a high curvature is comparable only to self-assembled microsphere cast arrays [ 90 ], or nanofabricated planar arrays [ 91 ]. Recent approaches aim to provide microlens arrays spherically distributed to cover an ultrawide field of view [ 92 ]. The significant difference between these biomimetic approaches and our results is that the bioarchitectonic array covers conformally an oval surface of very complex topography, as it is endowed by nature. In fact, the radius of curvature of this surface varies significantly along the axes of vision to provide the best possible field coverage and the vision processing required by the insect. The contour schematic of Figure 8 e presents anatomical details and the respective radii of curvature of the different parts, estimated here by SEM imaging and projection in the range of approximately 70 μm to 1068 μm. This clearly represents a step beyond the current state of the art. Using the refractive index value of fused silica n sil = 1.46 at λ = 500 nm, we estimate the microlens focal length to be about f f-s ~ 35 μm, at F/3.5 and FWHM = 1.32 μm for the diffraction-limited Gaussian. A polychromatic focal spot of FWHM ~ 3.5 ± 1 μm is observed in Figure 8 d. The deduced propagation factor M 2 ~ 3 indicates a deviation from the diffraction-limited performance, which could be due to refractive index and shape errors in the densified silica solid. 3.3. The Paradigm of Bioarchitectonic Microneedles In this section we demonstrate the replication of microtrichia arrays, a representative case in which the fabricated elements serve a function totally independent of their original biological counterparts. Microtrichia cover the forewings, elytra , and the hindwings of the scarab. To explore the optical properties, we employed APD processing to fabricate aerogel replicas of microtrichia found on different locations of the wings and body of the insect. Figure 9 shows the monolithic aerogel replicas of elytra microtrichia. These replicas were fabricated from different areas of the specimen, with an example depicted in Figure 9 a and its detail in Figure 9 b. The replicas are conical elements of an approximate height of 5 μm and a cone base diameter of about 2–3 μm. The apex radius of these needle-like structures is around 100 nm, indicating a slight shrinkage of approximately 20 %, attributed to the spring-back effect inherent to the APD process. Figure 9 c,d demonstrate the optical functionality of the aerogel microneedles under white light plane-wave transillumination. Figure 9 c presents an image focused on the cone base, while Figure 9 d shows the light emitted from the nanotips as it is detected at the resolution limits of the microscope. Light is guided through the transparent microneedle’s trunk and is decoupled producing the beamlet spots observed at the nanotip. The monolithic aerogel replicas of hindwing microneedles from three different locations have also been investigated and are presented in Figure 10 a–c. In this case, smaller structures are observed with an average height of 2–4 μm and a cone base diameter of ~2 μm or smaller. Figure 10 d demonstrates their optical performance. The noticeable variation in surface curvature results in differing focusing conditions. Areas of interest with the image focus at the microneedle’s cone base are encircled with solid line, while areas with image focus at the nanotip are encircled with a dotted line. We underline the color change at the center of the light circle, indicating a region of high intensity resolved at the limits of detection. The reproduction fidelity at this miniaturization level is particularly noteworthy. Further elaboration of the method addressed the xerogel replication of both elytron and hindwings. Figure 11 a,b present the replication results of two different areas of the elytron, corresponding to the specimens shown in Figure 2 a and 2b respectively. The radii of the microneedle apex are in the sub-100 nm range, with SEM images clearly reaching the resolution limits of the instrument. We note that due to experimental complexity, it is practically impossible to identify the exact one-to-one correspondence between the natural microtrichia and the replicas produced. However, the elytron area and associated replicas are distinguishable, allowing us to estimate an average systolic factor in the range of SF × 2.5. Figure 11 c,d illustrate the optical performance of xerogel replica the elytron. Examples of focusing on the base of the conical microneedles are encircled with solid line. Focal spots of the microneedle apex are encircled with dashed lines. The apparent focal spot, observed by the microscope is in the ~1 μm range, clearly reaching the resolution limits of the instrument. Examples of hindwing replicas in xerogel are also presented in Figure 12 . Figure 12 a–c are SEM micrographs of replicas from three different areas of the hindwings. The dimensions of the base of the conical microneedles vary depending on the location, ranging from approximately 1 μm to 2 μm, with heights around 2 μm to 3 μm. In Figure 12 d the image focus is at the cone base, while Figure 12 e shows the corresponding focal spots at the nanotips, which are about 1.5 μm in size and reach the diffraction limits of the instrument. In the next task, we applied the final systolic step in both aerogel and xerogel replicas. Thermal processing in the range of 1100 °C resulted in a total dimensional reduction of approximately SF × 4 and the fabrication of fused silica clones. Examples of vitrified xerogel microtrichia replicas are depicted in Figure 13 . Figure 13 a,b present fused silica replicas drawn from two different areas of the hindwing. The cone base diameter of the observed nanoneedles is about 1 μm, with the submicron cone trunk extending to nanotip apex radii estimated in the sub-50 nm range. Increasing SEM magnification produces image blurring, making the nanotip unresolvable. This blurring is likely due to inhomogeneous conductivity at the tip, charging and instabilities induced by electron beam irradiation. Figure 13 c shows the paradigmatic light guiding and focusing functions of the replicated nanoneedles. When the image focus is at the surface, the cone bases are encircled in a solid line. Focusing a few microns above the surface, emission from the nanotips, with spots approximately 1 μm in size is observed, as indicated by the dotted line, clearly reaching the limits of detection. Considering artificial nanotip probes, such as fiber tips used in SNOM or photon scanning tunneling microscope (PSTM) we note that the dimensions of the nanotips fabricated above have apex radii of similar size, and thus can provide the like operations and potentially multiple sensing applications in the nanoscale. Specifically, a nanotip array in nanometric proximity to a back-illuminated, totally reflecting surface can frustrate the total reflection and retrieve light from the evanescent field, thus enabling the detection of materials or agents on the sample surface. On the other hand, a significant portion of light waveguided in the nanoneedle reaches the propagation cut-off due to the nanotip dimensions and is backreflected. Approaching the nanotip on the surface frustrates total reflection and couples light into the sample. Furthermore, these nanotips can be transformed into plasmonic arrays by deposition of noble metal (Au, Ag, etc.) nanolayers, which will extend their interactions beyond the purely dielectric domain. Conclusively, the complex stereo-topography of the fabricated bioarchitectonic nanotip arrays can offer several advantages in sensing and light-matter interactions, and further work is in progress."
} | 8,960 |
27476989 | null | s2 | 7,670 | {
"abstract": "Cell-free metabolic engineering (CFME) is advancing a powerful paradigm for accelerating the design and synthesis of biosynthetic pathways. However, as most cell-free biomolecule synthesis systems to date use purified enzymes, energy and cofactor balance can be limiting. To address this challenge, we report a new CFME framework for building biosynthetic pathways by mixing multiple crude lysates, or extracts. In our modular approach, cell-free lysates, each selectively enriched with an overexpressed enzyme, are generated in parallel and then combinatorically mixed to construct a full biosynthetic pathway. Endogenous enzymes in the cell-free extract fuel high-level energy and cofactor regeneration. As a model, we apply our framework to synthesize mevalonate, an intermediate in isoprenoid synthesis. We use our approach to rapidly screen enzyme variants, optimize enzyme ratios, and explore cofactor landscapes for improving pathway performance. Further, we show that genomic deletions in the source strain redirect metabolic flux in resultant lysates. In an optimized system, mevalonate was synthesized at 17.6 g·L"
} | 280 |
29853554 | null | s2 | 7,671 | {
"abstract": "A common theme in the self-organization of multicellular tissues is the use of cell-cell signaling networks to induce morphological changes. We used the modular synNotch juxtacrine signaling platform to engineer artificial genetic programs in which specific cell-cell contacts induced changes in cadherin cell adhesion. Despite their simplicity, these minimal intercellular programs were sufficient to yield assemblies with hallmarks of natural developmental systems: robust self-organization into multidomain structures, well-choreographed sequential assembly, cell type divergence, symmetry breaking, and the capacity for regeneration upon injury. The ability of these networks to drive complex structure formation illustrates the power of interlinking cell signaling with cell sorting: Signal-induced spatial reorganization alters the local signals received by each cell, resulting in iterative cycles of cell fate branching. These results provide insights into the evolution of multicellularity and demonstrate the potential to engineer customized self-organizing tissues or materials."
} | 272 |
27896028 | PMC5119243 | pmc | 7,672 | {
"abstract": "Many cnidarians engage in a mutualism with endosymbiotic photosynthetic dinoflagellates that forms the basis of the coral reef ecosystem. Interpartner interaction and regulation includes involvement of the host innate immune system. Basal metazoans, including cnidarians have diverse and complex innate immune repertoires that are just beginning to be described. Scavenger receptors (SR) are a diverse superfamily of innate immunity genes that recognize a broad array of microbial ligands and participate in phagocytosis of invading microbes. The superfamily includes subclades named SR-A through SR-I that are categorized based on the arrangement of sequence domains including the scavenger receptor cysteine rich (SRCR), the C-type lectin (CTLD) and the CD36 domains. Previous functional and gene expression studies on cnidarian-dinoflagellate symbiosis have implicated SR-like proteins in interpartner communication and regulation. In this study, we characterized the SR repertoire from a combination of genomic and transcriptomic resources from six cnidarian species in the Class Anthozoa. We combined these bioinformatic analyses with functional experiments using the SR inhibitor fucoidan to explore a role for SRs in cnidarian symbiosis and immunity. Bioinformatic searches revealed a large diversity of SR-like genes that resembled SR-As, SR-Bs, SR-Es and SR-Is. SRCRs, CTLDs and CD36 domains were identified in multiple sequences in combinations that were highly homologous to vertebrate SRs as well as in proteins with novel domain combinations. Phylogenetic analyses of CD36 domains of the SR-B-like sequences from a diversity of metazoans grouped cnidarian with bilaterian sequences separate from other basal metazoans. All cnidarian sequences grouped together with moderate support in a subclade separately from bilaterian sequences. Functional experiments were carried out on the sea anemone Aiptasia pallida that engages in a symbiosis with Symbiodinium minutum (clade B1). Experimental blocking of the SR ligand binding site with the inhibitor fucoidan reduced the ability of S. minutum to colonize A. pallida suggesting that host SRs play a role in host-symbiont recognition. In addition, incubation of symbiotic anemones with fucoidan elicited an immune response, indicating that host SRs function in immune modulation that results in host tolerance of the symbionts.",
"introduction": "Introduction Cnidarians such as reef-building corals engage in an intimate mutualistic symbiosis with photosynthetic dinoflagellates in the genus Symbiodinium that together form the trophic and structural foundation of coral reef ecosystems. Symbiodinium spp. provide large amounts of reduced organic carbon to the host in exchange for inorganic nutrients, a high light environment and refuge from herbivory ( Yellowlees, Rees & Leggat, 2008 ). In the majority of cnidarian- Symbiodinium interactions, the symbionts are taken up by host cells via phagocytosis. Instead of being digested as food, the symbionts resist host destruction and persist in host cells by residing in vacuoles known as symbiosomes ( Davy, Allemand & Weis, 2012 ) The molecular interplay between host cnidarian and resident symbionts during both the establishment and ongoing maintenance of the symbiosis is critical for a healthy holobiont ( Weis & Allemand, 2009 ). Animal innate immune systems are central to managing microbes by both tolerating and promoting the survival of beneficial symbionts and resisting and destroying negative invaders ( Bordenstein & Theis, 2015 ; McFall-Ngai et al., 2013 ; Schneider & Ayres, 2008 ). With the increased availability of sequence resources, there is now ample evidence that innate immune pathways are ancestral and that basal metazoans including cnidarians possess many of these pathways originally described in mammals and flies ( Fuess et al., 2016 ; Miller et al., 2007 ; Yuen, Bayes & Degnan, 2014 ). Furthermore there are numerous examples of expansions of some innate immune gene families in invertebrates that are larger than those in vertebrate genomic repertoires, including NOD-like receptors, scavenger receptors, TIR-domain-containing proteins and ficolins ( Baumgarten et al., 2015 ; Buckley & Rast, 2015 ; Hamada et al., 2013 ; Pancer, 2000 ; Poole & Weis, 2014 ; Shinzato et al., 2011 ). A class of well-described host-microbe molecular interactions mediated by innate immunity are the PRR-MAMP interactions where microbe-associated molecular patterns (MAMPs) on the surface of microbes, such as lipopolysaccharide or glycans, are recognized by pattern recognition receptors (PRRs) on the surface of host cells ( Janeway & Medzhitov, 2002 ). These steric interactions launch a series of downstream signalling cascades in the host that serve to resist and destroy negative invaders or tolerate and nurture positive microbes. Genomic and transcriptomic studies of cnidarians are revealing the presence of many classical PRRs that have been extensively characterized in higher metazoans ( Fuess et al., 2016 ; Miller et al., 2007 ). One group of PRRs in the Metazoa are the scavenger receptors (SRs), so-named for their role in the scavenging and clearing of microbial invaders, modified host molecules, and apoptotic cell debris ( Areschoug & Gordon, 2009 ; Canton, Neculai & Grinstein, 2013 ). SRs have a high affinity for a wide range of ligands and this flexibility of ligand binding has led them to be described as ‘molecular fly paper’ ( Krieger, 1992 ). A key role of SRs in innate immune function is their action as PRRs on phagocytic cells where they mediate direct non-opsonic phagocytosis of pathogenic microbes ( Areschoug & Gordon, 2009 ) SRs are thought to engage in heteromultimeric signalling complexes, known as signalosomes, involving multiple PRRs and other molecules that together effect signal transduction in cells, thereby alerting them to microbes or modified host molecules ( Canton, Neculai & Grinstein, 2013 ). The SR superfamily is a large group of structurally diverse transmembrane cell surface glycoproteins, divided into nine classes SR-A through SR-I ( Canton, Neculai & Grinstein, 2013 ; Krieger, 2001 ). The classes have overlapping specificities that result in an enormous breadth of MAMP recognition ( Krieger, 1992 ). Members within a given class share some sequence homology, with little-to-no homology occurring between classes. The classes are grouped by their multiple domains with no single domain common to all ( Gordon, 2002 ; Gough & Gordon, 2000 ). SR domains occur on the extracellular portion of the protein; the proteins are anchored in the cell membrane with transmembrane domain(s) and contain short cytoplasmic tail(s). Figure 1 depicts the four SR classes that are relevant to this study. SRs are a potential target for manipulation by invading parasites, pathogens and potentially mutualists. Several pathogens have evolved mechanisms to evade SR-mediated recognition ( Areschoug & Waldemarsson, 2008 ; Faure & Rabourdin-Combe, 2011 ). Indeed, certain human pathogens exploit specific SRs for their own benefit. For example, the Hepatitis C virus (HCV) ( Catanese et al., 2007 ) and the malaria parasite Plasmodium falciparum ( Ndungu et al., 2005 ; Rodrigues et al., 2008 ) have surface ligands that are recognized by SR-B1, and both use this recognition to gain entry to host cells. 10.7717/peerj.2692/fig-1 Figure 1 Domain architecture of vertebrate SRs relevant to this study. All SR sequences are anchored in the membrane with one or two transmembrane domains. All have very short cytoplasmic tails and extensive extracellular ligand-binding domains. SR-As contain a collagen domain(s) and can include an SRCR or a CTLD. SR-Bs have two cytoplasmic tails on either side of a CD36 domain that forms an extracellular loop. SR-Es are defined by the presence of a CTLD. SR-Is have multiple SRCR repeats and no other identifiable extracellular domains. C, carboxy terminus; CTLD, C type lectin domain; LOX1, lectin-like oxidized low density lipoprotein receptor 1; MARCO, macrophage receptor with collagenous structure; N, amino terminus; SRCL, scavenger receptor with C-type lectin; SRCR, scavenger receptor cysteine-rich domain. SR-As and SR-Is contain the scavenger receptor cysteine rich (SRCR) domain, which consists of a 110 aa motif with conserved spacing of six to eight cysteines ( Hohenester, Sasaki & Timpl, 1999 ). The SRCR domain is found in a wide range of membrane and soluble proteins and often occurs in multiple repeats arrayed on the protein ( Hohenester, Sasaki & Timpl, 1999 ; Martinez et al., 2011 ; Sarrias, Grønlund & Padilla, 2004 ). Some SR-As and SR-Es contain C-type lectin domains (CTLDs), a common domain in many proteins, that are often involved in lectin-glycan interactions ( Cambi, Koopman & Figdor, 2005 ). SR-Bs contain the CD36 domain and have two cytoplasmic tails rooted in the membrane with two transmembrane regions, forming an extracellular loop ( Silverstein & Febbraio, 2009 ). SR genes encoding SRCR, CTLD and CD36 domains have been described in invertebrates ( Hibino et al., 2006 ; Lehnert et al., 2014 ; Pancer et al., 1997 ; Schwarz et al., 2007 ; Wood-Charlson & Weis, 2009 ). However a detailed bioinformatic characterization of cnidarian SR genes homologous to vertebrate SR-As, SR-Bs, SR-Es and SR-Is is lacking as are any studies exploring the function of these proteins. SRs are of interest in studies of cnidarian immunity and symbiosis. First, interactions between SR-E-like host lectin-like proteins and symbiont surface glycans play an important role in host-symbiont recognition during onset of symbiosis (reviewed in Davy, Allemand & Weis, 2012 ). In addition, SR-B homologues in two species of sea anemone, Anthopleura elegantissima ( Rodriguez-Lanetty, Phillips & Weis, 2006 ) and Aiptasia pallida ( Lehnert et al., 2014 ) were found to be highly expressed in symbiotic compared to aposymbiotic individuals. For A. pallida this was a dramatic difference in expression where symbiotic anemones had 28-fold greater expression than aposymbiotic animals. These studies suggest that SR-E and SR-B homologues are playing a role in host-symbiont communication. There were two aims for this study. The first was to identify SRs in six cnidarian species, all in Class Anthozoa (corals, sea anemones and others), using a variety of genomic and transcriptomic resources, and compare the repertoire to vertebrate SRs of known function. This provides a platform for identifying potential roles of cnidarian SR proteins in immunity and symbiosis. The second aim was to perform simple functional experiments to examine the role of SRs in symbiont recognition and uptake by the sea anemone A. pallida , a well-studied model system for the study of coral-dinoflagellate symbiosis. We hypothesized that if a symbiont is co-opting host SRs to initiate tolerogenic pathways that dampen or prevent an immune response, blocking SR-ligand-binding capabilities would induce an immune response.",
"discussion": "Discussion An expanded SRCR-domain-containing protein repertoire in cnidarians The SRCR-domain-containing protein repertoire in cnidarians, is expanded compared to that in humans, with the A. pallida genome containing the highest number at 36 genes ( Fig. 2 ). This finding is consistent with numerous other studies describing expansions of innate immune gene families in invertebrates (see ‘Introduction’). Other examples of SRCR-domain-containing protein repertoire expansion have been described in invertebrates, specifically in the sea urchin, Strongylocentrotus purpuratus and the cephalochordate, Branchiostoma floridae , which have 218 and 270 SRCR-containing sequences respectively ( Huang et al., 2008 ; Pancer, 2000 ; Pancer, Rast & Davidson, 1999 ; Rast & Messier-Solek, 2008 ). These numbers are high compared to the 16 genes present in humans. In addition, cnidarian SRCR-domain-containing proteins include a variety of genes with novel domain combinations that have not been found in other organisms ( Fig. 2 ). Identification of these novel domain combinations in cnidarian immune gene repertoires is consistent with other studies of basal metazoan immune genes ( Hamada et al., 2013 ; Poole & Weis, 2014 ; Ryu et al., 2016 ) The searches for SR genes in the three transcriptomes ( Table 1 ) likely revealed underestimates of the total SR repertoire, given that transcriptomes represent snapshots of the whole genome. CTLD-domain-containing SRs in cnidarians In contrast to the human genome, which contains a single LOX1 gene, all six cnidarian resources searched contained multiple LOX1-like SR-Es ( Fig. 2 ). These searches add to previous characterizations of lectin-like proteins in cnidarians, including in corals and sea anemones ( Jimbo et al., 2005 ; Jimbo et al., 2000 ; Kvennefors et al., 2010 ; Kvennefors et al., 2008 ; Meyer & Weis, 2012 ; Vidal-Dupiol et al., 2009 ; Wood-Charlson & Weis, 2009 ). Human LOX1 has a diversity of signalling functions, including in recognition of microbes via host CTLD-microbe glycan binding: a PRR-MAMP interaction ( Canton, Neculai & Grinstein, 2013 ). In cnidarians, previous studies have detailed a role for lectin-glycan interactions in the establishment of cnidarian-dinoflagellate symbioses (reviewed in Davy, Allemand & Weis, 2012 ). The identification of multiple LOX1-like proteins and several other CTLD-containing proteins with novel domain combinations across the six species examined further strengthens the hypothesis that host CTLD-symbiont glycan binding plays an important role in host innate immunity and host-symbiont recognition. Cnidarian CTLD-domain-containing proteins described here provide potential target proteins for future experimental investigation of the lectin-glycan interactions. CD36-domain-containing SRs in cnidarians Phylogenetic analysis of metazoan CD36 domains from SR-B homologues showed a well-supported clade of cnidarian sequences ( Fig. 3 ). A large analysis including additional sequences from basal metazoans is required to more definitively reveal deep branching patterns of this gene. The observed differing location of cysteine pairs within the CD36 domain in cnidarian sequences compared to vertebrate ones also occurred in other invertebrates ( Fig. S1 ). As with the cnidarians searched, C. elegans contained one differing pair and the three sponges, Oscarella carmella , S. domuncula , and Amphimedon queenslandica , and the ctenophore Mnemiopsis leidyi had no sequence pairs in common with vertebrates. These differences may explain why antibodies to human and mouse SR-B1 and CD36 failed to label proteins in A. pallida in immunoblot experiments (EF Neubauer, 2010, unpublished data). Functional experiments suggest that blocking SRs decreases colonization success and increases the stress response to immune challenge in A. pallida Colonization success in aposymbiotic A. pallida challenged with S. minutum CCMP830 displayed a dose-dependent response to incubation in the SR inhibitor fucoidan, exhibiting decreasing colonization success with increasing concentrations of fucoidan ( Fig. 4 ). In vertebrates, fucoidan blocks the positively-charged ligand binding sites on SR-As and SR-Bs and can thereby block phagocytic activity in macrophages ( Dinguirard & Yoshino, 2006 ; Hsu et al., 2001 ; Li et al., 2008 ) The observed inhibition of colonization in cnidarians suggests that phagocytosis of symbionts is likewise inhibited and provides evidence that one or multiple SRs with SRCR and/or CD36 domains function in host-symbiont recognition during onset of symbiosis. Previous transcriptomic studies in A. elegantissima and A. pallida have found SR-B homologues to be upregulated in symbiotic compared to aposymbiotic anemones, suggesting that they play a role in the symbiosis. Our experiments showing that incubation in fucoidan causes a dose-dependent immune response in symbiotic A. pallida ( Fig. 5 ), further implicates a role for SRs in immune tolerance and regulation of symbiosis. In previous work on A. pallida, we showed that symbiotic anemones produced significantly less NO in response to an immune challenge with LPS than did aposymbiotic animals, suggesting that symbionts are modulating the host immune response ( Detournay et al., 2012 ). The increase in this response in symbiotic anemones incubated in fucoidan suggests that this immune modulation involves an SR ligand-binding domain. Such a response is reminiscent of immune modulation by a variety of invading microbes ( Janeway & Medzhitov, 2002 ). In summary, this study provides the first description of the diversity of SRs in cnidarians. Members include proteins with domain combinations that are highly similar to those in vertebrates as well as those that possess novel combinations. Initial functional experiments using the SR inhibitor fucoidan suggest that SRs play a role in the regulation of cnidarian-dinoflagellate symbioses. Future functional studies on candidate SRs identified in this study can further explore their role in cnidarian immunity and symbiosis."
} | 4,290 |
33902739 | PMC8067657 | pmc | 7,675 | {
"abstract": "Background The cyanobacteria Prochlorococcus and Synechococcus are responsible for around 10% of global net primary productivity, serving as part of the foundation of marine food webs. Heterotrophic bacteria are often co-isolated with these picocyanobacteria in seawater enrichment cultures that contain no added organic carbon; heterotrophs grow on organic carbon supplied by the photolithoautotrophs. For examining the selective pressures shaping autotroph/heterotroph interactions, we have made use of unialgal enrichment cultures of Prochlorococcus and Synechococcus maintained for hundreds to thousands of generations in the lab. We examine the diversity of heterotrophs in 74 enrichment cultures of these picocyanobacteria obtained from diverse areas of the global oceans. Results Heterotroph community composition differed between clades and ecotypes of the autotrophic ‘hosts’ but there was significant overlap in heterotroph community composition across these cultures. Collectively, the cultures were comprised of many shared taxa, even at the genus level. Yet, observed differences in community composition were associated with time since isolation, location, depth, and methods of isolation. The majority of heterotrophs in the cultures are rare in the global ocean, but enrichment conditions favor the opportunistic outgrowth of these rare bacteria. However, we found a few examples, such as bacteria in the family Rhodobacteraceae, of heterotrophs that were ubiquitous and abundant in cultures and in the global oceans. We found their abundance in the wild is also positively correlated with that of picocyanobacteria. Conclusions Particular conditions surrounding isolation have a persistent effect on long-term culture composition, likely from bottlenecking and selection that happen during the early stages of enrichment for the picocyanobacteria. We highlight the potential for examining ecologically relevant relationships by identifying patterns of distribution of culture-enriched organisms in the global oceans.",
"conclusion": "Conclusions Although the culture collection being analyzed here was not designed with this study in mind, there are some generalizations we can extract from the heterotrophic “bycatch” that was selected for in the enrichment cultures, which may help guide experiments designed to unravel the co-dependencies in these micro-communities. First, as expected from their obligate oligotrophy, the heterotrophs that numerically dominate the open ocean habitats are not present in these xenic cultures. For most of them, including the abundant SAR11 group, high organic carbon conditions present an obstacle [ 64 , 66 ] and hence developing a defined medium for their growth was a challenge [ 65 ]. Although we previously designed a medium that sustains co-cultures of SAR11 and Prochlorococcus in log phase [ 5 ], SAR11 dies precipitously when Prochlorococcus enters stationary phase, suggesting that strict oligotrophs may not be able to tolerate the accumulation of substrates that occur during Prochlorococcus growth – a feature that would have eliminated them in the initial stages of isolating the cyanobacterial strains in our culture collection. This challenge during stationary phase might derive from deleterious effects of high nutrient concentrations on streamlined cellular physiology, with oligotrophs unable to regulate transport rates in the face of high nutrient concentrations [ 68 ]. The absence of oligotrophs in the enrichment cultures is in striking contrast to the presence of the numerous copiotrophic heterotrophic strains that thrive in these cultures (this study, and [ 26 – 28 , 30 , 33 ]) and even “bloom” when Prochlorococcus reaches stationary phase [ 5 ]. Studies of heterotroph community dynamics in cultures over the course of the exponential and sustained stationary-phase growth [ 42 ] have revealed shifts in heterotrophic abundances according to their differential capacities to utilize high- and low-molecular weight dissolved organic compounds. Further work as in that study (using a combination of transcriptomic, metagenomic, and metabolomic approaches) will enable detailed linkage between functional genes in heterotrophic bacteria for metabolism of cyanobacteria-derived photosynthate and the dynamics of these functions in the global oceans. Interestingly, most of the heterotrophic strains that are widespread among the cultures are not very abundant in the wild. These copiotrophs appear to thrive on high concentrations of organic compounds experienced in culture – an environment that must be patchily distributed in the open ocean habitat [ 15 ]. While it is likely that individual phytoplankton can selectively permit the growth of particular heterotrophic bacteria, the sharing of over 40% of OTUs between our dataset and each of the other phytoplankton datasets explored here suggests that phytoplankton may modify their environments in similar ways that lead to conservation of heterotrophic bacterial groups across cultures. It is possible that ubiquitous heterotrophic bacteria in cultures are similar to “broad-range taxa” that process simple metabolic intermediates [ 67 ], and thus grow on compounds that are released as generic byproducts of phytoplankton growth. Indeed, such compounds are likely abundant on nutrient rich particles, or ephemeral patches of high organic carbon in the phycosphere of larger phytoplankton [ 14 , 15 , 69 ]. Clearly, to understand the metabolic exchanges between picocyanobacteria and oligotrophic heterotrophs we will have to isolate sympatric strains from the same location. In contrast to standard enrichment approaches, which appear to favor the growth of fast-growing, phycosphere-enriched heterotrophic bacteria, using dilution-to-extinction techniques and low nutrient media is likely to yield more of the abundant, free-living bacteria characteristic of the oligotrophic oceans. Like all challenges with these microorganisms, it is only a matter of time and effort.",
"discussion": "Results and discussion The cyanobacterial culture collection The cyanobacterial isolates (co-isolated with unknown consortia of heterotrophs, the subjects of this investigation) used in the analysis span multiple clades of Prochlorococcus and Synechococcus obtained from many locations and times (between 1965 and 2013) in the global oceans (Fig. 1 a, Table S1 ). The isolates – most of which are from the oligotrophic oceans – have been maintained by serial transfer for many years, with cultures ranging in age from 6 to 55 years, and thus represent what we believe to be stable consortia of single algal isolates associated with diverse heterotrophic bacterial communities (see Materials and Methods , and Figure S1 ) (Fig. 1 b).\n Fig. 1 Microbial diversity in enrichment cultures from the global oceans. a Origin of Prochlorococccus and Synechococcus cultures obtained over several decades and maintained in serial transfer batch cultures for 100 s – 1000s of generations. The size of each point on the map represents the number of cultures obtained from a given location. For date and methods of isolation see Table S1 . b Richness of heterotrophic bacteria measured by the number of amplicon sequence variants (ASVs) exceeding a threshold relative abundance of 0.2% in Prochlorococcus (green) and Synechococcus (magenta) cultures. Richness is organized by the phylogeny of host organisms (built using the 16S–23S intertranscribed spacer (ITS) sequence, and collapsed to indicate monophyletic groups) indicated along the bottom. HL (high light) and LL (low light) designations in the triangles refer to light-adaptation features of Prochlorococcus ecotypes, as reviewed in (Biller et al. 2014 [ 39 ]), and 5.1a and 5.1b refer to subclusters of Synechococcus group 5.1 (Ahlgren and Rocap [ 40 ]). Colored boxes above each culture name indicate the years between when the culture was isolated and this analysis, as specified by the color bar in the upper left-hand corner. Inset: regression between culture age and richness Composition and diversity of heterotroph communities in the cultures We anticipated that heterotrophic culture richness – as defined by amplicon sequence variants (ASVs) of the V4 region of 16S rRNA gene – might decrease with age of the culture due to extinctions over time. However, the number of ASVs, which ranged from 1 to 23 (Fig. 1 b), was only weakly and not significantly anti-correlated with the age of the culture (Spearman’s rho = − 0.19, p -value = 0.11). NATL2A, for example, at nearly 30 years old, is one of the oldest cultures but has 22 ASVs, while P1344 is 6 years old and contains 23 ASVs – a difference of just one ASV for an age difference of 24 years. Further, we sampled three sets of cultures ( Prochlorococcus str. MED4, NATL2A, and MIT9313) in 2018 and 1 year later, and found a general correspondence in the community composition over time (Figure S1 ). Cultures of the same Prochlorococcus maintained by different individuals (MIT0604a & MIT0604b) also showed similar community composition, as well as cultures derived from the same starting enrichment culture (e.g. SS2, SS35, SS51, SS52, and SS120 were all derived from the LG culture (Table S1 )) (Fig. 2 ). These results suggest that the heterotroph communities in these cultures remain similar over time and independent maintenance, but exhibit slight drift over time (i.e. the cultures are similar but not identical in composition) (Figure S1 ).\n Fig. 2 Composition of heterotroph communities, as defined by class membership of ASVs, in long-term cultures of Prochlorococcus and Synechococcus hosts (tree on the left as defined in Fig. 1 ). The heterotroph communities in each culture are arranged using the phylogenetic tree of the host cyanobacterium based on the ITS sequence and collapsed into monophyletic groups as in Fig. 1 b. Relative abundance of heterotrophic community members, by class, is shown to the right of each strain name. To view the same data by hierarchical clustering see Figure S2 . We next explored whether the heterotroph community composition was related to the cyanobacterial “host” – i.e. Prochlorococcus or Synechococcus – in the cultures. Which taxonomic groups were common to both, and which were more common in one or the other? Grouping ASVs at the phylum level, heterotroph communities in the cultures were largely comprised of bacteria from the phyla Proteobacteria (primarily in the classes Alpha- and Gammaproteobacteria, with some Delta- and Betaproteobacteria), Bacteroidetes (classes Cytophagia, Flavobacteria, and Rhodothermi), and Planctomycetes (class Phycisphaerae), with a minor contribution from Spirochaetes (class Leptospirae) and candidate phylum SBR1093 (class A712011) (Fig. 2 , Figure S2 ). SBR1093 was only in cultures of the LLIV clade of Prochlorococcus (Fig. 2 ). With respect to heterotrophic phyla, Prochlorococcus and Synechococcus cultures contain Proteobacteria and Planctomycetes at similar frequencies, but Bacteroidetes are more common, though not significantly (Fisher Exact Test p -values > 0.05), in Synechococcus cultures (present in 94% of cultures versus 80%) (Fig. 3 a). At the class level, Alpha- and Gammaproteobacteria as well as Phycisphaerae have equal representation across the two cyanobacterial hosts (Fig. 3 b). Flavobacteria (present in 90% of Synechococcus cultures versus 56% of Prochlorococcus , Fisher Exact Test p -value = 0.01) and Rhodothermi (present in 56% of Synechococcus cultures versus 14% of Prochlorococcus , Fisher Exact Test p -value = 0.06), however, are more well-represented in Synechococcus cultures, suggesting that compounds in the exudate derived from Synechococcus may be better matched to their growth requirements. Indeed, Prochlorococcus and Synechococcus secrete distinct compounds during growth [ 12 ], few of which are characterized, but likely promote differential growth of associated heterotrophs as shown previously [ 41 , 42 ].\n Fig. 3 Proportion of cultures of either Prochlorococcus (green) or Synechococcus (magenta) containing ASVs belonging to a given bacterial ( a ) phylum, b class, or c genus (restricted to genera found in at least six cultures). Note that HTCC is listed in the Greengenes taxonomy, but is not formally recognized as a bacterial genus Minor differences in heterotroph communities between Prochlorococcus and Synechococcus hosts at the phylum and class level suggested that there might be more pronounced differences at the genus level. This was not the case, however; instead they were quite similar (no significant differences after multiple hypothesis correction) (Fig. 3 c). For instance, the most ubiquitous genus (present in over 60% of both culture types) is Marinobacter (Fig. 3 c), which is well-known to be associated with picocyanobacteria in culture [ 30 ]. In addition to Marinobacter , other genera present in at least 15% of cultures including Thalassospira , Methylophaga , Alteromonas , Alcanivorax , Maricaulis, Muricauda , and Hyphomonas (but not Shimia ) have been associated with metabolism of hydrocarbons or C1 compounds derived from lipid catabolism [ 10 , 43 – 54 ], suggesting that hydrocarbon metabolism might play an important role in their growth in these cultures. Indeed, previous work showed an upregulation of genes for fatty acid metabolism including lipid beta-oxidation in co-culture of Alteromonas macleodii with Prochlorococcus [ 28 ]. Prochlorococcus secretes vesicles (potentially a source of lipids) in culture and in the wild [ 39 , 55 ], which can outnumber cells by a factor of 10 or more, and marine heterotrophs like Alteromonas are capable of growing on these vesicles as a sole source of carbon [ 39 ]. Recent work also suggests that Alteromonas isolates derived from cultures of Prochlorococcus carry genes for the degradation of aromatic compounds produced by the cyanobacterium [ 47 ], which may provide a selective advantage for Alteromonas in culture with Prochlorococcus . Additionally, metagenomics on a Synechococcus -associated culture revealed high abundance of TonB-dependent transporters in Muricauda , potentially involved in lipid uptake, and proteins involved in the export of lipids in the proteome of the Synechococcus host [ 42 ]. Finally, we investigated whether specific ASVs were differentially represented between culture types ( Prochlorococcus versus Synechococcus ). To identify taxa associated with a specific habitat (here either a Synechococcus or Prochlorococcus culture), we used indicator species analysis (see Materials and Methods ), based on presence/absence data. We found that 20 ASVs were more prevalent among Synechococcus cultures than Prochlorococcus cultures, but only one ASV (a Marinobacter sequence) more prevalent in Prochlorococcus cultures (Table S2 ). These associations strengthen the possibility that the structure of heterotroph communities may arise in response to cyanobacterial host-specific secretion of compounds. Consistent with there being more indicator species in Synechococcus cultures, community composition was more similar across Synechococcus cultures than Prochlorococcus cultures (as measured by unweighted UniFrac distance at the ASV level (PERMANOVA, p -value < 0.01)) (Fig. 4 a, Figures S3 and S4 ). Further, heterotroph community composition using the same metric was significantly associated (PERMANOVA, p -value < 0.01, Table S5 ) with cyanobacterial ecotype (Figure S4 A) and clade (Figure S4 C), but there were no discernible patterns in these associations.\n Fig. 4 Ordination of heterotroph community composition at the ASV level overlaid with metadata pertaining to conditions of isolation of enrichment cultures. Non-metric multidimensional scaling (NMDS) of heterotroph communities using unweighted UniFrac as the distance metric overlaid with ( a ) the host cyanobacterium genus in the enrichment culture, b the cruise name from which the culture was isolated, c the isolation method (pre-filtered to exclude larger cells, pre-filtered and sorted via flow cytometry, pre-filtered and diluted-to-extinction (DTE), not filtered and sorted via flow cytometry, or cloned from a mother culture), and d the depth of the seawater sample. The closeness of two points in NMDS space reflects the distance between communities, with communities having more similar phylogenetic structure (as measured by unweighted UniFrac) grouping more closely together. Heterotroph communities for which metadata was not known are indicated as NAs. See also Figures S3 , and S4 for relationships between community structure and ecotype, clade, isolation location, and culture age Comparison of composition with similar studies of diatoms and Synechococcus To assess the similarity of other phytoplankton-associated microbiomes to those obtained in this study we re-processed – in the context of our dataset – 16S rDNA amplicon sequence data as 97% OTUs (see Materials and Methods ) to limit noise and batch effects from differences in sample processing and sequencing approaches – from published diatom (Behringer et al. 2018 [ 56 ]) and recently isolated Synechococcus -associated microbial communities (Zheng et al. 2018 [ 41 ]). Surprisingly, the heterotroph communities in diatom and Synechococcus cultures each shared almost half the OTUs found in our cultures: 45% (13/29) for the diatoms and 41% (29/70) for the Synechococcus cultures (Table S3 , Figure S5 ). Among these OTUs, 11 were present in cultures of all three groups of phytoplankton, potentially comprising a ‘core’ set of sequence clusters associated with phytoplankton isolates. These include three of the ubiquitous genera ( Roseivirga , Maricaulis , and Alteromonas ) from this study, which derive from three separate classes (Fig. 3 c). At a high level, the taxonomic diversity in cultures from A. tamarense and T. pseudonana in a separate study (Fu et al. [ 57 ]) exhibited many of the same marine heterotrophs seen here (Figure S6 ), which suggests that diverse phytoplankton may have overlapping selective effects on their associated microbial communities. In contrast to this core of shared heterotrophic bacteria, several additional classes (Acidimicrobiia, Actinobacteria, Planctomycetes class OM190, and Saprospirae) were represented in the Zheng et al. 2018 [ 41 ] Synechococcus dataset, but absent from our study; only the class Saprospirae was present in the diatom dataset and absent in ours (Figure S6 ). Notably, the Actinobacteria from the Synechococcus dataset were only found in isolates from eutrophic waters; by contrast, all except for two ( Prochlorococcus SB and Synechococcus WH8017) of our isolates were from oligotrophic waters, while the diatoms of Behringer et al. 2018 [ 56 ] were isolated from coastal waters. Further work should examine the extent to which these differences are driven by starting inoculum versus physiological features of the host phytoplankton. Features of the sample of origin in determining community composition We examined the extent to which differences in heterotroph communities – here measured at the level of ASVs – in the cultures were related to factors involved in isolation (Table S1 ). Specifically, we investigated whether the heterotroph composition between all pairs of communities varied with: cruise on which it was isolated, isolation method, sample location and depth, and date the culture was isolated. Each of the tested metadata variables showed a significant association (PERMANOVA, p -value < 0.01, Table S5 ) with differences in community composition as measured by unweighted UniFrac distance (a measure of the phylogenetic similarity between two communities), and there were no obvious cases in which the pattern of associations for one variable completely overlapped those for another (Fig. 4 , Figure S3 , Figure S4 ). Notably, richness (number of ASVs) in the cultures showed no association with any of the tested variables, as previously mentioned for culture age (Wilcoxon Rank Sum Test, p -value > 0.05). In other words, differences in the heterotroph membership of communities (unweighted UniFrac distance) associated with features of the sample of origin did not arise from differences in the total number of heterotroph ASVs (richness). While “cruise” itself is not a particularly informative variable on its own, we did see a reproducible tendency for cultures obtained on the same cruise to have similar heterotroph community composition (PERMANOVA, p -value < 0.01, Table S5 ). For example, cultures with names beginning in MIT13 or P13 (i.e. MIT13 XX or P13 X ) , were isolated by the same person on the same cruise (HOE-PhoR) in 2013 from a depth of 150 m, and include Prochlorococcus from multiple low light clades (LLIV, LLII/LLIII, and LLVII) (Table S1 ). They were isolated with a variety of methods including filtration, flow sorting, and dilution-to-extinction. With two exceptions (the LLIV cultures P1344 and MIT1313, the latter of which was dominated by a single heterotroph ASV after dilution-to-extinction) from the suite of eight cultures, the MIT13 XX and P13 X strains clustered together by unweighted UniFrac, suggesting a sensitive dependence of heterotroph community composition on the conditions – not solely the physical isolation methodology – under which the culture was first isolated (indeed, we see differences in beta-dispersion across several of the cruises, 10/39 pairwise comparisons with p -value < 0.05, 1/39 after Bonferroni correction) (Fig. 4 b and Figure S2 ). These conditions include the initial composition of heterotrophic bacteria in the collected seawater sample, the specific media formulation or light and temperature conditions used for a given enrichment attempt, and the duration of time an enrichment culture was maintained before derivation of individual algal strains. Together, these factors may drive the observation that cultures obtained from a given cruise frequently share similar heterotroph communities (seen also with the CoFeMUG and EqPac/IRONEX cruises, whose beta-dispersion differed significantly, Bonferroni-adjusted p -value = 0.039) (Fig. 4 b). We hypothesize that these similarities result from a bottlenecking of community diversity shortly after cultures are sampled and phytoplankton are enriched. To look into the effects of potential bottlenecking during the initial culturing phase, we examined the relationship between culture composition and physical isolation methods (Fig. 4 c). We see a tendency for cultures obtained using only filtration (not accompanied by dilution-to-extinction or flow sorting) to be more similar in heterotroph composition to each other than cultures also subjected to dilution-to-extinction or flow sorting (post-hoc tests on pairwise dispersion, Bonferroni adjusted p -values = 0.02) (Fig. 4 c). These findings suggest that stochastic loss of heterotrophs that might occur by dilution-to-extinction reduces the structural convergence of heterotroph communities. Finally, heterotroph community composition in the cultures differs by collection depth (Fig. 4 d). For example, the heterotroph communities of cultures isolated from 0 to 50 m and 50–100 m tend to cluster more tightly with each other than cultures isolated from 100 to 150 m and 150–200 m, consistent with the idea that community composition of heterotrophs varies with depth (post-hoc test on pairwise dispersion between 100 and 150 m and 0–50 m, Bonferroni adjusted p -value = 0.006) [ 58 – 60 ]. This finding is not surprising given that light and nutrient gradients in the water column lead to more complex biogeochemical regimes – and hence a gradient in dissolved organic carbon compounds [ 61 , 62 ] from the surface. In conducting the above analyses, we note that some of these variables are not independent (e.g. because ecotype depends on clade; or because strains were only isolated from one depth or one ecotype was isolated on a given cruise), so associations may arise from interdependencies between factors (Figure S7 , Table S5 ). These interdependencies may lead to associations with multiple variables which could be explained by a single variable. However, the overall trends show a clear linkage between initial conditions of culture isolation and the long-term composition of the culture. Comparison of cultured communities to distributions of their members in the wild Because all of the cultures are sourced from seawater samples, we wanted to determine how heterotrophs present in cultures were represented in the global oceans. We used the bioGEOTRACES metagenomics dataset, which spans 610 samples over time and at multiple depths in locations throughout the Atlantic and Pacific Oceans [ 63 ] to examine the spatial distributions of heterotrophs present in our cyanobacterial cultures (Fig. 5 ). We obtained reads mapping to the V4 region of the 16S gene, and clustered the ASVs from our culture collection with the global oceans data at 97% similarity (i.e. 97% OTUs) to accommodate differences in the nature of the data and processing (see Materials and Methods for details). The heterotrophs that dominate global ocean datasets (primarily belonging to the Pelagibacteraceae within the Pelagibacterales) are not well represented in our cultures (Fig. 5 b, Figure S8 ). This absence is not surprising as simply culturing oligotrophs like SAR11 is extremely challenging [ 64 – 66 ]; thus, we expected that the high nutrient concentrations and periodic dilution of cells from culture transfers would enrich for opportunistic copiotrophic organisms.\n Fig. 5 Representation of picocyanobacterial enrichment culture heterotrophic bacterial OTUs in global ocean surveys. a Location of the sampling sites of the bioGEOTRACES expeditions. b Example of a heterotrophic OTU (in the family Pelagibacteraceae) abundant in bioGEOTRACES, but absent from cultures. c Distribution of the most ubiquitous OTUs in cultures across the bioGEOTRACES sites. Such OTUs were present in almost all bioGEOTRACES sites (Cosmopolitan), some sites (Intermediate), or only a few sites (Sparse). Scale bar indicates the log10 relative abundance (number of reads normalized by number of non -Prochlorococcus, non- Synechococcus reads) at a given site. See also Figure S8 for the distribution of some OTUs abundant in bioGEOTRACES, but not prevalent in culture We next asked how well the prevalent heterotrophic OTUs in cultures were represented in the bioGEOTRACES database. While a few OTUs were readily detected at most sites (Fig. 5 c, Figure S8 ), this was not the general tendency; most of the heterotrophic OTUs that were prevalent in the cultures were rare or sparsely detectable in the wild (Fig. 5 c). Of the OTUs that dominated in culture, OTUs that were cosmopolitan across bioGEOTRACES sites included those classified as Maricaulis , Alteromonadaceae, and Rhodobacteraceae, while unclassified Gammaproteobacteria, Alcanivorax , and Marinobacter were intermediately distributed OTUs present in several sites, and OTUs in the genera Muricauda , Thalassospira , and Hyphomonas were generally sparse or absent across the bioGEOTRACES sites (Fig. 5 c). The sparsity of these heterotrophs in the global oceans coupled with their prevalence in cultures of Prochlorococcus and Synechococcus suggests that they are selected for by the rich culture conditions – conditions that must be patchily distributed in the oceans. As an example, the genus Muricauda (family Flavobacteriaceae), which had an OTU well-represented in cultures, but sparsely detectable in bioGEOTRACES (Fig. 5 c), is known to be particle-associated, potentially specializing in degradation of high molecular weight organic compounds [ 67 ]. Across all cultures in this study, more heterotrophic OTUs in culture were shared with bioGEOTRACES samples taken below the epipelagic zone (> 200 m) than above (0.64% of OTUs below versus 0.3% above, Fisher Exact Test p -value = 1e-4). The increase in OTUs shared with cultures at depth might be because of the reliance of heterotrophs on organic carbon as a source of energy in both systems: light energy for photoheterotrophy is unavailable at depth and likely restricted by algae in cultures. Given the cosmopolitan distribution of some of the prevalent culture heterotrophic OTUs across bioGEOTRACES sites, we expected that they might be positively correlated with picocyanobacterial abundance along the transects. Indeed, we find that across the bioGEOTRACES sites, there is a strong relationship (Spearman’s rho = 0.302, p -value = 0) between the abundance of the Rhodobacteraceae OTU, which is ubiquitous in cultures, and the combined abundance of Prochlorococcus and Synechococcus . Notably, none of the other prevalent culture OTUs showed this relationship, suggesting that this OTU in particular may be coupled to the dynamics of these cyanobacteria in oceans (Table S4 ). Further laboratory investigations of the interactions of these ubiquitous bacteria with marine picocyanobacteria should reveal interesting and relevant exchanges of matter and energy in the global oceans."
} | 7,337 |
30992508 | PMC6467887 | pmc | 7,676 | {
"abstract": "The application of fertilisers incorporated with plant residues improves nutrient availability in soils, which shifts the microbial community structure and favours plant growth. To understand the impact of wheat straw compost fertiliser on soil properties and microbial community structure, tobacco planting soils were treated with four different fertilisers using varied amounts of straw compost fertiliser and a no fertiliser control (CK). Results showed that different fertilisers affected available soil nutrient contents differently. Treatment of tobacco soil with application of combined chemical fertiliser/wheat straw compost led to improved soil chemical properties, and increased soil organic matter and available phosphorus and potassium content. Treatment with FT1 200 kg/mu straw was found to be superior in improving soil fertility. Metagenomic DNA sequencing revealed that different fertiliser treatments resulted in changes in the microbial community composition. In soil treated with FT2 300 kg/mu straw for 60 days, the predominant bacterial phyla were Proteobacteria, Actinobacteria, and Verrucomicrobia, whereas Cyanobacteria, Basidiomycota, and Chlorophyta were found in high abundance in soil samples treated with FT1 200 kg/mu straw for 30 days. Functional annotation of metagenomic sequences revealed that genes involved in metabolic pathways were among the most abundant type. PCoA analysis clearly separated the samples containing straw compost fertiliser and chemical fertiliser. A significant correlation between soil properties and the dominant phyla was identified.",
"conclusion": "Conclusion In conclusion, chemical fertiliser combined with wheat straw compost fertiliser improved soil fertility, tobacco quality, and economic traits. It also significantly shifted the microbial community structure. A significant correlation was identified between soil properties and the dominant phyla under different fertilisation treatments in our study. The chemical fertiliser combined with straw compost fertiliser can effectively improve the soil traits and microbial diversity. The straw compost fertiliser could be widely used in agricultural production. The application of straw compost fertiliser may cure soil contamination induced by chemical products.",
"introduction": "Introduction Environmental soil degradation and production sustainability in different agricultural systems have aroused public concern about soil fertility and quality 1 . As previous studies have shown favourable effects of soil nutrients (e.g. organic C, N, P and K) on the physical, chemical, and biological properties of soil, concentrations of these soil nutrients are good indicators of soil quality and productivity 2 . Long-term continuous cropping leads to lack of carbon in the soil, leading to an imbalance in carbon and nitrogen and a reduction in soil nutrient availability 3 . With the development of agricultural production, appropriate fertilisation has been widely used as an important management practice for maintaining soil fertility and quality 4 and improving crop yields 5 , 6 . The long-term overuse of chemical fertilisers induces soil compaction and acidification in tobacco planting soils 3 , 7 . Therefore, organic and inorganic amendments, including application of nitrogen fertilisers in combination with straw are returning as recommended approaches to increase the availability of nutrients and improve the yield and quality of plants, including tobacco 8 . Previous studies have shown that different fertilisation strategies influence carbon cycling in soil 9 , 10 , and therefore, influence the phylogenetic structure of the soil microbial community 11 . Results from previous studies have shown that nitrogen amendments affect a wide range of bacterial 12 and fungal communities 13 . Organic and inorganic fertiliser amendments can also affect the composition of soil microbial communities 14 , 15 . These shifts in community composition are likely associated with changes in the functional capabilities of the communities as well. Moreover, a recent laboratory-based study 16 supports the concept that soil microbial communities are neither functionally redundant nor similar. Recent research has demonstrated that prolonged elevations in nitrogen availability may directly or indirectly change the microbial carbon dynamics, thereby shifting the composition of soil microbial communities, their catabolic capabilities, and metagenomes 17 . A recent study indicated that the application of straw, with and without straw decomposer, could shift the soil bacterial community structure, specifically activating the copiotrophic bacteria and increasing the soil biological activity, thus, contributing to soil productivity and sustainability in agro-ecosystems 6 . Increasing number of research studies have demonstrated that chemical fertilisers with straw significantly enhance bacterial abundance 18 , 19 . Recent studies have demonstrated metagenomic, phylogenetic, and physiological responses of microbial communities in soil 17 . Indeed, advances in next-generation DNA sequencing approaches combined with traditional microbiological and chemical analyses of soil parameters may provide a biologically relevant assay to assess the potential effects of fertiliser applications and straw compost amendments on soil microbial communities. Thus, in this study, we investigated the impact of different fertiliser treatments on microbial communities in tobacco planting soils. Soil samples were collected from a long-term field experiment involving use of conventional fertiliser treatments alone or in combination with wheat straw compost. Tobacco traits and soil properties were analysed and a metagenomics sequencing approach was undertaken to examine the microbiological profile diversity.",
"discussion": "Discussion The crop yield and healthy food production is dependent on plant nutrients. The type of fertilisers used is essential for the increased production to provide sufficient nutrients for plants. The sound management of fertilisers is essential to ensure the plant production and environmental safety. In this study, we evaluated the effects of chemical fertiliser treatments with and without wheat straw compost fertiliser on microbial communities in tobacco planting soils by metagenomics sequencing. Previous studies have shown that environmental factors play important roles in shaping microbial community structure and composition 20 – 22 . Soil pH is found to be an important factor influencing the chemical reactions in soil, and it also affects the availability of soil nutrients impacting plant growth 23 . Notably, pH is reported as the most important factor for determining the microbial community structure in natural environmental systems 24 , 25 . Our results show that soil amended with wheat straw compost fertiliser maintained pH at 5.7, slightly lower than that with chemical treatment alone (FCK2 pure nitrogen). Although wheat straw compost fertiliser had a relatively small influence on soil pH, a slightly reduced pH relative to other soil treatments may significantly affect soil microbial properties, nutrient availability, root growth, and tobacco plant yields. Previous reports suggest that variation in pH may induce the abundance and composition of acidobacterial community 25 , 26 . Increased abundances of Acidobacteria versus other bacterial phyla have been observed in more acidic soils 14 . We found Acidobacteria as a predominant phylum and had greater relative abundances in soil treated with wheat straw compost fertilisers. In addition, variations in soil nutrient properties (e.g. available carbon, total nitrogen, available potassium) commonly cause shifts in soil microbial communities with short-term fertilisation treatments 27 – 29 . Fertilisation may be a contributing factor favouring a more active, copiotrophic microbial community, thus shifting the predominant microbial life history strategies 17 . Soil organic matter is a critical index of soil fertility due to its capacity to affect plant growth indirectly or directly 30 . Organic fertiliser is a source of organic components 13 , promoting the growth of microbes in soil. Dynamics of soil organic carbon and total nitrogen storage in soils determine microbial activity and nutrient cycles, improves soil physical properties, promotes water retention capacity, and decreases erosion 14 . A great number of genes involved in the function of carbon metabolism were enriched in the associated microbial communities. Proteobacteria, which is known to play critical role in the global cycles of carbon, nitrogen, iron, and sulphur 31 – 33 , was the most abundant group in soil. Actinobacteria were overrepresented with fertiliser treatments in this study. Actinobacteria are believed to contribute to the global carbon cycle by breaking down plant biomass 28 , and due to their ability to decompose organic matter in soils 34 , they are capable of producing cellulases, hemicellulases, chitinases, glucanases, and amylases 15 . In line with previous studies showing positive correlations between soil properties and microbial communities, our results identified correlations between soil properties and the dominant phyla. A recent study found that nitrogen, phosphorus, potassium, and combined NPK treatments shifted bacterial community compositions, and under these four fertiliser treatments, Actinobacteria were predominant compared to the control. Redundancy analysis of bacterial community profiles and soil parameters revealed that available trace elements (Mg, total N, Cd, and Al) were positively correlated with variations in community composition 15 . Furthermore, the application of different amounts of wheat straw compost fertiliser had a certain influence on the number of microbes in the rhizosphere soil of the tobacco field. Communities of bacteria and actinomycetes were highest in the FT1 200 kg/mu straw treatment and the fungal community reached a peak in the FT2 300 kg/mu straw treatment. The communities of bacteria and fungi reached a peak at 60 days post-transplantation whereas the actinomycete community reached its peak at 90 days. These microbial delineations were positively correlated with the amount of wheat straw compost fertiliser. The quality of the soil surrounding the root is dependent on the organic material from the vicinity of root and the microbial activity on nutrient cycling and plant growth. The composition of bacteria and fungi populations has positive, negative, or neutral role on plants depending on the soil conditions 35 . One goal of fertilisation management is to increase the beneficial effects of soil microbial community. However, the association between fertilisation and the population of bacteria and fungi has not been determined. Further analyses are urgently needed. The application of different wheat straw compost fertiliser concentrations had a certain influence on the economic traits of tobacco. With different applications of wheat straw compost fertiliser, the content of total sugars, reducing sugars, chlorine and potassium in roots and leaves and the content of nicotine in root system increased to certain extent. The increased total sugars and reducing sugars under the use of wheat straw compost fertiliser treatments did improve the performance of flue-cured samples. High levels of chlorine have been shown to affect the flammability of tobacco, while high levels of potassium can lead to a lack of coordination between potassium and chlorine 36 – 38 . In the present study, nicotine and potassium contents were not affected significantly but increased applications of wheat straw compost fertiliser did lead to an increase in chlorine content. It is reported that microbial metabolism is responsible for the fate of chemical components in the biosphere 39 . β-Proteobacterium has been found to be capable of degradation of chlorides and provide a critical link in the chain of microbial metabolism 40 . In this study, Proteobacteria was found to have high abundance in straw fertiliser treated soil, which may contribute to the increase of chlorine content in soils. Overall, the application of FT1 200 kg/mu straw showed a comprehensive superiority in the overall desirable tobacco traits. We speculated that the dominated microbial populations play key roles in the variety of soil properties. Therefore, to improve tobacco quality and its economic qualities, growers need to consider applying an appropriate amount of wheat straw compost fertiliser during growth."
} | 3,135 |
21255366 | PMC3158428 | pmc | 7,678 | {
"abstract": "Summary The global transcriptional regulator Hha of Escherichia coli controls biofilm formation and virulence. Previously, we showed that Hha decreases initial biofilm formation; here, we engineered Hha for two goals: to increase biofilm dispersal and to reduce biofilm formation. Using random mutagenesis, Hha variant Hha13D6 (D22V, L40R, V42I and D48A) was obtained that causes nearly complete biofilm dispersal (96%) by increasing apoptosis without affecting initial biofilm formation. Hha13D6 caused cell death probably by the activation of proteases since Hha‐mediated dispersal was dependent on protease HslV. Hha variant Hha24E9 (K62X) was also obtained that decreased biofilm formation by inducing gadW , glpT and phnF but that did not alter biofilm dispersal. Hence, Hha may be engineered to influence both biofilm dispersal and formation.",
"introduction": "Introduction Hha (high haemolysin activity, 72 aa) belongs to the Hha‐YmoA family of low‐molecular‐mass proteins (about 8 kDa) that regulate many genes in Gram‐negative bacteria ( Madrid et al ., 2007 ); Hha was identified as reducing haemolysin production by repressing the hly operon of Escherichia coli ( Godessart et al ., 1988 ). However, Hha does not bind specific DNA sequences; instead, Hha has a protein partner, H‐NS, and this complex binds specific sequences of the hly regulatory operon ( Madrid et al ., 2002 ) as well as 162 genes controlled by the Hha‐H‐NS complex ( Baños et al ., 2009 ). Hha also represses the pathogenicity locus of enterocyte effacement (LEE) set of operons in enterohemorrhagic E. coli by repressing transcription of ler , which encodes the activator of LEE ( Sharma et al ., 2005 ). We found that Hha is induced 30‐fold in E. coli biofilms relative to planktonic cells ( Ren et al ., 2004a ) and that Hha decreases initial biofilm formation by repressing the transcription of rare codon tRNAs and by repressing transcription of fimbrial genes ( García‐Contreras et al ., 2008 ). Hha is also toxic and leads to cell lysis and biofilm dispersal due to activation of prophage lytic genes appY and rzpD of DLP12 and alpA and yfjZ of CP4‐57, and due to the induction of protease ClpXP ( García‐Contreras et al ., 2008 ). In addition, Hha induces excision of prophages CP4‐57 and DLP‐12 of E. coli ( Wang et al ., 2009 ). Hence, Hha is a global transcriptional regulator in modulating cell physiology. Biofilms consist of a complex heterogeneity of cells ( Stewart and Franklin, 2008 ) that are the result of diverse signals and regulatory networks during biofilm development ( Prüß et al ., 2006 ). The final stage of the biofilm cycle is the dispersal of cells from the biofilm into the environment ( Kaplan, 2010 ). Like other stages of biofilm development, biofilm dispersal is a highly regulated process involving many sensory circuits ( Karatan and Watnick, 2009 ). There have been several biofilm dispersal signals proposed including the auto‐inducing peptide of the agr quorum sensing system of Staphylococcus aureus that triggers biofilm detachment ( Boles and Horswill, 2008 ) and changes in carbon sources ( Sauer et al ., 2004 ). To date, little is known about the intracellular molecular mechanisms of bacterial biofilm dispersal ( Kaplan, 2010 ); however, phosphodiesterases can decrease concentrations of the secondary messenger cyclic diguanosine‐monophosphate (c‐di‐GMP) ( Karatan and Watnick, 2009 ), which results in increased motility ( Kaplan, 2010 ), and surfactants such as rhamnolipid produced by Pseudomonas aeruginosa can cause dispersal of biofilms ( Boles et al ., 2005 ). In addition, cell lysis may accompany the dispersal process; hence, biofilm dispersal in P. aeruginosa may be achieved through prophage‐mediated cell death ( Webb et al ., 2003 ) and in Pseudoalteromonas tunicata biofilm dispersal involves autolysis via AlpP ( Mai‐Prochnow et al ., 2006 ). To control biofilm formation, nitric oxide was investigated as a dispersal agent not only for single‐species biofilms including Gram‐negative and Gram‐positive bacteria and yeasts but also for multi‐species biofilms of clinical and industrial relevance ( Barraud et al ., 2009 ). Also, T7 bacteriophage was engineered so that during bacteriophage infection dispersin B of Actinobacillus actinomycetemcomitans is produced to hydrolyse the glycosidic linkages of polymeric β‐1,6‐ N ‐acetyl‐ d ‐glucosamine found in the biofilm matrix ( Lu and Collins, 2007 ). Protein engineering and recombinant engineering are promising strategies to control biofilm formation, but they have not been applied for promoting biofilm dispersal. Previously, we created the first synthetic circuit for controlling biofilm formation via an external signal by manipulating indole concentrations via toluene o‐ monooxygenase in a consortium of Pseudomonas fluorescens and E. coli ( Lee et al ., 2007a ). In addition, we controlled E. coli biofilm formation by rewiring the quorum sensing regulator SdiA; upon addition of the extracellular signal N ‐acylhomoserine lactone, biofilm formation was increased with SdiA variant 2D10, and SdiA variant 1E11 increased concentrations of the biofilm inhibitor indole and thereby reduced biofilm formation ( Lee et al ., 2009 ). Furthermore, we engineered global regulator H‐NS to control biofilm formation via prophage excision and cell death ( Hong et al ., 2010 ); however, the engineered H‐NS has no effect on biofilm dispersal. These results showed that bacterial biofilm formation may be controlled by manipulating key regulatory proteins and enzymes. In this study, protein engineering was used to rewire the global transcriptional regulator Hha to control biofilms. The aims of this study were to engineer Hha to enhance biofilm dispersal, to engineer Hha to reduce biofilm formation, and to investigate the mechanism by which the Hha variants influence biofilm formation. Utilizing whole‐transcriptome analyses, random mutagenesis, saturation mutagenesis, site‐directed mutagenesis, double mutations, quantitative real‐time reverse transcription PCR (qRT‐PCR) and quantitative PCR (qPCR), we show Hha may be rewired to promote biofilm dispersal by interacting with protease HslV. We also show that Hha may be engineered to decrease biofilm formation.",
"discussion": "Discussion In this work, we show that the global transcriptional regulator Hha may be rewired to control biofilm dispersal, without affecting initial biofilm formation. In static biofilms, Hha13D6 removed 70% of the biofilm while there was little dispersal with wild‐type Hha ( Fig. 1A ), and in flow‐cells, the engineered Hha removed 96% of the biofilm ( Fig. 1B ), whereas wild‐type Hha removed 35% of the biofilm ( Fig. 1B ). Hha13D6 increased biofilm dispersal by causing apoptosis ( Fig. 3C ) by activating proteases (Table S1); hence, we have created one of the first protein switches to control cell lysis. Biofilm dispersal by the engineered Hha requires protease HslV since adding the hslV mutation reduced the dispersal activity of Hha13D6 ( Fig. 1D ). Cell death in biofilms is an important mechanism of dispersal since it leads to the creation of voids inside the biofilm, which facilitates dispersal of the surviving cells ( Webb et al ., 2003 ). Hence, programmed cell death is a social (altruistic) activity ( Wood, 2009 ). Cell death in biofilms is mediated by prophage ( Webb et al ., 2003 ) and autolysis proteins ( Mai‐Prochnow et al ., 2006 ). Since the biofilm matrix consists of structural proteins ( Karatan and Watnick, 2009 ), protease activity is necessary for biofilm dispersal; for example, agr ‐mediated detachment in Staphylococcus aureus biofilms does not occur in the presence of a protease inhibitor and is induced by extracellular proteases ( Boles and Horswill, 2008 ). Also, wild‐type Hha disperses E. coli biofilms by inducing protease ClpXP that activates toxins by degrading antitoxins ( García‐Contreras et al ., 2008 ). Hence, induction of protease activity is a key mechanism of Hha13D6 to trigger biofilm dispersal by causing cell death and so by disrupting biofilm structure. In addition to the evolution of Hha for biofilm dispersal, we also show that Hha may be reconfigured to control biofilm formation by inducing the activity of GadW, GlpT and PhnF ( Fig. 2B ). We showed previously that biofilm formation is connected to the acid resistance system, showing increased biofilm formation in the absence of acid resistance genes such as gadW , gadABC and hdeABD ( Domka et al ., 2007 ; Lee et al ., 2007d ), and GadW is dual regulator with GadX of the glutamate‐dependent decarboxylase acid‐resistance system of E. coli ( Sayed et al ., 2007 ). Hence, the engineered Hha24E9 uses the acid resistance regulatory system to inhibit biofilm formation. We found the carboxy‐terminus of Hha is important for controlling biofilm dispersal but not for biofilm formation. Hha is a non‐specific DNA binding protein, but the Hha‐H‐NS complex allows specific binding to target DNA sequences to regulate transcription ( Madrid et al ., 2002 ). Hha consists of four α‐helices separated by a loop: helix 1 (8–16 aa), helix 2 (21–34 aa), helix 3 (37–55 aa) and helix 4 (65–69 aa), and all four helices interact with H‐NS to cause conformational changes in Hha at both the surface and in the hydrophobic core ( Madrid et al ., 2007 ). The dispersal variant Hha13D6 has amino acid replacements on helix 2 (D22V) and helix 3 (L40R, V42I and D48A) ( Fig. 1C ), and all three residues on helix 3 are perturbed by the N‐terminus H‐NS ( García et al ., 2005 ), suggesting that the replacements in Hha13D6 cause conformational change of Hha structure to interact with H‐NS and results in the enhanced biofilm dispersal. However, loss of C‐terminal 11 amino acids by truncation at K62 of Hha13D6 abolished the effect of the other four replacements. Since helix 4 also is highly affected by the binding of H‐NS ( García et al ., 2005 ), complete deletion of helix 4 of Hha may change the conformation of Hha and inhibit biofilm formation. In addition, whole‐transcriptome profiles of Hha13D6 (Table S1) and Hha24E9 (Table S2) support that gene regulation pathways of Hha13D6 are quite different from those of Hha24E9. Biofilm dispersal is an essential process of biofilm cycles for disseminating cells into the environment for their survival ( Kaplan, 2010 ), and engineering approaches to disperse biofilms, like using engineered phage ( Lu and Collins, 2007 ), are promising ( Kaplan, 2010 ). The current study demonstrates that genetic switches exist for both biofilm formation and dispersal, and once they are discerned, as in the case of Hha ( Ren et al ., 2004a ; García‐Contreras et al ., 2008 ), they may be manipulated to control bacterial activity. We envision that since biofilms are robust, i.e. biofilm cells withstand stress better than their planktonic counterparts ( Singh et al ., 2006 ), biofilms will be used for diverse applications and some of these applications will involve controlling biofilm formation including replacing some existing engineered biofilms with other engineered strains; this will require control of biofilm dispersal. This capability is important for patterning biofilms in microdevices as well as for creating sophisticated reactor systems that will be used to form bio‐refineries where various engineered strains produce a plethora of chemicals as function of position and depth in biofilms."
} | 2,873 |
39014373 | PMC11253385 | pmc | 7,681 | {
"abstract": "Background Komagataella phaffii , a type of methanotrophic yeast, can use methanol, a favorable non-sugar substrate in eco-friendly bio-manufacturing. The dissimilation pathway in K. phaffii leads to the loss of carbon atoms in the form of CO 2 . However, the ΔFLD strain, engineered to lack formaldehyde dehydrogenase—an essential enzyme in the dissimilation pathway—displayed growth defects when exposed to a methanol-containing medium. Results Inhibiting the dissimilation pathway triggers an excessive accumulation of formaldehyde and a decline in the intracellular NAD + /NADH ratio. Here, we designed dual-enzyme complex with the alcohol oxidase1/dihydroxyacetone synthase1 (Aox1/Das1), enhancing the regeneration of the formaldehyde receptor xylulose-5-phosphate (Xu5P). This strategy mitigated the harmful effects of formaldehyde accumulation and associated toxicity to cells. Concurrently, we elevated the NAD + /NADH ratio by overexpressing isocitrate dehydrogenase in the TCA cycle, promoting intracellular redox homeostasis. The OD 600 of the optimized combination of the above strategies, strain DF02-1, was 4.28 times higher than that of the control strain DF00 (Δ FLD , HIS4 + ) under 1% methanol. Subsequently, the heterologous expression of methanol oxidase Mox from Hansenula polymorpha in strain DF02-1 resulted in the recombinant strain DF02-4, which displayed a growth at an OD 600 4.08 times higher than that the control strain DF00 in medium containing 3% methanol. Conclusions The reduction of formaldehyde accumulation, the increase of NAD + /NADH ratio, and the enhancement of methanol oxidation effectively improved the efficient utilization of a high methanol concentration by strain ΔFLD strain lacking formaldehyde dehydrogenase. The modification strategies implemented in this study collectively serve as a foundational framework for advancing the efficient utilization of methanol in K. phaffii . Supplementary Information The online version contains supplementary material available at 10.1186/s12934-024-02475-1.",
"conclusion": "Conclusions In a previous investigation [ 15 ], we knocked out the first important enzyme Fld in the dissimilation pathway to reduce the loss of methanol in the form of CO 2 through the dissimilation pathway of K. phaffii . Transcriptomic and metabolomic analyses of the resulting strain ΔFLD revealed a down-regulation in the assimilation pathway, elucidating the growth impairment observed in ΔFLD when cultivated in methanol. In this study, we analyzed the strain ΔFLD in 1% methanol, revealing subpar methanol utilization, substantial formaldehyde accumulation, and a diminished intracellular NAD + /NADH ratio. To address these issues, we employed a multifaceted approach. We assembled Aox1 and Das1 into multifunctional enzyme complexes using the Spytag/Spycather protein scaffold, creating substrate channels to reduce formaldehyde diffusion within the cell. Simultaneously, we augmented the catalytic rate of formaldehyde and enhanced Xu5P regeneration by overexpressing key enzymes in the XuMP pathway, thereby fostering strain growth in methanol and mitigating formaldehyde accumulation. Notably, the NAD + /NADH ratio determines the metabolic fluxes of many intracellular pathways and the transcription levels of many genes [ 28 ]. As expected, the strategy of overexpressing IDH of the TCA cycle and MDH of the malate transport shuttle system improved the growth of the strain in methanol and the NAD + /NADH ratio. The high concentration methanol weakened the methanol utilization capacity of the strain, nevertheless, the heterologous expression of MOX could improve the transformation of methanol by the strain. Ultimately, by combining multiple strategies—Aox1/Das1 double enzyme assembly, overexpression of Fbp in the XuMP pathway, Idh in the TCA cycle, and heterologous expression of MOX —the resulting recombinant strain DF02-4 exhibited a remarkable OD 600 , 4.08 times higher than that of the ΔFLD strain in a medium with 3% methanol. These findings establish a solid research foundation for achieving the economic and efficient utilization of methanol in K. phaffii .",
"introduction": "Introduction Methanol, an important bulk chemical, is industrially produced from natural gas and various renewable resources through an intermediate syngas [ 1 ]. Ongoing research explores the synthesis of methanol from CO 2 , considered a promising avenue for mitigating global warming and achieving global carbon neutrality on a global scale [ 2 ]. As a non-food organic C1 feedstock, methanol avoids competition with human food sources and stands out as a carbon substitute for sugar in eco-friendly bio-manufacturing processes [ 3 ]. Its appeal lies in both its low cost and abundant sources [ 4 ]. Furthermore, owing to its higher degree of reduction compared to the majority of sugars [ 5 ], methanol can serve as a primary or supplementary carbon source for the production of reducing chemicals, including alcohols, organic acids, and hydrocarbons, with the expectation of higher yields. Nature encompasses two main categories of methylotrophic microorganisms: methylotrophic bacteria and methylotrophic yeasts. These organisms possess the natural ability to utilize C1 compounds, such as methanol, as substrates for growth and metabolism [ 6 ]. Numerous studies have been conducted on the utilization of methanol by various industrial microorganisms, including Escherichia coli [ 7 ], Corynebacterium glutamicum [ 8 ], Saccharomyces cerevisiae [ 9 ], and Yarrowia lipolytica [ 10 ]. These investigations involve the introduction and optimization of heterologous methanol assimilation pathways. Komagataella phaffii (syn Pichia pastoris ) [ 11 ], a native methanotrophic yeast, is widely used in the industry, utilizing methanol as a carbon source for the production of high value-added products, such as heterologous proteins and biochemicals. This preference is attributed to advantages like strain stability and high cell density fermentation [ 12 , 13 ]. Despite its widespread use, K. phaffii encounters limitations in methanol-based bioindustry due to the inherent toxicity of methanol and its intermediate metabolite formaldehyde to cells, coupled with the loss of carbon atoms in the form of CO 2 formed through the dissimilation pathway [ 12 ]. Addressing these challenges, Cai et al. [ 14 ]. overexpressed the endogenous gene DAS2 in K. phaffii , which further drives formaldehyde assimilation, reduces formaldehyde accumulation, and increases biomass fatty acid yield. In general, there is a relative scarcity of studies focused on the efficient methanol utilization in K. phaffii . In a previous study [ 15 ], our efforts to mitigate the loss of carbon atoms in the dissimilation pathway of K. phaffii involved knocking out the first key enzyme, Fld, in the methanol dissimilation pathway of strain GS115. The resulting dissimilation pathway-blocking strain, ΔFLD, exhibited pronounced growth defects in methanol-containing medium compared to the control strain GS115. Transcriptome analysis indicated that the blocked dissimilation pathway led to the downregulation of the assimilation pathway. In this study, we aimed to improve the utilization of methanol in strain ΔFLD through metabolic pathway modification. Our investigation revealed that the growth defect in strain ΔFLD was partially attributed to the excessive accumulation of formaldehyde and a decrease in the NAD + /NADH ratio in the presence of methanol. Consequently, we focused on reducing formaldehyde accumulation and increasing the NAD + /NADH ratio to improve methanol utilization in strain ΔFLD (Fig. 1 ). To address the toxicity of formaldehyde, we implemented strategies such as limiting formaldehyde diffusion through the self-assembly of key enzymes Aox1 and Das1 involved in methanol metabolism. Additionally, we promoted formaldehyde assimilation by augmenting the amount of formaldehyde co-reactive substrate, Xu5P. In order to increase the NAD + /NADH ratio, NADH production was primarily increased by overexpression of the isocitrate dehydrogenase (Idh) in the TCA cycle. Simultaneously, we facilitated intracellular NADH translocation to mitochondria by increasing the amount of malate dehydrogenase (Mdh) in the malate-aspartate shuttle (also known as malate shuttle) system. Furthermore, the heterologous expression of methanol oxidase (Mox) from Hansenula polymorpha was introduced to improve the methanol utilization capacity and growth of the strain in 3% high concentration methanol. In this study, the efficient utilization strategy of methanol provided a valuable base for the application of K. phaffii in industrial biotechnology. \n Fig. 1 Sketch of methanol metabolic pathway modification in K. phaffii. Fld, formaldehyde dehydrogenase; Fdh, formate dehydrogenase; Fgh, S-Formylglutathione Hydrolase; Mox, methanol oxidase from Hansenula polymorpha ( Ogataea polymorpha ); Aox, alcohol oxidase; Cat, catalase; Das, dihydroxyacetone synthase; Dak, dihydroxyacetone kinase; Fba fructose-bisphosphate aldolase; Fbp, fructose bisphosphatase; Tpi, triosephosphate isomerase; Tal, transaldolase; Rpi, ribose-5-phosphate isomerase; Rpe ribulose phosphate 3-epimerase; Mdh, malate dehydrogenase; Idh, isocitrate dehydrogenase; GAP, glyceraldehyde 3-phosphate; DHA, dihydroxyacetone; DHAP, dihydroxyacetone phosphate; F1,6BP, fructose-1 6-bisphposphate; F6P, fructose 6 phosphate; E4P, erythrose-4-phosphate; SBP, sedoheptulose 1,7-bisphosphate; S7P, sedoheptulose 7-phosphate; Xu5P, xylulose 5-phosphate; R5P, ribose 5-phosphate; Ru5P, ribulose 5-phosphate; MAL, malate; OXA, oxaloacetate; ISO, isocitrate; OXAL, oxalosuccinate",
"discussion": "Results and discussion Weak growth of ΔFLD in methanol In the dissimilation pathway of methanol metabolism in K. phaffii , formaldehyde undergoes initial oxidized by Fld and Fgh to formic acid, subsequently further oxidized to CO 2 [ 25 ]. During the dissimilation of formaldehyde, the conversion of 1 molecule of formaldehyde is coupled with the transformation of 2 molecules of NAD + to NADH. However, excessive dissimilation of methanol leads to a significant loss of C1 substrate in the form of CO 2 , thereby endangering the yield of biomass and target chemicals and reducing the economics of the methanol carbon atom [ 26 ]. In the previous research, a strain ΔFLD with a blocked dissimilation pathway was generated [ 15 ]. The ΔFLD strain exhibited severe growth defects in medium containing 1% methanol and demonstrated limited methanol utilization (Fig. 2 A and C). Through the experiment, we found that there was a large amount of formaldehyde accumulated in the supernatant of the fermentation broth of the ΔFLD strain, with the accumulated formaldehyde levels being 2 ~ 3 times higher than those in the control strain GS115 (Fig. 2 B). Formaldehyde is known to be non-specifically toxic to intracellular DNA and proteins [ 27 ]. Strains of K. phaffii with compromised integrity face challenges in normal growth in medium where methanol serves as the carbon source, similar to the trend of inhibition of the assimilation pathway due to knockdown of the dissimilation pathway found in previous studies [ 15 ]. Measurement of intracellular NAD + /NADH, it was found that the intracellular NAD + /NADH revealed a lower ratio in the ΔFLD strain compared to GS115 (Fig. 2 D). As a cofactor, NAD (NAD + and NADH) participates in over 300 intracellular redox reactions, playing an important role in cellular metabolism [ 28 ]. The imbalance in the supply of NAD + /NADH caused disrupted intracellular metabolism, affecting the growth of the K. phaffii strain in methanol. The methanol dissimilation pathway is a rapid metabolic pathway for the toxic substance formaldehyde and a source of NADH and ATP [ 29 , 30 ]. Impairment of the dissimilation pathway led to formaldehyde accumulation and intracellular NAD disruption, causing increased intracellular ROS levels (Fig. 2 E), which affected methanol metabolism in microbial cell factories. Hence, our studies focused on improving the utilization of methanol by the ΔFLD strain by reducing formaldehyde accumulation and increasing NAD balance. \n Fig. 2 A comparison between strains ΔFLD and GS115 in 25 mL of BMMY medium with 1% methanol. A Measurement of growth curve; The x-axes of the culture plots start at 18 h. B Measurement of fermentation supernatant formaldehyde; C Measurement of fermentation supernatant methanol; D Measurement of intracellular NAD + /NADH; E Measurement of intracellular ROS; F Schematic diagram of strain ΔFLD. Error bars represent the standard deviation of 2 or 3 biological replicates \n Reduction of formaldehyde diffusion by self-assembly of Aox1 and Das1 The excessive accumulation of formaldehyde exerts a toxic effect on proteins and nucleic acids, impeding the efficient utilization of methanol by strains. Therefore, the swift metabolism of formaldehyde is crucial to enhance methanol utilization in strains. Fan [ 31 ]et al. fused and expressed Mdh, Hps and Phi in Synthetic methanotrophic E. coli by flexible linker (GGGGS) n , resulting in improved methanol biotransformation. Thus, preventing the diffusion of toxic intermediates into other intracellular pathways by improving the spatial proximity of enzymes in cascade reactions and the construction of substrate channels through enzyme complexes [ 32 ], offering be a promising strategy for enhancing methanol biotransformation [ 3 ]. Methanol is oxidized to formaldehyde by Aox in K. phaffii , and formaldehyde is further catalyzed by Das in the carbon metabolism assimilation pathway. Aox and Das are two key enzymes for methanol assimilation. Intracellular assembly of Aox1 and Das1 was performed using the protein scaffold Spytag/Spycather in the strain ΔFLD, with Spycather attached to the C-terminus of Aox1 and Spytag attached to the C-terminus of Das1. Experimental results revealed that, in 1% methanol, the OD 600 of the recombinant strain DF02 at 48 h was 2.15 times higher than that of DF00 (ΔFLD back-complemented HIS4 + ) (Fig. 3 A). Notably, assembled strain DF02 exhibited a similar methanol utilization capacity compared to unassembled strains DF00 and DF01 (Fig. 3 B). However, DF02 demonstrated superior growth and lower formaldehyde accumulation, with a nearly 52.6% decrease in formaldehyde accumulation in the fermentation supernatant compared to DF00 (Fig. 3 C). These results showed that the supramolecular enzyme complex formed by the self-assembly of Aox1 and Das1 helped reducing the diffusion and accumulation of formaldehyde in cells, thereby promoting the growth of the strain in methanol. Moreover, it also indicates that formaldehyde is more toxic than methanol and has a great inhibitory effect on cell growth. \n Fig. 3 Analysis of strains with the Aox1/Das1 dual enzyme assembly strategy was incubated in 25 mL of BMMY medium containing 1% methanol. A Measurement of growth curve; The x-axes of the culture plots start at 18 h. B Measurement of fermentation supernatant formaldehyde; C Measurement of fermentation supernatant methanol; D Schematic diagram of double enzyme assembly. Aox, alcohol oxidase; Das, dihydroxyacetone synthase; Dak, dihydroxyacetone kinase; GAP, glyceraldehyde 3-phosphate; DHA, dihydroxyacetone; Xu5P, xylulose 5-phosphate. Error bars represent the standard deviation of 2 or 3 biological replicates \n Promoted formaldehyde assimilation by increased formaldehyde receptor Xu5P In the process of methanol metabolism, the generated formaldehyde necessitates further metabolized with a co-reaction substrate. The metabolism of formaldehyde is constrained by the availability of the co-reaction substrate, making it imperative to enhance the regeneration of the formaldehyde receptor for efficient methanol assimilation. However, the regeneration of the formaldehyde receptor poses a significant barrier to formaldehyde assimilation. Woolston et al [ 13 ] found that activation of the SBPase pathway and reduction of GAPDH in the RuMP cycle significantly enhanced the regeneration of Ru5P. In the methanol assimilation metabolic pathway of K. phaffii , it is necessary for formaldehyde to produce GAP and DHA of Xu5P catalyzed by Das. Consequently, the metabolism of formaldehyde is also limited by the amount of receptor Xu5P. Comparative analysis of transcriptomic data (Table S1 ) based on ΔFLD strains in glucose or methanol media showed a significant increase in transcript levels of DAK , FBA2 , FBP1 , RPIA , and TAL2 , key enzymes of the Xu5P regeneration cycle pathway under methanol culture conditions. The intracellular distribution of these enzymes has been analyzed using online protein localization simulations [ 33 ], and most of these enzymes are located in peroxisomes (Table S2). This was in contrast to the previously reported notion of an independent set of non-oxidative phosphorylation pathways in peroxisomes to regenerate Xu5P [ 34 ]. Genes for these enzymes were amplified from the K. phaffii GS115 genome using PCR technology and overexpressed under the Aox1 methanol-induced promoter in the ΔFLD strain (Fig. 4 B). The engineered strains exhibited approximately 30% higher growth compared to the control strain DF00 (Fig. 4 C), an increased methanol utilization rate (Fig. 4 D), and reduced formaldehyde accumulation in methanol (Fig. 4 E). Notably, strain DF05, overexpressing Fbp1, demonstrated a 32% increase in growth at 48 h and a 21.9% reduction in the accumulation of formaldehyde in the supernatant compared to the control strain DF00, showcasing maximal enhancement in formaldehyde receptor regeneration. \n Fig. 4 Analysis of strains with overexpression of key enzymes of the XuMP pathway were incubated in 25 mL of BMMY medium containing 1% methanol. A Schematic diagram of the methanol assimilation metabolic pathway, Aox, alcohol oxidase; Das, dihydroxyacetone synthase; Dak, dihydroxyacetone kinase; Fba fructose-bisphosphate aldolase; Fbp, fructose bisphosphatase; Tpi, triosephosphate isomerase; Tal, transaldolase; Rpi, ribose-5-phosphate isomerase; Rpe ribulose phosphate 3-epimerase; GAP, glyceraldehyde 3-phosphate; DHA, dihydroxyacetone; DHAP, dihydroxyacetone phosphate; F1,6BP, fructose-1 6-bisphposphate; F6P, fructose 6 phosphate; E4P, erythrose-4-phosphate; SBP, sedoheptulose 1,7-bisphosphate; S7P, sedoheptulose 7-phosphate; Xu5P, xylulose 5-phosphate; R5P, ribose 5-phosphate; Ru5P, ribulose 5-phosphate; B Schematic diagram of the overexpression plasmid; C Measurement of 48 h growth OD 600 ; D Measurement of 48 h fermentation supernatant methanol; E Measurement of fermentation supernatant formaldehyde. Error bars represent the standard deviation of 2 or 3 biological replicates \n Optimization of intracellular NAD + /NADH balance promote strain growth In methanol-grown methylotrophic yeast, cellular energy is primarily derived through two main pathways: the TCA cycle reaction via the respiratory chain and the methanol dissimilation pathway [ 35 ]. Within the dissimilation pathway, each molecule of formaldehyde is accompanied by 2 molecules of NADH production, which then passes H + through the NADH shuttle system to the respiratory chain inside the mitochondria to produce ATP for cell growth [ 26 , 36 ]. FLD knockdown resulted in a reduction in the intracellular NAD + /NADH ratio. The NAD + /NADH ratio in cells represents the redox state, which is influenced by and in turn regulates metabolic activity, and redox homeostasis is necessary for optimal cellular health throughout the life cycle [ 37 , 38 ]. Overexpression of Idh in the TCA cycle and Mdh in the NADH malate transport shuttle system was performed individually and in combination in the ΔFLD strain. Experimental results demonstrated improved intracellular NAD + /NADH ratio and methanol utilization rates in 1% methanol, leading to a 30% increase in OD 600 growth. The DF10 strain, expressing Idh and Mdh in combination, exhibited the most substantial increase in NAD + /NADH ratio, reaching approximately 10, followed by the DF08 strain expressing Idh (Fig. 5 D). However, the growth and methanol utilization capabilities of DF10 strains were not as strong as DF08 strains (Fig. 5 C and E), potentially linked to the metabolic stress tendency associated with strains expressing multiple proteins. After 48 h of incubation in 1% methanol, the strain DF08 increased the growth OD 600 by 34.9% compared with the strain DF00 (Fig. 5 C), and methanol residue in the fermentation supernatant decreased by nearly 33.9% ((Fig. 5 E). Some studies have observed a decrease in the NAD + /NADH ratio during cellular senescence [ 39 , 40 ]. By enhancing the TCA cycle and NADH shuttle system, the intracellular NAD + /NADH ratio can be optimized, improving the intracellular redox state, and promoting methanol utilization in ΔFLD strains. \n Fig. 5 Analysis of strains with enzyme overexpression with NAD + as cofactor was incubated in 25 mL of BMMY medium containing 1% methanol. A Schematic diagram of TCA cycle and malate shuttle system, Mdh, malate dehydrogenase; Idh, isocitrate dehydrogenase; MAL, malate; OXA, oxaloacetate; ISO, isocitrate; OXAL, oxalosuccinate; B Schematic diagram of expression plasmid; C Measurement of growth curve; The x-axes of the culture plots start at 18 h. D Measurement of intracellular NAD + /NADH; E Measurement of methanol in fermentation supernatant. Error bars represent the standard deviation of 2 or 3 biological replicates \n Systematic metabolic modification portfolio to improve methanol utilization According to the analysis of the results of the multi-strategy described above, the Aox1/Das1 dual enzyme assembly strain DF02 underwent individual and combined overexpression of FBP (XuMP pathway), IDH (TCA cycle) and MDH . The results, shown in Fig. 6 A, revealed that strain DF02-1 exhibited optimal growth performance, with a 4.28-fold increase in OD 600 over the DF00 control strain after 48 h in 1% methanol. In contrast, strains DF02-2 and DF02-3 displayed lower growth rates compared to DF02. Subsequent comparisons between the final strain DF02-1 and strain DF00 in 1% methanol (Fig. 6 B) demonstrated a 20% reduction in methanol residue and a 65.7% decrease in formaldehyde accumulation in DF02-1, accompanied by a 81.1% reduction in cell death compared to DF00 at 48 h (Fig. 6 D). The observed excessive formaldehyde accumulation in the DF00 strain over time, attributed to the FLD deletion, led to increased toxicity to cells. This, coupled with a decreased methanol utilization capacity, resulted in elevated methanol levels in the fermentation supernatant, contributing to the dual toxicity of methanol and formaldehyde and an escalating number of dead cells. Transcript levels analysis of key methanol metabolizing enzymes in strains DF00 and DF02-1 in 1% methanol (Fig. 6 C) showed that the recombinant strain DF02-1 exhibited 2.5-7 times higher overall transcript levels of methanol metabolizing pathway genes compared to DF00. Notably, the overexpressed FBP and IDH genes in DF02-1 showed 41.9 and 203.6 times higher transcript levels, respectively (Fig. S3). In conclusion, the construction of the Aox1/Das1 dual enzyme assembly, along with enhanced Xu5P regeneration and increased intracellular NAD + /NADH, successfully promoted methanol assimilation, leading to improved strain growth in methanol. \n Fig. 6 The analysis of strains with an integrated strategy was incubated in 25 mL of BMMY medium containing 1% methanol. A Growth OD 600 of recombinant strains at 48 h; B Measurement of OD 600 , methanol content and formaldehyde content of recombinant strains DF02-1 and DF00 grown in 1% methanol; The x-axes of the culture plots start at 12 h. C Transcript levels of key enzymes of methanol metabolism; D Mortality of recombinant strains in 1% methanol by flow cytometry. Error bars represent the standard deviation of 2 or 3 biological replicates \n Improved utilization of high methanol concentrations in recombinant strain DF02-1 In order to determine the methanol utilization ability of recombinant strain DF02-1 in high concentrations of methanol, we found that the residual amount of methanol in the fermentation supernatant of strain DF02-1 and DF00 were similar in the medium containing 3% methanol, and the methanol utilization ability of the strain was impaired by high concentrations of methanol [ 35 ]. In previous studies [ 41 ], heterologous expression of MOX derived from Hansenula polymorpha in K. phaffii promoted the methanol utilization in high concentration methanol by recombinant strains. Here we heterologously expressed the enzyme Mox on the basis of the DF02-1 strain, and the designated recombinant strain DF02-4 further improved the utilization of methanol. In BMMY medium containing 3% methanol, the growth OD 600 of DF02-4 incubated for 48 h was 1.12 times higher than that of DF02-1. At 48 h, the growth OD 600 of DF02-4 was 4.08 times higher than that of the initial strain DF00, and methanol utilization increased by 10.26%. (Fig. 7 ). This result indicated that the heterologous expression of MOX could effectively improve the growth of the strain in 3% methanol and utilization of methanol. \n Fig. 7 Analysis of recombinant strain DF02-4 in 3% methanol. Error bars represent the standard deviation of 2 or 3 biological replicates. The x-axes of the culture plots start at 14 h"
} | 6,383 |
34160856 | PMC8456938 | pmc | 7,682 | {
"abstract": "Abstract Despite the importance of soil microorganisms for ecosystem services, long‐term surveys of their communities are largely missing. Using metabarcoding, we assessed temporal dynamics of soil bacterial and fungal communities in three land‐use types, i.e., arable land, permanent grassland, and forest, over five years. Soil microbial communities remained relatively stable and differences over time were smaller than those among sites. Temporal variability was highest in arable soils. Indications for consistent shifts in community structure over five years were only detected at one site for bacteria and at two sites for fungi, which provided further support for long‐term stability of soil microbial communities. A sliding window analysis was applied to assess the effect of OTU abundance on community structures. Partial communities with decreasing OTU abundances revealed a gradually decreasing structural similarity with entire communities. This contrasted with the steep decline of OTU abundances, as subsets of rare OTUs (<0.01%) revealed correlations of up to 0.97 and 0.81 with the entire bacterial and fungal communities. Finally, 23.4% of bacterial and 19.8% of fungal OTUs were identified as scarce, i.e., neither belonging to site‐cores nor correlating to environmental factors, while 67.3% of bacterial and 64.9% of fungal OTUs were identified as rare but not scarce. Our results demonstrate high stability of soil microbial communities in their abundant and rare fractions over five years. This provides a step towards defining site‐specific normal operating ranges of soil microbial communities, which is a prerequisite for detecting community shifts that may occur due to changing environmental conditions or anthropogenic activities.",
"conclusion": "5 CONCLUSIONS Soil bacterial and fungal community structures of 30 sites from three different land‐use types (i.e., arable land, permanent grassland and forest) were land‐use‐ and site‐specific and stable over 5 years. Consistent community shifts over time were largely absent, which further supports the high temporal stability of soil bacterial and fungal communities. Normal operating ranges of bacterial and fungal community structures depend on site, land management and possibly soil properties. Partial communities composed of abundant or rare OTUs were highly correlated to entire communities, revealing that subsets of a few hundred OTUs can be highly representative of entire communities composed of several thousand OTUs. Focusing on a few hundred OTUs may facilitate the establishment of references for soil identification and long‐term soil quality monitoring. Finally, we showed that scarce OTUs, which account for about a third of all OTUs, have little impact on the assessment of community structures, but may be valuable scarce biota performing important soil functions in scarce habitats.",
"introduction": "1 INTRODUCTION Soil bacterial and fungal communities impact ecosystem services such as crop production (Hu et al., 2018 ) or nutrient cycling (Regan et al., 2017 ), and therefore have a major influence on soil quality. Maintaining soil quality requires the maintenance of stable soil microbial communities, because changes in their structures may induce disturbances in ecosystem processes. However, soil quality assessment and especially long‐term soil monitoring rarely include high‐throughput DNA sequencing of soil microbial communities in addition to soil physicochemical analyses (van Leeuwen et al., 2017 ). A major reason for this gap in soil quality assessments is the largely missing information on temporal dynamics of soil bacterial and fungal communities over multiple years. The temporal stability of microbial communities in soil has been shown to be higher as compared to the stability of microbial communities living in water, air or host‐associated environments (Shade et al., 2013 ). The first evidence showing temporal stability of soil bacterial communities over 1 year was obtained using fingerprinting techniques (Gelsomino et al., 1999 ). Since then, high‐throughput DNA sequencing has been developed enabling the assessment of entire soil bacterial and fungal communities to a depth that reflects their diversity. A 3‐year survey using high‐throughput sequencing revealed a stable fungal community until a heavy rainfall occurred in the last year and resulted in soil water saturation (Barnes et al., 2018 ). Larger differences were also observed for soil bacterial communities among sites as compared to temporal variability over 6 months (Carini et al., 2020 ; Lauber et al., 2013 ). In general, soil bacterial and fungal communities appear to be relatively stable over several months to a few years, but empirical data are scarce. Temporal stability of communities depends on the one hand on their resistance and resilience to environmental fluctuations, where resistance is defined as the insensitivity of a community to a disturbance, and resilience as the ability of a community to return to its initial, predisturbed state (Shade et al., 2012 ). Furthermore, temporal stability depends on ecological stochasticity including ecological drift and random dispersal (see Zhou & Ning, 2017 for a review), which may cause communities to shift over time. Monitoring programmes may allow to identify community shifts over longer time periods in natural habitats. Long‐term community shifts continuously drive a community away from its initial state (Figure S1 ), which can be assessed by linking community dissimilarities to the elapsed time between two sampling time points. This may be achieved by analysing time–decay relationships (e.g. Berg & Bengtsson, 2007 ; Chow et al., 2013 ; Shade et al., 2013 ), which are analogous to distance–decay relationships (Nekola & White, 1999 ). Time–decay analyses allow identification of steady shifts in microbial communities. However, they do not allow identifying specific time points when changes occur, because time lags rather than single time points are assessed in these analyses. The identification of time points when community shifts occur is of particular importance for long‐term monitoring systems. Therefore, long‐term biomonitoring of macro‐organisms often relies on comparison to a reference point for identifying community shifts (Magurran et al., 2010 ). Due to the lack of long‐term surveys of soil microbial organisms, it is currently unknown whether continuous shifts of soil microbial communities occur over extended time periods. Assessment of temporal variability of soil microbial communities in ecosystems that are not subject to exceptional disturbances allows definition of their normal operating ranges (NORs). An NOR describes the multivariate space in which the states of an undisturbed ecosystem occur (Kersting, 1984 ; van Straalen, 2002 ). For soils, Semenov et al. ( 2014 ) established NORs for sandy and clay soils based on 21 parameters including chemical (e.g., pH and nitrogen content) as well as biological (e.g., gene abundances and bacterial diversity) parameters that were determined at eight agricultural sites over 3 years. In a second step, these authors applied weak and strong temperature and flooding stresses to these soils in microcosms and compared the measured parameters against the NOR established based on undisturbed soils. While control microcosms remained within the NOR, the stressors caused increasing distances of soil parameters to the NOR. However, due to the lack of long‐term surveys using high‐throughput sequencing of soil microbial communities, NORs of soil microbial community structures are currently unknown. Along with Semenov et al. ( 2014 ), who showed different NORs for different soil types (i.e., clay and sandy soils), Lauber et al. ( 2013 ) have shown that arable soils harbour temporally less stable bacterial communities as compared to grassland, which suggests that the NOR of soil microbiomes depends on environmental settings and soil management. Furthermore, the NOR may be influenced to different degrees by abundant and rare operational taxonomic units (OTUs), as rare OTUs are hypothesized to over‐proportionally affect temporal variability (Shade et al., 2014 ). In changing environments rare microbial OTUs may become more abundant (Aanderud et al., 2015 ; Barnes et al., 2018 ), and thus temporal variability may be more strongly influenced by conditionally rare as compared to abundant OTUs (Shade et al., 2014 ). Among the rare taxa, some are present in a dormant state and build a seed bank from which components of a community can be recruited in case conditions would become favourable (Shade et al., 2014 ). Other rare microbial taxa can be continuously active despite their rarity (Campbell et al., 2011 ; Hausmann et al., 2019 ). These would therefore be stably detected at low abundances. Finally, parts of a community appearing as rare OTUs may be due to analytical artefacts, such as polymerase chain reaction (PCR) amplification errors (Potapov & Ong, 2017 ). With the exception of sequence‐dependent errors leading for instance to chimera formation during PCR (Haas et al., 2011 ), these will occur randomly and represent analytical background noise. Therefore, two groups of rare OTUs may conceptually be distinguished: (i) consistent, rare OTUs that include microorganisms living in rare microniches, and (ii) rare OTUs that occur randomly in highly fluctuating soil conditions (e.g., flooding) or very infrequently not yielding information to differentiate microbial systems. The second group, termed scarce OTUs here, includes biological as well as erroneous sequences, which cannot be reliably distinguished due to their infrequent occurrence. To distinguish consistently rare OTUs from scarce OTUs within a community, spatial and temporal replication is needed. Here, we assessed the temporal dynamics of soil bacterial and fungal communities over 5 years at 30 different long‐term monitoring sites of the Swiss Soil Monitoring Network (NABO). The sites represented the three different land‐use types arable land, permanent grassland and forest, with 10 sites each. The overarching goal of this study was to identify NORs of soil bacterial and fungal communities in these three land‐use types and thus to provide the basis for the development of reference baselines for variations of soil microbial communities. We defined four main research objectives, which focused exclusively on the microbial community level, rather than on individual microbial taxa, their identities and their functions. Moreover, the analytical focus was on temporal shifts of microbial communities and differences among land‐use types as well as sites rather than detailed effects of soil properties or soil textural classes. The objectives were: (i) to identify factors that influence the long‐term dynamics of soil microbial communities, (ii) to screen microbial communities for consistent community shifts over 5 years, (iii) to compare community structures and notably temporal stability of their abundant and rare fractions, and (iv) to differentiate rare from scarce OTUs.",
"discussion": "4 DISCUSSION 4.1 Temporal stability of soil bacterial and fungal communities Our survey of 30 long‐term monitoring sites over 5 years revealed a high land‐use‐ and site‐specificity of soil bacterial and fungal community structures, along with comparatively small temporal variability (Figure 1 ; Table 1 ). A higher spatial variability relative to the temporal variability was also reported by Carini et al. ( 2020 ), who analysed the development of soil bacterial and fungal communities at two mountain slopes over 6 months. Interestingly, consistent differences not only between the two sites but also within sites at the metre scale were detected over time. To prevent spatial variability at the metre scale from masking temporal effects in our study, we took advantage of the sampling design of the Swiss long‐term soil monitoring network (NABO) and collected bulk samples of 25 cores from exactly the same locations of the 100‐m 2 plots every year (Figure S2 ). Despite the stability and high site‐specificity of soil microbial community structures, temporal variability was detect at the site level, with arable sites showing significantly more temporal variability when compared to permanent grassland and forest sites (Figure 2 ). In agreement with our findings, Lauber et al. ( 2013 ) detected more variable bacterial communities in arable land as compared to a grassland plot at a single site over 6 months during one growing season. The authors attributed the increased temporal variability of bacterial communities in arable land to the land management and to the plant community, which developed over the growing season. All arable sites we assessed were managed with crop rotations, which included three to six different crops, and with one exception they were conventionally tilled. Land management of arable sites, such as crop rotations (Peralta et al., 2018 ), tillage (Degrune et al., 2017 ), fertilization (Hartmann et al., 2015 ) and plant protection (Rivera‐Becerril et al., 2017 ), has been shown to affect the structures of soil microbial communities, and may therefore reduce their temporal stability. Beside the influence of land management, temporal variability of soil microbial communities may also be caused by environmental factors. Lauber et al. ( 2013 ) hypothesized, for example, that environmental selection of microorganisms with different life‐strategies may lead to varying temporal stability of soil microbial communities. However, environmental factors that correlated with temporal stability of soil bacterial and fungal communities in our survey were also significantly different between arable sites and the other land‐use types (Tables S2 and S4 ). Within land‐use types no significant correlations were detected, and temporal variability of soil bacterial and fungal communities was always highest in arable soils regardless of the soil texture class (Figure S6 ). Together, this indicates that the increased temporal variability of bacterial and fungal communities at arable sites is likely explained by land management, and that environmental factors appear to have a minor effect on temporal stability of soil microbial communities. Interestingly, the increase in temporal variability at arable sites as compared to the other two land‐use types was larger for fungal than for bacterial communities (Figure 2 ; Table S3 ). This may be due to a stronger dependence of certain fungal taxa on specific plant species (Ai et al., 2018 ; Fox et al., 2020 ) or a stronger disturbance of fungal communities in tilled systems (Schmidt et al., 2019 ). 4.2 Temporal community shifts Communities may follow various trajectories over time. On the one hand, communities may be disturbed but return to their initial state due to a high resilience (Lamothe et al., 2019 ). If this trajectory were to occur repeatedly, the community could be considered stable, although with a larger NOR. On the other hand, communities may shift to another state, which may occur rapidly or over longer time periods. Various mechanisms could cause community shifts, including deterministic processes such as short‐term (pulse) or long‐term (press) disturbances (Shade et al., 2012 ), and possibly stochastic processes such as ecological drift or random dispersal (Zhou & Ning, 2017 ) . Notably, continuously changing environmental factors, for instance caused by land management‐induced soil compaction (Hartmann et al., 2014 ), by climate change (Isobe et al., 2020 ) or by increasing atmospheric nitrogen depositions (Leff et al., 2015 ; Peñuelas et al., 2012 ), represent press disturbances (Shade et al., 2012 ), which could alter the structures of soil microbial communities. Subsequently, ecosystem functions and services provided by soil microbial communities could also be altered (Allison & Martiny, 2008 ) and might affect soil quality. To assess, whether indications for community shifts over 5 years were detectable in our system, we determined for each site the distance of the communities to the one of the first year (Figures S3 and S4 ). Only soil microbial communities from three arable sites showed indications for consistent community shifts over at least 2 years. As shifts in a single year were more common (Figures S3 and S4 ), this may show the high resilience of soil bacterial and fungal community structures to pulse disturbances that may occur throughout the years (e.g., tillage at arable sites or meteorological changes). In agreement with the high temporal stability detected in our study, it has been demonstrated that experimental warming of forest soils lead to shifts in bacterial communities only after 20 years, while no treatment effect was found for warming periods of 5 or 8 years (DeAngelis et al., 2015 ). Furthermore, by correlating past and present climate data to current soil microbial diversity, Ladau et al. ( 2018 ) identified a lag of ~50 years between changes of climate variables and soil microbial diversity. This indicates that shifts of soil microbial communities may occur over time periods in the order of decades rather than a few years, but particular taxa (populations) responsive to daily varying environmental conditions such as soil moisture, temperature or nutrient levels may be subject to more frequent changes. However, it becomes increasingly evident that soil microbial communities show a high temporal stability, despite their sensitivity to changing environmental factors, and despite their short generation times, which may increase random diversification (Zhou & Ning, 2017 ). Consequently, long‐term experiments and monitoring are clearly needed to assess long‐term effects of changing environmental and anthropogenic factors on soil microbial communities. 4.3 Normal operating ranges of soil microbial community structures With the absence of lasting community shifts over 5 years, the temporal variability assessed for most sites can be considered as within the NOR of these soil microbial communities. The temporal variability differed between land‐use types and sites (Figure 2 ), which indicated that the extent of NORs depend on land‐use as well as on specific environmental factors. Therefore, a better understanding of these factors is necessary to reliably define NORs of microbial communities. In the present study, we used the first year as the reference to detect community shifts over time. The results could, therefore, be affected by unusually large differences between the first and the second year. Our analyses also showed that significant differences of a single year were relatively common within 5 years (i.e., this was the case at eight of the 30 sites for bacteria and at two sites for fungi; Figures S3 and S4 ). Consequently, reference baselines or NORs based on longer time periods will provide more balanced and stable values and thereby enhance the ability to detect relevant community shifts over time. The number of years needed for the definition of a NOR or a significant community change will, however, depend on empirical data of site characteristics. In arable land, it may be necessary to consider entire crop rotations, while reference data sets with lower temporal resolution may be sufficient for permanent grassland or forest sites. 4.4 Similarities of entire communities with abundant and rare fractions The high temporal stability of soil bacterial and fungal communities raises the question of whether all fractions throughout the community from abundant to rare remain temporally stable over 5 years. Therefore, we first compared groups of 500 randomly selected OTUs with entire communities, which all revealed a high structural similarity (Mantel test, Figure 3 ). This suggested that subsets of a few hundred OTUs may be representative of entire soil microbial communities composed of several thousand OTUs. In agreement with our findings, 511 dominant bacterial OTUs from a global dataset have been shown to be highly correlated (Mantel test, r = .92) to the rest of the bacterial community (Delgado‐Baquerizo et al., 2018 ) and, in an experimental warming experiment of forest soils, the 155 most abundant OTUs were highly correlated (Mantel test, r = .98) with the entire community structures and also represented the detected warming effects (DeAngelis et al., 2015 ). Furthermore, the removal of rare OTUs with increasing abundances had little effects on community structures of bacterial and fungal communities (Botnen et al., 2018 , Zinger et al., 2014). These results suggest that community structures and their changes may already become evident based on the most dominant members of soil microbial communities. In cases where coarse differences among land‐use types and sites are analysed, soil bacterial and fungal community structures may be correctly assessed using earlier culture‐independent techniques such as fingerprinting approaches that assess their dominant members. To assess whether rare OTUs were also representative of entire communities we used a sliding window analysis. This revealed that structures of entire communities were best represented by abundant OTUs, but interestingly, relative abundance dropped much faster as compared to the structural similarity of partial and entire communities (Figure 4 ). Thus, correlated community structures were consistently detected throughout the majority of the community ranging from abundant to rare OTUs. This suggests that similar processes were driving the assembly of abundant and rare community fractions at the surveyed sites. The high stability and the strong differences among sites and land‐use types could indicate that environmental filtering (Yan et al., 2019 ) was a major determinant of the recovered community structures. The heterogeneity within a soil habitat with interconnected major and minor niches, where abundant and rare organisms thrive, may lead to concerted community structures of abundant and rare fractions of soil bacterial and fungal communities. This may indicate that rare OTUs inhabit small niches, which are specific to a given land‐use type or site. As observed for the entire communities, temporal variability remained lower than variability among sites or land‐use types for communities composed of rare OTUs (Table 3 ). Therefore, rare fractions of soil bacterial and fungal communities also remained generally stable over 5 years. This may reflect the stability of rare microniches within soils, which offer a habitat for specialized rare taxa. However, temporal variability was slightly higher in communities composed of rare as compared to abundant OTUs (Table 3 ), which may be due to OTUs that are occasionally more abundant but that are generally present in low numbers (Shade et al., 2014 ). Experimentally, this was demonstrated in microcosms, where an initially rare OTU thrived in response to hydrocarbon pollution (Fuentes et al., 2016 ). Rare OTUs may therefore be more sensitive to environmental changes, on the one hand because they may thrive due to environmental changes, or on the other hand because the microniches they inhabit may be more easily disturbed compared to larger niches, where abundant taxa live. 4.5 Identification of scarce OTUs Rare OTUs are distinguished from abundant OTUs by relative abundance thresholds (e.g., 0.01%), but no thresholds exist for the definition of scarce OTUs. Scarce OTUs are extremely rare and infrequently detected OTUs, which provide little information on a system. These OTUs may include real biological sequences originating for instance from organisms that could not establish in a particular habitat, but also erroneous sequences that represent analytical failure. The continuous decrease of the structural similarity between entire and partial communities from the most to the least abundant taxa (Figure 4 ) impedes the definition of a threshold for scarce OTUs. Here, we defined indicative and core OTUs to empirically separate rare from scarce OTUs. This resulted in 23.4% and 19.8% of bacterial and fungal OTUs that were classified as scarce. Despite this relatively large number of scarce OTUs, analyses of community structures were only marginally affected by the inclusion or exclusion of scarce OTUs (Table 3 ). This is in agreement with analyses of marine and freshwater bacterial communities, where the exclusion of 45% of the OTUs with the lowest abundance showed correlations of at least .95 to entire community structures (Gobet et al., 2010 ; Liu et al., 2015 ). Furthermore, our analysis revealed that bacterial communities were composed of 9.4% abundant and 67.3% rare OTUs (excluding scarce OTUs), while these numbers were 15.3% and 64.9% for fungi. Soil microbial communities are therefore mainly composed of rare and temporally stable OTUs (Table 3 ). The temporal stability of rare OTUs (excluding scarce OTUs) suggests that these can be robustly assessed using metabarcoding. Finally, our analysis revealed that a relative abundance of <0.0002% (Table 2 ) may represent a first conceptual basis for an operational threshold to define scarce OTUs. 4.6 Consequences for soil quality monitoring Metabarcoding of soil bacterial and fungal communities has been shown to sensitively detect responses to a multitude of factors such as heavy metal pollution (Frossard et al., 2018 ) or salinization (Rath et al., 2019 ). Metabarcoding also had a higher power to discriminate between different soils (i.e., subject to periodic waterlogging or different management), as compared to physicochemical soil analyses (Gschwend et al., 2020 ). The reproducible detection of soil microbial communities in soils from defined environmental conditions and land‐use is a major requirement for the implementation of metabarcoding in long‐term monitoring programmes. Therefore, the stability of soil bacterial and fungal community structures over 5 years builds further support for metabarcoding‐based soil quality monitoring. Experimental studies are needed to assess the extent of the NOR and its relationship to pulse and press disturbances. The high structural similarities between entire and partial communities also showed that defined partial communities can be used for surveys of effects on soil microbiota at the community level. This may facilitate long‐term monitoring as it allows for a robust comparison among recurring sampling campaigns."
} | 6,632 |
32606383 | PMC7327058 | pmc | 7,683 | {
"abstract": "We use a unique set of terrestrial experiments to demonstrate how soil management practises result in emergence of distinct associations between physical structure and biological functions. These associations have a significant effect on the flux, resilience and efficiency of nutrient delivery to plants (including water). Physical structure, determining the air–water balance in soil as well as transport rates, is influenced by nutrient and physical interventions. Contrasting emergent soil structures exert selective pressures upon the microbiome metagenome. These selective pressures are associated with the quality of organic carbon inputs, the prevalence of anaerobic microsites and delivery of nutrients to microorganisms attached to soil surfaces. This variety results in distinctive gene assemblages characterising each state. The nature of the interactions provide evidence that soil behaves as an extended composite phenotype of the resident microbiome, responsive to the input and turnover of plant-derived organic carbon. We provide new evidence supporting the theory that soil-microbe systems are self-organising states with organic carbon acting as a critical determining parameter. This perspective leads us to propose carbon flux, rather than soil organic carbon content as the critical factor in soil systems, and we present evidence to support this view.",
"introduction": "Introduction Soil—the basis of terrestrial life on Earth—continues to defy our comprehensive understanding despite the evident catastrophic consequences of mismanagement, such as the North American “Dust Bowl” of the 1930s which was exacerbated by poor stewardship of agricultural soils 1 . Faced with the multiplicity of processes which constitute soil, scientific reductionism has led to studies which have advanced our knowledge of soil’s biological, chemical or physical components predominantly in isolation. However, soil—in common with many biological phenomena—is more appropriately considered a hierarchical assemblage of interacting processes, stabilized and actively maintained at different timescales 2 : soil is processual and not comprehensible based on single-discipline experimentation. Tisdall and Oades’ pioneering conceptual model 3 linking microbial activity to soil structural development advanced the importance of interaction between biotic and abiotic phenomena in the process of generating soil structural complexity (topology and connectivity). Soil organic matter (SOM) is the fundamental causative agent generating structural complexity, as it acts to bind mineral particles and colloids together. Plant and animal residues are processed by microbes before joining the SOM pool 4 , 5 : this step is an important facet of both the Tisdall and Oades model, and its subsequent extension 6 . SOM may take the form of microbial polysaccharidic and proteinaceous exudates as well as cell debris and is chemically structurally diverse 4 ; in effect, SOM is a continuum of progressively more extensively oxidized compounds 7 . Much of this SOM is associated with pores of 30–100 μm diameter 8 , scales comparable to the 12–13 μm distances observed in soil between microbial cells 9 . As a result, the effect of microbial processes—metabolism, extracellular degradation of compounds, polymer secretion and cell lysis—on soil structure is particularly evident at the scales < 50 μm responsible for regulating convective and diffusive flow rates, as well as the balance of air and water at any given matric potential 6 . These hierarchical processes exhibit characteristic properties of self-organizing and emergent systems 10 , 11 . Such experiential and theoretical approaches are formulating a new understanding of how microbial activity controls soil structure—in effect, how soil should be viewed as an expression of biological process. They also provide evidence supporting a view of soil as a product of genes, manifest through the combined effects of multiple organism phenotypes: in essence, an extended composite phenotype (Phillips 12 , after Dawkins 13 ). The identifying features of this phenomenon are a strong influence of at least one organism upon the form or structure of a soil environment—termed a process-form relationship ; demonstrable synchrony between the activity of influencing organisms and form development; selective pressure arising from form development acting, in Dawkins’ strict sense upon alleles 13 , 14 and in Phillips’ broader concept upon soil organisms 12 ; which results in positive feedback where selective pressure favours alleles (or organisms) associated with the process-form state, manifest as the influence of microbial turnover of SOM upon soil structural development, discussed above. There is compelling evidence implicating plant-derived organic carbon inputs in the soil extended composite phenotype 15 , 16 . However, complete description of such a phenotype requires, in turn, a well-developed understanding of the consequences of evolving soil structure for the genetic manifestation of on-going microbial processes—such feedback is necessary for emergence of organisation, observable at the whole-system level in complex biological, chemical and physical systems. Currently, few studies present comprehensive description of the influence of soil structure upon microbial processes, and those that do, typically address only the association of metabolically defined bacterial groups with soil aggregate or particle size, rather than soil structure per se (see Lensi et al. 17 and Chotte et al. 18 ). The principal influence of soil structural complexity is predicted to be on diffusion processes dictating the microenvironments surrounding surface-associated cells 19 . Observation of anaerobic regions of soil aggregates associated with denitrification processes 20 , and the influence of anaerobic microsites in ostensibly oxygen-rich soils upon microbial respiration and carbon compound oxidation rates 21 provide indirect evidence for such metabolic constraints arising from soil structure. However, this view of soil as an extended composite phenotype requires two specific conditions to be met. The first we term the Process-Form Condition , where the biological structures and functions that emerge from interactions between individual genotypes and their microenvironments should result in soil structural changes beyond the scale of individual cells. The second we term the Allelic Response Condition , where the process-form interaction should be reflected in significant modification at the level of individual alleles in soil microbiomes (i.e., fundamental changes in gene abundance patterns and whole metabolic pathways) such that alleles that correspond with specific processes are preferentially selected for—extending beyond short-term quantitative changes in specific gene expression profiles. In this paper, we integrate biological and physical data relating to dynamics of the soil system with mathematical modelling to explore these conditions. This approach is used to interpret results from a unique long-term field-experiment within the context of the proposed view of soil as an extended composite phenotype: linking organic carbon inputs to soil with emergence of key soil structural properties; and describing the gene-level microbiome responses to contrasting emergent soil structural complexity arising from long-term carbon input regimes. The experiment uses the Highfield Ley-Arable Experiment at Rothamsted Research, Harpenden, U.K.",
"discussion": "Discussion We have presented data consistent with the conditions which should be met if soil is an extended composite microbial phenotype: the emergent physical states are organised at several orders of magnitude above the scale of individual microbes; different physical states are associated with different genetic states at the level of individual alleles rather than organisms. The data also demonstrates that both physical and biotic states of the system can be manipulated by nutritional interventions, particularly relating to the flux of energy through the system. The data do not prove the associations are causal, however comparison with simpler systems that are amenable to deeper theoretical analysis and direct manipulation provides additional evidence of a causal feedback between allelic abundance, process and form. Specifically, the soil-microbe system exhibits behaviour seen in physical systems that display spontaneous (and endogenously driven) emergence of large-scale self-organisation as a result of such feedback. In a subset of such systems, there is a critical point at which the state of the system changes discontinuously (a phase transition ) with continuous change in one of the system parameters. The rate of change of the system with respect to that parameter is characterised by a power law close to the phase transition, reflecting the emergence of coherence across a wide range of scales (often referred to as fractal scaling). In Fig. 3 we observe such behaviour between connected porosity and hydraulic conductivity. The state of the soil system changes from one with a disconnected pore space to one with a connected pore space where C org (energy) flux is a critical parameter. There is a power law relation between conductivity and porosity, consistent with the emergence of large-scale spatial coherence in soil structure at a critical value of C org flux. In this sense, soil displays many of the properties of self-organising systems 39 . Results presented here provide further evidence for a causal feedback between allelic abundance, process and form. We have previously posited a mechanism for this in soil and shown how soils with and without plants are capable of spontaneously generating emergent structures at important scales 6 compared with sterile soils, which do not. This interpretation predicts that soils which are more self-organising will be more metabolically active in any given situation than a soil where the interaction between biological process and form is weak or non-existent. We see that, after a minimum of 52 years, each soil in our study is a different expression of its multiple biotic components; a phenomenon termed an extended composite phenotype 12 . With plants present, such as land managed as long-term mixed grass sward, the extended composite phenotype has an increased capacity to store water and soluble nutrients, a property which may confer a degree of resilience to the soil–plant–microbe system during periods of low rainfall or nutritional inputs. Independent analysis of these same soils has demonstrated greater water storage capacity in the grassland soil 40 . In addition, the more extensive and more connected pore network selects for assimilatory, and against dissimilatory, processes by permitting greater flux of O 2 through the system: it thus improves the efficiency of metabolic processes and C org conversion into biomass while reducing potential losses of plant nutrients arising from leaching or emission to the atmosphere. Thus, the extended phenotype interacts with plants to increase the flux, resilience and efficiency of nutrient transport to plants (including water). The finding that soil under grassland management has significantly higher capacity, efficiency and resilience compared with arable or bare fallowed management is associated with greater C org inputs and turnover. Furthermore, the rate of recovery of degraded soil is also linked to stocks and flows of C org (Fig. 1 ). Our experiments cannot distinguish between C org flux or storage as the dominant mechanism supporting improved soil function. However, interpreting results in terms of soil remodelling through self-organizing processes, we predict that the biophysical state of soil and rate of change of that state will both be related to cumulative metabolic activity. Our data are consistent with recovery rate being limited by cumulative soil metabolism: soil C org content acts as a diagnostic for this. This raises the important questions of what limits soil metabolism and incorporation of C org in soil 41 , and how it can be manipulated in each context to maximise the rate of soil recovery. We know both anaerobic niches and physical dislocation of microbes from resources result from low pore connectivity, and both significantly limit microbial metabolism. We also know soil recovery is associated with more voluminous and better-connected pore space and significantly lower levels of anaerobic respiration. We speculate that the rate-limiting factor in recovery of degraded soil is the process of microbially-mediated micro-structure remodelling, and that this is soil texture dependent 25 . Sandy-textured soil would be less able to recover compared to soils with higher fractions of silt and clay, where remodelling fine-scale structure is inherently more feasible due to a greater proportion of “raw materials” to enable such fine-scale architecture to be manifest. It is also likely to be dependent on the quality and quantity of organic inputs to soil, especially in relation to the latent energy contained in them. This is apparent in our data, though we are not able to distinguish the relative importance of each. Tillage is known to contribute to decreases in soil C org , and the most effective recovery rate and highest metabolizing end-state in our data was achieved with management under grassland without tillage. Tillage has the effect of significantly changing the distribution of microenvironments in soil through increased aeration and exposure of previously physically protected prey organisms and soil C org . This results in the immediate release of physical and chemical constraints on metabolism and therefore to loss of soil C org . More importantly, rearrangement of microenvironments—i.e. within and between soil macro- and micro-aggregates—will have the effect of “re-setting” microbial remodelling of soil microarchitecture, slowing down establishment of connected pore space and longer-term cumulative metabolism. This new interpretation of the role of nutritional and physical management of soil is a step towards a more general theory of soil. Such a theory is needed as a framework upon which to synthesize data and knowledge on biological, chemical and physical properties of soil that are typically studied in isolation. Theory leading to quantitative prediction is also essential in seeking synergistic interventions that recognise the interplay between capacity, efficiency and resilience of soil, and to avoid the unintended consequences of land management that are directing us towards systemic collapse of productive land and an amenable climate."
} | 3,677 |
20662386 | null | s2 | 7,684 | {
"abstract": "Molecular characterization of subsurface microbial communities in the former Homestake gold mine, South Dakota, was carried out by 16S rDNA sequence analysis using a water sample and a weathered soil-like sample. Geochemical analyses indicated that both samples were high in sulphur, rich in nitrogen and salt, but with significantly different metal concentrations. Microbial diversity comparisons unexpectedly revealed three distinct operational taxonomic units (OTUs) belonging to the archaeal phylum Thaumarchaeota, typically identified from marine environments, and one OTU belonging to a potentially novel phylum that fell sister to Thaumarchaeota. To our knowledge this is only the second report of Thaumarchaeota in a terrestrial environment. The majority of the clones from Archaea sequence libraries fell into two closely related OTUs and were grouped most closely to an ammonia-oxidizing, carbon-fixing and halophilic thaumarchaeote genus, Nitrosopumilus. The two samples showed neither Euryarchaeota nor Crenarchaeota members that have often been identified from other subsurface terrestrial ecosystems. Bacteria OTUs containing the highest percentage of sequences were related to sulphur-oxidizing bacteria of the orders Chromatiales and Thiotrichales. Community members of Bacteria from individual Homestake ecosystems were heterogeneous and distinctive to each community, with unique phylotypes identified within each sample."
} | 359 |
25992546 | PMC4551410 | pmc | 7,685 | {
"abstract": "Summary A major challenge in theoretical ecology is understanding how natural microbial communities support species diversity 1 - 8 , and in particular how antibiotic producing, sensitive and resistant species coexist 9 - 15 . While cyclic “rock-paper-scissors” interactions can stabilize communities in spatial environments 9 - 11 , coexistence in unstructured environments remains an enigma 12 , 16 . Here, using simulations and analytical models, we show that the opposing actions of antibiotic production and degradation enable coexistence even in well-mixed environments. Coexistence depends on 3-way interactions where an antibiotic degrading species attenuates the inhibitory interactions between two other species. These 3-way interactions enable coexistence that is robust to substantial differences in inherent species growth rates and to invasion by “cheating” species that cease producing or degrading antibiotics. At least two antibiotics are required for stability, with greater numbers of antibiotics enabling more complex communities and diverse dynamical behaviors ranging from stable fixed-points to limit cycles and chaos. Together, these results show how multi-species antibiotic interactions can generate ecological stability in both spatial and mixed microbial communities, suggesting strategies for engineering synthetic ecosystems and highlighting the importance of toxin production and degradation for microbial biodiversity."
} | 362 |
31827245 | PMC7031335 | pmc | 7,686 | {
"abstract": "Hadal trench bottom (>6000 m below sea level) sediments harbor higher microbial cell abundance compared with adjacent abyssal plain sediments. This is supported by the accumulation of sedimentary organic matter (OM), facilitated by trench topography. However, the distribution of benthic microbes in different trench systems has not been well explored yet. Here, we carried out small subunit ribosomal RNA gene tag sequencing for 92 sediment subsamples of seven abyssal and seven hadal sediment cores collected from three trench regions in the northwest Pacific Ocean: the Japan, Izu-Ogasawara, and Mariana Trenches. Tag-sequencing analyses showed specific distribution patterns of several phyla associated with oxygen and nitrate. The community structure was distinct between abyssal and hadal sediments, following geographic locations and factors represented by sediment depth. Co-occurrence network revealed six potential prokaryotic consortia that covaried across regions. Our results further support that the OM cycle is driven by hadal currents and/or rapid burial shapes microbial community structures at trench bottom sites, in addition to vertical deposition from the surface ocean. Our trans -trench analysis highlights intra- and inter-trench distributions of microbial assemblages and geochemistry in surface seafloor sediments, providing novel insights into ultradeep-sea microbial ecology, one of the last frontiers on our planet.",
"introduction": "Introduction The abyssal plain extends from the continental slope to the rim of deep trenches (3000–6000 m below sea level [mbsl]) and covers 85% of the global seafloor area, while the hadal zone (>6000 mbsl) comprises 1–2% of it [ 1 , 2 ]. In general, abyssal water and sediments are usually oligotrophic, and physical and chemical conditions (e.g., salinity, temperature, dissolved oxygen, and nutrient concentrations) in hadal water are similar to the overlying abyssal water despite the higher hydrostatic pressure [ 1 – 3 ]. However, cell abundance and microbial carbon turnover rates are significantly higher at hadal trench bottom compared with abyssal plain sediment below the surface layer, while those in outermost surface layer are sometimes comparable between hadal and abyssal sites [ 4 ]. This could be hypothesized to be attributed to factors apart from the vertical downward flux of sinking organic matter (OM) from the ocean surface and hydrostatic pressure. Hadal zones are generally located in oceanic trenches that are formed along plate boundaries by the movement of oceanic plates, and thus experience episodic and/or regular landslides [ 5 , 6 ]. These landslides cause downward transportation of surface sediments along with relatively fresh OM via the funnel effect of trench geomorphology [ 7 – 11 ]. Moreover, higher sedimentation rates and concentrations of subseafloor organic compounds in hadal trench bottom sediments compared with neighboring abyssal plain sediments have been reported in multiple trench regions under oligotrophic and eutrophic oceans [ 4 , 11 – 14 ]. Therefore, the labile organic carbon deposition in hadal zone supported by the high sedimentation is considered to facilitate establishment of distinct faunal and prokaryotic community observed at global scales [ 15 – 19 ]. Recently, the influence of physicochemical features on hadal biospheres were reported for microbial communities in the Mariana and Kermadec Trench regions [ 20 , 21 ], where under oligotrophic and relatively eutrophic (intermediate) ocean, respectively [ 22 , 23 ]; the pioneering studies reported microbial biodiversity in hadal trench bottom, trench slope, and adjusted abyssal plain sediments using culture-independent high-throughput sequencing techniques, and consequently demonstrated the structural similarity among each abyssal and hadal site. However, relations between geochemistry and microbial composition in hadal trench bottom and adjacent abyssal sediments have been still uninvestigated. Here, we evaluated prokaryotic community structure in 92 sediment subsamples of 14 sediment cores collected from four hadal trench bottom and seven adjacent abyssal plain sites located in three trenches in the northwest Pacific Ocean; the Japan, Izu-Ogasawara (Izu-Bonin), and Mariana Trenches (Fig. 1 and Table S1 ). The Japan and Mariana Trenches lie under eutrophic and oligotrophic oceans, respectively, while the primary productivity of the ocean above the Izu-Ogasawara Trench presents intermediate features. We performed geochemical analyses of the sediments and culture-independent microbial analyses including direct cell count, quantitative polymerase chain reaction (qPCR), and tag sequencing for small subunit ribosomal RNA (SSU rRNA) gene, to investigate intra- and inter-trench diversities of prokaryotic communities and potential metabolic interactions using co-occurrence networks. Our analyses illuminated the distinctive prokaryotic assemblages that spread through the trenches in each of abyssal and hadal zones, providing new insights into the microbial ecology in deep ocean, where one of the least understood aquatic habitats. Fig. 1 Maps of the sampling stations of sediment cores and their geographic locations. Plane (a) and three-dimensional (b, c) maps are shown. Stations with green and blue rectangles represent abyssal and hadal sites, respectively. The dashed line of JC station indicates that the station is located behind the seamount.",
"discussion": "Results and discussions Organic geochemistry TOC and TN in the studied sediments ranged from 0.10 to 3.28 and 0.02 to 0.42 wt%, respectively (Figs. S1 and S2 ). TOC concentrations were in general lower at the abyssal (0.10–0.55 wt%) than hadal stations (0.29–3.28 wt%), although large variations were observed at the hadal stations. At the abyssal sites, TOC concentrations of the outermost layers were 0.42 ± 0.09 wt% ( n = 7), and gradually decreased to 0.22 ± 0.10 wt% around 15 cm below seafloor (cmbsf) (Fig. S1 ). At the hadal sites TOC concentrations also decreased with sediments depth from 1.39 ± 0.95 wt% ( n = 6) at the outermost layers to 0.95 ± 0.61 wt% on average around 15 cmbsf. However, variations between the stations were substantially large and decreasing trend is less clear than abyssal sites. These trends are concordant to the previous findings that reported rapid sediment deposition and burial at hadal trench bottom compared with adjacent abyssal plain [ 4 ]. Layers with high TOC concentration were found in JC and IO1-2 hadal cores, which may be explained by event deposit (e.g., [ 9 ]). Among the hadal sites, the highest TOC and TN values were detected in the Japan Trench sites (1.63–3.28 and 0.22–0.42 wt%, respectively) and the lowest were observed in the Mariana Trench sites (0.16–0.59 and 0.02–0.08 wt%, respectively), which differed by an order of magnitude. The C/N ratio was concordant with past observations that concentrations of protein, carbohydrate, and lipid were generally higher in the Izu-Ogasawara Trench region compared with the Mariana Trench [ 45 ]. The δ 15 N values of surface sediments may reflect nutrient availability at the ocean surface due to isotopic fractionation during nutrient consumption by phytoplankton [ 46 ]; nutrient-rich conditions can lead to lower values (5.1–5.4‰ and 2.4–8.2‰ at the Japan and Izu-Ogasawara Trenches, respectively), whereas nutrient-poor conditions cause higher values (8.9–11.7‰ at the Mariana Trench). Thus, the sedimental OM concentrations and traits likely reflect the different geographical settings and productivity of the investigated station. It is generally expected that the organic carbon deposition and subsequent diagenetic process reflects the surface productivity of the overlying ocean. The Japan Trench (station JC) is located under the relatively eutrophic north-western Pacific Ocean, where nutrient-rich Oyashio currents encounter warm Kuroshio currents; In addition, the close distance to Honshu island, Japan, may contribute terrestrial OM to the seafloor [ 47 , 48 ]. In contrast, the other stations are located under the oligotrophic Pacific Ocean, far from continents or large islands. In particular, the Mariana Trench region is located near a subtropical gyre known to have one of the lowest surface ocean productivities [ 49 ]. Even if definitive conclusions could not be made from the available data, the differences in C/N ratios of each site could reflect differences in OM sources as well as stable isotopic signatures. Porewater chemistry We measured concentrations of dissolved oxygen (O 2 ) and porewater nutrients (NO 3 – , NO 2 – , NH 4 + , and PO 4 ) in the obtained cores (Figs. 2 and S3 ) to study microbial decomposition of sedimentary OM using oxygen and/or nitrate as electron acceptors [ 50 ]. O 2 concentrations decreased rapidly with increasing sediment depth in most sediment cores and depleted above 30 cmbsf, whilst those of abyssal cores collected from the Mariana Trench showed moderate decreases. The oxygen concentrations and decreasing trends at the Mariana trench region (~120–180 and 50–150 μM at the abyssal and hadal sites, respectively, in 0–16 cmbsf depth) were generally similar to those previously reported (130–180 and 50–180 μM) [ 4 ]. The NO 3 – concentrations in outermost layers at all stations except JC were ~35 μM and this was concordant with those in seawater at hadal zone in Izu-Ogasawara [ 8 ]. In all hadal cores, NO 3 – concentrations drastically decreased with sediment depth to less than 5 μM above 30 cmbsf with a concomitant increase in NH 4 + concentrations, especially at the Japan and Izu-Ogasawara Trenches. In contrast, no apparent depletion of NO 3 – ( > 27 μM) and lower NH 4 + concentrations (<11 μM) were observed throughout the sediment depths in all abyssal stations from the Izu-Ogasawara and Mariana Trench regions, which also exhibited low TOC concentrations. Notably, NO 3 − profiles showed large variations among hadal sediments compared with abyssal sediments. Among the NO 2 – profiles, clear subsurface peaks up to 6.6 μM were found in only three sediment cores (3, 7, and 7 cmbsf of cores from JC, IO1-1, and IO1-2, respectively). PO 4 concentrations generally increased with sediment depth, except for stations IO1-1 and IO1-2. These profiles of dissolved oxygen and nitrogen compounds are concordant with previous studies [ 4 , 51 ]. The profiles suggested that microbial nitrate respiration was relatively modest in abyssal sediments down to 50 cmbsf, whereas the respiration was active in hadal sediments above 30 cmbsf, probably reflecting higher concentrations of fresh OM. Interestingly, the increase rate of NH 4 + in hadal sediments along with sediment depth were gradually changed along with latitude of the sites (i.e., higher increasing rates at the northern site (JC) while lower in southern sites (MC-1 and MC-2)), suggestive of variance among stations in microbial populations and functions involved in nitrogen cycles. Fig. 2 Porewater chemistry of the surface sediments. Data down to 50 cmbsf are presented in this figure and entire sediment data are presented in Fig. S3 . Microbial abundances The direct cell counting and qPCR technique showed similar trends in microbial abundance (Figs. 3 and S4 ). Cell abundances by cell counting ranged from 4.2 × 10 5 to 9.6 × 10 7 cells/mL sediment, with a general decrease with sediment depth, while scattered profiles were found in the hadal sites of the Japan and Izu-Ogasawara Trenches (Figs. 3a and S4a ). The cell densities were similar to or even higher than those reported in other works [ 4 , 52 ]. Cell abundances in the hadal stations were generally higher than those in the adjacent abyssal stations. When comparing maximum cell abundances between cores, the largest and smallest cell abundances in hadal sites were found in the Japan and Mariana Trenches, respectively; for abyssal sites, abundance at the Izu-Ogasawara Trench was larger than that of the Mariana Trench. For qPCR analysis, the copy numbers of prokaryotic and archaeal SSU rRNA gene in each station ranged from 3.4 × 10 5 to 3.0 × 10 9 and 9.9 × 10 4 to 5.7 × 10 8 copies/mL sediment, respectively (Figs. 3b and S4b ). In addition, in the Mariana Trench region only, copy numbers from the shallow abyssal sediments were higher than those from the hadal sediments. The SSU rRNA gene copy numbers were 2–197-fold higher than the cell abundances by direct counting method, likely resulting from biases associated with direct cell counting (e.g., staining) [ 28 ] and molecular analyses (e.g., primers, probes, extracellular DNA, and/or SSU rRNA gene copy numbers on prokaryotic genomes [ 53 , 54 ]. However, cell abundance and SSU rRNA gene copy number was significantly correlated (Fig. S5 ). Interestingly, the ratio of archaea to prokaryotes rapidly decreased at approximately under 20 cmbsf of the hadal cores, while higher values were observed through the vertical profile in abyssal cores (Figs. 3c and S4c ), which may be explained by lower nutrient availability in abyssal than hadal sites. Overall, these trends were consistent with previous studies [ 4 , 52 , 55 , 56 ], supporting more vigorous microbial activity in hadal trench bottom sediments, especially in the subsurface under 5 cmbsf. Fig. 3 The abundance of microbes and ratios of archaea. Abundances were measured using ( a ) cell counting and ( b ) qPCR techniques. The X -axis represents cell counts and SSU rRNA gene copies per milliliter of sediment, respectively. The error bars represent standard deviation. ( c ) Ratios of archaea/prokaryotes were calculated using qPCR data. Data from layers ranging between 0 and 50 cmbsf are presented in this figure, and full data are shown in Fig. S4 . OTU-level compositions of microbial communities in sediment samples Based on the geochemical profiles in sediments, especially dissolved oxygen and nitrate concentrations, we selected four to ten layers from each sediment core for SSU rRNA gene tag sequencing. A total of 8,286,508 high-quality SSU rRNA gene sequences with 414 bp average length were obtained from the 92 sediment subsamples. The sequences comprised of 80,478 OTUs with 1587–13,181 (5478 average) OTUs per sample (Table S1 ). A number of OTU per sample were similar to or even higher than those reported in previous studies [ 21 , 57 ]. Based on rarefaction curves, the obtained OTUs in each sediment sample well represented their microbial communities (Fig. S6 ). To investigate compositional similarity between samples, we performed OTU-based NMDS analysis. OTU compositions were related to water depth (Fig. 4a ) and generally similar along with sediment cores at each station (Fig. 4b ). The OTU compositions were significantly differed between the abyssal and hadal sediments ( A = 0.11, p < 0.001, MRPP) (Fig. 4c ) and structured along the stations ( A = 0.31, p < 0.001) (Fig. 4d ). Significant associations were also observed with each of the pair of trench and zonation. ( A = 0.21, p < 0.001) (Fig. 4e ). Unexpectedly, the compositions in the abyssal sediments from the Izu-Ogasawara and Mariana Trench regions largely overlapped. The separation among the hadal sediments related with the different depression trends in porewater nitrate, TOC, and TN concentrations, but little with porewater oxygen concentration and cell abundance (Fig. S7 ). Fig. 4 Nonmetric multidimensional scaling (NMDS) plots for OTU compositions. The distance matrix was calculated based on the Bray–Curtis dissimilarity. The stress value of the final configuration was 18.2%. The sediment samples were colored depending on ( a ) water depth, ( b ) sediment depth, ( c ) zonation, ( d ) sampling station, and ( d ) trench and geomorphology. c–e Ellipsoids represent a 95% confidence interval surrounding each group, and each sediment core, zonation, and geography are coded by color. Vertical profiles of community diversity in the hadal sediment cores were distinct from those of the abyssal cores (Fig. S8 ). The Shannon, Simpson, and Chao1 index values of all abyssal cores gradually decreased with sediment depth, while those in hadal samples fluctuated. Overall, the Shannon and Chao index values in the upper most layers at the abyssal stations were higher than those at the hadal stations. In hadal cores, index values were decreased at ~5–20 cmbsf and retained or increased below these layers. In all hadal cores from the Izu-Ogasawara Trench, the Shannon index plot showed clear peaks at 8–25 cmbsf and the values were higher than those of the most surface layers. Conversely, in hadal cores from the Mariana Trench, the index values were slightly increased in deep sediment sections (>10 cmbsf), indicating that peaks were possibly located in layers deeper than 30 cmbsf. Interestingly, the diversity was decreased at layers close to those where depletion of oxygen and nitrate occurred (Fig. 2 ). These trends were concordant with the scattered microbial cell abundances observed in most of the hadal sediments in contrast to the abyssal sediments (Fig. 3 ). However, those trends were opposed to the general trends that prokaryotic growth and bioactivity are restricted according to sediment depth in energy-limited subseafloor sediments [ 48 , 58 , 59 ]. Thus, those trends of microbial diversity and cell abundance should be a unique feature of hadal subsurface biosphere. In the cases of gut and freshwater environments, it has been discussed that the supply of fresh nutrient resources generally correlate with microbial biomass and diversity [ 60 , 61 ]. According to the proposed theory, the feature of the hadal subsurface biosphere also could be explained by nutrient supply (i.e., the deposition of relatively fresh organic compounds with high sedimentation rate as discussed above). Besides, recent deposition of sediments via landslide could be other potential source of microbial cells that lead to varying diversity with sediment depth. Taxonomic composition of microbial communities in sediment samples Among the retrieved OTUs, 76,881 (99.7%) were taxonomically assigned to eleven Bacterial and two Archaeal phyla. Only 153 OTUs were assigned to Eukarya and the remaining 243 OTUs were taxonomically unassigned. The top three and ten most abundant phyla accounted for >58% and >88%, respectively, of the sequence pool of all sediment samples. The most abundant OTU in the sequencing pool belonged to Thaumarchaeota, which represent seven of the top ten OTUs (Fig. S9 and Supplementary Data 1 ). Overall, Thaumarchaeota (average 23.8%) was the most abundant phylum, followed by Proteobacteria (23.7%), Planctomycetes (10.6%), Chloroflexi (9.6%), Bacteroidetes (8.0%), Nanoarchaeaeota (6.2%), Acidobacteria (2.5%), Candidatus ( Ca.) Atribacteria (1.7%), and Ca . Marinimicrobia (1.7%) (Fig. 5 ). Within sequencing reads assigned to Nanoarchaeaeota, 99.6% were assigned to class Woesearchaeia, currently proposed as novel phylum Ca . Woesearchaeota. Thus, we assigned Nanoarchaeaeota as Woesearchaeota in this study. Within Proteobacteria, Gammaproteobacteria (10.3%) is the most abundant class, followed by Alphaproteobacteria (8.4%) and Deltaproteobacteria (5.1%) (Fig. S10 ). In general, the dominant phyla were similar to those of previous studies of abyssal and hadal sediments [ 20 , 21 , 52 , 57 , 62 – 64 ]. The most abundant OTU within Eukarya belonged to Alveolata (41.5%), followed by Stramenopiles (21.0%), Nucletmycea (19.4%), Holozoa (11.7%), and Rhizaria (2.2%). Fig. 5 Relative abundances of sequences at the phylum level. Groups with <5% abundance were summarized as “Other.” Sediment samples retrieved from deep sediment layers (>50 cmbsf) are indicated by surrounding black rectangles. Although dominant phyla were shared among subsamples within each of the sediment cores, their relative abundance gradually changed with sediment depth (Fig. 5 ). The relative abundances of Chloroflexi, Woesearchaeota, and Marinimicrobia generally increased in deeper sections (e.g., >8 cmbsf in IO1-2 and >2.5 cmbsf in IOC-2) along with the gradual decreases in oxygen and nitrate concentrations (Fig. 2 ). Conversely, Proteobacteria and Thaumarchaeota dominated in shallower sections (e.g., 0–7 cmbsf in IO1-2 and 0–22 cmbsf in MC-2). These trends are similar to previous studies [ 20 , 21 , 52 , 57 , 62 , 63 ] and likely depend on the concentrations of dissolved oxygen and nitrate in sediments. Supporting this, correlation analysis between taxa and geochemistry showed that the abundances of Proteobacteria and Thaumarchaeota were significantly positively correlated with oxygen and nitrate concentrations, respectively (Fig. S11 and Supplementary Data 2 ). Woesearchaeota has been detected from diverse benthic and anaerobic environments [ 54 , 65 – 68 ] and harbors genomic capability for fermentation-based metabolism [ 69 ]; hence, they may contribute to anaerobic carbon and hydrogen cycles in the deep seafloor sediments. We also observed some notable differences in prokaryotic communities likely associated with geographical location. Distinct community compositions were observed in station JC; Bacteroidetes dominated the communities (average 33.7%), while Thaumarchaeota was relatively scarce (4.1%). Within the Bacteroidetes, Flavobacteriaceae (belonging to class Flavobacteriia) is the most abundant family. The Flavobacterial OTUs abundant in station JC showed low similarity (88.0–91.4%) against Flavobacteriaceae sp. PRS1, whose genome was reconstructed from the Maria Trench water sample via single cell technique [ 56 ]. Because members of Flavobacteriia were reported to be abundant in eutrophic oceans [ 70 ], our results likely reflect eutrophic productivity in the Japan Trench. The predominance of phyla Atribacteria was found only in deeper sections of IO1-1 (15–29 cmbsf), IO1-2 (8–25 cmbsf), and IOC-1 (20–155 cmbsf) hadal stations at the Izu-Ogasawara Trench. Atribacteria is a common lineage in organic rich anaerobic sediments and probably grow with fermentation [ 71 – 73 ]. The higher abundances of Flavobacteriia and Atribacteria may represent a substantial deposition of degradable organic compounds into hadal sediments. The substantial differences revealed in the taxonomic analysis may also be explained by the geomorphological variations among sampling stations. Marinimicrobia showed significant unevenness between cores, with higher abundances observed in hadal (2.6%) verses abyssal sediments (0.73%) ( p < 0.05, U -test, Bonferroni correction). Marinimicrobia is known to be a dominant population in deep sea sediments and seawater, especially in oxygen-minimum zones, and harbors genetic potential of diverse energy metabolic processes using sulfur and nitrogenous compounds as electron donor and acceptor [ 74 – 78 ]. In contrast, Ca . Schekmanbacteria was detected in all abyssal samples (average 0.82%), while there was significantly lower abundance (0.007%) in hadal samples ( p < 0.05, U -test, Bonferroni correction). Although several draft genomes of the recently proposed phyla Schekmanbacteria were reconstructed by metagenomic approach [ 79 ], their biological and geochemical importance remains unclear. The relative abundance of Thaumarchaeota showed drastic decrease in hadal stations below 6–15 and 20–30 cmbsf in core(s) from the Izu-Ogasawara (IO1-1, IO1-2, IOC-1, and IOC-2) and Mariana (MC-2) Trenches, respectively, where nitrate was consumed and ammonium emerged (Fig. 2 ). In contrast, low abundance of Thaumarchaeota was continuously observed in sediments from the Japan Trench (JC), where nitrate concentration was depleted through the sediment core (Fig. 2 ). This was concordant with qPCR analyses (Fig. 3c ), as well as previous observations that Thaumarchaeota frequently dominated in aerobic sediment columns and radially decreased across the oxic–anoxic transition layer [ 63 , 80 ]. The most predominant family of Thaumarchaeota in the sediments was Nitrosopumilaceae (92%), which are known to be ammonia oxidizing archaea (AOA) [ 81 – 83 ], and thus may contribute markedly to nitrification processes in trench surface sediments. The co-existence of Thaumarchaeota and Marinimicrobia at relatively high abundance suggests the co-existence of nitrification and denitrification processes, respectively, as described previously [ 52 , 84 ]. Although abundance of functional genes related with nitrification (e.g., amoA ) was not analyzed in this study, we should note that abundance of the amoA gene decreased with sediment depth in trench bottom sediments from the Mariana and Izu-Ogasawara Trenches [ 52 , 84 ]. Co-occurrence network structure of OTUs Many prokaryotic lineages are known to establish consortia with specific prokaryotic members, who inhabit same environment and sometimes sharing similar ecological niches and biological interactions (e.g., sharing metabolic compounds via fixation and translocating process) [ 85 ]. Since co-occurrence patterns can be useful for revealing such concrete but mostly hidden relationships from complex community datasets, co-occurrence network analyses have been widely applied to various SSU rRNA tag sequencing datasets of marine and other environments [ 86 – 89 ]. Here, we conducted co-occurrence analysis to understand core metabolic interactions among microbes in trench subseafloor sediments. The co-occurrence network showed six clusters composed of 3–36 OTUs (66 OTUs in total) and 2–247 edges (Fig. 6 ). The 66 OTUs represented 23.6% of community compositions in each sample on average. Interestingly, most subnetworks linked to oceanographic zonation (i.e., abyssal and hadal) and sediment depth (Fig. 7 and Supplementary Data 3 ); OTUs belonging to the largest group A (composed of two subgroups A-1 and A-2) were abundantly detected in the abyssal sediments through the Izu-Ogasawara and Mariana trenches, whilst abundance OTUs in other groups were higher in hadal sediments (i.e., under 10 cmbsf for group B, in shallow sections (0–30 cmbsf) for group C, and above 6 cmbsf for group D). The members of groups E and F were detected from both abyssal and hadal cores. Although OTUs of groups C and F were spread among the three trenches, those of groups B, D, and E were rare in cores from the Japan Trench. Fig. 6 Architecture of co-occurrence OTU networks. Nodes represent OTUs and edges (blue lines) represent statistically significant positive correlations of each OTU pair. The size of nodes represents relative abundance of OTUs in the data set. Nodes are colored by taxonomy at the phylum level. Fig. 7 Bubble plot showing comparative OTU profiles belonging to each co-occurrence network. Bubbles are colored by sampling station and bubble sizes correspond to relative abundances. The white and gray backgrounds represent abyssal and hadal sediment samples, respectively. A to F written in the right side represent co-occurrence network groups summarized in Fig. 6 . OTUs with asterisks indicate statistical significance of localization in either the abyssal or the hadal samples ( p < 0.05, U -test, Bonferroni correction). Groups A and D were dominated by OTUs assigned to order Nitrosopumilales (Fig. S12 ). Nitrosopumilales are considered as aerobic nitrifying [ 81 – 83 ], and abundance of these OTUs were concordant with the high oxygen and nitrate concentrations in the sediments (Fig. 2 ). OTUs of group D belonged to one small clade (showing 95.6–96.1% identity with 16S rRNA gene of Nitrosopumilus maritimus SCM1 [DQ085097]) while those of group A were spread throughout the order. The niche separation of AOA subgroups is regulated by the availability of electron donors like ammonia [ 19 , 90 , 91 ]. Supporting this, the dominance of the group D clade (up to 78.1% in Thaumarchaeota in the hadal samples) is consistent with the enrichment of labile OM in the hadal sediments. Unfortunately, further subgroup assignment of Thaumarchaeota OTUs using short 16S rRNA gene tag sequences were not technically feasible in contrast to the previously described amoA gene [ 43 ]. Although most proteobacterial OTUs in group A were assigned to lineages that currently less well-understood, two OTUs were assigned to genus Woeseia in Chromatiales; this genus likely possesses denitrification pathway-related genes [ 92 ], and may consume nitrates that provided by nitrifiers including Thaumarchaeota. Intriguingly, group B consisted of 13 OTUs belonging to five phyla (Planctomycetes, Atribacteria, Chloroflexi, Bacteroidetes, and Ca . Aerophobetes), and OTUs were abundant in the deeper sections where oxygen and nitrate were depleted (Fig. 2 ). All 13 OTUs showed highest sequence similarities with uncultured lineages, most of which were found in anaerobic environments with low oxygen concentrations. Atribacteria partake in fermentation metabolisms that produce acetate, hydrogen and carbon dioxide [ 71 – 73 ]. Similar to Atribacteria, recently defined bacterial phylum Aerophobetes were reported to harbor saccharolytic and fermentative metabolism capabilities [ 93 ]. All Planctomycetes OTUs were assigned to order MSBL9. Metagenome sequencing analyses identified genes encoding pyruvate formate-lyase from a member of MSBL9 [ 94 ], indicative of fermentation capabilities. All three Chloroflexi OTUs were assigned to class Dehalococcoidia, which is frequently detected from seafloor sediments and possesses genes related to hydrogen and sulfur compound oxidation with reductive dehalogenation of halogenated organic compounds [ 95 – 97 ]. Notably, MSBL9 and Dehalococcoidia possess potential of flavin secretion in marine sediments [ 98 ], implying that these lineages also contribute to maintaining extracellular metabolite pools in hadal sediments. The single Bacteroidetes OTU was affiliated with class BD2-2, which may interact with methanotrophic archaea and sulfate-reducing bacteria in methane seep sediments [ 99 ]. While knowledge of group B OTUs is still limited, they may cooperatively establish anaerobic metabolic networks; e.g. products of fermentation by Atribacteria, Aerophobetes, and MSBL9 are used as electron donors by BD2-2 and Dehalococcoidia OTUs, and then the fresh labile OM will be used as energy resources again in hadal sediments following necromass turnover recycling, as discussed previously [ 100 ]. Overall, the network structure analyses highlighted associations between prokaryotic consortia and geochemical conditions (geomorphological zonation and sediment depth). Part of these consortia may represent potential metabolic interactions with habitat transition, although further experimental validation should be required. In addition, the consortia structures were widespread among the trenches in the northwest Pacific Ocean. While three groups (B, C, and D) were selectively abundant in the hadal sediments, only one group (A) showed high preference in the abyssal sediments (Fig. 7 ). Factors impacting hadal subseafloor ecosystem Here, we conducted culture-independent molecular analyses of trans-trench prokaryotic communities in the abyssal plain and hadal trench bottom sediments collected from three different trench systems under different oceanographical settings to understand the general role of hadal environments on subsurface geochemical cycles and microbial ecosystems. Microbial cell abundance showed greater biomass in the hadal sediments verses abyssal sediments especially in deeper layers, which is consistent with previous studies [ 4 , 52 , 55 ]. Although we cannot exclude an impact of water pressure on benthic microbes [ 101 ], the clear relations between geochemistry and microbial community at the hadal sediments likely indicating more importance of factors related to geomorphology rather than merely water pressure. Overall, the microbial composition suggested that development of prokaryotic communities depends on ocean geomorphological zonation (i.e., abyssal vs. hadal), geographic regionality (i.e., productivity of overlying ocean surface), and factors associated with sediment depth. We also observed different vertical fluctuation of microbial community diversity between the hadal and abyssal sediments and identified potential prokaryotic consortia that spread among inter–trenches habitats and likely share energy-conserving metabolic processes in both abyssal and hadal sediments. In general, deposition of OM and subsurface microbial cell abundance is related to productivity of the surface ocean and distance from continents or islands [ 48 ]. Indeed, in our abyssal sites, greater microbial cell abundance and oxygen consumption were observed in the Izu-Ogasawara compared with Mariana trench regions (Figs. 2 and 3 ). However, microbial compositions and some geochemical parameters (nitrate, TOC, and TN) were unexpectedly similar between these two regions, indicating that the impact of surface productivity on subsurface microbial community is much smaller than expected in the abyssal plains under oligotrophic to ultraoligotrophic oceans. In the hadal sites, we found variations in geochemical parameters, cell abundances, and microbial compositions along with surface productivities. The differences between the abyssal and hadal sediments cannot only be explained by the vertical flux of sinking organic particles influenced by the ocean surface productivity. To explain the variations, we have two hypotheses. One of them is a presence of hadal currents that could supply intrinsic OM on hadal sites apart from those sinking directly from ocean surface. The other is a difference in sedimentation rates between hadal and abyssal sites that change the impact of ocean surface productivity on subsurface microbes. Lateral transport along the trenches is one of the possible sources of OM in hadal trench bottom sediments apart from sinking OM. There are north- and south-ward currents along the trench axis at the Izu-Ogasawara and Japan Trenches, respectively [ 102 , 103 ]. These currents may contribute to transportation of suspended particles with relatively high OM contents from the north to south. A part of the latitudinal gradients in TOC values among hadal sites (Fig. S1 ) could be explained either by this lateral OM supply, or benthic microbial populations that prefer subsurface ecosystems under eutrophic oceans represented by Atribacteria (Fig. 5 ). However, we could not find clear geochemical signatures supporting OM delivery along the trenches. Also such currents may contribute for microbial dispersal, growing up the importance to elucidate the water current in abyssal and hadal zones. OM degradation process at the surface sediments may differ between hadal and abyssal sites due to differences in sedimentation rates, which are very high at hadal while low at abyssal sites. Extremely high sedimentation rates at hadal trench bottom, driven by landslides on trench slope via the funnel effect, cause the rapid deposition of labile OM into the subsurface [ 9 , 55 ]. This burial prevents oxidative degradation of OM at surface sediments, allowing semilabile OM to be available to subsurface microbes. In contrast, abyssal plains generally have slower sedimentation rates. In our studied Izu-Ogasawara Trench sites, estimated sedimentation rates were 25 and 2.9 cm per 1000 years at the hadal (IOC-2) and abyssal (IOA) stations, respectively, based on bulk 14 C-age analysis (Nomaki et al. unpublished data). The slower sedimentation rates at abyssal sites have allowed continuous oxic degradation of OM at surface sediments for over 1000 years, and labile OM are likely more diminished than those at hadal sites. Consequently, the differences in TOC concentrations (0.11–0.55%) were small across the abyssal plains in our sites, while those at the hadal sites varied substantially (0.16–3.28%) (Fig. S1 ). The variations in TOC concentration among the hadal trench bottom sediments subsequently influenced the dissolved oxygen and nitrate profiles through sediment depths (Fig. 2 ). Higher diversity of microbial communities among hadal than abyssal sediments (Fig. 4c ) may reflect variations in TOC and porewater chemistries. Moreover, the differences in labile OM deposition may be reflected in the microbial cell abundance in the abyssal stations (Fig. 3 ) instead of microbial community structures. Our findings provide new perspectives into hadal biospheres under different oceanographic regions, displaying contrasting properties to abyssal biospheres. We also identified novel insights into abyssal geochemistry and microbial communities whereby variations in surface productivity at abyssal sites are not profound in oligotrophic to ultraoligotrophic areas because the OM buried into subsurface sediments are extensively degraded before burial. However, we could not specify the factors controlling microbial ecosystems and biogeochemical cycles in this study. To further understand such controlling factors, investigation of microbiological processes in intra- and inter-cell scales and elucidation of biological mechanisms of trench systems that impact hadal biospheres are necessary. In addition, ecological functions and phylogenetic classifications of most predominant lineages in deep seafloor sediments remain largely unknown. Also, it should be noted that there is still difficulties in estimation of definitive phylogenetic and functional characters in genus or strain level based on the partial SSU rRNA gene sequencing reads. Moreover, although bacteria and archaea may account for a major part of ecosystems, viruses could facilitate biogeochemical cycles through biological interactions with prokaryotes in oligotrophic deep sea environment [ 100 , 104 – 106 ]. Therefore, further taxonomic composition analyses, gene- and genome-centric approaches (e.g., metagenomics, metatranscriptomics, and metaepigenomics [ 107 ]), and integrative analyses with viruses (e.g., viromics) will provide further insights into microbial ecology and associated biogeochemical cycles."
} | 9,532 |
34887550 | PMC7612208 | pmc | 7,688 | {
"abstract": "Mako is a software tool that converts microbiome data and networks into a graph database and visualises query results, thus allowing users without programming knowledge to carry out network-based queries. Mako is accompanied by a database compiled from 60 microbiome studies that is easily extended with the user’s own data. We illustrate mako’s strengths by enumerating association partners linked to propionate production and comparing frequencies of different network motifs across habitat types."
} | 124 |
36330267 | PMC9623246 | pmc | 7,689 | {
"abstract": "Symbiotic relationships with microbes may influence how plants respond to environmental change. In the present study, we tested the hypothesis that symbiosis with the endophytes promoted salt tolerance of the native grass. In the field pot experiment we compared the performance of endophyte-infected (E+) and endophyte-uninfected (E−) Leymus chinensis , a dominant species native to the Inner Mongolia steppe, under altered neutral and alkaline salt stresses. The results showed that under both neutral and alkaline salt stresses, endophyte infection significantly increased plant height, leaf length and fibrous root biomass. Under neutral salt stress, endophyte infection decreased Na + content and Na + /K + ratio ( p =0.066) in the leaf sheath while increased Ca 2+ and Mg 2+ content in the rhizome. Under alkali salt stress, endophyte infection tended to increase K + content in the fibrous root, enhance Mg 2+ content in the fibrous root while reduce Na + /K + ratio in the leaf blade in the 100 mmol/L alkali salt treatment. Although endophyte-infected L. chinensis cannot accumulate Na + high enough to be halophytes, the observed growth promotion and stress tolerance give endophyte/plant associations the potential to be a model for endophyte-assisted phytoremediation of saline-alkaline soils.",
"conclusion": "Conclusion There are over 932 million hectares of land suffering salinization and alkalization around the world ( Rengasamy, 2006 ). It is estimated that soil salinization caused a total loss of US$27.3 billion, with a direct impact on the global economy ( Meena et al., 2017 ). This study demonstrated that endophyte infection could enhance host tolerance to both neutral salt and alkali salt stresses. The significant effect of endophyte infection was decreasing Na + /K + ratio, increasing Ca 2+ and Mg 2+ contents, and thus promoted leaf and fibrous root growth. It is estimated that at least 30% of approximately 3000 pooid grass species harbor systemic endophytes, and most can develop this mutualistic association ( Leuchtmann, 1992 ). In this respect, the potential application of endophytes to mitigate negative salt stress impact would be a more favorable choice. Certainly, this experiment was conducted using soil in pots. Further comparisons of E+ and E− plants under natural field stress conditions will help to verify the potential use of grass-endophyte symbiota in phytoremediation of saline-alkaline soils.",
"introduction": "Introduction Soil salinization is a widespread issue throughout the world. It is estimated that more than 6 percent of the world’s land and 30 percent of the world’s irrigated areas suffer from salinity problems ( Chaves et al., 2009 ). Soil salinization generally includes two kinds of stresses: neutral salt stress resulted from NaCl and Na 2 SO 4 , and alkali salt stress resulted from NaHCO 3 and Na 2 CO 3 . Plants grown in neutral salt-affected soils are mainly subjected to water deficit, ion toxicity, and disorders of mineral nutrients ( Reza Sabzalian and Mirlohi, 2010 ). Excess amount of salt ions in soils can damage plant roots as well as the aboveground plant parts, and appear to be a major constraint to plant and crop productivity ( Parihar et al., 2015 ; Liu et al., 2022 ). Alkali salt stress causes not only the detrimental salt stress, but also induces high pH stress. The high pH soil surrounding the roots directly reduces the root activity, interferes with ion uptake, and breaks intracellular ion balances in plants ( Guo et al., 2009 ; Kaiwen et al., 2020 ). Saline soil rehabilitation can be performed by non-plant based, environment-friendly modifications (e.g., structural engineering alterations, leaching of salts) and plant-based remediation (e.g., phytoremediation) ( Gutierrez-Gines et al., 2016 ). Besides halophytes, the use of plant varieties with elevated tolerance to salinity is another alternative in phytoremediation ( Yin et al., 2014 ). Several studies have demonstrated that the local adaptation of plants to their environment is usually driven by closely associated microbes ( Rodriguez and Redman, 2008 ; Porcel et al., 2011 ) and it has been documented that mutualistic symbiosis with rhizobia ( Franzini et al., 2019 ) and arbuscular mycorrhizal fungi ( Rodriguez et al., 2004 ; Ma et al., 2021 ) can reduce the negative effects of salinization and improve the salt tolerance of plants grown in saline soils. Mutualistic microbiota symbiosis may improve water absorption and osmotic regulation of host plants by alleviating the adverse effects of excess salt ion accumulations in host plants ( Franzini et al., 2019 ). \n Epichloë endophyte is a class of symbiotic fungi and mainly exists in the aboveground part of the plants. Plants serve as hosts and provide nutrients to their endophytes, and endophytic fungus protect them from biotic and abiotic stresses ( Xia et al., 2015 ; Bastias et al., 2017 ; Shi et al., 2020 ), especially drought ( Ren et al., 2011 ; Wang et al., 2015 ; Decunta et al., 2021 ; Manzur et al., 2022 ). Endophytic fungus confer drought tolerance to the host grass by increasing root growth, enhancing photosynthesis and osmotic adjustment ( Richardson et al., 1992 ; Malinowski et al., 1998 ; Decunta et al., 2021 ). When growing in saline soils, plants are also exposed to drought stress. But limited studies about the effect of endophyte infection on salt tolerance of host grasses have focused on the neutral salt stress in a hydroponic system. In our previous study, we found that endophyte infection could significantly improve the tolerance of tall fescue to NaCl solution, a common lawn grass, by enhancing plant biomass and Na + uptake ability of the host grass ( Yin et al., 2014 ). Recently, Chen et al. (2021) found that endophyte infection ameliorated adverse effects of NaCl solution on Hordeum brevisubulatum , an important forage crop, by increasing the conducting tissues and endodermis thickness, which may help inhibiting water loss and the decrease of transport capacity. Endophyte-infected species occur in almost all habitats where grasses are common, including a large number of wild grasses, pasture grasses, lawn grasses, cultivated grains and their wild relatives, as well as weed grasses ( Bacon, 1993 ; White, 1993 ; Clay and Schardl, 2002 ). The expression of salt tolerance in a saline hydroponic system could be different from expression in a saline soil-based system ( Tavakkoli et al., 2012 ). Moreover, salinization and alkalization frequently co-occurred in soils, and the alkali stress expressed more serious growth inhibition than the neutral salt stress. However, little is known about whether endophyte infection can also improve the alkali tolerance of host plants. \n Leymus chinensis , an important perennial rhizome grass, is widely distributed at eastern Eurasian steppe zone, from North Korea westward to Mongolia and northern China, and northwestward to Siberia. Due to excellent stress tolerance, rapid growth, high palatability and herbage production, L. chinensis is an economically and ecologically important forage grass ( Lin et al., 2017 ; Liu et al., 2017 ). In the present study, endophyte-infected (E+) and uninfected (E−) L. chinensis were planted under neutral salt ranging from 0 to 400 mmol/L treatments or alkali salt ranging from 0 to 300 mmol/L treatments. The changes in the plant growth and photosynthesis, biomass allocation, and the accumulation of Na + , K + , Ca 2+ , Mg 2+ and Na + /K + in various plant parts were tested. Specifically, the following questions were addressed: (1) does neutral salt tolerance also exist in endophyte-infected wild grasses in soil-based system? And (2) does the endophyte improve alkali tolerance in the host grasses?",
"discussion": "Discussion Both alkali salt and neutral salt stresses involved in deleterious effects of salinity on plant growth; the adverse effects of high pH alkaline solutions on plant growth were more severe than those of low pH salt treatments. High pH is the key characteristic of alkali salt stress that is different from salt stress. The responses of L. chinensis were significantly correlated not only with salinity but also with the pH of the alkalinity treatment. It has been reported that alkali salt produced a stronger stress than neutral salt in both halophytes and glycophytes ( Yang et al., 2009 ; Li et al., 2010 ; Gong et al., 2014 ). In the present study, L. chinensis survived under 400 mmol/L neutral salt treatment, yet died under 300 mmol/L alkali salt treatment, which was consistent with previous reports. It has been well documented that Epichloë endophyes could ameliorate drought stress for both agronomically important forage species ( Clay and Schardl, 2002 ; Saikkonen et al., 2006 ) and wild grass species ( Ren and Clay, 2009 ; Liu et al., 2017 ). In the hydroponic experiment, the beneficial effects of Epichloë endophytes on NaCl tolerance has been demonstrated in several studies ( Reza Sabzalian and Mirlohi, 2010 ; Yin et al., 2014 ; Chen et al., 2021 ). However, expressions of salinity tolerance in hydroponic systems might different with the performance in the soil. Tavakkoli et al. (2010) ; Tavakkoli et al. (2012) compared salt tolerance of barley under three different growing conditions (hydroponics, potted soil and a naturally saline field) and found that salt tolerance differences between genotypes were expressed when grown in the soil, but cannot be discerned when grown in the hydroponic system. As for plants grown in the potted soil or in the naturally saline field, they expressed similar degrees of salt tolerance. In the present study, we found that endophyte infection significantly increased plant height, leaf length and fibrous root biomass under both neutral and alkali salt stresses, indicating that Epichloë endophytes can improve the host resistance to both neutral salt as well as alkali salt grown in the potted soil, only their beneficial effect was more obvious under neutral salt stress than under alkali salt stress. The mechanisms involved in endophyte-associated salt tolerance of the host have not been documented. According to reported studies and our research, the following possible reasons were proposed. Firstly, endophyte infection could improve root absorption that may alleviate drought and nutrients deficiency during salt stresses. Arbuscular mycorrhizal fungi has been reported to enhance plant growth under salt stress mainly through an extensive hyphal network which allowed enhancement of water and nutrient acquisition ( Plenchette and Duponnois, 2005 ). Epichloë endophytes can also increase root absorption ( Malinowski et al., 1999 ; Crush et al., 2004 ) in response to water stress. In the present study, L. chinensis produces two structurally and functionally different organs belowground: fibrous roots that play a central role for root absorption, and rhizomes that serve mainly as translocation conduits and overwinter storage. We found that endophyte infection significantly enhanced fibrous root biomass under both neutral and alkali salt stresses. Although we did not find that endophyte infection improved leaf water content of the host, we did find that endophyte infection increased LNC under neutral salt stress, which was usually negatively influenced by salt stress ( Frechilla et al., 2001 ). Another mechanism used by Epichloë endophytes to promote salt tolerance of the host may be the regulation of plant nutrition. The Na + ion is the main toxic ion in salinized soil. Unlike Na + , K + plays a key role in many physiological processes vital to plant growth. Under salt stress, a good balance of Na + /K + ratio is important for maintaining ion balance and a number of enzymatic processes, and thus is a potential indicator of salt tolerance in the plants ( Munns and Tester, 2008 ; Abdelhamid et al., 2010 ; Tomar and Agarwal, 2013 ). Arbuscular mycorrhizal fungi has been found to play a significant role in sustaining a high K + /Na + ratio in plants that were exposed to salt stress ( Selvakumar and Thamizhiniyan, 2011 ; Zhang et al., 2011 ; Kadian et al., 2013 ; Fariduddin et al., 2014 ). In the present study, we found that endophyte infection tended to decrease Na + /K + ratio in leaf sheath under neutral salt stress, and also reduced Na + /K + ratio in leaf blade in the 100 mmol/L alkali salt treatment. The accumulations of Ca 2+ and Mg 2+ in plants are usually inhibited by salt stress ( Khan et al., 1999 ; Aziz and Khan, 2001 ). Arbuscular mycorrhizal fungi has been documented to increase the uptake and concentration of Ca 2+ in different plants ( Elhindi et al., 2017 ; Cui et al., 2019 ). Rahman and Saiga (2005) reported that endophyte-infected tall fescue had a greater ability to take up Ca 2+ and Mg 2+ , compared to uninfected plants under normal growth conditions. Bayat et al. (2009) found that endophytes increased Ca 2+ content in tall fescue under drought stress. In the present study, endophyte infection increased Ca 2+ and Mg 2+ contents in the rhizome under neutral salt stress while increased Mg 2+ content in the fibrous root in 100 mmol/L alkali salt treatment. It is well known that Mg 2+ is the key component of chlorophyll. The Ca 2+ ion can maintain membrane stability, help to form cell walls, and take part in signal transduction. Here, increased Mg 2+ in response to endophyte infection may be related to alleviate chlorophyll breakdown, while increased Ca 2+ may be related to membrane stability and signal transduction ( Cui et al., 2019 ; Thor, 2019 )."
} | 3,403 |
35189333 | PMC8942843 | pmc | 7,690 | {
"abstract": "Syntrophus aciditrophicus is a model syntrophic bacterium that degrades fatty and aromatic acids into acetate, CO 2 , formate, and H 2 that are utilized by methanogens and other hydrogen-consuming microbes. S. aciditrophicus benzoate degradation proceeds by a multistep pathway with many intermediate reactive acyl-coenzyme A species (RACS) that can potentially N ε -acylate lysine residues. Herein, we describe the identification and characterization of acyl-lysine modifications that correspond to RACS in the benzoate degradation pathway. The amounts of modified peptides are sufficient to analyze the post-translational modifications without antibody enrichment , enabling a range of acylations located, presumably, on the most extensively acylated proteins throughout the proteome to be studied. Seven types of acyl modifications were identified, six of which correspond directly to RACS that are intermediates in the benzoate degradation pathway including 3-hydroxypimeloylation, a modification first identified in this system. Indeed, benzoate-degrading enzymes are heavily represented among the acylated proteins. A total of 125 sites were identified in 60 proteins. Functional deacylase enzymes are present in the proteome, indicating a potential regulatory system/mechanism by which S. aciditrophicus modulates acylation. Uniquely, N ε -acyl-lysine RACS are highly abundant in these syntrophic bacteria, raising the compelling possibility that post-translational modifications modulate benzoate degradation in this and potentially other, syntrophic bacteria. Our results outline candidates for further study of how acylations impact syntrophic consortia.",
"discussion": "Discussion PTMs can modulate protein function in response to cellular or environmental changes and may occur spontaneously due to cellular conditions or simply over time ( 17 , 54 , 55 , 56 ). Acylation is one class of modification that can be spontaneously induced in the presence of acyl phosphate or RACS or can be enzyme-mediated by a lysine acyltransferase (KAT) ( 57 ). Regardless of its origin, acylation can affect enzymatic activity profoundly. In vitro work has shown that modifying lysine side chains near catalytic regions of bacterial enzymes can directly alter function ( 58 , 59 , 60 ). Work in other systems has shown more subtle but equally significant acylation effects: acetylation can disrupt enzyme complexes, thereby altering activity ( 61 , 62 , 63 , 64 ). Acetyl-lysine has been identified as a ubiquitous modification in bacteria ( 65 ). Most acetylated proteins in our system were involved in the synthesis of secondary metabolites ( Fig. 3 A ), which is sensible given that acetyl-CoA is an intermediate metabolite in many of those pathways. Benzoyl- ( 51 ), glutaryl- ( 66 ), hydroxybutyryl- ( 67 , 68 ), crotonyl- ( 69 , 70 ), and butyryl-lysine ( 71 , 72 ) have shown function in some systems under certain conditions, although the studies have been less extensive than those for acetyl-lysine, and few, if any, bacterial studies have reported their occurrence. A 3- hydroxypimelylation has previously been reported in S. aciditrophicus by our laboratory ( 22 ). Detecting these modifications across biological replicates without pre-enrichment suggests that they are uniquely prevalent and abundant in this system. While butyryl-CoA is not expected to be an intermediate in the degradation pathway, S. aciditrophicus has genes for butyrate dehydrogenase activity ( 3 , 8 ) that could synthesize butyryl-CoA from excess crotonyl-CoA or reduced cofactors ( 8 , 12 ). To begin to understand how these modifications affect global cellular function requires a comprehensive and unbiased catalog of acylations. Recording a biological system’s acyl modifications comprehensively presents many challenges. Low stoichiometries of acyl modifications in many model systems have required enrichment to identify and characterize acylated peptides ( 66 , 73 , 74 , 75 , 76 , 77 , 78 , 79 , 80 , 81 ). Pan-specific antibodies are valuable for enriching specifically modified peptides but have some limitations. Antibody cross-reactivity may lessen the modification specificity of enriched peptides, a problem for Western blotting, but not mass spectrometry. Sequence-dependent binding efficiencies can bias results and challenge quantification strategies. Our syntrophic system provides a unique opportunity to investigate the array of acylations without enrichment, bypassing technical challenges in other systems. The experimental approach described herein using RACS as a predictor for lysine acylations complements antibody enrichment and utilizes diagnostic marker ions to validate putative modifications proposed from an enlarged search space. Many of the S. aciditrophicus proteins decorated with acyl modifications function to degrade benzoate, a process that could result in carbon or reductive stress, if unregulated. Modulating enzymatic activity through protein acylation would be a fitting and elegant mechanism for global metabolic regulation, given the limited energy available to these cells and the high concentrations of acyl-CoA intermediates, a feature also evidenced by the unusual ability of this species to use acetyl-CoA to form ATP ( 29 ). The benzoate degradation pathway is a likely candidate for this level of regulation as its constituents frequently interact with RACS. While we have identified a wide range of acyl modifications in this system, the biological role that these modifications play has yet to be explored. We can, however, provide some insight into the functional effect of certain acylation sites by drawing on similar modifications in other systems. BCL acylation, specifically acetylation, has been identified as a means of negative feedback inhibition, stopping benzoate-CoA ligase activity in R. palustris by directly inhibiting enzymatic catalysis ( 48 ). The large number of acyl modifications we report at this site may indicate that when its respective acyl-intermediate builds, the PTM can act as a brake to slow or stop degradation, thereby acting as a negative feedback inhibitor to mitigate carbon and reductive stress. BCL also consumes ATP, creating a direct link between the energetic state of the cell and aromatic degradation. Additionally, the multiple sites of acetylation found on ATP synthase subunits further hint at a link between bioenergetics and acylation as it, too, demonstrates a relationship between the reactive metabolites and oxidative phosphorylation. Beyond energetics, the cellular reductive state may also be regulated by the acylation of enzymes. Benzoate degradation generates nicotinamide adenine dinucleotide reduced and flavin adenine dinucleotide reduced, but excess reducing capacity stresses cells. A buildup of intermediates that subsequently modifies and slows the rates of enzymatic catalysis would safeguard against damaging reductive stress ( 15 , 82 , 83 ). This link is further enforced by the presence of sirtuins, whose activity is non–energy-conserving, and relies upon the oxidized cofactor NAD + for activity ( 84 ), innately linking the extent of protein acylation to cellular redox state. Interestingly, the sirtuin assayed in this system appears to have a bias for longer chain acylations, particularly glutarylation, a trait that is shared with Sirt4 and Sirt5 in mammalian systems ( 66 , 85 , 86 ). Although S. aciditrophicus sirtuins were identified by homology to CobB, their activities deviate notably from those of the E. coli enzyme, as illustrated by their action on only longer acyl groups, not acetyl. Previous studies indicated that sirtuin specificity across species including E. coli CobB, mammalian Sirt5, and Archaeoglobus fulgidus Sir2Af2 demonstrates that while also having deacetylase activity, their selectivity toward long-chain deacylation is dependent on a YxxR motif present in the binding pocket ( 87 , 88 ). However, in the S. aciditrophicus sirtuins, no such motif exists ( supplemental Fig. S4 ), suggesting a different substrate selection mechanism. The PTMs identified derive from acyl-CoA intermediates at steps which require the loss of 2 [H] (release of two reducing equivalents) ( 16 ). Hydrogen buildup is a limiting factor in the syntrophic metabolism, as is evidenced by the need for a hydrogen-scavenging partner ( 4 ). It is therefore expected that some metabolites would accumulate under high hydrogen concentrations, providing another link between the metabolic conditions of the cell and the observed acyl modifications. Interestingly, the lower benzoate pathway degrading 3- hydroxypimelyl-CoA ( Fig. 2 C ) is conserved across many anaerobes that degrade benzoate and other aromatic compounds ( 89 , 90 ). The acyl modifications we identified on these proteins suggest that the modifications are likely present in other bacteria sharing the same or similar metabolic pathways, especially under cellular conditions that induce an accumulation of intermediates. S. aciditrophicus has demonstrated a wide array of acyl modifications within its proteome. Further investigation into the function of these acylations presents an opportunity to understand how these essential environmental microbes regulate their metabolism. Other syntrophs also contain critical pathways with RACS intermediates ( 91 , 92 , 93 ), and a wider investigation of other syntrophs may further reveal the role RACS play in bacterial metabolism."
} | 2,375 |
34504098 | PMC8429590 | pmc | 7,691 | {
"abstract": "Wet and dry foams are prevalent in many industries, ranging from the food processing and commercial cosmetic sectors to industries such as chemical and oil-refining. Uncontrolled foaming results in product losses, equipment downtime or damage and cleanup costs. To speed up defoaming or enable anti-foaming, liquid oil or hydrophobic particles are usually added. However, such additives may need to be later separated and removed for environmental reasons and product quality. Here, we show that passive defoaming or active anti-foaming is possible simply by the interaction of foam with chemically or morphologically modified surfaces, of which the superamphiphobic variant exhibits superior performance. They significantly improve retraction of highly stable wet foams and prevention of growing dry foams, as quantified for beer and aqueous soap solution as model systems. Microscopic imaging reveals that amphiphobic nano-protrusions directly destabilize contacting foam bubbles, which can favorably vent through air gaps warranted by a Cassie wetting state. This mode of interfacial destabilization offers untapped potential for developing efficient, low-power and sustainable foam and froth management.",
"introduction": "Introduction Defoaming is the process of destabilizing existing foam while antifoaming aims to prevent the formation of foam 1 – 4 . This is achieved respectively, by depositing chemicals onto bulk foams or within the target liquid. These chemicals are also known as defoaming or antifoaming agents and include oils 4 , hydrocarbons or waxes 4 , 5 , microparticles 2 , 3 , 6 , 7 , or mixtures 1 , 8 of these. The agents enhance the coalescence of foam bubbles by speeding up the disintegration of bubble films 3 , 9 . The effects of size, shape, and degrees of hydrophobicity of microparticles on defoaming have also been intensively studied to optimize defoaming 2 , 3 . However, hydrophobic particles quickly become inactivated by surface-active surfactants, which are always present in foams. To circumvent this, oils are used as carrier fluids to deliver hydrophobic particles directly to the thin film separating neighboring bubbles 5 , 6 . Alternatively, oils can also be used independently for defoaming 2 , 3 , 6 . Although efficient, oils and/or particles can be environmentally harmful while also altering the properties of the final product. Therefore, they might need to be removed afterward, requiring subsequent energy-intensive separation processes 1 , 5 , 6 . These standing issues make alternative approaches highly desirable. Surprisingly, the consequences of liquid-repellent coatings on defoaming and antifoaming have not been explored. Here, we demonstrate that liquid-repellent coatings show excellent antifoaming properties. Their antifoaming potential has likely been underestimated because the typical contact area between foam and surface is small. This can be rectified by the use of liquid-repellent coatings on three-dimensional surfaces, thus enabling bulk interaction and defoaming. In this work, we first show that superamphiphobic 10 and liquid-infused 11 surfaces, so termed liquid-repellent surfaces, can speed up defoaming. Liquid-infused surfaces are composed of a rough surface, which is infused with a lubricant. Although efficient in the short term, they suffer from depletion of lubricant after repeated use. Superamphiphobic surfaces consist of an amphiphobic, hierarchically rough surface which traps a layer of air 12 – 20 . We use the term air in the airgaps to distinguish it from gas in foam bubbles. Superamphiphobic surfaces show remarkable defoaming and antifoaming properties because surface protrusions destabilize and rupture contacting foam bubbles. Gases released from burst bubbles escape through these surfaces’ airgaps. This technique does not cause leaching and damage of the surfaces: thus preventing contamination of the target liquid’s composition. Superamphiphobic surfaces remain functionally stable (Supplementary Movie 1 ) over multiple cycles while demonstrating enhanced performance in both defoaming (50%) and antifoaming (larger than 100%) compared to controls.",
"discussion": "Results and discussion Bulk defoaming via liquid-repellent surfaces Cylindrical glasses (internal diameter: 5.5 cm, height: 12 cm) were used as test surfaces for defoaming. Unfunctionalized soda-lime glass represents the control. The inner surfaces of glasses were modified by coating the walls with a superamphiphobic or a liquid-infused layer. The superamphiphobic surface (SA) was synthesized by depositing surface-functionalized nanoparticles 10 onto a polystyrene binder (“Methods”). Superamphiphobicity was verified by low sliding angles (3° ± 1°) for hexadecane ( γ = 27.5 mN m −1 ) drops (Supplementary Fig. 1 ). The slippery liquid-infused porous surface (SLIPS) variant comprises of a 2-µm-thick nanoparticle layer (Glaco) infused with various oils (“Methods”) 21 . SLIPS are also termed liquid-infused surfaces. Liquid-like surfaces were synthesized from PDMS brushes, liquid-like PDMS (LL-PDMS) 22 (“Methods”). Photos of the foaming dynamics by beer foam in these glasses are represented in Fig. 1a . The wettability of a drop of beer on these surfaces greatly differs. The beer wets the glass control and contacts the other liquid-repellent surfaces, whereas a drop shows a high contact angle and easily rolls off a superamphiphobic coating (Fig. 1b and Supplementary Fig. 1 ). Fig. 1 Evaluation of super liquid-repellent surfaces for defoaming. a Unmodified glass (control), superamphiphobic surface (SA), slippery liquid-infused porous surfaces (SLIPS), and liquid-like PDMS (LL-PDMS) 5 min after deposition. All surfaces were coated inside glass cups. As the wet liquid foam, commercially procured beer (Bitburger Pilsner) was used. b Schematic and optical image of a drop of beer (schematized as yellow) on test surfaces. The static, advancing, and receding contact angles were: (1) Control: 58°, 71°, 15°; (2) superamphiphobic: 164°, 180° 45 , 162°; (3) SLIPS: apparent contact angle of 74°, (4) liquid-like surface: 95°, 100°, 15°. Defoaming of foam generated from 1.0 bar pressure-dispensed beer. Analysis of ( c ) foam and ( e ) beer volume with test surfaces. d A comparison of various liquid-infused surfaces. f Volume of foam for three dispensing pressures using control, superamphiphobic and hexadecane-infused SLIPS glasses after 5.5 min. HD refers to hexadecane, SO refers to silicone oil. All data are presented in mean ± standard deviations. Foamy beer was dispensed into the functionalized cups at dispensing pressures of 0.1–1.0 bar. We determined the initial height of the beer-foam mixture, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${H}_{{{{{{\\rm{foam}}}}}}}\\left(t=0\\right)$$\\end{document} H foam t = 0 , after filling the glasses. This height changes with time, as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${H}_{{{{{{\\rm{foam}}}}}}}\\left(t\\right)$$\\end{document} H foam t . After 10 min, the height of the liquid beer (without foam) remained constant within our experimental accuracy, defined as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${H}_{{{{{{\\rm{beer}}}}}}}(\\infty )$$\\end{document} H beer ( ∞ ) . The dynamic volume of foam, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{f}}}}}}}$$\\end{document} V f is defined as, 1 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{f}}}}}}}=\\frac{{H}_{{{{{{\\rm{foam}}}}}}}\\left(t\\right)-{H}_{{{{{{\\rm{beer}}}}}}}({{\\infty }})}{{H}_{{{{{{\\rm{foam}}}}}}}\\left(t=0\\right)-{H}_{{{{{{\\rm{beer}}}}}}}({{\\infty }})}\\times 100.$$\\end{document} V f = H foam t − H beer ( ∞ ) H foam t = 0 − H beer ( ∞ ) × 100 . Analogously, we defined the volume of liquid beer, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{b}}}}}}}$$\\end{document} V b as, 2 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{b}}}}}}}=\\frac{{H}_{{{{{{\\rm{beer}}}}}}}\\left(t\\right)}{{H}_{{{{{{\\rm{beer}}}}}}}({{\\infty }})}\\times 100.$$\\end{document} V b = H beer t H beer ( ∞ ) × 100 . The beer volume increases over time primarily because of the gravitational drainage of liquid from the foam 1 . To increase contrast, backlighting (also termed shadowgraphy) was used. This made the foam column appear black (Supplementary Fig. 2 and Supplementary Movies 2 – 4 ). The height of the beer and the foam on the glass walls (contact lines) were computationally tracked using automated image processing techniques (Supplementary Fig. 2 and Supplementary Movie 5 ). Defoaming occurred fastest with SLIPS infused with hexadecane (Fig. 1c , SLIPS-HD, red hexagons). This was followed by SA (Fig. 1c , purple diamonds), LL-PDMS (Fig. 1c , red spheres), and the control (Fig. 1c , black squares). To test the influence of viscosity and interfacial tension, liquid-infused surfaces with silicone oils (SO, 5 cSt and 500 cSt) and sunflower oil were compared to hexadecane, which shows the fastest defoaming rate (Fig. 1d ). Defoaming rates decreased with increasing viscosity. This is likely caused by the slower imbibition of oil into foams if the viscosity of oil increases. The presence of multicomponent impurities (sunflower oil) may also influence defoaming. For comparison, we investigated the use of LL-PDMS brushes. However, they perform significantly worse at lower dispensing pressure (Supplementary Fig. 3 ). The lower dispensing pressure resulted in a foam head with comparatively lower stability despite a similar initial liquid fraction. With the use of liquid-like PDMS brushes, remnants are always left on the glass walls (Fig. 1a ). The presence of persistent foam remnants disqualifies them from being the ideal surface variant for defoaming/antifoaming. Figure 1f shows the foam volume at 5.5 min after dispensing. Superamphiphobic and liquid-infused glasses enhance defoaming at various dispensing pressures, albeit at different timescales (Supplementary Fig. 3 ). The foam volume remaining at 5.5 min decreased with decreasing dispensing pressure for all surfaces. At 1.0 bar, foam volume decreased by 15% for the superamphiphobic surface, relative to the control glass. For the liquid-infused glass, this difference was 25%. The gains in beer volume in the SLIPS-HD system (Fig. 1e , red hexagons) are marginally higher, complementing the fastest defoaming. Bubble size distribution and coalescence/rupture events To gain insight into the defoaming mechanisms, we determined the number and radii of bubbles in close contact with surfaces (Fig. 2 and Supplementary Fig. 4 ). At a dispensing pressure of 1.0 bar, the whole glass was filled with a foam having a liquid fraction, ϕ ≈ 0.45 at \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$t=0$$\\end{document} t = 0 . We imaged the formation and evolution of bubbles 2.5 cm below the top of the original foam head (investigated area ≈3.5 mm 2 ). For detailed image analysis, we excluded the first 12 s because the small bubble radii and fast-dispensing motion blurred the images and led to poor automatization of image analysis (Videos M6 and M7). During drainage, liquid fraction in the foam can be verified by observing bubble geometries. Bubbles remain spherical and we did not observe jamming, hinting that the influence of osmotic pressure 23 is small over the timespan of observations (see Supplementary Fig. 4 and Supplementary Movie 6 ). Fig. 2 Bubble dimensions with respect to time. Imaging at 2.5 cm below the maximum original foam height. Bubbles were analyzed over a surface area of 3.5 mm 2 , at the bubble-to-surface interface. The foam line recedes from view beyond 96 s for the fastest defoaming superamphiphobic surface. Control refers to unfunctionalized glass, SLIPS-HD refers to a slippery liquid-infused porous surface infused with hexadecane, and SA refers to superamphiphobic glass. a Average bubble radii with respect to time, showing a gradual rise in the control (inset, gray squares). Bubbles remained smallest for the superamphiphobic system. b Coalescence events observed in SLIPS. c Bursting events caused by the dissipation of a bubble during bubble-to-surface interaction were observed for the superamphiphobic surface. Observation of the first bursting event is dependent on the region of observation. All bubbles were computationally tracked (see “Methods”). d Analysis of the percentage count of bubble radii and areas over the entire 96 s duration (minus the first 12 s) revealed persistently small bubbles in the control and superamphiphobic surfaces. The largest bubble radii in each system are represented with a dashed line. In all systems, the average bubble radius, R , increased over time (Fig. 2a ). In the control, bubble radii increased almost linearly with an average rate of approximately 0.3 µm s −1 (Fig. 2a , inset). Bubble growth is likely caused by the diffusion of CO 2 from the beer into the bubbles 24 because no coalescence events were detected in the control despite growth (Fig. 2b ). Here, coalescence is defined as the merging of two originally separate bubbles through inter-bubble film rupture. For the superamphiphobic surface, bubble radii reached a maximum ( ca . 37 ± 3 µm) at around 40 s and remained constant thereafter. Bubbles that are in contact with the superamphiphobic surface burst immediately and disappeared (Fig. 2c , purple diamonds and Supplementary Movie 6 ). Thus, superamphiphobic surfaces destabilize bubbles. Considering the total number of bursting events over time, a time delay of ca . 40 s was noted before a linear increase in bursting events occurs (Fig. 2c ). During the first 40 s, we observed excess liquid flowing along the glass wall driven by gravitational drainage (Supplementary Movie 6 at 40 s) 25 . This liquid prevented bubbles from contacting the wall. For SLIPS infused with low-viscosity hexadecane oil, a strongly fluctuating increase in bubble radii was observed. This is caused by the coexistence of a few large bubbles with a larger number of smaller bubbles after a few tens of seconds of contact (Supplementary Fig. 4 ). Bubbles interacting with a hexadecane-infused surface reached radii of up to 1 mm, as compared to the ca . 200 and 100 μm with the control and superamphiphobic surfaces, respectively (Fig. 2d ). This is caused by the series of coalescence events that resulted in spontaneously large bubble sizes (Fig. 2a ). These large singular bubbles can dominate the entire field of view, causing the peak in the evolution of bubble radii (Fig. 2a ). However, they are quickly driven out of the field of view by buoyant forces, and replaced by relatively smaller bubbles (Fig. 2a–d ). This is a universal observation for all oil-infused surfaces assessed, i.e., independent of viscosity and surface tension (Supplementary Movie 6 , SLIPS variants). To understand the defoaming mechanism, we should consider the mass transfer of CO 2 in the wet liquid foam. CO 2 diffuses from the liquid beer into the bubbles. Bubbles grow in size and their volumes increase. This occurs through two potential routes: (1) nucleation of smaller bubbles that diffuse into larger bubbles through Ostwald ripening. (2) Direct diffusion of gases from the liquid phase into the bubbles. The increasing volume presses bubbles in the foam head against the glass wall. Bubbles in close contact with the test surfaces can deform 26 . Defoaming mechanisms For superamphiphobic surfaces, nanoprotrusions which are composed of nanoparticles ( ca . 30 ± 10 nm diameter) stabilize the airgap (Fig. 3a, b ). The bubble’s average radius ( ca . 37 ± 3 µm) is much larger than the nanoprotrusions. When the bubble bursts, gases from the ruptured bubbles flow into the airgaps (Fig. 3c, d ). The coating’s air channel transports and releases the gas into the ambient environment. This continuous replenishment of gas also helps to preserve the stability of the airgap/channel in the superamphiphobic layer. Fig. 3 Foam decomposition by merging and bursting events. a Superamphiphobic coating (low-magnification side-profile) showing micro-agglomerates. b High-magnification image (top-profile) showing distinct nanoparticles (termed nanoprotrusions). c Sketch of a superamphiphobic surface. The bordered red particles depict functionalized agglomerates (fluorinated). d Macro-imaging of bubbles at the interface highlights a bursting event. Scale bar: 200 µm. A larger bursting bubble was chosen for image clarity. Bursting typically occurs in the range of bubble radii measured in Fig. 2 . e Sketch of a liquid-infused surface. The green particles depict hydrophobized agglomerates (Glaco coating). The coating is infused with oil (light pink). f The foam bubble films (yellow) flatten close to the oil-infiltrated coating. Scale bar: 200 µm. Macro-imaging at the interface reveals a coalescence event of two bubbles (silicone-oil assisted). Schematics are not to scale. In contrast, when a bubble makes contact with a liquid-infused surface, a wetting ridge forms (Fig. 3e , Supplementary Fig. 5 and Supplementary Discussion 2 , Curvatures and Pressures in the Wetting Ridge ) . As soon as the wetting ridges of neighboring bubbles overlap, bubbles experience a long-range attractive force (Supplementary Fig. 5 and Supplementary Movie 6 , SLIPS) 27 . The attractive force depends on the size of both bubbles (Supplementary Fig. 5 ), the height of the wetting ridge, the interfacial tension, and the viscosity of the oil 28 . The bubbles merge upon contact with one another, increasing buoyancy (Fig. 3e, f ). Coalesced bubbles move to the top of the foam column, thereafter destabilize and release trapped gases. When observed on a macroscopic scale, rapid decomposition of the foam column follows. After destabilization of the foam column, oil remains at the air–liquid interface. This reveals the unavoidable depletion of oil from within the slippery liquid-infused porous surface, hence limiting long-term performance (Supplementary Fig. 6 ). Thus, superamphiphobic surfaces are more promising in the long run for extended operations and are hereafter further investigated. Defoaming in bulk foams stems from a combination of (1) gravitational drainage of liquid, (2) Ostwald ripening, (3) interfacial force-driven film thinning, and (4) spontaneous film rupture. Gravitational drainage typically requires several minutes before completion 25 . Previous simulation and experimental results reveal that the gravitational drainage of bulk wet foams is much slower compared to bubble film thinning from interfacial forces 29 – 31 . The dynamics and stability of bubbles in the close vicinity of the superamphiphobic layer depends mainly on interfacial forces between the solid–liquid interface at the top of protrusions and the liquid–gas interface of the bubble. Both may be coated by adsorbed layers of proteins or surfactants 31 , 32 . To gain insight into the timescales of bursting, we let single bubbles rise (buoyancy-driven), contacting a superamphiphobic surface. The bursting process was imaged with a high-speed camera (Fig. 4a ) (Fastcam AX10, Photron, Japan). For deionized water (Fig. 4a , blue circles) and ethanol–water mixtures (Fig. 4a , green circles), bubbles burst within 1–10 ms after contacting the surface. In beer (Fig. 4a , orange circles), bubbles experienced a spread of rupture times of over three orders of magnitude (1 ms to 1 s) after making contact with the superamphiphobic surface. This delay in rupture timing suggests increased repulsive interactions in a protein-laden liquid as compared to water and ethanol–water mixtures. Fig. 4 Mechanism of defoaming by bubble bursting. a Temporal analysis of bubble rupture using high-speed cameras. The approach velocities (bubble rise) varied between 1 and 25 cm s −1 using different release distances from the superamphiphobic surface. Liquids: water, ethanol–water mixture, and beer. Scale bar: 500 µm. b Spatial analysis of bubble rupture using holographic microscopy. The film height, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$h\\left(t\\right)=\\delta h\\left(t\\right)+{h}_{c}$$\\end{document} h t = δ h t + h c , of a captive bubble during a slow controlled approach (5 µm s −1 ). The thickness of spontaneous rupture, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${h}_{c}$$\\end{document} h c is within the order of 100 nm for beer (orange circles) due to hydrophobic interactions. Inset: Phase maps monitoring the variation of film height. Height variation between dark and white lines is ≈ 300 nm. Scale bar: 100 µm. Schematics: c Approach of a bubble encountering a superamphiphobic surface. d Bulk drainage of liquid between the bubble and the nanoprotrusions. e Film thinning by interfacial and hydrodynamic forces. \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${h}_{c}$$\\end{document} h c , is the critical height of spontaneous rupture, taking into account the surface-penetrated depth determined using interference microscopy ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle h$$\\end{document} △ h , Supplementary Fig. 7 ). f Momentary film stabilization due to repulsive interactions. g Hydrophobic interactions induce spontaneous rupture of a sufficiently thin film. Schematics are not to scale. To monitor film thinning with an improved spatial resolution (Fig. 4b ), single bubbles encountering a superamphiphobic surface were analyzed using transmission holographic microscopy (T-DHM, T-1000, Lyncee-tec, Switzerland). In this case, we moved immobilized bubbles using a micromanipulator. Fixed bubbles approach the surface at 5 µm s −1 in both water (Fig. 4b , blue circles) and beer (Fig. 4b , orange circles). The shape of the bubble and the changing film height (or thickness), δh , were analyzed down to the point of film rupture (rupture height, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{h}}}}}}}_{{{{{{\\rm{c}}}}}}}$$\\end{document} h c ). In water, bubbles approach the surface without experiencing a slowdown (Fig. 4b , blue circles). Bubbles rupture within the time resolution of a single frame (5 ms). In beer, the bubble deforms when approaching the superamphiphobic layer (Fig. 4b , inset: 0–805 ms). Deformation (deviation from spherical cap) reaches a maximum of ≈ 6 µm (Fig. 4b , orange circles, schematized Fig. 4c, d ). The thin film was quasi-stable (Fig. 4b , orange circles, inset: 805–1305 ms) on the superamphiphobic surface for ca . 0.5 s before rupturing. Flattening of the interface is reflected in an increasing distance between interference fringes and the formation of an almost smooth area in the contact zone (Fig. 4b , inset: 805–1305 ms). The height varies by less than 200 nm. No dimple formation was noted. However, the intrinsic roughness of the superamphiphobic layer resulted in local changes in film thickness. Just before rupture, an agglomerate can be discerned. As observed in scanning electron micrographs in Fig. 3a, b , agglomerates are composed of nano- and micro-structured silica nanoparticles. This particular agglomerate is found, per the holographic image (Fig. 4b , inset: 1305 ms) in the bottom left quadrant as a black-colored distortion, indicative of a sharp change in profile. The scale of this distortion is ca . 50 μm in lateral dimensions. Rupture is likely initiated by the interaction of the thin film with the hierarchical profile. Our current phenomenological observations allow for several possibilities on how exactly the rupturing may proceed. The corresponding timescale appears to be statistical, largely because information on flow boundary conditions and liquid rheology remains stochastic and incomplete. To start, beer contains ~4.5 wt% proteins which very likely populate the liquid–air interfaces 33 . The adsorption of a protein layer at the interface controls what happens during thin-film drainage. Two extremities bracket the situation: 34 First, (1) no proteins assemble at the interface, mirroring the unique situation of pure water or water/ethanol mixtures 31 , 34 . Alternatively, (2) proteins pack densely at the interface, resulting in an effectively solid shell 35 . In reality, the assembly is stochastic in nature, and falls in between both extremes, creating an intermediate situation 35 , 36 . This is supported by the observations that while the film does not rupture immediately upon surface contact (per Case 1 as in pure liquids), it is also not indefinitely stable (per Case 2 as in stabilized foams). Continued film thinning is likely to facilitate improved packing of the protein layer 35 , moving the dynamically changing interface toward Case 2. When the thinning film is viewed from a continuum perspective after suitable coarse graining, the assignment of effective flow boundary conditions becomes rather difficult, as a rapidly changing rheology must be factored into the analysis, which is beyond the scope of this study. In Supplementary Discussion 4 , Thin Film Behavior described by Stokes–Reynolds equation, we work out one scenario where no-slip boundary conditions are assumed to facilitate an analysis utilizing the Stokes–Reynolds equation. In this way, the magnitude of the timescale for film rupture (within the order of 100 ms) can be rationalized. Hydrodynamic effects are negligible because of the slow approach and the low radius of curvature of the protrusions. The drainage of a bubble’s thin film near a superamphiphobic surface results from the pressure difference ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varDelta P$$\\end{document} Δ P ) between the film at the protrusion (Fig. 4e , point A, film height \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$h(A)$$\\end{document} h ( A ) ) and the film away from the protrusion. \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varDelta P\\approx$$\\end{document} Δ P ≈ \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varPi \\left(h\\right)$$\\end{document} Π h , where \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varPi (h)$$\\end{document} Π ( h ) is the disjoining pressure due to interfacial forces. Away from the protrusion, we assume that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varPi \\left(h\\right)$$\\end{document} Π h (Fig. 4e , point B, film height \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${h}_{c}$$\\end{document} h c + \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${dh}$$\\end{document} d h ) is negligible since the film thickness is larger (excess depth, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varDelta h$$\\end{document} Δ h = 350 ± 220 nm, Supplementary Fig. 7 ) than the effective range of interfacial forces 37 . \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varPi \\left(h\\right)$$\\end{document} Π h has contributions from van der Waals ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\varPi }_{{{{{{\\rm{vdW}}}}}}}$$\\end{document} Π vdW ) forces, electrical double layer ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\varPi }_{{{{{{\\rm{EDL}}}}}}}$$\\end{document} Π EDL ), and steric \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$({\\varPi }_{{st}})$$\\end{document} ( Π s t ) forces due to adsorbed proteins. Disjoining pressure from van der Waals interactions \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$({\\prod }_{{{{{{\\rm{vdW}}}}}}}={A}_{{{{{{\\rm{H}}}}}}}/6\\pi {h}^{3})$$\\end{document} ( ∏ vdW = A H / 6 π h 3 ) is the distance-dependent ( h ) parameter governing interactions between two phases (separating a medium) that can be repulsive (pulling liquid) or attractive (expelling liquid) 37 . The Hamaker constant ( A H ) may be approximated as fluoro–water–air, A H = +1.6 × 10 −20 J (attractive) 37 . In this instance, liquid films likely experience repulsive electrical double layer and attractive van der Waals forces 38 , 39 once they have drained to thicknesses within the order of 10 and 100 nm 37 , respectively (Fig. 4c–e ). However, attempts to model our experiments (Fig. 4b ) without \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\varPi }_{{st}}$$\\end{document} Π s t failed (see Supplementary Discussion 4 – 6 , Thin Film Behavior described by Stokes–Reynolds Equation). During film drainage of liquids such as beer, proteins likely adsorb at the interfaces (liquid–air, and liquid–fluoro). As we know that rupture eventually occurs, the adsorption is likely metastable 35 . During this quasi-stable equilibrium (Fig. 4b , orange circles, 900–1300 ms), adsorbed proteins may spontaneously move within the thin film. During this phase, steric repulsion by momentarily adsorbed proteins stabilizes the thin film (Fig. 4f ). The film thickness hardly changes but rupture is delayed (Fig. 4b , orange circles). During this time, spontaneous motion of the adsorbed proteins 35 along and between the interfaces (liquid–air and liquid–fluoro) allows for stochastic time windows through which hydrophobic forces 40 , 41 act (as they would immediately if the liquid was pure and clean, per Fig. 4b , blue circles). The mobility of any adsorbed proteins should be present to some extent, without which the thin film will experience indefinite stability. Once the hydrophobic forces momentarily act across the interface, the film ruptures and a three-phase contact is formed (Fig. 4f, g ). Thereafter, the gases escape. The extent of remnant proteins adsorbed onto fluoro-functionalized nanoparticles is likely to be minimal, per previous anti-fouling studies using bovine serum albumin (BSA) 42 . Moreover, they may also be easily washed away during future interactions (with water or with more foam liquids). Functionally, the presence of airgaps/-channels facilitates the easy escape of a film-destabilized bubble, enabling cycling of the process. As soon as a bubble has burst, a neighboring bubble moves to the vacated area (Supplementary Movie 6 ). Bursting of bubbles at the interface results in fast defoaming. However, in the center of the glass, defoaming is still dominated by slower buoyancy-induced processes. The differences in surface and volumetric defoaming rates reshape the foam column. The foam column forms a truncated cone, receded from the superamphiphobic coating (Fig. 5a , shadowgraphs). After the foam column has delaminated from the sidewalls at ca . 6.5 min, the slope of the foam column buckles, forming a kink (Fig. 5b ) with two distinctive slopes (angles). This occurs due to the finite elasticity of the cone’s solid-like structure, resulting in it collapsing (Fig. 5b , inset) on its own weight. A receded foam-to-surface contact line takes place alongside the formation of the cone-shaped foam column. This results in a shrunken volumetric bulk that does not contact the functional superamphiphobic surface. As observed in Fig. 5a, b , at beyond 6 min, defoaming of the stable wet foam is still dominated by the volumetric bulk, thus questioning the relevance of bubble bursting induced only by the surrounding surface. Fig. 5 Importance of foam-to-surface contact for defoaming. a The foam-to-surface contact line recedes from the superamphiphobic (SA) surface due to rapid destabilization. An angular cone-like structure first evolves, bearing a single angle. a , b The single angle degrades into two distinctive angles during continued defoaming of the volumetric bulk. Inset—top view, scale bar: 2 cm. c A three-dimensional volumetric defoamer was synthesized using a high surface area template, i.e., a brush coated with the superamphiphobic coating. This was compared to the original superamphiphobic and control systems. c , d The superamphiphobic volumetric defoamer experienced up to 40–50% defoaming enhancement vs . the control. Dispensing pressure: 1.0 bar using contrasted backlighting. All data are presented in mean ± standard deviations. To circumvent the reduction of defoaming rates caused by a shrinking volumetric bulk, we introduce a three-dimensional volumetric defoamer. This consists of the use of the superamphiphobic coating on a laboratory brush. The bristles increase the net effective surface area within the bulk. The entire brush was rendered superamphiphobic by spray coating under identical conditions used for the cups. The coating deposited on brush bristles gives rise to a three-dimensional interconnected air layer (Fig. 5c and Supplementary Movie 7 ). The model mechanical defoamer maintained high foam-to-surface contact during the entire defoaming process, thereby increasing the overall rate of defoaming by up to 40–50% (Fig. 5c, d , half-filled purple diamonds). More interestingly, the presence of the bristles at the foam-to-liquid contact line enabled complete defoaming down to the liquid level. A negligible decrease in defoaming rate was observed throughout the process. The superamphiphobic layer induced bubble bursting while the interconnected air layers ensured rapid removal of escaping CO 2 . Antifoaming properties To investigate the use of superamphiphobic surfaces for antifoaming (suppression of foam formation), bubbling of soapy water (0.25 bars through a tube) was performed within the superamphiphobic cups. Two soaps were tested, the nonionic surfactant pentaethylene glycol monododecyl ether (C12E5) (Fig. 6a, b and Supplementary Movie 8 ) and commercial dishwashing soaps (Supplementary Fig. 8 and Supplementary Movie 9 ), which are a mixture of cationic, anionic, and nonionic surfactants of various molecular weights. Soap foams are composed of dry foam cells (Fig. 6c ), unlike the stable liquid foam bubbles observed in beer, Fig. 3d . Soap foams can also be continuously aerated to simulate uncontrollable foam formation and growth. Fig. 6 Demonstration of antifoaming and froth control. Antifoaming was quantified using a bubbling setup that delivers ca . 5 mL s −1 of nitrogen at 0.25 bars through a U-tube at an inner orifice diameter of 0.5 mm. The soap demonstrated here was ( a , b ) pentaethylene glycol monododecyl ether (C12E5, critical micelle concentration of 0.03 g L −1 ). The soap was mixed into water at a concentration of 0.4 g L −1 . c Photo of foam bubble close to the surface. The Plateau borders refer to the channels where films meet. The nodes refer to places where four channels meet. The formation of ( d ) dry foams in the superamphiphobic glasses (SA) was suppressed (<10% of total available volume) compared to the control glass (glass). Beyond 30 s, foam in the control system spills over, and further observation was halted. All data are presented in mean ± standard deviations. Superamphiphobic surfaces suppressed the formation of foams above the liquid line. In uncoated glasses, foams grew with an almost linear behavior, resulting in spillover after 30 s (Fig. 6a, b, d ). Compared to controls (glass), the superamphiphobic surfaces suppressed foaming up to a measured level of larger than 100%. In fact, superamphiphobicity appears to be capable of limiting the maximum height and volume of a spontaneously foaming column by interfacial destabilization. Dry foams burst almost immediately (at ≪1 s after surface interaction) upon contact with the superamphiphobic surface. This is expected since no liquid drainage (as in wet foams) is necessary before the formation of unstable films (Fig. 6d ). This holds for all pressures investigated, i.e., up to 1.0 bar. In summary, superamphiphobic surfaces are capable of actively defoaming in situ generated wet liquid foams as well as inhibiting dry foam formation. This additive-free and energy-efficient method for defoaming and antifoaming processes can be particularly important for the food and chemical industries. Advantages of superamphiphobic surfaces for defoaming, antifoaming, and froth control are: (a) superamphiphobic surfaces are not affected by repeated use with foaming liquids. (b) They do not require or result in the leakage/release of material into the two-phase foam. (c) They are easily scalable with current methods."
} | 10,547 |
37156822 | PMC10167324 | pmc | 7,692 | {
"abstract": "Contrary to a photographer, who puts a great effort in keeping the lens still, eyes insistently move even during fixation. This benefits signal decorrelation, which underlies an efficient encoding of visual information. Yet, camera motion is not sufficient alone; it must be coupled with a sensor specifically selective to temporal changes. Indeed, motion induced on standard imagers only results in burring effects. Neuromorphic sensors represent a valuable solution. Here we characterize the response of an event-based camera equipped with fixational eye movements (FEMs) on both synthetic and natural images. Our analyses prove that the system starts an early stage of redundancy suppression, as a precursor of subsequent whitening processes on the amplitude spectrum. This does not come at the price of corrupting structural information contained in local spatial phase across oriented axes. Isotropy of FEMs ensures proper representations of image features without introducing biases towards specific contrast orientations.",
"introduction": "Introduction Human vision rapidly adapts to unchanging retinal input up to experiencing a real perceptual fading when retinal image motion is artificially compensated or eliminated. Also for this reason, vision is an active process and this need leads our eyes to constantly move for keeping motionless parts of the visual scene visible. A particular class of involuntary eye movements, known as fixational eye movements (FEMs), serves this purpose 1 . In the attempt of understanding how brain exploits eye’s jitter, neuromorphic engineering and event-based vision sensors 2 , 3 provide a natural “learning-by-doing” framework to investigate the early stages of visual processing in active (i.e., real world) conditions 4 . These cameras convert a visual scene into a stream of asynchronous ON and OFF events based on positive or negative temporal contrast; as opposed to frame-based and clock-driven acquisitions of luminance. These continuous-time sensors functionally emulate the key features of the human retina and represent a major shift from conventional cameras, by transmitting only pixel-level changes at microsecond precision. It is therefore not surprising that, as for neurons in the retina, no visual information can be gained in the absence of relative motion between the sensor and the environment. An active vision mechanism based on FEMs can be implemented on a bio-inspired robotic system for making visual perception of static objects feasible by event-based sensors. Besides being a means for refreshing neural activity and preventing perceptual fading (retinal adaptation), FEMs have been pinpointed to play a key role in terms of efficiency coding 5 . The efficiency principle 6 states that one of the goals of early vision processing is to maximize the information that is encoded about relevant sensory variables, given constraints on the available (neural) resources (e.g., the limited capacity of the optic nerve), by reducing uninformative correlations typical of natural scenes. Before advancing this hypothesis, the spatio-temporal behavior of retinal bipolar and ganglion cells (RGCs) has long been considered as the only responsible for this signal decorrelation. At first approximation, RGCs act as linear spatio-temporal filters that implement lateral and temporal inhibition to generate receptive fields with antagonist center-surround spatial organization and transient (i.e., biphasic) response in time. In this way, they seek to reduce redundancy between parallel channels in space, and within each single channel along time. In addition to that, contributions of non-linear stimulus-response relationships 7 (such as synaptic rectification, depression, gain control, spiking threshold and refractory) refine the job, eventually permitting retinal neurons to transmit information with nearly optimal efficiency. However, this view lacks to consider the observer’s motor activity 8 , relying on the simplifying assumption that the input to the retina is a stationary image, or—at the most - a sequence of stationary frames. In living animals, the retina receives unstable visual inputs caused by movements of body, head and eyes. Even when an animal is fixating an object, the whole image on the retina is shifted by the presence of incessant microscopic albeit continuous and erratic eye movements. In Segal et al. 8 , authors proved that the response of RGCs alone still exhibits strong and extensive spatial correlations in the absence of fixational eye movements (e.g., with stimulus flash). In the presence of FEMs, instead, the levels of correlation in the neural responses dropped significantly, resulting in effective decorrelation of the channels streaming information to the brain. These observations demonstrate that microscopic eye movements act to reduce correlations in retinal responses and contribute to visual information processing. Similar conclusions have also been drawn in 9 . They demonstrated that the statistics of FEMs matches the statistics of natural images, such that their interaction generates spatiotemporal inputs optimized for processing by RGCs. This spatiotemporal reformatting is crucial for neural coding, as it matches the range of peak spatiotemporal sensitivity of retinal neurons in primates. As a consequence, jittery movements of a sensor can emphasize edges, as postulated and formalized in the fascinating theory of the Resonant Retina 10 , and very recently further examined in 11 . In the present research, we investigate the effects of fixational eye movements on neuromorphic sensors. Given the low occurrence and supposed minor significance of micro-saccades 9 as compared to the major effects due to slow fixational drifts (which are the two main components of FEMs), we target non-saccadic FEMs only. We extend an existing model of biological fixational movements 12 to increase their isotropy. We use this model to move a neuromorphic camera in a biological fashion while acquiring data from synthetic and natural stimuli. The resulting event stream is analyzed for characterizing the role of FEMs. The main focus is on understanding how it preserves structural information of the input natural images while decorrelating their amplitude in order to reduce redundancy.",
"discussion": "Discussion and conclusions Despite the name, human fixation is a highly dynamic process. In biology, some roles of fixational eye movements have already been pointed out and discussed, persuading the scientific community that FEMs are far from being a nuisance, as originally believed. In this work, we investigate the role of FEMs in neuromorphic vision, i.e on the output signal of silicon circuits that emulate some primary functionalities of the human retina (namely, transient dynamics, no spatial filtering). We started our investigation by examining the overall spectral response of the system following a similar procedure as in 9 . While the power spectrum in natural scenes is highly concentrated at low spatial frequencies (with an amplitude falloff of 1/ k ), a neuromorphic and actively-fixating system intrinsically enhances higher spatial-frequency contents by amplifying its response to them. Therefore, such a system tends to oppose to the power-law falloff, counterbalancing the latter and enabling an equalized response to all discernible frequencies when the stimulus has such statistical properties. The investigation with natural image stimuli also proved that our neuromorphic system equipped with FEMs starts an early stage of redundancy suppression as a precursor of subsequent whitening processes 23 – 25 . Since no explicit spatial filtering is actually implemented from the DAVIS sensor, the origin of the observed whitening effect should be ascribed to the combination of three main characteristics of the sensing strategy: (1) the peculiar motion used 8 , (2) the transient response of the camera, and (3) some non-linear behavior 7 in the acquisition process of single pixels. However, when the neuromorphic camera recorded the same natural image flashing on the monitor, the resulting signal was still highly correlated despite intrinsic non-linearity of the sensor and trial-to-trial noise in the recordings. Therefore, much of the decorrelation in the FEM-based signal is ascribable to the combination of movements and sensitivity to brightness transitions, with a small contribution of additive noise and non-linear behaviors. In other words, the small image displacements induced by FEMs—given retina/sensor temporal DC removal—help discarding redundant spatial correlations of natural visual input, hence boosting the decorrelation induced by subsequent center-surround filtering of RGCs (not sufficient alone to disrupt the strong correlations of natural images 7 ). A mere weak correlation of the input signal does not yet imply a highly efficient coding system. For instance, if two signals are affected by independent noise, this decorrelates them without improving coding efficiency. In order to efficiently encode a visual scene, it is necessary that the decorrelation procedure does not compromise the preservation of its structure-related information. Despite being commonly related to coding efficiency, second-order statistical moments—such as the autocorrelation function or the power spectrum of an image—consider amplitude information only. They are by definition insensitive to local phase, which is essential to fully convey information about image structure. By analyzing local phase content, we proved that most structural information of the original natural scene is not lost after FEM-based whitening. By comparing results from FEM and flash-based acquisitions, we noticed that the active fixational strategy makes the event-based sensor to extract less redundant and still informative content. As a final beneficial consequence of whitening, reliability of phase information is expanded in a wider range of spatial frequencies 21 . We can hence conclude that fixational instability encourages redundancy minimization in neuromorphic vision by boosting the equalization of natural-images’ amplitude spectrum while preserving its phase spectrum and increasing its reliability at high spatial frequencies. In other words, FEMs contribute to an efficient encoding of the visual scene providing a pre-whitened signal to RGCs for further processing. Finally, we analyzed the effects of possible biases in the direction of FEMs. Isotropy in the motion strategy was reflected on the acquired signal, leading to an equalization of sensor response to all oriented edges in the image. These results could possibly suggest an additional role of FEMs in biological systems—beyond those already postulated in the literature—related to their erratic nature, for which no theory has been advanced so far. Specifically, this equalization strategy could underlie an unbiased representation of image features carried out by orientation-selective cortical neurons, which is believed as one of the most important functionality of early vision, supporting subsequent object recognition and scene understanding. For a long time, the idea of the eye operating as a standard camera (i.e. taking discrete spatial snapshots of the scene) has dominated visual neuroscience. However, alike other sensory modalities, vision is an active process in which the eye palpates external objects by means of motion. Movements transform spatial features into specific temporal modulations on the retina 26 , which consequently shape neural dynamic patterns in cortical regions. However, the results here presented concern on spatial information only, following a well-established framework for analyzing spatial coding efficiency (which mainly refers to the old “camera model” of the eye). To fully appreciate the functional role of FEMs, the organization of information in time should be addressed as well, since distribution of events in each sensor pixel is strongly structured by the motion sequence. Pure spatial information could be finely encoded in time as precisely synchronized activity of retinal neurons 27 , or phase-locked firing patterns across nearby cells 28 . Specifically, fine details of shapes, texture, and motion could be encoded by inter-cell temporal phases, instantaneous intra-burst rates, and inter-burst temporal frequencies of individual RGCs, respectively. Therefore, temporal dynamics as provided by FEMs could similarly benefit neuromorphic vision applications. Similar conclusions were also pinpointed by Akolkar and colleagues 29 : they observed that precise timings, produced by the combination of dynamic viewing and asynchronous sensing, carry important visual information that is useful for high-level computation (e.g. for pattern recognition). Hence, by productively spreading visual information in time, FEMs could ultimately aid subsequent brain-inspired spike-based processing stages—able to learn complex temporal codes—to effectively extract rich informative content."
} | 3,256 |
31370164 | PMC6722658 | pmc | 7,693 | {
"abstract": "Controllably tuned infrared emissivity has attracted great interest for potential application in adaptive thermal camouflage. In this work, we report a flexible multilayer graphene based infrared device on a porous polyethylene membrane, where the infrared emissivity could be tuned by ionic liquid intercalation. The Fermi level of surface multilayer graphene shifts to a high energy level through ionic liquid intercalation, which blocks electronic transition below the Fermi level. Thus, the optical absorptivity/emissivity of graphene could be controlled by intercalation. Experimentally, the infrared emissivity of surface graphene was found to be tuned from 0.57 to 0.41 after ionic liquid intercalation. Meanwhile, the relative reflectivity R v /R 0 of surface graphene increased from 1.0 to 1.15. The strong fluorescence background of Raman spectra, the upshift of the G peak (~23 cm −1 ), and the decrease of sheet resistance confirmed the successful intercalation of ionic liquid into the graphene layers. This intercalation control of the infrared emissivity of graphene in this work displays a new way of building an effective thermal camouflage system.",
"conclusion": "4. Conclusions In summary, we demonstrated a flexible multilayer graphene-based device with tunable infrared emissivity. Upon ionic liquid intercalation, the Fermi level shifts to a higher energy level, which blocks optical transition below the Fermi level. Thus, the reflectivity/absorption of graphene increased/reduced, which further reduces the infrared emissivity. Actually, the infrared emissivity of the surface multilayer graphene could be tuned between 0.57 and 0.41 through ionic liquid intercalation/deintercalation. The in-situ Raman and resistivity measurements present a reversible intercalation of anions/cations of the ionic liquid into graphene layers and heavy doping of the graphene layers. These results demonstrate a new strategy for a wide range of applications, such as thermal camouflage, infrared sources, and thermal management.",
"introduction": "1. Introduction All matter at a temperature above absolute zero emit infrared waves, the intensity of which is dependent on the temperature (T) and infrared emissivity (ε) of the surface materials [ 1 ]. This is widely used nowadays for night vision [ 2 ], infrared source [ 3 , 4 ], and temperature measurement [ 5 ]. Since thermal cycling is usually quite slow [ 6 ], mere temperature control could not satisfy the need for the further development of the infrared source [ 3 , 4 ], thermal management [ 7 ], and thermal camouflage [ 6 , 7 , 8 ]. Thus, it would be of great significance to controllably tune infrared emissivity (ε) no matter whether for fundamental research or for real applications. Recently, it was found that infrared emissivity could be modulated in situ through the design of surface structure [ 7 ], controlling of carrier density [ 7 , 8 ] or phase transition by external stimuli [ 6 ]. These results indicate that proper tuning of the electronic structure of the surface materials is an effective way to modulate infrared emission. However, these materials are usually on a rigid substrate and the tuning range is quite limited [ 7 ]. A flexible material with a tunable infrared emissivity is highly desirable. Due to linear band dispersion, graphene has a very broad band optical absorption/emissivity (Kirchhoff’s radiation law shows that the infrared absorption and emissivity are equal to each other at thermodynamic equilibrium) [ 3 , 9 , 10 , 11 , 12 ]. However, the interband electronic transition below the Fermi level is blocked in graphene due to the Pauli exclusion principle. Thus, doping, which shifts the Fermi level to a high energy level [ 7 , 9 , 10 , 13 , 14 ], yields tunable optical absorbance [ 9 , 10 ] or light emission [ 7 , 11 , 15 , 16 ] of graphene in a very broad spectral range from visible to far infrared frequencies. Unfortunately, the absorption/emissivity of monolayer graphene is limited to 2.3% by the fundamental fine structure constant [ 12 ], which is obviously not acceptable for real applications. Increasing the number of graphene layers increases the optical absorption/emissivity [ 17 , 18 , 19 ], which is an achievable way to increase the modulation depth of absorption/emissivity. However, electrostatic doping, the common way of doping in monolayer graphene, is not possible for tuning the interband transition for multilayer graphene, because of the shielding effect of the surface layers. Recently, intercalation has been demonstrated to be an effective way to dope multilayer two-dimensional (2D) materials [ 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ]. The intercalation process is reversible and compatible with the current semiconductor fabrication process [ 17 , 19 , 22 , 26 , 27 ], which makes it very promising to tune the infrared emissivity of multilayer graphene. In this work, by intercalating nonvolatile ionic liquid [DEME][TFSI] (98.5%, diethyl methyl (2-methoxyethyl)—ammonium bis (trifluoromethyl sulfonyl) imide, Sigma-Aldrich, Catalog No.727679, St. Louis, MO, USA) into multilayer graphene layers, we demonstrated a flexible thermal surface, which can electrically control the infrared emission without changing the surface temperature. Upon ionic liquid intercalation, the Fermi energy of graphene upshifts from the Dirac point to a high energy level, causing the emissivity of graphene to reduce from 0.57 to 0.41. The increase of the infrared relative reflectivity (R v /R 0 of graphene with the sample on copperplate is increased from 1.0 to 1.15) confirms the decrease of emissivity. The in-situ Raman spectra with an electronic voltage above 3 V have a strong fluorescence background and an upshift of the G peak of about 23 cm −1 , which indicates a strong doping effect of graphene layers with intercalation. In addition, the sheet resistance of graphene is reduced from 11 to 4 Ω/□ after intercalation due to the charge transfer between the ionic liquid and graphene layers. What is more, the intercalation process is fast and reversible; the film is light, flexible, and thin (<20 μm); and the four-layer architecture device architecture (see Figure 1 a) is compatible with modern roll-to-roll transfer processes. This makes the device in our work a promising material for designing tunable thermal camouflage systems.",
"discussion": "3. Results and Discussion For thermal imaging measurements, the multilayer graphene device was put on a polished copper plate ( Figure 2 a), which prevented the transmission of background infrared emission [ 8 , 17 , 22 ]. Due to the low infrared emissivity of polished copper substrate (~0.1) and the porous polyethylene membrane (<0.1), the thermal radiation was mainly from the surface multilayer graphene. To analyze the infrared behavior of the multilayer graphene surface, the device with the polished copper substrate was put on a hot plate. A thermocouple was used to monitor the temperature of the surface graphene. The temperature of the hot plate was adjusted to keep the surface graphene temperature at 35 °C. A Tix500 thermal camera (Fluke, Everett, WA, USA) was adopted to record the thermal images (emissivity was set as 1) at different intercalation bias voltages between 0 and 4 V. The voltages bias range is mainly limited by the electrochemical window of the ionic liquid at room temperature. Additionally, for bias voltages larger than 5 V, the surface graphene l turns to dark black and cannot be reversed. This is likely due to the oxidation of surface graphene, which usually becomes sensitive after heavily doping. Figure 2 b shows that the infrared temperature of the device is reduced from 30.5 °C at 0 V to 28.1 °C at 4 V, although the surface temperature has not changed at 35 °C. This implies the emissivity of the device is suppressed by ionic liquid intercalation. At thermodynamic equilibrium, the infrared emission energy is described by the Stefan−Boltzmann law, P = εσT 4 , where ε is the emissivity of the surface, σ is the Stefan−Boltzmann constant, and T is the temperature of the surface. Thus, the integrated emissivity ε of the surface could be determined by ε = ε I (T I /T) 4 , where ε I is the emissivity used for thermal imaging, T I is the infrared temperature, and T is the real temperature measured by thermocouple. Thus, the integrated emissivity for different intercalation bias voltages could be determined from the thermal images. The integrated emissivity at different bias voltages is summarized in Figure 2 c, and can be tuned from 0.57 to 0.41 through intercalation. What is necessary to mention is that the modulation of infrared emissivity is reversible and the switch time between different states is very fast with a response time less than 1s. Due to the ionic liquid intercalation, the Fermi level of graphene shifts to a higher energy level. The electronic transition below the Fermi level is suppressed due to Pauli blocking ( Figure 3 a), resulting in a decrease of the emissivity/absorption. Since the polished copper plate has a very high infrared reflectivity (~100%) and the porous polyethylene membrane is infrared transparent, the transmission of the multilayer graphene device on polished copper plate is 0. Thus, the emissivity of the surface multilayer graphene can be written as ε = α = 1 − R, where ε, α and R are the emissivity, absorption and reflectivity of surface multilayer graphene on polished copper plate. Figure 3 d presents the in-situ reflectance (R V /R 0 ) measurement of the multilayer graphene device on polished copper plate. The reflectance measurement shows that above 3 V there is an obvious increase of reflectance. This implies that the decrease of absorption/emissivity above 3 V, is consistent with Figure 2 c. In addition, we found that the reflectance below 500 nm remained almost unchanged with intercalation. This indicates that ionic liquid intercalation is more efficient for modulating the optical response at long wavelength range. Additionally, the multilayer graphene device was put on a xenon lamp ( Figure 3 c), which produced a round shape of white light illumination. With the voltage bias between two multilayer graphene layers increasing from 0 to 4 V, the light illumination became visible. In other words, the transmittance (absorption) of multilayer graphene is increased (decrease), due to the increase of Fermi energy through ionic liquid intercalation. However, the light illumination on the graphene device is red not white, which indicates that the optical modulation is more effective for long wavelengths, such as the infrared range, agreeing with the reflectance measurement ( Figure 3 d). The modulation of infrared emissivity is clearly due to the intercalation of the ionic liquid [DEME][TFSI] into the graphene layers. To further characterize the intercalation process of surface multilayer graphene, in-situ Raman measurement was carried out ( Figure 4 a). Figure 4 b presents the Raman spectra for surface graphene under different bias voltages. For a pristine multilayer graphene, there are three Raman modes: D (1321 cm −1 ), G (1580 cm −1 ), and 2D (2688 cm −1 ) modes, consistent with previous research [ 28 , 29 , 30 ]. The D Raman modes indicates the defect in graphene, which is probably induced by the etching and transfer process [ 22 ]. For an intercalation bias voltage below 2 V, the Raman spectrum remains similar to that of the pristine sample. However, when the applied voltage is above 3 V, the intensity of the G mode and D mode is increased significantly and the G mode is found to shift from 1580 to 1603 cm −1 with the bias voltage increase to 3 V. The increase of the G mode intensity is an indication of the doping effect through intercalation [ 17 , 22 ]. While, the upshift of 23 cm −1 of the G mode implies the successful intercalation [ 27 ]. The increase of D mode intensity shows the increase of defects in the graphene layers during the intercalation process, which enhances the intercalation process. For a bias voltage above 4 V, a strong fluorescence background appears with the diminishing of the 2D Raman modes. This further confirms the strong doping effect of intercalation. After removal of the applied voltage, the surface graphene shows a similar Raman spectrum to the pristine sample ( Figure 4 c). In other words, the intercalation process is reversible. The sheet resistance under the intercalated bias voltage of multilayer graphene was also measured by the four-point resistivity method ( Figure 5 a). The weak Van der Waals forces between the graphene layers allows atom or small molecules to intercalate into the Van der Waals gap [ 20 ]. In this case, the anions/cations of the ionic liquid intercalate into the layers under a voltage bias. As a result, the charge density on the graphene increases significantly and the sheet resistance of the multilayer graphene has a sharp drop from 11 Ω/□ below 2 V to 4 Ω/□ above 3.5 V ( Figure 5 b). This is consistent with the Raman measurements. These results suggest that there is a threshold bias voltage around 2 V. At a voltage below 2 V, the ions accumulate at the graphene—ionic liquid interface while above 2 V, the ions are intercalated onto the surface graphene layers. The variation of the integrated emissivity and sheet resistance is similar. However, the integrated infrared emissivity could also be modulated below 2 V ( Figure 2 c). This is attributed to the electrostatic doping effect of the accumulating ions at the surface graphene sample, resulting in a small infrared emissivity modulation. Additionally, the intercalation process is reversible, with the deintercalated graphene sample having similar I-V curves to that of pristine graphene samples ( Figure 5 c). In addition, we noted that the change of the infrared temperature of the surface multilayer graphene started at the edge and then propagated across the device at the very beginning. However, the intercalation process induces defects on sthe urface multilayer graphene, which was demonstrated by the increase of D peak intensity in the in-situ Raman spectra ( Figure 4 b). Thus, for the following cycle, the change of the infrared temperature of the surface multilayer graphene alters homogeneously across the whole sample, i.e., defects are induced across the whole sample. To test the long-term stability of the multilayer graphene device, we tested the infrared and Raman behaviors of the devices for 50 cycles in ambient conditions. The behavior of these devices was stable up to 30 cycles, then they started to degrade. The degradation was probably due to the hydration of the ionic liquid in ambient atmosphere, which can be avoided by the passivation of the device. Finally, we found that the infrared behavior of the graphene device was similar for positive and negative bias, which indicated the co-intercalation of anions and cations of the ionic liquid into graphene layers."
} | 3,738 |
30775638 | PMC6373447 | pmc | 7,695 | {
"abstract": "Microfluidic devices with integrated valves provide precise, programmable fluid\nhandling platforms for high-throughput biological or chemical assays. However, setting up\nthe infrastructure to control such platforms often requires specific engineering expertise\nor expensive commercial solutions. To address these obstacles, we present a Kit for\nArduino-based Transistor Array Actuation (KATARA), an open-source and low-cost\nArduino-based controller that can drive 70 solenoid valves to pneumatically actuate\nintegrated microfluidic valves. We include a python package with a GUI to control the\nKATARA from a personal computer. No programming experience is required.",
"discussion": "9. Discussion The KATARA provides a user-friendly solution to control solenoid valves at low\ncost. A complete microfluidic platform must also include the pneumatic infrastructure to\nrelay pressure to the solenoid valves and operate the microfluidic device. In this issue,\nBrower et al. present a comprehensive pneumatic platform for microfluidic large-scale\nintegration [ 16 ]. The KATARA may be\nused as an alternate control module in this platform, as it serves as a low-cost alternative\nto the Wago controller. The KATARA shield also has the capability to control microfluidic\ndevices remotely without a computer. This opens up the possibility to use microfluidic\ndevices with integrated valves in field settings. The KATARA shield also maintains the\nArduino’s ability to use its digital pins for purposes other than driving solenoid\nvalves when valves are not connected to their amplifying circuits. Additionally, the KATARA\nmay be suitable for purposes other than microfluidics including soft robotics, driving\nmotors, and powering light sources. As we continue to develop the KATARA, we will post\nhardware and software updates on the Streets Lab website ( http://streetslab.berkeley.edu/tools/katara/ )."
} | 468 |
24411456 | null | s2 | 7,696 | {
"abstract": "Increasing the production of fatty acids by microbial fermentation remains an important step toward the generation of biodiesel and other portable liquid fuels. In this work, we report an Escherichia coli strain engineered to overexpress a fragment consisting of four dehydratase domains from the polyunsaturated fatty acid (PUFA) synthase enzyme complex from the deep-sea bacterium, Photobacterium profundum. The DH1-DH2-UMA enzyme fragment was excised from its natural context within a multi-enzyme PKS and expressed as a stand-alone protein. Fatty acids were extracted from the cell pellet, esterified with methanol and quantified by GC-MS analysis. Results show that the E. coli strain expressing the DH tetradomain fragment was capable of producing up to a 5-fold increase (80.31 mg total FA/L culture) in total fatty acids over the negative control strain lacking the recombinant enzyme. The enhancement in production was observed across the board for all the fatty acids that are typically made by E. coli. The overexpression of the DH tetradomain did not affect E. coli cell growth, thus showing that the observed enhancement in fatty acid production was not a result of effects associated with cell density. The observed enhancement was more pronounced at lower temperatures (3.8-fold at 16 °C, 3.5-fold at 22 °C and 1.5-fold at 30 °C) and supplementation of the media with 0.4% glycerol did not result in an increase in fatty acid production. All these results taken together suggest that either the dehydration of fatty acid intermediates are a limiting step in the E. coli fatty acid biosynthesis machinery, or that the recombinant dehydratase domains used in this study are also capable of catalyzing thioester hydrolysis of the final products. The enzyme in this report is a new tool which could be incorporated into other existing strategies aimed at improving fatty acid production in bacterial fermentations toward accessible biodiesel precursors."
} | 491 |
25182323 | PMC4173780 | pmc | 7,697 | {
"abstract": "ABSTRACT Due to the increasing concerns about limited fossil resources and environmental problems, there has been much interest in developing biofuels from renewable biomass. Ethanol is currently used as a major biofuel, as it can be easily produced by existing fermentation technology, but it is not the best biofuel due to its low energy density, high vapor pressure, hygroscopy, and incompatibility with current infrastructure. Higher alcohols, including 1-propanol, 1-butanol, isobutanol, 2-methyl-1-butanol, and 3-methyl-1-butanol, which possess fuel properties more similar to those of petroleum-based fuel, have attracted particular interest as alternatives to ethanol. Since microorganisms isolated from nature do not allow production of these alcohols at high enough efficiencies, metabolic engineering has been employed to enhance their production. Here, we review recent advances in metabolic engineering of microorganisms for the production of higher alcohols.",
"conclusion": "CONCLUSION Limited fossil fuel resources and increasing environmental concerns have been urging us to develop platform technologies for the sustainable and economical production of alternative fuels. In order to develop economically competitive bioprocesses for their production, the metabolic pathways need to be optimally engineered by designing the best pathways to increase the metabolic flux toward the desired product, improving the kinetics and substrate specificities of the enzymes involved, and balancing the cofactors and redox. As described above, several higher alcohols can be efficiently produced by employing metabolically engineered microorganisms. It is expected that more successful examples of microbial higher-alcohol production will appear through the strain development integrated with bioprocess engineering. Metabolic engineering will keep playing a key role in developing such economically competitive bioprocesses.",
"introduction": "INTRODUCTION Increasing concerns on climate change and inevitable depletion of fossil resources are urging us to develop fuels and energy that are independent of fossil resources. Microbial production of biofuels from renewable biomass has been considered one of the solutions ( 1 ). Currently, ethanol is a major biofuel produced worldwide, mainly because it can be produced by fermentation technology that has been available for a long time. However, ethanol is not such a great biofuel due to its inferior fuel characteristics, such as low energy density, high vapor pressure, hygroscopy, and incompatibility with current infrastructure. Therefore, there has recently been much interest in producing advanced biofuels possessing fuel characteristics similar to those of petroleum-derived fuels, such as hydrocarbons and higher alcohols. In this paper, we review recent advances in the production of higher alcohols, with a focus on metabolic engineering strategies employed for the development of microbial strains efficiently producing them."
} | 739 |
35095828 | PMC8795815 | pmc | 7,699 | {
"abstract": "Crop plants are more often exposed to abiotic stresses in the current age of fast-evolving climate change. This includes exposure to extreme and unpredictable changes in climatic conditions, phytosanitary hazards, and cultivation conditions, which results in drastic losses in worldwide agricultural productions. Plants coexist with microbial symbionts, some of which play key roles in the ecosystem and plant processes. The application of microbial biostimulants, which take advantage of symbiotic relationships, is a long-term strategy for improving plant productivity and performance, even in the face of climate change-associated stresses. Beneficial filamentous fungi, yeasts, and bacteria are examples of microbial biostimulants, which can boost the growth, yield, nutrition and stress tolerance in plants. This paper highlights recent information about the role of microbial biostimulants and their potential application in mitigating the abiotic stresses occurring on crop plants due to climate change. A critical evaluation for their efficient use under diverse climatic conditions is also made. Currently, accessible products generally improve cultural conditions, but their action mechanisms are mostly unknown, and their benefits are frequently inconsistent. Thus, further studies that could lead to the more precisely targeted products are discussed.",
"conclusion": "Conclusion and Future Prospects Microbial biostimulants have the potential to be a long-term and successful method for reducing the abiotic stressors that climate change has exacerbated. Furthermore, the application of microbial biostimulants may help to maintain the ecological balance of agro-ecosystems, reducing the usage of pesticides and/or heavy metals for agricultural practices. Nonetheless, several concerns should be considered both at the regulatory level and throughout the development and research, to achieve greater efficacy of the product and wider adoption. Plant biostimulant has a claim-based definition, which means that the function is used to establish the product. Many potent ingredients with various activities and objectives can be found in a single product. As a result, the inherent heterogeneity of microbial biostimulants could elude regulatory categorization (e.g., fungicide, fertilizer, and amendment). Products may be subjected to lengthy and costly trial procedures depending on the country of registration. The lack of a consistent worldwide regulatory framework creates a barrier to product marketing and may deter the development of novel products. In agroecological and biological research, the adoption of microbial biostimulants still has certain drawbacks, mostly due to their lesser efficacy and greater environmental sensitivity when compared to synthetic products (e.g., pesticides, fertilizers, and growth regulators). Furthermore, the efficacy of microbial biostimulants varied widely according to the crop and environmental conditions. Future research should focus on developing better-targeted products, such as delving deeper into interactions of the microbial biostimulant with indigenous plant-associated microbiomes. The application of microbial biostimulants might provide a long-term and cost-effective solution to plant productivity losses caused by changing climatic factors, as well as aid in the optimization of human inputs in agro-ecosystem. Results from preliminary experiments involving microbial biostimulants should be disseminated by all relevant policymakers and stakeholders, such as extension services and growers, to ensure that this methodology can be largely applied to a variety of crops, regions, and under different environmental conditions.",
"introduction": "Introduction Global climate records in the last decades have revealed a rise in global temperature alongside changes in rainfalls, resulting in various serious implications on environmental and agricultural aspects ( Füssel et al., 2012 ). Crop plants are more frequently exposed to abiotic stresses caused by climate change because, aside from direct implications of abiotic stresses on plants, climate change could increase the number of pests and diseases, as well as increase the severity and frequency of the outbreak of diseases ( De Wolf and Isard, 2007 ; Garrett et al., 2016 ). According to recent estimates, abiotic stresses are anticipated to cause up to 50% losses, or higher, in worldwide agricultural productivity, depending on the region ( Kumar and Verma, 2018 ). These losses, coupled with the continual rise in the human population, revealed that about 60% boosting of agricultural production is needed to meet the world food needs ( Wild, 2003 ), with a concrete risk of dramatic deforestation increment and loss of natural ecosystems ( Byerlee et al., 2014 ). Increased plant resilience to mitigate climate change-associated stresses is a sustainable method for ensuring food security with a restricted increase in agricultural surface, and the use of microbial biostimulants is one of the best options to achieve this goal ( Calvo et al., 2014 ; Yakhin et al., 2017 ). Plants are associated with a diverse group of microorganisms (the microbiome) in their endosphere (internal compartments), rhizosphere (attached soil to roots), and phyllosphere (aboveground parts), making microbial biostimulants particularly fascinating ( Compant et al., 2019 ; Babalola et al., 2020 ). Crop plants coexist with microbial symbionts, which play key roles in plant production, performance, nutrition, and tolerance to abiotic stress ( Vandenkoornhuyse et al., 2015 ; Enebe and Babalola, 2018 ; Ojuederie et al., 2019 ). For instance, geological evidence indicates that the relationship between microbes and plants predates the emergence from the water, suggesting that symbiosis involving arbuscular mycorrhizal was important in the process of terrestrialization ( Selosse and Le Tacon, 1998 ). Moreover, microorganisms are involved in multiple biogeochemical cycles, such as nitrogen and carbon cycling, nitrogen fixation, soil formation and plant nutrition acquisition in the ecosystems ( Wagg et al., 2014 ; Igiehon et al., 2019 ). As a result, many microbial symbionts can be used as a biofertilizer, releasing additional nutrients to the plant through synergistic mechanisms, which include nitrogen fixation (e.g., Mesorhizobium loti , Rhizobium etli , Azotobacter vinelandii , and Azospirillum brasilense ), phosphate solubilization (e.g., Arbuscular mycorrhizal fungi, Azospirillum spp., Bacillus spp., and Pseudomonas spp.), cellulolytic activity ( Aspergillus spp., Trichoderma spp., Bacillus spp., and Penicillium spp.), soil acidification ( Bacillus spp. and Arthrobacter spp.), and production of siderophores (e.g., Pseudomonas spp.) ( Bhattacharyya and Jha, 2012 ; Orozco-Mosqueda et al., 2021 ). Also, thanks to its abilities to boost plant development, defenses, antibacterial compounds, combat pathogen infections and feed on nematodes, Trichoderma spp. is a well-studied symbiotic fungal genus ( Adnan et al., 2019 ; Szczałba et al., 2019 ). Despite the beneficial effects exhibited on their hosts, some of which include increased protection from abiotic stresses and nutritional efficiency, some weaknesses may limit the use of Trichoderma spp. as commercial biostimulant products, such as difficulties of in vitro cultivation and escalating of bioproduction, the lack of understanding on host specificity and population dynamics in the agroecosystem ( Du Jardin, 2015 ). Other types of fungi can form part of the beneficial microbiome associated with plants, such as yeasts belonging to Brettanomyces naardensis , Candida oleophila , Aureobasidium pullulans , Metschnikowia fructicola , Cryptococcus albidus , and Saccharomyces cerevisiae ( Freimoser et al., 2019 ; Nafady et al., 2019 ). Foliar infections can be controlled by yeasts that could colonize the leaf of a plant using direct antagonism ( Preininger et al., 2018 ) or through the induction of systemic resistance ( Lee et al., 2017 ). Likewise, yeast inhabiting the soil can enhance the growth of the plant through phosphate solubilization, digestion of organic materials, soil aggregation and stimulation of root development, and suppressing root infections ( Sarabia et al., 2018 ). Plant growth-promoting bacteria (PGPB), which includes rhizobacteria or bacterial endophytes, are known to majorly populate the plant rhizosphere and the most studies genera are Azospirillum , Azotobacter , Arthrobacter , Burkholderia , Gluconacetobacter , Pseudomonas , Bacillus , Streptomyces , and Serratia ( Kour et al., 2020a ). Additional genera are more recently proposed as possible bioinoculants with biocontrol and/or plant growth-promoting activities, such as Rouxiella badensis and Rahnella spp. ( Ulloa-Muñoz et al., 2020 ; Morales-Cedeńo et al., 2021 ). Physiological, molecular, and biochemical investigations of the interactions that exist between plant and beneficial microorganisms have shown that the presence of microbe-induced plant stress responses ( Farrar et al., 2014 ; Igiehon et al., 2021 ; Igiehon and Babalola, 2021 ), which may trigger induced systemic tolerance (IST) against abiotic stressors ( Yang et al., 2009 ; Vacheron et al., 2015 ). Microbial biostimulants are a viable alternative for supporting plants exposed to abiotic stresses in the current context of fast-developing climate change ( Santoyo et al., 2021b ). While recent advancements and laboratory studies have revealed the positive activities of plant-associated microorganisms, the efficacy of microbial biostimulants is yet to be successfully validated in field experiments. As a result, microbial biostimulants are often used as supplemental therapies rather than being used to their full potential in crop management. The goal of this paper is to summarize current information about microbial biostimulants, especially the current commercially available products, examine their applications in enhancing plant tolerance to abiotic stresses caused by climate changes, and forecast the creation of novel products that may be used in adverse conditions. Also, limitations in the application of microbial biostimulants under field conditions alongside further studies required for the better development of targeted products are discussed."
} | 2,587 |
38392768 | PMC10889539 | pmc | 7,705 | {
"abstract": "Wood-rotting fungi’s degradation of wood not only facilitates the eco-friendly treatment of organic materials, decreasing environmental pollution, but also supplies crucial components for producing biomass energy, thereby reducing dependence on fossil fuels. The ABC gene family, widely distributed in wood-rotting fungi, plays a crucial role in the metabolism of lignin, cellulose, and hemicellulose. Trametes gibbosa , as a representative species of wood-rotting fungi, exhibits robust capabilities in wood degradation. To investigate the function of the ABC gene family in wood degradation by T. gibbosa , we conducted a genome-wide analysis of T. gibbosa ’s ABC gene family. We identified a total of 12 Tg-ABCs classified into four subfamilies (ABCA, ABCB, ABCC, and ABCG). These subfamilies likely play significant roles in wood degradation. Scaffold localization and collinearity analysis results show that Tg-ABCs are dispersed on scaffolds and there is no duplication of gene sequences in the Tg-ABCs in the genome sequence of T. gibbosa . Phylogenetic and collinearity analyses of T. gibbosa along with four other wood-rotting fungi show that T. gibbosa shares a closer phylogenetic relationship with its same-genus fungus ( Trametes versicolor ), followed by Ganoderma leucocontextum , Laetiporus sulphureus , and Phlebia centrifuga in descending order of phylogenetic proximity. In addition, we conducted quantitative analyses of Tg-ABCs from T. gibbosa cultivated in both woody and non-woody environments for 10, 15, 20, 25, 30, and 35 days using an RT-qPCR analysis. The results reveal a significant difference in the expression levels of Tg-ABCs between woody and non-woody environments, suggesting an active involvement of the ABC gene family in wood degradation. During the wood degradation period of T. gibbosa , spanning from 10 to 35 days, the relative expression levels of most Tg-ABCs exhibited a trend of increasing, decreasing, and then increasing again. Additionally, at 20 and 35 days of wood degradation by T. gibbosa , the relative expression levels of Tg-ABCs peak, suggesting that at these time points, Tg-ABCs exert the most significant impact on the degradation of poplar wood by T. gibbosa . This study systematically reveals the biological characteristics of the ABC gene family in T. gibbosa and their response to woody environments. It establishes the foundation for a more profound comprehension of the wood-degradation mechanism of the ABC gene family and provides strong support for the development of more efficient wood-degradation strategies.",
"conclusion": "5. Conclusions In this study, we systematically analyzed the ABC gene family in T. gibbosa , examining its structure, predicting its functions, and identifying the characteristics of Tg-ABCs . Additionally, we also explored how Tg-ABCs respond to woody environments. Our findings revealed that the ABC gene family plays a crucial role in the degradation of poplar wood by T. gibbosa . This study lays the foundation for a deeper understanding of the wood degradation mechanism of the ABC gene family and provides strong support for the development of more efficient strategies for the biological treatment of wood and other organic materials.",
"introduction": "1. Introduction Poplars, known for their rapid growth, wide distribution, and short genome, among other advantages, have found extensive applications in various fields. Their fast growth makes their wood suitable for biomass energy production, and their lightweight, durable, and easy-to-process characteristics contribute to their widespread use in furniture manufacturing, construction, and the paper industry [ 1 ]. Populus simonii Carr × Populus nigra L., a hybrid variety of poplar, inherits favorable traits from both parent species, exhibiting excellent fast-growing properties and features such as drought resistance, cold resistance, and insect resistance. As a result, it has become a key afforestation species in northern regions [ 2 , 3 , 4 ]. Known for its rapid growth, it typically reaches commercial timber size in a few years. Its wood generally contains a moderate amount of cellulose, making it well-suited for paper and pulp production. Therefore, it has significant potential in areas such as timber production, the paper industry, and biomass energy production [ 5 , 6 ]. Wood degradation is the gradual breakdown of wood into smaller compounds and organic substances in natural surroundings or specific conditions. This process plays a crucial role in the organic material cycle, contributing to the preservation of ecological balance. Furthermore, it serves as a source of raw materials for biomass energy production, offering a means to reduce reliance on fossil fuels [ 7 , 8 , 9 , 10 ]. However, wood, consisting of intricate compounds like lignin, cellulose, and hemicellulose, encounters numerous challenges during degradation. Lignin, the most challenging component to degrade in wood, possesses a complex structure and resists degradation by most biological enzymes due to its multiple chemical bonds. Cellulose, which is relatively more susceptible to degradation than lignin, benefits from its linear structure, facilitating enzymes in breaking β-1,4-glucosidic bonds. Hemicellulose presents a degradation difficulty that falls between lignin and cellulose. These components intertwine, forming a robust and less easily destructible structure, thereby heightening the complexity of wood degradation [ 11 , 12 , 13 ]. Wood-rotting fungi excel at degrading wood, and wood decomposition constitutes their primary way of life. They efficiently break down wood and plant residues, converting organic matter into forms more readily absorbed by other organisms. This process contributes to the natural recycling of organic materials. Furthermore, these fungi can be employed to treat waste from the wood and pulp industry, reducing environmental pollution. They present a potential for low-energy, pollution-free wood degradation and sustainable energy production [ 14 , 15 ]. Trametes gibbosa , as one of numerous wood-rotting fungi, possesses significant wood-degrading capabilities. It can release extracellular enzymes such as manganese peroxidases, lignin peroxidase, and laccase, facilitating the degradation of lignin in wood. Additionally, it produces intracellular glucanases, extracellular glucanases, and xylanases, enabling the breakdown of cellulose and hemicellulose in wood. These enzymes assist T. gibbosa in breaking down wood into small molecular compounds, achieving wood degradation [ 16 , 17 , 18 , 19 ]. The ABC gene family is widely distributed among plants, insects, and microorganisms. Genes in this family encode ABC transporters, which play a pivotal role in the metabolism of lignin, cellulose, and hemicellulose [ 20 , 21 ]. ABC transporters facilitate the movement of the intermediates or degradation products of lignin and cellulose between intracellular and extracellular spaces. They can create channels on the cell membrane, transporting various degradation enzymes involved in breaking down lignin and cellulose from the intracellular space to the extracellular space or other cell structures, thereby enhancing degradation efficiency. Additionally, they can expel potentially toxic substances from the cell’s products into the external environment, maintaining the stability of the intracellular environment [ 22 , 23 , 24 ]. Researchers have conducted thorough investigations into the relationship between the ABC gene family and lignin and cellulose. For example, de Lima et al. found that the ABC gene ( SvABCG17 ) is a potential transporter of lignin monomers [ 25 ]. Yu et al. suggested the potential role of the ABC gene ( PgrABCG14 ) in promoting plant growth and lignin accumulation [ 26 ]. Xie et al. identified ABC family genes in Phanerochaete chryso-sporium and suggested that the transporter protein identified in the P. chrysosporium secretomes could play a role in cellobiose and/or cellodextrin uptake [ 27 ]. Based on these studies, it is speculated that the ABC gene family may have a significant impact on wood degradation. However, at present, no research has explored the specific functions of ABC family genes in wood degradation. Here, we focused on the degradation process of P. simonii Carr × P. nigra L. wood by T. gibbosa , aiming to investigate the functions and responses of the ABC gene family, conducting analyses on its structure, predicting functions, and identifying characteristics. Additionally, we conducted a quantitative analysis of T. gibbosa ’s ABC genes at different time points in both woody and non-woody environments using the RT-qPCR analysis, to elucidate the response of ABC gene family in the process of degrading poplar wood of T. gibbosa . This study establishes a solid foundation for wood-rotting fungi to contribute to the production of low-energy, environmentally friendly organic matter and sustainable energy.",
"discussion": "4. Discussion The investigation of wood degradation holds substantial ecological and biological importance. Efficient wood degradation is vital for carbon cycling in ecosystems and provides essential technical support for sustainable biofuel production [ 35 , 36 ]. The ABC gene family plays a crucial role in lignin, cellulose, and hemicellulose metabolism, regulating vital substrate transport during this process [ 37 ]. T. gibbosa , as one of the numerous wood-rotting fungi, demonstrates efficient wood degradation capabilities [ 38 ], presenting a promising opportunity to investigate the role of the ABC gene family in wood degradation. We identified a total of 12 Tg-ABCs by analyzing the genomic sequence of the ABC gene family in T. gibbosa . The results of the DNA binding domain alignment reveal that all of these genes contain conserved ABC transporter domains, exhibiting a high degree of sequence conservation. Tg-ABCs are classified into four subfamilies, ABCA, ABCB, ABCC, and ABCG. These subfamilies likely play significant roles in wood degradation, while the absence of five subfamilies [ 39 ] may have a relatively minor impact on this process. Scaffold localization results indicate that Tg-ABCs are dispersed on scaffolds. A collinearity analysis reveals that there is no duplication of gene sequences within the Tg-ABCs in the genome sequence of T. gibbosa . These findings collectively suggest that the functions of T. gibbosa ’s ABC gene family in the wood degradation process may exhibit diversity. Phylogenetic and collinearity analyses of T. gibbosa with four other wood-rotting fungi show that T. gibbosa shares a closer phylogenetic relationship with its same-genus fungus ( T. versicolor ) [ 40 ], followed by G. leucocontextum , L. sulphureus , and P. centrifuga in descending order of phylogenetic proximity. Additionally, the branches of ABC genes within the same subfamily are closer than those within the same species, indicating a higher degree of homology within the same ABC subfamily across different species. This suggests that ABC protein sequences exhibit a higher level of conservation during the evolutionary process, and their wood degradation characteristics may also show a higher degree of similarity [ 41 ]. To gain a deeper understanding of how Tg-ABCs response to wood degradation by T. gibbosa , we conducted quantitative analyses of the expression of Tg-ABCs from T. gibbosa cultivated in both woody and non-woody environments for 10, 15, 20, 25, 30, and 35 days using an RT-qPCR analysis. The results reveal a significant difference in the expression levels of Tg-ABCs between woody and non-woody environments, suggesting an active involvement of the ABC gene family in wood degradation [ 21 , 42 , 43 ]. During the wood degradation period of T. gibbosa , spanning from 10 to 35 days, the relative expression levels of most Tg-ABCs exhibited a trend of increasing, decreasing, and then increasing again. It is speculated that this trend is associated with the different products generated during various stages of wood degradation by T. gibbosa . In the early stages of wood degradation, T. gibbosa produces a significant amount of extracellular enzymes, and ABC transporters facilitate the transport of these enzymes from the extracellular space into the cell, leading to an increase in their expression levels. In the mid-phase of wood degradation, these enzymes catalyze the oxidation and decomposition of large molecules such as lignin, cellulose, and hemicellulose. At this stage, ABC transporters are not required, resulting in a decrease in their expression levels. In the later stages of wood degradation, the generation of small molecular substances like acids, alcohols, and esters prompts ABC genes to transport these substances, causing a resurgence in their expression levels [ 16 , 44 , 45 ]. The fluctuation in the relative expression levels of Tg-ABCs corresponds to the metabolic trajectory of wood degradation, indicating the pivotal role of Tg-ABCs in the process. Additionally, at 20 and 35 days of wood degradation by T. gibbosa , the relative expression levels of Tg-ABCs peak, suggesting that at these time points, Tg-ABCs exert the most significant impact on wood degradation by T. gibbosa ."
} | 3,330 |
25421463 | null | s2 | 7,706 | {
"abstract": "We investigated ion transport limitations on 3D graphite felt electrodes by growing Geobacter sulfurreducens biofilms with advection to eliminate external mass transfer limitations. We characterized ion transport limitations by: (i) showing that serially increasing NaCl concentration up to 200 mM increased current linearly up to a total of +273% vs. 0 mM NaCl under advective conditions; (ii) growing the biofilm with a starting concentration of 200 mM NaCl, which led to a maximum current increase of 400% vs. current generation without NaCl, and (iii) showing that un-colonized surface area remained even after steady-state current was reached. After accounting for iR effects, we confirmed that the excess surface area existed despite a non-zero overpotential. The fact that the biofilm was constrained from colonizing and producing further current under these conditions confirmed the biofilms under study here were ion transport-limited. Our work demonstrates that the use of high surface area electrodes may not increase current density when the system design allows ion transport limitations to become dominant."
} | 280 |
36133584 | PMC9417734 | pmc | 7,707 | {
"abstract": "Networks based on nanoscale resistive switching junctions are considered promising for the fabrication of neuromorphic computing architectures. To date random networks of nanowires, nanoparticles, and metal clusters embedded in a polymeric matrix or passivated by shell of ligands or oxide layers have been used to produce resistive switching systems. The strategies applied to tailor resistive switching behavior are currently based on the careful control of the volume fraction of the nanoscale conducting phase that must be fixed close to the electrical percolation threshold. Here, by blending laboratory and computer experiments, we demonstrate that metallic nanostructured Au films fabricated by bare gold nanoparticles produced in the gas phase and with thickness well beyond the electrical percolation threshold, show a non-ohmic electrical behavior and complex and reproducible resistive switching. We observe that the nanogranular structure of the Au films does not evolve with thickness: this introduces a huge number of defects and junctions affecting the electrical transport and causing a dynamic evolution of the nanoscale electrical contacts under the current flow. To uncover the origin of the resistive switching behavior in Au cluster-assembled films, we developed a simple computational model for determining the evolution of a model granular film under bias conditions. The model exploits the information provided by experimental investigation about the nanoscale granular morphology of real films. Our results show that metallic nanogranular materials have functional properties radically different from their bulk counterparts, in particular nanostructured Au films can be fabricated by assembling bare gold clusters which retain their individuality to produce an all-metal resistive switching system.",
"conclusion": "Conclusions We demonstrated that nanostructured Au films fabricated by bare gold nanoparticles produced in the gas phase and with thickness well beyond the electrical percolation threshold, show a non-ohmic behavior and reproducible complex resistive switching. This is ascribed to the nanogranular structure of the films that does not evolve with thickness causing the presence of a huge number of defects and junctions. The highly defective structure affects the electrical transport and dynamically rearranges under the current flow. A simple computational model based on a variable resistor network qualitatively reproduce the observed resistive switching behavior. Our results show that metallic nanoscale materials have functional properties radically different from their bulk counterparts, in particular nanostructured Au films can be fabricated by assembling bare gold nanoparticles which retain their individuality to produce an all-metal resistive switching system. The presence of recurrent features in the switching behavior such as a threshold voltage for the appearance of the switching activity, recurrently and discrete explored resistance values and their dependence from the structural features of the cluster-assembled film (average thickness, resistance value reached on the percolation curve), offers the possibility to exploit such class of systems for the fabrication of complex networks suitable for reservoir computing, 5,18 using a straightforward bottom-up approach. Our results on nanostructured gold films opens new and important perspectives for the investigation of the fundamental mechanisms of electrical conduction in nanoscale metallic materials.",
"introduction": "Introduction The looming end of Moore's scaling laws is stimulating the development of systems and architectures able to overcome the roadblock of CMOS-based digital computing. 1 The speed, spatial scale, circuit size, and energy dissipation required to maintain the ever increasing data flux between the memory and the processor is not any longer sustainable at extreme miniaturization: 2 a substantial paradigm shift in the design of computing hardware is mostly needed. An approach aiming at reproducing in hardware the human brain architectural and dynamical properties has been proposed as an alternative overcoming these limitations. 3 Networks based on nanoscale resistive switching (RS) junctions are currently investigated for the fabrication of neuromorphic computing architectures, where the processing of cognitive and distributed data intensive tasks is performed similarly to synapses networks. 4–6 RS refers to physical phenomena where the resistance of a dielectric material changes reversibly in response to the application of a strong external electric field. RS has been reported in several systems including oxides, nitrides, chalcogenides, semiconductors, and organic materials. 6,7 For instance, random networks of nanowires, nanoparticles, and clusters, embedded in a polymeric matrix or passivated by shell of ligands or oxide layers (insulator-conductor nanocomposites), show RS phenomena resulting in features typical of neuromorphic systems. 6,8 This has been attributed to either polymer breakdown between junctions or conducting filament formation between individual metallic nanoobjects. 9–11 The evolution of the electrical properties of such nanocomposites, upon the application of an electric field, strongly depends on the volume fraction of the conductive phase and it can be described by the percolation theory. 12,13 Alternatively, networks of metallic nanoparticles produced in the gas phase and subsequently deposited on a substrate also exhibit RS, 14 provided that the thickness of the resulting film is close to the percolation threshold. Switching events are argued to depend on the electric-field-induced formation/breaking of atomic wires in tunnel gaps between neighboring clusters. 15 Recently, we reported that resistive switching networks can be fabricated by supersonic cluster beam deposition (SCBD) of Au clusters near the percolation threshold. 16 In particular, we have studied the role of substrates in determining the characteristics of the observed switching behavior. 17 Tailoring the conductivity of networks of nanoobjects showing RS is typically achieved by engineering the junctions between the different components and/or by the careful control of the volume fraction of the conducting phase that must be fixed close to the electrical percolation threshold. 11,15,18 Here, by blending laboratory and computer experiments, we report that nanostructured gold films fabricated by assembling bare gold nanoparticles and with thickness well beyond the percolation threshold, are in fact conductive, however show non-linear I – V curves together with reversible and reproducible resistive switching. Cluster-assembled gold films are of particular interest for the study of nanostructured metallic layers since they are not affected by oxidation of the nanoparticles during and after the deposition process. We characterize the evolution of the structure of the films with increasing thickness and provide robust evidence that the granularity, despite the evolving transport properties of the film, does not evolve since the nanoscale building blocks maintain their individuality. To uncover the origin of the resistive switching behavior in Au cluster-assembled films, we developed a simple computational model for determining the evolution of the conduction properties of a model granular film under bias condition. The model exploits the information provided by experimental investigation about the nanoscale granular morphology of real films. As a result, the main electrical quantities ( i.e. current and resistance) are calculated and monitored during the film evolution. Comparison with experimental data is eventually drawn.",
"discussion": "Results and discussion To test the electrical behavior of cluster-assembled Au films with different thicknesses, we fabricated two-terminal devices consisting in a pair of rectangular (2 × 7 mm) gold electrodes separated by a gap of 1 mm and deposited on an oxidized silicon substrate by standard thermal evaporation of gold in vacuum. ( Scheme 1 ). Cluster-assembled films bridging the gap between the two gold electrodes have been deposited at room temperature by Supersonic Cluster Beam Deposition (SCBD) (see Methods). 19 The experimental apparatus of the SCBD is schematically depicted in Scheme 1 , the sample holder is equipped with electrical contacts for the in situ characterization of the evolution of the electrical properties of the film during the deposition process. Cluster-assembled films are also deposited on silicon substrates without electrodes or with a single electrode covering half of the substrate in order to perform structural and electrical characterization of the films by atomic force microscopy (AFM) (see Methods). Scheme 1 Schematic representation (not to scale) of the apparatus for the deposition of cluster-assembled Au films. It consists of a pulsed microplasma cluster source mounted on the axis of differentially pumped vacuum chambers, the PMCS produces a supersonic expansion of an inert gas seeded with metallic clusters to form a cluster beam. The beam is intercepted by a substrate placed on a mobile holder (manipulator) in the deposition chamber. Substrates with gold electrodes previously deposited by thermal evaporation are mounted on the sample holder. A quartz microbalance attached to the manipulator is periodically exposed to the cluster beam to monitor the amount of deposited material. In situ electrical characterization during the cluster-assembled film growth is performed. Au clusters, with a bimodal log-normal mass distribution peaked at 0.5 nm and 6 nm, were deposited using an apparatus equipped with a Pulsed Microplasma Cluster Source (PMCS) as described in detail in ref. 20 (see also Methods). \n Fig. 1a shows the evolution of the electrical properties of a typical cluster-assembled film (red curve) compared to that of an atom-assembled one (blue curve) measured in situ during deposition. For different growth stages of the cluster-assembled film we associate a scanning electron microscopy (SEM) image ( Fig. 1B, C and D ) with its electrical behavior. For the atom-assembled film we report a SEM micrograph of a continuous layer ( Fig. 1A ). Fig. 1 Top: percolation curves of an atomic-assembled gold film (blue) and of a cluster-assembled film (red), with the conductance (the inverse of the measured film resistance) on the y -axis in logarithmic scale and the film thickness on the x -axis; bottom: SEM images of the film morphology are associated to different film thicknesses and electrical behaviour. (A) continuous atom-assembled film (scale bar 200 nm). (B–D): Images of the principal steps of growth of a cluster-assembled metallic films are reported: (B) insulating stage; (C) close to percolation; (D) conducting regime: a fully connected thick-film (scale bar 100 nm). We followed the evolution of the electrical properties of the different nanostructured film morphologies monitoring the evolution of the resistance as a function of thickness during the film growth. To describe the observed behavior we used the percolation theory based on the variation of the connectivity of elements, in our case clusters or nanoaggregates, in a random system. 12 In this framework, three principal growth stages of a random assembling of nanoparticles are easily identified from the conductance-thickness curve (percolation curve). 21,22 The evolution of the conductivity of cluster-assembled films with thickness ( Fig. 1 , red curve) shows that for a film characterized by isolated aggregates, an insulating behavior is observed. 13 By increasing the thickness, the first percolation paths are formed (geometrical percolation stage) the conductivity abruptly increases, while the film is still in sub-monolayer regime. The critical thickness for the electrical percolation threshold, corresponding to the film morphology in Fig. 1C , can be determined by the occurrence of the maximum slope of the conductivity vs. thickness curve. 23 After the percolation transition, the cluster-assembled film is fully connected ( Fig. 1D ) and an ohmic behavior typical of a metallic regime is expected, as observed in atom-assembled metallic films. 24 We point out that in order to determine the thickness at which the transition to a ohmic behavior takes place, we consider the minimum of the product between the resistance and the squared thickness as function of the thickness. 25,26 For the sake of comparison, the evolution of the electrical properties of an atom-assembled film (blue curve) is also reported in Fig. 1 , showing the percolation threshold always below 10 nm, whereas for cluster-assembled films such a threshold is found at thickness ∼10–15 nm. The different surface mobilities of atoms and clusters are the origin of the different threshold thickness in the two systems. 27 For the atom-assembled film the transition to an ohmic behavior occurs at a thickness around 12 nm, while the cluster-assembled film shows the same transition in a range between 18 nm and 25 nm. We note also a faster increase of conductance in the case of the atom-assembled films; this also can be explained by the morphologic properties determined employing different building blocks for the two types of film fabrication. 28 Multilayer cluster-assembled films are highly porous and characterized by a granular structure strongly reminiscent of the dimension distribution of the primeval building blocks. In order to observe the grain size distribution and the growth dynamics characterizing their structure, we performed a morphological analysis with atomic force microscopy (AFM) of the sub-monolayer cluster-assembled films. Fig. 2a reports the height distribution of deposited clusters (diameter in z direction, i.e. normal to the substrates) for a sub-monolayer film (coverage ∼8%). We observe very small aggregates with a broad distribution with height peaked at 0.4 ± 0.1 nm, while the median value of the population of the largest clusters is around 6 ± 2.5 nm. The equivalent radius distribution is bimodal, with the main two peaks around 0.7 and 4.4 nm. Fig. 2 From left to right: (a) histogram in logarithmic scale of the height of the clusters measured by AFM on the smallest coverage sample. (b) relative island heights (measured by AFM) and the relative radius (measured from SEM micrographs, see Materials and methods) as a function of the coverage (the subscript ‘0’ refer to the smallest coverage sample). (c) histogram of the equivalent radius of the grains obtained segmenting a SEM micrograph of the thick film, in logarithmic scale. By increasing the surface coverage from 0.08 to 0.45 ( Fig. 1B and C ), the mean value of the whole clusters heights increases of only 50%, on the other hand the equivalent radius increases of 160%. This suggests a preferential growth in x – y directions instead of z one in the first growth stages caused by the highest mobility of the small clusters. 27 By continuing the deposition we produce a fully connected film ( Fig. 1D ) with a thickness of 57 nm and characterized by a grain distribution substantially similar to that of the sub-monolayer films (quantitative granulometric characterization of SEM images is reported in Fig. 2c ). 29,30 We underline the fact that the grain distribution nearly perfectly overlaps to that of the sub-monolayer film, thus suggesting that no significant grain growth is present even for very high thickness and that the nanoscale building blocks retain their individuality as reported in previous investigations. 31–33 This is at odd to what observed for the growth of gold films assembled by atoms: in this case the grain size increases linearly as films become thicker. 34–36 In Fig. 3 we report a comparison between two gold films assembled by atoms ( Fig. 3a ) and by clusters ( Fig. 3b ) respectively. The resistance values of an atom-assembled film (2 mm × 7 mm, 100 nm thickness) results in a resistivity ρ atomic ∼ 10 −8 Ωm, the same order of magnitude than the bulk gold ρ bulk = 2.44 × 10 −8 Ωm, while a cluster-assembled film 65 nm thick shows a resistivity ρ cluster ∼ 10 −7 − 10 −6 Ωm, well above the bulk values. Fig. 3 (a) electrical resistance and current of an atom-assembled Au film 100 nm thick as a function of time under the application of 0.5 V. (b) electrical resistance and current of a cluster-assembled Au film 65 nm thick as a function of time under the application of 12 V in the proximity of the forming step; the current is approximately equal to that circulating in the atom-assembled film (c) I – V curve of a cluster-assembled film after the activation of the switching activity in semilog- y scale. \n Fig. 3b reports the temporal evolution of a cluster-assembled film with an initial resistance of about 33 Ω under a bias of 12 V: after few seconds we observe a switching towards substantially lower values of the circulating current and a corresponding steep increase of the resistance, followed by a sequence of switching events. The coexistence of two reversible switching regimes characterized by very different resistance ranges has already been reported in cluster-assembled gold films. 17 The behavior reported in Fig. 3b is similar to what observed in silicon oxide-based devices used to emulate the generation of action potentials in neuronal assemblies 37 The cluster-assembled film is characterized by a higher initial resistance compared to the atom-assembled; for this reason we observe a larger voltage drop in the former for the same current circulating in both systems. Due to its extremely low resistance, the atom-assembled film has been tested at voltages lower than or equal to 0.5 V to keep the delivered power at a reasonable level. Nevertheless the atom-assembled film does not show any anomalous change in its electrical resistance under to flow of the same current travelling in the cluster-assembled film (see Scheme 1 ). Current has been limited in order to avoid macroscopic joule heating effects both in the atom-assembled electrodes and in the cluster-assembled film. The difference in the resistivity values between atom-assembled and cluster-assembled films is due to the presence of an extremely large number of defects and grain boundaries typical of the low-energy cluster beam deposition regime 38–41 resulting in highly porous films. 42,43 The nanogranular and porous structure of the Au cluster-assembled films affects their electrical behavior, for thickness far beyond the percolation threshold, in a dramatic and unexpected way: we observe either a remarkable departure from an ohmic behavior and the onset of a resistive switching feature. The I – V curve of the cluster-assembled film is reported in Fig. 3c showing a non-linear behavior and the presence of hysteresis. We characterized several cluster-assembled films with thicknesses beyond the percolation threshold and different initial resistances: all the films showed stable resistive switching behavior, with well-defined resistance levels that are recurrently explored under the application of a constant voltage bias, and whose order of magnitude depends by the applied voltage (see Fig. 3c and 4 ). This behavior is at the origin of the non-ohmic I – V curves. In all cases we observed a non-linear behavior after a given threshold voltage which reproducibly depends on several factors, namely: the initial resistance, the thickness, and the history of the sample (the applied voltage values before applying the electroforming one). Usually, the threshold voltage corresponds to a power of 3–4 W. Since the electrodes have a negligible resistance compared to the cluster-assembled film, the total areal power is about 4 × 103 W m −2 . After the first switching event, we observe a switching activity for a wide voltage range, above and below the critical value. Fig. 4 (a) electrical resistance of a cluster-assembled film with thickness 30 nm under the bias of 20 V; several switch events in the interval time of 20 s are evident. (b) Histogram of the resistance values assumed by the sample under the application of 20 V in a time window of 200 s. (c) Electrical resistance of the same sample under the bias of 0.5 V. (d) Histogram of the resistance values assumed by the sample under the application of 0.5 V in a time window of 200 s. We report in Fig. 4a on the evolution of the resistance for a cluster-assembled film with a thickness of 30 nm under the application of 20 V bias after the activation: typical switching features are present, consisting in reversible switches in different resistance ranges. This is quantitatively represented in Fig. 4b reporting the histogram of the resistance values under a bias of 20 V in a time window of 200 s. The histogram shows different Gaussian type peaks around each resistance values reached after the switch events occur. It is to be noted that each level can be explored several times. In Fig. 4c and d we show the evolution of resistance with the histograms for the same sample under a lower voltage bias (0.5 V). The switching activity is always present although with a lower number of explored levels. A weakening of the switching activity can be observed, with a minor number of well distinct explored resistance values. This demonstrates the possibility to tune the switching activity of the device controlling the applied voltage in a reproducible manner. The observed switching activity has a low amplitude with a dynamics and temporal behaviour that shows periods of activity separated by periods of inactivity. 37 This suggests that cluster-assembled gold films can electrically emulate the generation of action potentials typical of neural systems. 37 The observed behavior is radically different if compared to what observed in the atom-assembled films ( Fig. 3a ): here an ohmic regime is always found under similar current conditions consistently with all the literature reporting the electrical behavior of atom-assembled metallic films beyond the percolation threshold. 34 To date the characterization of the electrical behavior of cluster-assembled metallic films is limited to systems below or close to the percolation transition; 44–46 to the best of our knowledge no reports about the electrical properties of cluster-assembled gold films beyond the percolation threshold are available. This is mainly due to the fact that electrical characterization has been used to monitor the evolution of the cluster-assembled systems towards percolation (assuming that structural percolation and electrical percolation coincide) 45 and by the assumption that, after reaching percolation, the systems is characterized by a ohmic behavior, although with a very high number of defects, as described, for example, in the Sondheimer 38 or Mayadas–Shatzkes model. 39 To investigate in greater detail the link between the film nanogranular structure and the electrical transport properties, we performed measurements by AFM in PeakForce Tapping conductive mode, using the current-sensing scheme described in ref. 47 . By applying a continuous voltage between an all-platinum conductive tip (Rocky Mountain Nanotechnology, tip radius <8 nm, k = 18 N m −1 and resonance frequency 20 kHz) and another electrode connected to the sample it was possible to measure the continuous current which flows between them. This approach allows for recording both a topographic and an electrical current map, with the identical high spatial resolution. In the left column of Fig. 5 we report the morphology of an atom-assembled Au thin film (a), and of a cluster-assembled Au film with a thickness beyond the percolation transition (b). Each topographical image has a high imaging quality; the cleaning condition of the AFM tip is good and the acquired topographies reproduce accurately the morphologies of the different Au samples. Fig. 5 Topographical AFM images of different Au films (left column) with the matching current map superimposed to the topographical one (right column), where the red regions are the conductive one. (a) Atom-assembled Au film. (b) Cluster-assembled Au film 37 nm thick. In the right column of Fig. 5 , we report the topographical maps with a modified color code: conductive regions (identified by applying to the current maps the threshold method described in the Materials and methods section) are uniformly colored in red, in order to directly associate the nanotopography to the electrical conductivity. By inspecting the maps reporting in red the conductive regions of the atom-assembled films ( Fig. 5a right), we observe that the red areas are disconnected: this is due to the fact that the dimension of the AFM tip radius (approximately 8 nm) does not allow to follow the entire profile of the rough film. The cluster-assembled film with a thickness of 37 nm beyond the percolation threshold ( Fig. 5b right) shows also another non-uniform electrical behaviour, since the large uppermost nanoparticles appear no conductive, irrespective of those forming the film core. This evidence suggests that the clusters sitting on the film surface are in bad electrical contact with their neighbours, even if they are morphologically connected with them and well in contact with the AFM tip. We can conclude that the cluster-assembled film is not electrically homogeneous, also with a thickness beyond the percolation threshold. Based on the above picture, we understand that the granular structure of cluster-assembled Au films largely determines the non-ohmic electrical properties of the system. Although the elemental building blocks can be considered metallic, their assembly does not result in an overall ohmic conductor. 48 Clusters are in fact physically connected ( i.e. they form a porous but connected system, whose structure is beyond the percolation threshold), nevertheless the conduction features of the inter-cluster contacts is affected by the presence of defects and by the mismatch in the crystalline orientation of the touching faces. 49 These factors represent a barrier for the electric charge flow determining a distribution of different “resistances” among the film. 50 This distribution can dynamically change under the flow of the current due to the formation of local “hot spots” that induce atomic rearrangement, formation/destruction of contact depending of the power dissipated locally. 8 We argue that a possible mechanism responsible for the resistive switching behaviour is represented by the modification/breaking of the physical contacts among nanoparticles and by the dynamical rearrangement of defects. More specifically, we guess that high electric fields could trigger the local rearrangement of matter so as to create new connection paths. 51 On the contrary, a suitable power dissipation due to current flow along a percolative path could determine the breaking of a connection (a sort of local melting). The overall resistive switching naturally follows from the interplay of such counteracting mechanisms. To test the validity of the above arguments, we developed a simple computer experiment aimed at modelling the microstructure evolution of a model granular film under bias condition. This model aims at a coarse-grained picture, with little insight on the actual details at the atomic scale. In this respect, the basic mechanisms driving the structural modifications of the system are introduced through a probabilistic description. The approach we followed consists of a multi-step procedure. At first, the morphological features of real samples are extracted from experimental SEM images (see Methods section) and used to generate a corresponding computer model of granular films. Then, a resistive network is setup and all of the electrical quantities of the system are accordingly calculated. Finally, a probabilistic algorithm ruling over the evolution under suitable bias condition is elaborated and eventually applied to determine the related evolution of the system morphology and electrical characteristics. All steps are properly included in an iterative self-consistent scheme. Details are provided in the Methods section. We assumed that a resistive model based on direct inter-grain conduction can describe the transport mechanisms. 52 We did not consider quantum tunnelling between neighbouring grains, and activated conduction since we are addressing electrical transport beyond the percolation threshold, where electron transport is known to mainly occur by ordinary conduction between metallic islands in physical contact. 53 In order to effectively build an electrical model, a regular grid is superimposed to binary images obtained by SEM micrographs, thus establishing a one-to-one correspondence between the nodes of the grid and the pixels of the digital image. Next, a resistance is associated to each bond (connecting two neighbouring nodes) according to the following rules: (i) if the bond connects two pixels occupied by “Au matter”, then a finite resistance R is associated to the bond; (ii) if any of the neighbouring nodes connected by the bond lies in a “void” pixel, then an infinite value of resistance is used. The electrical network obtained by this approach is analysed by solving the Kirchoff circuit equations. 54 More specifically, calculations are performed using a matrix formulation reading in the form , where is the conductance matrix as obtained from the actual resistive network implemented as above, while and are the nodal potential and the bond current matrix, respectively. The solution of such a linear system straightforwardly provides the current and potential maps over the simulated sample. In addition, the Joule dissipated power in each resistor and the electric field in the “void” pixels are as well calculated. As a validation of our model, we preliminary performed a systematic analysis of the static electrical properties ( i.e. no evolution occurring) of a large set of model systems with different coverage, here defined as the ratio between the number of pixels representing matter and void, respectively (see ESI † ). The various model systems have been elaborated from a real sample by adopting different binarization processes of the original SEM micrographs. This procedure is beneficial in tuning the resulting connectivity of the model 2D granular system. \n Fig. 6a shows the behaviour of the overall conductance as a function of the coverage. The results provide robust evidence of the percolative behaviour: very low conduction is found in low-coverage systems, while beyond the percolation threshold, occurring at a coverage 0.54, the conductance of the system rapidly increases. As expected, the maximum value is found only at maximum coverage: this situation indeed corresponds to the case of homogeneous ( i.e. non-granular) metal conductor. Fig. 6 (a) Percolation curve of conductance as a function of the coverage value. In the inset, images of some structural configurations are depicted. (b) Resistance as a function of simulation steps in a system of 50 × 70 pixels; the resistance values at the time steps corresponding to the shown configurations are highlighted. (c–f) The structure (c) and the maps of electrostatic potential (d), current (e) and dissipated power (f) in a system of 50 × 70 pixels, (coverage = 0.64). Values at three unlike microstructure arrangements are shown. Modifications in the structural parameters determine the continuous creation and destruction of percolative paths. A deeper understanding of conduction processes can be achieved by investigating the electrostatic potential, current and dissipated power through the space maps (see ESI † ). It is here evident how in very dilute systems conduction is suppressed due to absence of percolative paths within the sample. By increasing coverage, paths start to form and conductance rapidly increases. Finally, for very large values of coverage, the transport is mostly diffusive and current tends to flow uniformly along the sample. Starting from the real sample of Fig. 1 , we simulated the evolution of the system as described above. The resulting resistance is shown in Fig. 6b during the system evolution scanned by the synthetic events of modification/breaking of inter-cluster contacts described above. The system presents a value of coverage of 0.62, well beyond the percolation threshold for our systems. A resistive switching behaviour is indeed observed with the resistance assuming different values during the evolution of the contact network in the system, characterized by single variations opposite in sign. The dimension of the binary image used for the above analysis was 70 × 50 pixels: dense enough to capture the main morphological features, while resulting in an affordable computational cost. We can now have a look at the maps of the electrostatic potential, current and dissipated power. In Fig. 6d–f these quantities are shown for three different sequential arrangements (corresponding to the three columns) collected during the system evolution under bias. An evolving network of current paths is observed, inducing rapid oscillations in the instantaneous values of the net through current and total resistance. It is important to remark that such evolving behaviour does not qualitatively change upon a modification of the coverage (beyond the percolation threshold) and construction and destruction probabilities (pC and pD). This makes us confident that the present computer model is trustworthy and the corresponding physical picture reliable and robust. This model is too schematic to aim to quantitatively reproduce the electrical features of the cluster-assembled film, however it provides indications about the possible mechanisms of the observed behaviour. First, we highlight that the overall evolution of the resistance is reproduced as shown in Fig. 7a , where simulated and experimental normalized data are compared. Both the experiment and the model show discrete and reversible change in the resistance under a constant voltage bias and switching events. Both the sample and the model explore different resistance levels that are quite comparable in amplitude relatively to the initial resistance value and in frequency, as shown by the histogram in Fig. 7b , where two different peaks are clearly observed with a broadening determined by the switching events. This demonstrates that local random events of breaking/reconnection in the resistance network can determine a global behaviour of the systems like that observed for the cluster-assembled film. Fig. 7 (a) Top: the resistance as function of simulation steps (the same data as in Fig. 6b ) normalized to the initial resistance of a system 50 × 70 pixels; bottom: resistance evolution normalized to the initial resistance of the cluster-assembled film under a bias of 5 V as function of time. (b) The histogram of the resistance values both for the cluster-assembled film and the simulation. We highlight also some differences: cluster-assembled films explore a larger number of resistance level ( Fig. 4 ) compared to simulations and experimental resistance patterns are characterized by the presence of a larger number of switches. This reflects in the histogram that usually shows more distinct peaks with different broadening (see Fig. 3b and 7b ). The higher complexity in the morphology of the cluster-assembled films causes different events of local breaking and reconstruction with different degree of influence on the global electrical resistance. The computational cost to simulate complex networks with a higher number of resistors and for a larger time interval, under different voltage bias is prohibitive. Anyway, this model gives a reasonable insight for the understanding of the resistive switching behaviour and how the nanostructure in a complex metallic film can give rise to the observed electrical properties."
} | 9,004 |
37230971 | PMC10212962 | pmc | 7,710 | {
"abstract": "Memristors, a cornerstone for neuromorphic electronics, respond to the history of electrical stimuli by varying their electrical resistance across a continuum of states. Much effort has been recently devoted to developing an analogous response to optical excitation. Here we realize a novel tunnelling photo-memristor whose behaviour is bimodal: its resistance is determined by the dual electrical-optical history. This is obtained in a device of ultimate simplicity: an interface between a high-temperature superconductor and a transparent semiconductor. The exploited mechanism is a reversible nanoscale redox reaction between both materials, whose oxygen content determines the electron tunnelling rate across their interface. The redox reaction is optically driven via an interplay between electrochemistry, photovoltaic effects and photo-assisted ion migration. Besides their fundamental interest, the unveiled electro-optic memory effects have considerable technological potential. Especially in combination with high-temperature superconductivity which, in addition to facilitating low-dissipation connectivity, brings photo-memristive effects to the realm of superconducting electronics.",
"introduction": "Introduction The search for faster, energy-efficient memories and novel computation schemes has fostered the exploration of resistive switching effects 1 . Observed in a variety of systems that span from magnetic 2 , 3 or ferroelectric 4 , 5 tunnel junctions to transition-oxide capacitors 6 , 7 and strongly correlated materials 8 – 10 , the term resistive switching denotes a “jump” between non-volatile electrical resistance states (high/low resistance, 0/1 for logics) generally triggered by a voltage or a current pulse. Memristors 11 constitute a particular class of two-terminal resistive switching devices whose functionality is beyond that of a binary memory: they show a continuum of states - instead of only two - and the switching between them is driven by the time-integrated current across the device (or by the history of applied voltages). Thus, one can think of memristors as multi-state memories switchable by cumulative stimuli. Depending on the resistance states’ lifetime and dynamics, memristors can mimic the function of either synapses or neurons 11 , 12 , thus constituting a cornerstone for the nascent field of neuromorphic computing 13 – 16 . A related, tantalizing idea is the development of memristors sensitive to light 17 – 25 −particularly, in which illumination can trigger a switching across non-volatile resistance states. This type of optical memory may be game-changing in applications such as photonic neural networks 26 and open the door to novel optoelectronic devices 27 , 28 −for example, neuromorphic vision sensors 21 , 29 . Various systems have been recently explored in search of photomemristive effects. Those include complex nanostructures based on optically active polymers 17 , metal-oxide capacitors 21 , 22 , 24 , all-oxide 19 and semiconductor 18 heterostructures, as well as ferroelectric tunnel junctions 23 among others 30 , 31 . Their conductance shows photosensitivity due to mechanisms that span from light-induced polymer contraction/expansion 17 and electron trapping/detrapping 18 , 19 , 21 , 22 , 24 to photoinduced ferroelectric switching 23 . In those systems, the resistance states’ lifetimes range from typically seconds in electronic processes 18 , 19 , 21 , 22 , 24 to minutes in organic systems 17 , up to the virtual non-volatility of ferroelectric devices 23 . In addition to the common challenges associated with memristors, e.g. obtaining large resistance variations to facilitate readout and finding geometries enabling the high connectivity required for neuromorphic circuits 11 , 14 – 16 , photo-memristors pose additional specific ones. Namely, in memristors, electrical excitation often allows for bidirectional switching −the polarity of the electrical stimulus determines whether the resistance increases or decreases. Such property is crucial for mimicking the so-called depression and potentiation phenomena characteristic of synaptic plasticity 14 , 15 , 17 , 18 . An analogous function is generally absent in the optical response (in most cases illumination only produces a resistance decrease 19 , 21 – 23 ) except for a few realizations that require the combination of various light sources. For instance, in polymer-based devices the opposite effects of circularly and linearly polarized light respectively lead to a resistance increase or decrease, thus mimicking depression and potentiation 17 . Illumination under variable wavelength has also been exploited to that end, particularly in systems based on electron trapping/detrapping 24 , in which the natural relaxation of visible-light excited conductance states is potentiated via infrared illumination, thus allowing for a virtually bidirectional optical switching. Here we report on a new class of photo-memristor that exploits a distinct microscopic mechanism: a controllable oxygen exchange between the two materials that constitute the device –a superconducting cuprate and a semiconducting oxide. Crucially, such a mechanism allows for giant resistance-switching effects that can be driven both optically and electrically. Indeed, a remarkable specificity of our ionic photo-memristor is that its response to a given optical stimuli depends on the electrical history. Due to this entanglement, and at variance with other approaches, a single light source can controllably produce bidirectional switching. This behaviour results from an unusual, competing interplay between electrochemistry, photon-activated oxygen diffusion and strong photovoltaic effects. Interestingly, the present demonstration is based on a high-temperature superconducting cuprate. As further discussed below, although superconductivity is not a necessary ingredient for the photo-memristive effects, it greatly multiplies their technological potential, not only because it enhances resistive switching effects and facilitates the connectivity required for neuromorphic circuits, but also because electro-optical memory is a novel, game-changing function in the thriving field of superconducting electronics 32 – 37 . The key specificities of the electro-optical switching behaviour observed here are schematically summarized in Fig. 1 . The application of voltage pulses \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e (of the order of a few Volts) enables a conductance switching \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle {G}_{E}$$\\end{document} △ G E across a continuum of levels, whose non-destructive readout is possible with a much lower \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{read}}$$\\end{document} V r e a d of the order of mV. The conductance levels span between two extreme states −hereafter called ON (high conductance) and OFF (low conductance) − that can be up to four orders of magnitude afar. The electrical switching is hysteretic, bidirectional, and reversible. The ON state and intermediate resistance levels are metastable, and their lifetime depends on temperature: they are virtually non-volatile at low T (tens of K), and slowly relax into the OFF state at higher T , at a rate that increases with temperature. Overall, this behaviour is reminiscent ( mutatis mutandis ) of the tunnelling electroresistance (TER) of ferroelectric tunnel junctions 23 , 38 – 40 . Strikingly, illumination with visible or UV light also leads to conductance switching, and it does so in a very characteristic fashion: the conductance level shifts in opposite directions depending on the previous electrical junction’s state. Namely, optical stimuli lead to either an enhancement or a decrease in the conductance \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle {G}_{{Op}}$$\\end{document} △ G O p depending on whether the device had been previously set in (or nearby) the OFF or ON state. While the electric history determines the sign of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle {G}_{{Op}},$$\\end{document} △ G O p , crucially the amplitude of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle {G}_{{Op}}$$\\end{document} △ G O p cumulatively depends on the number of photons shone on the device, that is, it is controlled by the optical history. In addition to the electro-optical switching summarized in Fig. 1 , the devices studied here show unusually large photovoltaic effects which, as discussed thereafter, play a key role in the photo-memristive behaviour. Fig. 1 Dual optical-electrical conductance switching. Sketch and experimental data of a typical hysteresis loop showing the differential conductance measured on YBCO/ITO tunnel junctions as a function of the writing pulse voltage. The black arrow indicates the maximum amplitude of the electrical switching \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\triangle G}_{E}$$\\end{document} △ G E . The red and blue arrows respectively indicate the amplitude of the optical switching \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\triangle G}_{{Op}}$$\\end{document} △ G O p in the OFF and ON states respectively. Optical stimuli drive the junction towards intermediate states. Junction’s fabrication and interface characterization The scheme of the photo-memristor is shown in Fig. 2a . It consists of a micrometric junction between the archetypal high-T C superconductor \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{YB}}}}}}{{{{{{\\rm{a}}}}}}}_{2}{{{{{\\rm{C}}}}}}{{{{{{\\rm{u}}}}}}}_{3}{{{{{{\\rm{O}}}}}}}_{7-{{{{{\\rm{\\delta }}}}}}}$$\\end{document} YB a 2 C u 3 O 7 − δ (YBCO, bottom electrode) and the transparent semiconductor indium-tin-oxide (ITO, top electrode). The junctions are fabricated on 30 nm thick c-axis oriented YBCO films grown epitaxially on (001) \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{SrTiO}}}}}}}_{3}$$\\end{document} SrTiO 3 (STO) substrates. A \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\sim$$\\end{document} ~ 1 μm-thick photoresist is spin-coated on the YBCO film and micrometric openings are photolithographed. The ITO subsequently deposited contacts the YBCO surface across the resist openings, forming the junctions (for further details see 41 and Methods). The as-grown YBCO/ITO interface was characterized by Scanning Transmission Electron Microscopy (STEM) and Electron Energy Loss Spectroscopy (EELS), displayed in Fig. 2b–d . The atomic resolution Z-contrast STEM image (Fig. 2b ) demonstrates highly epitaxial YBCO growth on the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${SrTi}{O}_{3}$$\\end{document} S r T i O 3 (001) substrate. The ITO layer displays an amorphous matrix with nanocrystalline regions of a few tens of nm in size and different crystalline orientations. Evidence of oxygen exchange between both materials is provided by the EELS analysis. On the one hand, the Cu L 2,3 edge shows a chemical shift towards higher energies at the interface with ITO (Fig. 2c ), which indicates 42 , 43 a local decrease of the Cu oxidation state (electron doping). On the other hand, and consistently, the analysis of the oxygen K edge (Fig. 2d ) reveals that the pre-peak feature (highlighted by the Gaussian fits) decreases in intensity when moving towards the interface, indicating a reduced hole carrier density 44 . In summary, the EELS analysis of both the Cu L 2,3 and O K edges indicate electron doping of the interfacial YBCO, which is consistent with local oxygen depletion. That is as we observed earlier in YBCO/Mo 80 Si 20 junctions 41 , and is explained by the high reduction potential of copper in YBCO, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${E}_{0}\\left({Cu}\\right)=2.4$$\\end{document} E 0 C u = 2.4 \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{V}}}}}}$$\\end{document} V , as compared to that of indium in ITO, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${E}_{0}({Cu})=-0.49$$\\end{document} E 0 ( C u ) = − 0.49 \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{V}}}}}}$$\\end{document} V (Supplementary Table 1 ). From this, a spontaneous reduction of the interfacial YBCO is expected upon deposition of ITO. Based on the profiles in Fig. 2c , the oxygen depletion in YBCO extends over \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\sim$$\\end{document} ~ 5-6 nm from the interface. According to earlier experiments 41 , the doping gradient observed in Fig. 2c yields a gradient of physical properties as sketched in Fig. 2a (under the “OFF” label), having insulating YBCO at the interface, followed by YBCO with depressed superconducting properties (low T c YBCO) and by optimal oxygenation (and T C ) farther from the interface. As we discuss in the manuscript, the conductance switching triggered by applying \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e or by illumination reflect changes in the oxygen content at both sides of the YBCO/ITO interface. Fig. 2 Junction’s structure and electrical switching. a Scheme of the photo-memristor composed of a c-axis \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{Y}}}}}}{{{{{{\\rm{Ba}}}}}}}_{2}{{{{{{\\rm{Cu}}}}}}}_{3}{{{{{{\\rm{O}}}}}}}_{7-{{{{{\\rm{\\delta }}}}}}}$$\\end{document} Y Ba 2 Cu 3 O 7 − δ /ITO tunnel junction grown on \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{SrTiO}}}}}}}_{3}$$\\end{document} SrTiO 3 (001). The micrometric junction is defined by depositing ITO into an opening through the insulating photoresist. Sketches of the YBCO/ITO interface in the OFF and ON states are displayed below. Solid and hollow circles respectively represent oxygen atoms and vacancies. b Atomic resolution Z-contrast STEM image of a YBCO/ITO bilayer grown on STO. c Cu L 3 peak position along the 30 nm YBCO thin film shows a chemical shift towards higher energies at the ITO interface. d Electron energy loss spectra at the O K edge along the YBCO thin film. Gaussian fits of the oxygen pre-peak are also shown. The EELS intensity is normalized and shifted vertically to show the peak variation along the line scan. The different spectra are obtained from a line scan and correspond to 6 nm averages of the O K edge centred at (from top to bottom) 3, 9, 15, 21 and 27 nm from the interface. e Differential conductance \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$G\\equiv {dI}{{{{{\\boldsymbol{/}}}}}}d{V}_{{read}}$$\\end{document} G ≡ d I / d V r e a d as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{read}}$$\\end{document} V r e a d after applying a voltage pulse \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}( \\sim {{{{{\\rm{V}}}}}})$$\\end{document} V w r i t e ( ~ V ) at T = 3.2 K to access the ON \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$(V_{write}\\, > \\,0)$$\\end{document} ( V w r i t e > 0 ) and OFF \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$(V_{write}\\, < \\,0)$$\\end{document} ( V w r i t e < 0 ) states. The inset displays the temperature dependence of the differential conductance at zero-bias \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$G_{0}$$\\end{document} G 0 (squares) and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V_{read}\\,=\\,100 \\,{{{{{\\rm{mV}}}}}}$$\\end{document} V r e a d = 100 mV at \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$G_{100}$$\\end{document} G 100 (circles) in the ON (solid) and OFF (hollow) states. At zero bias ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$G_0,$$\\end{document} G 0 , squares) the conductance drops at a higher pace below a certain temperature which is close to the superconducting transition. f Hysteretic behaviour of the differential conductance as a function of the writing voltage \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V_{write}$$\\end{document} V w r i t e for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V_{read}\\,=\\,100 \\,{{{{{\\rm{mV}}}}}}$$\\end{document} V r e a d = 100 mV ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{100},$$\\end{document} G 100 , grey, left scale) and for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V_{read}\\,=\\,0$$\\end{document} V r e a d = 0 ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$G_0,$$\\end{document} G 0 , black, right scale), showing the maximum electrical switching amplitude \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle {G}_{E}$$\\end{document} △ G E . The junction was cycled twice from positive to negative voltages as shown by the spinning arrow. g Relaxation of the normalized conductance switching \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\Delta G_{E} (t)/G_{on}(t\\,=\\,0)$$\\end{document} Δ G E ( t ) / G o n ( t = 0 ) as a function of time after the junction is set in the ON state ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V_{write}\\,=\\,6\\,{{{{{\\rm{V}}}}}}\\, {{{{{\\rm{at}}}}}}\\, {{{{{\\rm{T}}}}}}\\,=\\,3.2\\,\\,{{{{{\\rm{K}}}}}}$$\\end{document} V w r i t e = 6 V at T = 3.2 K ) measured at different temperatures (see legend), along with the best fits to a stretched exponential as discussed in Supplementary Fig. S4 .",
"discussion": "Discussion Origin of electrical switching effects The electrical switching behaviour observed in Fig. 2 is analogous to that of YBCO/Mo 80 Si 20 41 and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{NdNi}}}}}}{{{{{{\\rm{O}}}}}}}_{3}$$\\end{document} NdNi O 3 /Mo 80 Si 20 47 junctions, for which we demonstrated that the underlying mechanism is a voltage-driven oxygen exchange between the junctions’ electrodes. Sketches of the interface in the ON and OFF state are shown in Fig. 2a . OFF is the ground state, in which YBCO is severely oxygen depleted at the interface due to the reduction of YBCO in favour of ITO oxidation, as it is demonstrated by the STEM and EELS analysis of the pristine YBCO/ITO interface (Fig. 2b–d ). The oxygen-depleted YBCO layer is expectedly semiconducting with a 1.25 eV gap 51 , and thus behaves as a tunnelling barrier under low voltage bias (a hundred meV). That layer’s thickness is estimated at 5 ± 1 nm from the fits of the differential conductance to the BDR model for electron tunnelling (see supplementary Fig. S1 ). That is consistent with the STEM-EELS observations in Fig. 2b–d . The presence of this relatively thick oxygen-depleted YBCO yields the lowest conductance state (hollow symbols in Fig. 2e ). The application of a sufficiently high \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}} > 0$$\\end{document} V w r i t e > 0 , comparable to the different between electrochemical potentials \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle {E}_{0}={E}_{0}\\left({Cu}\\right)-{E}_{0}\\left({In}\\right)\\approx 2.9\\,{{{{{\\rm{eV}}}}}},$$\\end{document} △ E 0 = E 0 C u − E 0 I n ≈ 2.9 eV , reverses the redox reaction, driving oxygen back into the interfacial YBCO and hence thinning down the oxygen-depleted YBCO layer 41 . This increases the conductance by orders of magnitude, leading to the ON state (solid symbols curve in Fig. 2e ). The system can be driven again into the OFF state by the application of a negative voltage. Notice that the magnitude of the voltage required to trigger the ON→OFF switching is lower than for the OFF→ON one, since for the former the underlying reaction is spontaneous from an electrochemical standpoint. Thus, the finite \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}} < 0$$\\end{document} V w r i t e < 0 is required only to accelerate oxygen ion diffusion. This is consistent with the spontaneous ON →OFF relaxation observed in the presence of thermal activation (Fig. 2g and supplementary Fig. S4 ). It is worth noting that the activation energy extracted from the relaxation analysis \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${E}_{a} \\sim$$\\end{document} E a ~ 0.1 eV is well in the range of the values found in oxygen exchange reactions in YBCO 52 , which further supports the redox scenario. Another piece of evidence supporting that the large conductance switching is produced by oxygen exchange between YBCO and ITO is that it is suppressed when ITO is replaced by a material which does not oxidize (see supplementary Fig. S3 ). Origin of the photovoltaic effect The key to understanding this effect, sketched in Fig. 5 , is that the strongly oxygen-depleted YBCO is a p-type semiconductor with a \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\sim$$\\end{document} ~ 1.25 eV gap 51 , and the adjacent ITO is a degenerate n-type semiconductor with a \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\sim$$\\end{document} ~ 3.5 eV gap 53 . Thus, the junction interface can be seen as a p-n junction. This is supported by the diode behaviour observed in the I(V) of the junctions under high voltage bias (see supplementary Fig. S10 ). In consequence, a photovoltaic effect is naturally expected. The different work functions \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\phi }_{{ITO}}=4.5$$\\end{document} ϕ I T O = 4.5 eV 54 and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\phi }_{{YBCO}}\\approx 5.5-6$$\\end{document} ϕ Y B C O ≈ 5.5 − 6 eV 55 leads to a space charge layer (SCL) with a built-in electric field pointing towards the YBCO, as represented Fig. 5a . In this scenario, upon illumination, electrons and holes photogenerated in the SCL respectively flow towards the YBCO and the ITO, leading to the observed photocurrent (photovoltage in open-circuit configuration). We can obtain a rough estimate of the SCL thickness using 56 \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$W=\\sqrt{\\left({N}_{A}{+N}_{D}/{N}_{A}{N}_{D}\\right){2\\varepsilon }_{s}{V}_{{bi}}{\\backslash e}}$$\\end{document} W = N A + N D / N A N D 2 ε s V b i \\ e where \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${N}_{A} \\sim 8\\,1{0}^{19}{{{{{\\rm{c}}}}}}{{{{{{\\rm{m}}}}}}}^{-3}$$\\end{document} N A ~ 8 1 0 19 c m − 3 and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${N}_{D} \\sim 5\\,1{0}^{20}{{{{{\\rm{c}}}}}}{{{{{{\\rm{m}}}}}}}^{-3}$$\\end{document} N D ~ 5 1 0 20 c m − 3 are respectively the carrier densities in oxygen-depleted YBCO 57 and in ITO 58 , and an upper limit for built-in voltage \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{bi}} < 1.5\\,{{{{{\\rm{V}}}}}}$$\\end{document} V b i < 1.5 V can be inferred from the difference between work functions. From this, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$W < 5\\,{{{{{\\rm{nm}}}}}}$$\\end{document} W < 5 nm which is consistent with the fact that the conduction is dominated by electron tunnelling in the low-bias regime, as well as with the tunnel barrier thickness \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\sim$$\\end{document} ~ 5 nm obtained from the fitting to the BDR 45 model (supplementary Fig. S1 ). The scaling of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{oc}}$$\\end{document} V o c with the logarithm of the junction’s conductance state \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${G}_{100}$$\\end{document} G 100 shown in Fig. 4b is also consistent with the discussed scenario. Indeed, if one considers that the current across the junction and the conductance are roughly proportional in the low \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{read}}$$\\end{document} V r e a d regime, the behaviour of Fig. 4d is as expected from the Shockley relation 59 \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{oc}}=\\frac{{k}_{b}T}{\\eta q}{{{{{\\rm{ln}}}}}}(\\frac{{I}_{{sc}}}{{I}_{0}}-1)$$\\end{document} V o c = k b T η q ln ( I s c I 0 − 1 ) where \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\eta$$\\end{document} η is the ideality factor, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$q$$\\end{document} q is the electric charge, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{sc}}$$\\end{document} I s c is the photocurrent and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{0}$$\\end{document} I 0 is the reverse saturation current in the dark. This explains why \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{oc}}$$\\end{document} V o c is the largest in the low conductance (OFF) state, in which \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{0}$$\\end{document} I 0 is the lowest, as well as the logarithmic decay of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{oc}}$$\\end{document} V o c as the conductance (and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\,{I}_{0}$$\\end{document} I 0 ) increases. Notice finally (Supplementary Fig. S8 ) that a finite \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{oc}}$$\\end{document} V o c appears for wavelengths in the range 365–400 nm (3.4–3 eV), i.e. for photon energies above the YBCO charge transfer gap. This is as expected since metal-ligand charge transfer excitations into CuO 2 plane states at energy losses >3 eV have been reported in oxygen-depleted YBCO 60 . Fig. 5 Microscopic Model. a Energy band diagram of the YBCO/ITO interface where ITO (blue) is a degenerate n-type semiconductor and oxygen-depleted YBCO (green) is a p-type semiconductor. A space charge layer, characterized by a built-in electric field at the interface, is indicated by the light green region. \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\triangle }_{{ITO}}$$\\end{document} △ I T O and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\triangle }_{{YBCO}}$$\\end{document} △ Y B C O represent the electronic band gaps of ITO and YBCO respectively. b , c Schematic representation of the ITO/YBCO interface during optical switching. The full and hollow circles indicate oxygen atoms and vacancies respectively. In the OFF state, see b , the interfacial YBCO is highly oxygen deficient. Conversely, the ITO oxygen content is optimal. Upon illumination, the photovoltage V OC leads to an accumulation of holes (red crosses) in the interfacial YBCO, which promotes local oxidation by migration of O 2− atoms. When the junction is set in the ON state, see c , there is no photovoltage. Oxygen has been driven from the ITO into the YBCO, leaving an interfacial, oxygen-depleted ITO. This is an out-of-equilibrium state because ITO has a lower reduction potential than YBCO. Thus, oxygen tends to migrate into ITO, leading to a natural relaxation from the ON towards the OFF state. This is accelerated by illumination. Origin of optical switching effects After the junction is set in the ON state by the application of a positive \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e , illumination leads to a strong conductance decrease (Fig. 3a ) which is persistent at low temperatures. To understand this behaviour, we start by recalling that the ON state (sketch in Fig. 2a ) is out of equilibrium from the electrochemical point of view, since the material with the lowest reduction potential (ITO, see Supplementary Table 1 ) presents a strong oxygen deficiency near the interface while most of top YBCO layers are fully oxygenated. This explains why the ON state naturally relaxes towards a lower conductance one, as demonstrated in Fig. 2g and Supplementary Fig. S4 . Microscopically, relaxation occurs as oxygen migrates back into ITO, driving the system towards its ground state, as shown Fig. 5c −with the interfacial YBCO oxygen-depleted in favour of ITO oxidation (see STEM in Fig. 2b, c ). However, this requires overcoming the barrier for ion diffusion. While this is thermally activated in the dark (see Supplementary Fig. S4 ), at low temperatures (below ~ 90 K) oxygen mobility is low and the relaxation is very slow, making the ON state virtually non-volatile (Fig. 2g ). Upon illumination, the relaxation is dramatically accelerated: as demonstrated by Supplementary Figure S5 , illumination makes the system relax as it would do \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\sim$$\\end{document} ~ 200 K above the actual temperature. That is, illumination activates ion diffusion, expectedly due to phonon excitation by light absorption. This is consistent with the fact that the photoinduced conductance decrease does not show a marked wavelength dependence in the visible range (Supplementary Figure S6 ), and that IR light produces strong effects (Supplementary Figure S7 ). As oxygen migrates from YBCO into ITO thanks to optical activation (see the sketch in Fig. 5c ), the severely oxygen-deficient YBCO layer at the interface thickens, leading to the decrease of the tunnelling conductance (Fig. 3a )—which, at low T, is stopped as soon as illumination is halted. In summary, the photoinduced conductance decrease in the ON state reflects light-activated oxygen mobility. It is worth noting that this mechanism was proposed earlier in the literature of cuprates 61 , 62 to explain a different type of persistent photoconductivity in thin films. Once the junction is set in the OFF state by a negative \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e , the interfacial YBCO is severely oxygen depleted and ITO is oxygen-rich (sketch in Fig. 2a ). In this state, and contrary to the ON state, a conductance enhancement is observed after illumination (Fig. 3b ). That is explained by a key specificity of the OFF state: the large positive photovoltage \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{OC}}$$\\end{document} V O C measured during illumination (see Fig. 4 ). Indeed, a causal relationship exists between \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{OC}}$$\\end{document} V O C during illumination and the conductance enhancement observed after illumination. This is documented in the supplementary Figs. S7 and S8 , which demonstrate that the conductance enhancement is observed only in the presence of a photovoltage \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{OC}}$$\\end{document} V O C , and scales with its size. As discussed above and sketched in Fig. 5a , the photovoltage reflects a hole accumulation at the YBCO side of the space charge layer. This accumulation changes the doping of the interfacial YBCO, locally enhancing the number of holes per copper atom and thus taking the oxidation state Cu +2 + h + →Cu +3 beyond the level corresponding to the actual oxygen content. This favours the re-oxygenation of the interfacial YBCO by migration of O 2− ions (sketch in Fig. 5b ). This mechanism is analogous to the well-known photocatalytic effect produced as photocarriers promote redox reactions in the vicinity of pn-junctions 63 , 64 . O 2− ions can migrate from the oxygen-rich interfacial ITO, where the photo-induced electron accumulation (Fig. 5a ) allows O 2− to be released without changing the In oxidation state (In +3 + e − → In +2 ). Notice that, consistently, the O 2− migration from ITO is also favoured by the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{OC}}$$\\end{document} V O C polarity, which is the same required to trigger that migration by application of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e pulses. In addition, we cannot discard that some O 2− ions diffuse into the interfacial YBCO from deeper YBCO layers that present higher oxygen content, given the sharp gradient of oxygen content expected from \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e < 0. In summary, the positive photovoltage drives oxygen ions towards the oxygen-depleted interfacial YBCO, similarly as an applied \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e > 0 does. Re-oxygenation leads to a thinning of this oxygen-depleted YBCO layer, which is the electron tunnelling barrier, thus enhancing the electrical conductance (Fig. 3b ). To summarize, light promotes oxygen migration, either in or out of the interfacial YBCO depending on whether the junction is previously set in the ON or the OFF state by \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{write}}$$\\end{document} V w r i t e pulses. Oxygen migration is favoured by optical activation over the barrier for ion diffusion and driven either by (i) (ON state) the redox reaction dictated by the higher reduction potential of YBCO when it is oxygen-rich and ITO is oxygen deficient; or (ii) (OFF state) by the positive photovoltage and ensuing hole/electron accumulation at both sides of the interface, which reverses the redox reaction. While (i) and (ii) are obviously antagonistic, (ii) is not present in the ON state and becomes relevant gradually as the junction is driven into the OFF state and the photovoltage \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{OC}}$$\\end{document} V O C increases (Fig. 4b ). This explains why light cannot drive the junction all the way from the ON into the OFF state (nor vice versa), but only into an intermediate level that depends on the predominance of either (i) or (ii), their balance being controlled by temperature (Fig. 4d ) and/or the optical wavelength (supplementary Figs. S6 , S7 and S8 ). Aside from their fundamental interest, the described effects are also technologically relevant for various reasons. In addition to the unique photo-memristive behaviour, the devices studied here −made of a single interface− are of ultimate simplicity when compared to approaches that involve more complex geometries 17 , 20 and photo-active materials 17 , 23 . It is worth noting that photo-memristive effects similar to the ones reported here for superconducting YBCO should be observed if this is replaced by other conducting oxides, provided that the junction is formed by materials having a different reduction potential (to promote oxygen exchange) and that a p-n junction is formed at the interface (allowing for photovoltages). This is the case, for example, of junctions based on \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{NdNi}}}}}}{{{{{{\\rm{O}}}}}}}_{3}$$\\end{document} NdNi O 3 , for which we have recently reported 47 electrical switching effects analogous to those demonstrated here. The fact that the hereby demonstrated photo-memristive effects are based on a high-temperature superconductor enhances their technological potential. First, while the switching effects are observed at any temperature, they become much stronger below the superconducting T C . This, together with the vanishing Joule dissipation, expectedly facilitate miniaturization and the layout of dense memristor arrays with high interconnectivity, as required for neuromorphic computing 13 , 15 —an emergent area also within superconducvitiy 65 , 66 . Furthermore, the photo-memristors realized here could be naturally implemented in conventional superconducting electronics 32 , a thriving field, particularly after the advent of Josephson devices based on high-temperature cuprates 33 – 35 as the one used here, which allow operation above liquid nitrogen temperatures. One could for instance cite applications such as quantum antennas 36 and logic circuits 37 , in which non-volatile memory is a grail 67 and photo-sensitivity was up to now unavailable. The incorporation of these functions should refashion and greatly broaden the range of applications of such Josephson circuits."
} | 13,253 |
36945052 | PMC10031880 | pmc | 7,718 | {
"abstract": "Background Candidate phyla radiation (CPR) constitutes highly diverse bacteria with small cell sizes and are likely obligate intracellular symbionts. Given their distribution and complex associations with bacterial hosts, genetic and biological features of CPR bacteria in low-nutrient environments have received increasing attention. However, CPR bacteria in wastewater treatment systems remain poorly understood. We utilized genome-centric metagenomics to answer how CPR communities shift over 11 years and what kind of ecological roles they act in an activated sludge system. Results We found that approximately 9% (135) of the 1,526 non-redundant bacterial and archaeal metagenome-assembled genomes were affiliated with CPR. CPR bacteria were consistently abundant with a relative abundance of up to 7.5% in the studied activated sludge system. The observed striking fluctuations in CPR community compositions and the limited metabolic and biosynthetic capabilities in CPR bacteria collectively revealed the nature that CPR dynamics may be directly determined by the available hosts. Similarity-based network analysis further confirmed the broad bacterial hosts of CPR lineages. The proteome contents of activated sludge-associated CPR had a higher similarity to those of environmental-associated CPR than to those of human-associated ones. Comparative genomic analysis observed significant enrichment of genes for oxygen stress resistance in activated sludge-associated CPR bacteria. Furthermore, genes for carbon cycling and horizontal gene transfer were extensively identified in activated sludge-associated CPR genomes. Conclusions These findings highlight the presence of specific host interactions among CPR lineages in activated sludge systems. Despite the lack of key metabolic pathways, these small, yet abundant bacteria may have significant involvements in biogeochemical cycling and bacterial evolution in activated sludge systems. \n Video Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s40168-023-01494-1.",
"conclusion": "Conclusion Overall, our longitudinal metagenomic analyses demonstrate the high bacterial diversities and abundances of CPR bacteria in AS system. The observed striking fluctuations in CPR communities and similarity-based networks collectively reveal that CPR bacteria might interact with multiple bacterial hosts with a specific taxonomic distribution. The limited metabolic and biosynthetic capabilities are also observed in the CPR guilds in AS systems, suggesting that the dynamics of CPR are directly driven by the available hosts. Although the studied WWTP was designed to treat municipal wastewater, proteome content similarities between human-associated and environmental CPR MAGs indicate that dominant CPR bacteria (Saccharibacteria) in AS systems may originate from environmental samples. Furthermore, the significantly enriched genes for oxygen stress resistance in AS CPR MAGs might enable them to become more adapted to high oxygen conditions compared to the human-associated ones, supporting the observed niche differentiation of CPR bacteria in AS system. Notably, our analyses highlight that CPR bacteria in AS might have involvements in carbohydrate hydrolysis and fermentation in wastewater treatment systems, as well as affect bacterial evolution via horizontal gene transfer. As the findings of this study are obtained based on in silico analyses, the real host range of CPR bacteria in AS systems and how they interact with their bacterial hosts are required to be answered in future investigations.",
"introduction": "Introduction Advances in metagenomic techniques have enabled a more efficient pathway to get the missing pieces in the intricate puzzle of the tree of life. The discoveries of bacterial and archaeal genomes represented previously unknown lineages have constantly broadened microbial diversity and amended the phylogenetic structure of the tree of life, pushing the common ancestor of bacteria and archaea deeper into the past [ 1 , 2 ]. Among these newly reported bacterial lineages, a group of candidate phyla of mainly uncultivated bacteria that have been identified with metagenomics forms monophyletic radiation [ 3 – 5 ]. This radiation was defined as candidate phyla radiation (CPR, also referred to as Patescibacteria). In addition to the important role of CPR in studying bacterial and cellular evolutionary history, CPR has been inferred to represent > 15% of all bacterial diversity and contains over 70 different phyla [ 4 ]. The widely reported candidate phyla, e.g., Saccharibacteria (TM7), Parcubacteria (OD1), Gracilibacteria (BD1-5), and Microgenomates (OP11), are assigned to the CPR phylum (Patescibacteria) and placed into class-level in Genome Taxonomy Database (GTDB) [ 6 ]. These four CPR classes represent > 75% of all reported CPR members in GTDB. While the first CPR organism ( Candidatus Nanosynbacter lyticus strain TM7x) was co-cultivated from human oral [ 7 ], 16S rRNA gene sequence analyses have demonstrated that CPR bacteria could be found in a wide range of environments, including terrestrial [ 3 , 4 ], freshwater [ 2 , 8 – 10 ], and marine [ 11 , 12 ] ecosystems. CPR bacteria predominate in groundwater, lake, and other aquifers with limited nutrients and oxygen [ 13 , 14 ]. However, when considering the fact that a single copy 16S rRNA gene is typical for CPR lineages [ 4 ], the relative abundance of CPR organisms estimated using 16S rRNA gene sequences without gene copy number correction may be underestimated. Genome-resolved metagenomics provides new insights into the ecological distribution and roles of CPR organisms. He et al. [ 15 ] recovered 540 CPR bacterial genomes from groundwater metagenomes and demonstrated that CPR organisms accounted for up to ~ 40% of microbial communities of groundwater (bulk biomass onto 0.1 µm filter). Compared with the groundwater environment, the activated sludge (AS) system in wastewater treatment plants (WWTPs) is a eutrophic and aerobic engineered ecosystem with much higher biomass concentration and complex microbial diversity. Despite the documentation of CPR bacterial genomes recovered from AS [ 11 ], the microbial diversity and temporal variation pattern of CPR bacteria in AS systems are barely discussed. A common feature of CPR organisms is their small cell sizes (200–300 nm) [ 7 , 8 ] and extremely reduced genome sizes (0.85 ± 0.23 Mb). CPR bacteria often lack the complete pathways for the biosynthesis of amino acids, lipids, and nucleotides [ 2 , 4 , 15 , 16 ]. Moreover, some CPR bacteria (e.g., Ca . Katanobacteria and Ca . Dojkabacteria (WS6)) cannot even de novo synthesize cell envelopes due to incomplete lipids and/or peptidoglycan synthesis [ 13 ]. Despite the reduced metabolic platforms of CPR organisms, analyses of gene repertoires and metabolic capacities revealed the highly divergent genome content among CPR lineages, even within the same lineage. Moreover, the divergence between CPR genome content showed complex relationships with correlated microbial community members and environmental types. For example, the comparative genomic analysis revealed that animal-associated Saccharibacteria have smaller gene repertoires than their environmental counterparts [ 9 ]. In contrast to the numerous genes that cope with oxidative stress in soil-associated CPR organisms, genomes recovered from the anoxic groundwater environment lack genes relate to oxygen metabolism [ 9 , 16 ]. While these genetic comparisons between CPR lineages from different environments provide important clues to exploring their niche adaptation strategies, it is still poorly understood how CPR bacteria survive and interact with other organisms in AS. Despite a generally limited biosynthetic potential, these ubiquitous CPR bacteria probably contribute to biogeochemical cycling [ 2 , 17 – 19 ]. Many members of CPR have been reported to encode genes involved in lactate, formate, and/or ethanol production. Several CPR genomes recovered from groundwater encode copper nitrite reductase ( nirK ) and/or an NADPH nitrite reductase ( nirB ). The co-culture experiments have confirmed the obligate symbiotic lifestyle of CPR organisms. In addition to getting essential compounds from its hosts, one epiparasitic bacterium from the Saccharibacteria had been reported to act as a bacteriophage and lyse foaming bacteria in WWTPs [ 20 ]. Given the roles of CPR organisms in biogeochemical cycling and microbial interactions, it is of particular interest to explore what ecological role they play in wastewater treatment systems. In this study, we took advantage of our previously reported nine-year time-series AS metagenomes (97 samples) from Shatin (ST) WWTP in Hong Kong, China [ 21 ], and sequenced 22, 13, and 13 newly collected AS samples taken monthly from ST, Shek Wu Hui (SWH), and Stanley (STL) WWTPs, respectively. Using these AS metagenomes, we recovered non-redundant bacterial and archaeal metagenome-assembled genomes (MAGs), including 135, 31, and 28 CPR bacterial MAGs from ST, SWH, and STL WWTPs, respectively. The long-term time-series data of ST AS enabled the characterization of the temporal dynamics of CPR communities and the inference of their putative hosts. Comparative genomic analysis among CPR from different environments was performed to predict the origin of abundant CPR in WWTPs. The roles of CPR organisms in the carbon cycling and microbial evolution process were further discussed based on the genome-resolved analyses. Overall, these results advanced our understanding of the ecological roles of CPR bacteria in bioengineered systems that have not yet been fully resolved.",
"discussion": "Discussion When considering the wide distribution of CPR across human-associated and environmental ecosystems, it is of critical importance to answer how the CPR communities changed and what kind of ecological roles they might play in a given environment. Our long-term longitudinal metagenomic analyses enabled the high-resolution characterization of changes in CPR communities over time. It is well accepted that CPR bacteria predominate in groundwater [ 16 , 52 ] and lakes [ 14 ]. The temporal dynamic profile of CPR bacteria in AS system demonstrated that CPR bacteria could be particularly abundant in engineered systems under eutrophic and high dissolved oxygen conditions. A previously reported study focusing on the recovery of high-quality MAGs from AS metagenomes also reported the high relative abundance of CPR organisms in AS at a spatial scale (relative abundance of > 7%) [ 53 ]. With the newly recovered CPR MAGs, we demonstrated that AS system harbored high CPR bacterial diversity with a large proportion of novel bacteria lineages. Saccharimonadia and Paceibacteria were the most dominant CPR taxa in the studied AS system, while the routinely detected lineages in groundwater, lake, and other aquifer environments (e.g., Parcubacteria and Microgenomatia) [ 54 , 55 ] were rarely identified as abundant populations in AS systems. More importantly, the time-series metagenomic analysis in this study provided a long-term temporal profile (> 10 years) charting how CPR communities change over time. The striking fluctuations observed with respect to the high taxonomic and MAGs levels both suggested the absence of generalists for CPR bacteria in AS system. As reported in our previous study, the overall microbial communities in ST AS changed from an Actinobacteriota to a Proteobacteria-dominated community due to the addition of bleach solution [ 21 ]. This shift of microbial communities may explain the abrupt decrease of Saccharibacteria because bacteria within Actinobacteriota have been experimentally demonstrated as hosts for Saccharibacteria [ 7 , 56 ]. The associations between Saccharibacteria and Actinobacteriota were also confirmed by the similarity-based network analysis. On the one hand, multiple network connections between given CPR bacterium and non-CPR bacteria with distinct taxonomic affiliations indicated the multiple hosts association of CPR bacteria. On the other hand, we should be noted that CPR bacteria within the same lineage (e.g., class) typically conserved a similar bacterial host distribution, suggesting the specificity of interactions between CPR and host bacteria. Together with the nature of limited metabolic potentials possessed by CPR bacteria in AS systems, we speculated that changes in CPR communities in AS systems were directly driven by the available bacterial hosts. Saccharibacteria were widely identified in human oral and gut microbiomes [ 57 ]. It is expected that the human-associated Saccharibacteria should be identified and/or enriched in AS systems when considering the human wastes will be collected by sewage collection system and finally received by WWTPs. However, our analyses revealed that CPR bacteria in AS shared higher similarities of proteome content with environmental CRP bacteria than human-associated ones. Since microbial communities from different types of habitats are divergent from each other, this might to some degree explain the observed niche differentiation between CPR organisms in AS and the human body. Even for the same type of ecosystem, Christine et al. observed little shared CPR species across different groundwater sites [ 15 ]. These findings further confirmed that CPR community compositions and dynamics may be determined by the host populations. Comparative genomic analysis further provided gene signs of the adaptive response of CPR bacteria in AS systems. Genes for oxidative stress resistance were significantly enriched in AS CPR MAGs may enable CPR bacteria to adapt to the high oxygen condition in aeration tanks. Despite the patchy metabolic capabilities of CPR bacteria, our results suggested these small, yet abundant bacteria might have involvements in the wastewater treatment process. The widely distributed polysaccharide lyase, glycoside hydrolase and carbohydrate-binding modules in CPR MAGs indicated that these bacteria might be involved in cooperative biogeochemical cycling, especially considering the ultrasmall cellular size may increase the surface area relative to cytoplasm volume of the bacterial host [ 52 ]. We also observed that Dojkabacteria, Gracilibacteria, Microgenomatia, and Saccharimonadia may act as fermenters in AS and contribute to the production of acetate, lactate, and formate, which might support the growth of CPR bacterial hosts [ 13 ]. These genetic findings collectively suggested that the exchange of metabolic productions between CPR bacteria and their hosts might not be just one way. The observed horizontal gene transfer events between CPR bacteria and prokaryotic organisms or phages were evident for the impacts of CPR bacteria on microbial evolution. As reported in the experiments, saccharibacterium TM7x has a broad bacterial host range [ 56 ]. Therefore, CPR bacteria might be a key driver of bacterial evolution, with consideration of the broad range of CPR bacterial hosts. The enrichment of CPR bacteria in WWTPs effluent highlighted the importance of disinfection steps for conventional biological wastewater treatment systems, particularly considering the observed significant enrichment of tetracycline and multidrug resistance genes in WWTP-associated CPR bacteria. Unattached TM7x remains viable and could re-infect new bacterial hosts when available [ 58 ], suggesting the discharged CPR bacteria might contribute to the spread of ARGs in the wastewater effluent-receiving ecosystems."
} | 3,897 |
35557859 | PMC9086191 | pmc | 7,720 | {
"abstract": "According to classical heterogeneous nucleation theory, the free energy barrier (ΔG c ) of heterogeneous nucleation of vapor condensation ascends dramatically as the substrate nanostructure diameter (R s ) decreases. Based on this idea, we fabricated two types of superhydrophobic surfaces (SHSs) on an aluminum substrate by different roughening processes and the same fluorization treatment. Water vapor condensation trials by optical microscope and ESEM confirmed that on SHSs with submicron rectangle structures, a typical self-propelled motion of condensates or jumping condensation occurred. However, on SHS with coral-like micro/nano-structures, vapor nucleation occurred tardily, randomly, and sparsely, and the subsequent condensation preferentially occurred on the nuclei formed earlier, e.g., the condensation on such SHS typically followed the Matthew effect. Higher vapor-liquid nucleation energy barrier caused by smaller fluorinated nanostructures should be responsible for such a unique “anti-condensation” property. This study would be helpful in designing new SHSs and moving their application in anti-icing, anti-fogging, air humidity control, and so on.",
"conclusion": "Conclusion In summary, we found an interesting phenomenon, e.g., “Matthew effect condensation” or “anti-condensation”, on SHS with much smaller nanostructures. Different to the classical self-propelled motion or “jumping” behavior of condensate droplets on SHS with relatively larger structures, condensate droplets on SHS with smaller nanoflakes appear slowly (∼50 s delay), dispersedly, and sparsely in the whole condensation procedure. Condensation started from random nucleation and the subsequent nucleation preferentially occurred on these primary nuclei. As a result, most of the SHS area appears dry during the condensation procedure. A much higher vapor–liquid nucleation energy barrier caused by much smaller nanostructures should be responsible for such a unique “anti-condensation” property. This study would be helpful in designing new SHSs and moving their application in anti-icing, anti-fogging, air humidity control, and other relative fields.",
"introduction": "Introduction Many plants exhibit remarkable water repellency owing to their rough surface. The textured surface traps air underneath water droplets and the air cushioning gives rise to superhydrophobicity ( Barthlott and Neinhuis, 1997 ; Neinhuis and Barthlott, 1997 ). However, biomimetic superhydrophobic surfaces (SHSs) generally do not retain water repellency when exposed to a condensing environment ( Zhao and Yang, 2017 ; Chen et al., 2018 ; Zhao et al., 2018 ; Orejon et al., 2019 ). Water condensates proceeding from nanoscale nuclei tend to penetrate into the surface texture and displace the trapped air, forfeiting the superhydrophobicity. Along with the condensation proceeds, arrays of visible, glittering, transparent, and adhesive large Wenzel drops (3–5 mm in diameter) cover the SHSs gradually. This seriously limits their applications in sustained dropwise condensation ( Hao et al., 2018 ; Wang et al., 2020 ), water collection ( Zheng et al., 2010 ), anti-icing ( Kreder et al., 2016 ; Caldona et al., 2017 ; Zhu et al., 2020 ), and anti-corrosion ( Xue et al., 2020 ). In some cases, these SHSs even represent a worse performance than general hydrophobic surfaces do such as increasing ice adhesion strength once the ice forms ( Kreder et al., 2016 ; Caldona et al., 2017 ). Recently, the self-propelled motion of condensate drops on some SHSs has attracted increasing attention due to its potential applications in delaying frost growth ( Hao et al., 2014 ; Jiang et al., 2020 ; Mohammadian et al., 2020 ), enhancing condensation heat transfer ( Hao et al., 2018 ; Sarode et al., 2020 ), stronger self-cleaning ( Geyer et al., 2020 ), and breathable anti-condensation coating on buildings ( Wu et al., 2021 ). Wang et al. (2021) demonstrated that the vapor molecules can be intercepted by oblique nanowires and preferentially nucleate at near-surface locations, avoiding the penetration of vapor into the microscale gaps. In our earlier studies ( Feng et al., 2012a ; Feng et al., 2012b ), we have confirmed that nuclei formed within the nanogaps of SHSs would grow and coalesce into micro-droplets. Then the micro-droplets derive themselves upwards and form into Cassie droplets. It is such a Cassie state that causes the spontaneous motion of drops after coalescence. A nanostructure with sufficiently narrow spacing and high perpendicularity is favorable to form such a Cassie condensation. According to classical heterogeneous nucleation theory ( Liu, 1999 ), the free energy barrier (ΔG c ) of heterogeneous nucleation of vapor condensation ascends dramatically as the substrate nanostructure’s diameter (R s ) decreases. No nucleation would bring none condensation. Based on this principle, new types of SHS with a more obvious anti-condensation property may be created by designing fine nanostructures. In this study, we fabricated two types of SHSs on an aluminum substrate by two different roughening processes and the same fluorization treatment. One was only by HCl etching and the other was by HCl etching and by further immersing in hot water. Water vapor condensation trials confirmed that although both two surfaces were superhydrophobic and supported Cassie condensation, only SHS by HCl etching and further by hot water treatment showed an obvious anti-condensation property, e.g., the condensate droplets appeared tardily, randomly, and sparsely on it. Most of the SHS areas appeared dry. A much higher nucleation energy barrier caused by much smaller nanostructures should be responsible for such phenomena. This study opens a new door for designing new SHSs and moving their applications in fields such as anti-icing, anti-fogging, anti-corrosion, and air humidity control.",
"discussion": "Results and Discussion Morphology and Superhydrophobicity of as-Fabricated Surfaces Similar to the results of Yin et al. (2012) and Z. Zhang Yang et al. (2011) , rectangle-shaped submicron-structure ( Figure 1A1–3 ) and coral-like micro/nano-hierarchical structures ( Figure 1B1–3 ) were obtained on the aluminum surface after HCl etching and HCl etching combined with hot water treatment, respectively. Vulnerable dislocation sites inside the crystalline aluminum should be responsible for such a submicron rectangle structure (∼0.5–1 μm) ( Yin et al., 2012 ). While the reaction of aluminum with hot water starting from the dissolution of aluminum and followed by the deposition of aluminum hydroxide colloidal particles on the aluminum surface should be responsible for the coral like micro/nano-hierarchical structures ( He et al., 2012 ). The average width of nano-flakes is ∼10 nm and the average space is ∼100 nm ( Figure 1B3 ). FIGURE 1 FESEM images of the textured aluminum surface obtained by 12 min HCl etching (9 wt%) at 20°C (A) and further immersing in 50°C water for 40 min (B) . Magnification from A1 to A3 or B1 to B3 is increased. The insets were profiles of 4 μL water droplets showing WCA both at ∼155°. CAs and SAs measurement showed that the sessile CAs of two types of as-prepared surfaces were both larger than 150° ( Figure 1 , insets) and the SAs were both less than 2°. This demonstrated that both two types of surfaces were typical superhydrophobic and the intrinsic surface energy was sufficiently low. The latter is one of two key factors affecting the vapor condensation nucleation energy barrier ( Liu, 1999 ). Compared with the anodization method, simple hot water immersing supplied a facile process in creating dense nanostructures and narrow nanogaps on the aluminum substrate, which is necessary for forming larger upward Laplace pressure to the droplets condensed within the gaps (if they could form there) and thus bringing Cassie condensation and rapid self-propelled motion phenomenon to condensate droplets ( Feng et al., 2012a ; Feng et al., 2012b ). Condensation Under Ambient Condition \n Figure 2 shows the time-lapse top-view optical images of dropwise condensation on aluminum surfaces prepared by two different etching methods. It clearly demonstrates that different surface structures do bring different condensation behaviors. On a rectangle submicron structured SHS, condensate droplets appeared in a classical self-propelled motion or “jumping” behavior, e g., condensation, is continuously, covering all areas and homogeneous ( Figure 2A ). The spontaneous motion frequency began at the high level (>100 drops/s), changed a little in 1 min and then gradually decreased, and finally balanced at 70 drops/s. Re-nucleation and growth of condensate droplets appeared on any region of the SHS especially including bare areas caused by droplet move-away. However, on coral-like micro/nano-structured SHS, condensate droplets appeared slowly (∼50 s delay), dispersedly, and sparsely in the whole condensation procedure. Most of SHS was always bare and dry. Primary nucleation occurred randomly and the subsequent nucleation occurred preferentially on the droplets formed by these former nuclei. Because the distance between the droplets was so far, the coalescence opportunity was so low that no self-propelled motion or “jumping” appeared throughout the condensation process ( Figure 2B ). FIGURE 2 Water vapor condensation behavior on SHS with submicron rectangle microstructures (A) series (implying Cassie or jumping condensation), and on SHS with coral- like micro/nano-structures (B) series, implying an anti-condensation character). The scale bar is 60 μm. The temperature of SHSs was 0–1°C. The environmental RH was 80 ± 2% (28 ± 1°C). The time scale in the images is minute, second, and millisecond. Supplementary Video S1,S2 corresponding to A and B are available in the Supporting Information. Condensation in ESEM To better understand the aforementioned preferential condensation phenomenon, we further apply ESEM to observe the condensation dynamics on both types of SHSs. Figure 3 shows time-lapse images of condensation on the SHSs with submicron rectangle microstructures and coral like micro/nano-structures, respectively. It clearly proved the results of the microscopy video: both SHSs presented spherical Cassie state condensate droplets, however, only SHS with coral-like micro/nano-structures appeared to have an obvious anti-condensation property. That is, on SHSs with submicron rectangle microstructures, spherical droplets emerged continuously and coalesced successively thus forming droplets with dispersed diameters ( Figure 3A ). However, on SHS with coral-like micro/nano-structures, vapor nucleation occurred tardily, randomly, and sparsely, and the subsequent condensation preferentially occurred on these earlier formed nuclei. In the microscopy visual field, only several large drops grew up through their growth and asymmetric coalescence, while most areas were always dry ( Figure 3B ). FIGURE 3 Time-lapse ESEM images of vapor condensation on SHSs with submicron rectangle structures (A1-2) and with coral-like micro/nano-structures (B1-2), respectively. The scale bar is 50 μm. The temperature of the sample stage was fixed at ∼1°C. The vapor pressure was gradually increased to ∼800 Pa, at which the vapor started to nucleate on the sample surface, and then maintained at ∼800 Pa during imaging. The images were taken every 1.6 s. The time scale in the images is minute, second, and millisecond. Anti-Condensation Mechanism Analysis As it showed in Figures 2B , 3B , when homogeneous dropwise condensation continuously occurred on SHS with submicron rectangle microstructures, no condensate droplets appeared on most districts of SHS with coral-like micro/nano-structures. This phenomenon can be explained by the classical nucleation theory. Essentially, vapor condensation at least includes the process of nucleation and growth. Nucleation is the process of vapor molecules clustering together. It was generally triggered by supersaturation and with or without preferential sites such as dust and surface nanostructures (so called homogeneous or heterogeneous nucleation). Critical nucleation radius is the minimum size that must be formed by vapor molecules clustering before a droplet is stable and begins to grow. It mainly depends on supersaturation caused by dew point, supercooling temperature, and RH. According to the classical nucleation theory ( Liu, 1999 ), the radius of the critical nucleus (r c ) in vapor condensation can be estimated from: \n R T In P r P = − 2 γ V m r \n Where R is ideal gas constant (8.314 J⋅mol −1 K −1 ), T is the temperature of condensation (273.15 K), \n P r \n is the vapor pressure over a curved interface of a droplet with radius r, P is the equilibrium vapor pressure above a flat surface of the condensed phase at T ( P is 0.61129 kPa at 0°C), γ ≈ 7.56 × 10 –2 J/m 2 is the water (0°C) interfacial tension, and ν ≈ 1.8 × 10 –5 m 3 /mol is the water molar volume. When the \n P r \n ≤ P (28°C) (3.7818 kPa) × 80 % (RH at 28°C), nucleation occurs and the corresponding radius is at an equilibrium or critical radius (here named \n r c \n , which is 0.75 nm after calculation). The critical radius is the minimum droplet radius for the formation of stable nuclei. However, it is a concept suitable to homogeneous nucleation. In most cases, nucleation occurs at nucleation sites on surfaces contacting the vapor, and thus results in heterogeneous nucleation. Comparing with critical radius, the free energy barrier of nucleation is another index being developed to describe the difficulty of nucleation especially those that occur on the surfaces with nano or micro-structures (heterogeneous nucleation). According to the classical heterogeneous nucleation theory ( Liu, 1999 ), the effect of the surface structure on the free energy barrier of heterogeneous formation of condensate droplet ( \n Δ G c \n ) can be readily estimated as: \n Δ G c = Δ G c hom o f ( m , x ) \n Where ΔG c \n homo is the free energy barrier forming a droplet in a homogeneous way, \n f ( m , x ) \n is the ratio of free energy barrier for nucleation around a spherical particle relative to that in the bulk, e.g., a factor that reduces the energy barrier of heterogeneous nucleation. m = cosθ flat with θ flat = 108° for FAS treatment ( Feng et al., 2012a ), and x = R s /r c . Since r c is certain (0.75 nm), \n f ( m , x ) \n only changes with the radius of nucleating substrate structures (R s ). As Liu et al. (1999) derived, the \n f ( m , x ) \n ascends dramatically as the nanostructure diameter decreases till it approaches 1. When the condensation conditions (RH, supercooling, et al.) are same, the smaller nanostructure on SHS would bring a more difficult nucleation. On the SHS with coral-like micro/nano-structures ( Figure 1B ), the average width of the nano-flakes is ∼10 nm and the average space is ∼100 nm. This means that the corresponding apparent radius of nucleating structures (R s ) are ∼5 nm and ∼50 nm, respectively, both larger than the critical nucleus r c (0.75 nm). However, on SHS with submicron rectangle structures, the average apparent radius of nucleating structures (R s ) are ∼0.5–1 μm, which is much larger than the critical nucleus r c (0.75 nm). This means that the \n Δ G c \n on the submicron rectangle structures should be lower than that on the coral-like micro/nano-structures. A similar result had also been obtained by Lo et al. (2014) , where they found that water vapor preferentially condenses on the designed microgrooves on the Si nanowire surface (both with CA∼145°). The structure of the SHS has a strong effect on the nucleation rate J via the inverse exponential dependence on \n Δ G c \n , \n J = J 0 exp ( − Δ G / k T ) \n ( Varanasi et al., 2009 ). As a result, on the SHS with coral-like micro/nano-structures, due to higher vapor–liquid nucleation energy barrier caused by the finer nano-structures, vapor nucleation is difficult (∼50 s delaying) comparing with that on SHS with submicron rectangle structures (immediately, continuously, and densely). However, difficulty does not mean impossible. Vapor nucleation may also occur at some gaps of the surfaces. Once the primary nucleation is completed on the SHS with coral- like micro/nano-structures, the subsequent nucleation preferentially occurs on these primary nuclei ( Figure 4 This is because they are hydrophilic, which dramatically decreases the \n Δ G c \n ( Liu, 1999 ; Varanasi et al., 2009 ). As a result, the condensation occurred tardily, randomly, and sparsely. The condensation proceeded along the typical Matthew effect, e.g., always occurred on special sites (primary nuclei or defects). If we could fabricate SHS with homogeneous hydrophilic nano sites, a new type of condensation would be expected ( Xing et al., 2020 ). FIGURE 4 Scheme of water vapor condensation on SHS with submicron rectangle structures (A) , condensate droplets appeared immediately, continuously, and densely, and on the SHS with much smaller nanostructures (B) , condensate droplets appeared tardily, randomly, and sparsely, and the subsequent condensation preferentially occurred on the nuclei formed earlier, respectively."
} | 4,308 |
40295731 | PMC12038015 | pmc | 7,721 | {
"abstract": "Photonic accelerators have risen as energy efficient, low latency counterparts to computational hungry digital modules for machine learning applications. On the other hand, upscaling integrated photonic circuits to meet the demands of state-of-the-art machine learning schemes such as convolutional layers, remains challenging. In this work, we experimentally validate a photonic-integrated neuromorphic accelerator that uses a hardware-friendly optical spectrum slicing technique through a reconfigurable silicon photonic mesh. The proposed scheme acts as an analogue convolutional engine, enabling information preprocessing in the optical domain, dimensionality reduction, and extraction of spatio-temporal features. Numerical results demonstrate that with only 7 photonic nodes, critical modules of a digital convolutional neural network can be replaced. As a result, a 98.6% accuracy on the MNIST dataset was numerically achieved, with an estimation of power consumption reduction up to 30% compared to digital convolutional neural networks. Experimental results using a reconfigurable silicon integrated chip confirm these findings, achieving 97.7% accuracy with only three optical nodes.",
"conclusion": "Conclusions This work proposes an analog accelerator based on an integrated photonic reconfigurable platform, circumventing the issues that arise in hardware architectures adopting the MVM approach. Utilizing a single processing pass of the photonic chip, minimal external/internal memory, and minimal electro-optic/opto-electronic conversions, the hardware-friendly technique enables convolutional operations to be performed by optical filter-nodes without requiring fine control or elaborate training. Additionally, the OSS accelerator implements other key CNN operations, such as nonlinear transformation and pooling, in the analog domain without processing latency. Numerical results confirmed the scheme’s ability to achieve a classification accuracy of 98.6%, closely matching the performance of full-scale digital networks. Experimental validation using iPronics’s SmartLight reconfigurable photonic mesh achieved an accuracy of 97.7%, outperforming state-of-the-art photonic implementations on the MNIST dataset. More importantly, the proposed scheme confirms its accelerator properties by drastically reducing the total power consumption by 30% compared to its respective standalone digital neural network. The experimental performance and the reduced power consumption is substantiating the role of photonic accelerators as promising sub-systems for large-scale hybrid machine learning schemes.",
"introduction": "Introduction The exploding growth of the Internet of Everything (IoE) ecosystem 1 has unleashed the generation of a tremendous amount of raw data, demanding high-speed, low-power processing. In this landscape, typical von Neumann computers, characterized by their processing architecture, have met an efficiency road-block 2 , struggling with data bottlenecks and energy constraints. Consequently, bio-inspired computing has emerged as an unconventional and promising route, aiming to circumvent inherent limitations of traditional systems 3 . This approach, drawing on the complex mechanisms of biological organisms, offers parallel processing capabilities and adaptability, potentially transforming the way we handle the escalating data challenges in the IoE era. Towards this route, photonic integrated circuits can offer a proliferating hardware platform for neural network implementation, based on merits such as multiplexing-assisted parallelism, low power consumption, large-scale integration, and low latency 4 – 6 . One of the most common types of neural networks used for processing IoE generated visual data is convolutional neural networks (CNNs) 7 . Highly acclaimed for their ability to extract features from large datasets in a hierarchical manner, CNNs owe much of their effectiveness to their unique distinctive layered structure. Typically, CNN architectures consist of three types of layers: convolutional, pooling, and fully-connected layers (FCLs). The convolutional layers apply multiple sets of weights (kernel filters) to the input for feature detection, transforming the data into a feature map. Subsequently, the pooling layers reduce the dimensionality of the feature maps by computing the maximum or the mean of a local patch of units in one or several feature maps. Finally, the FCLs integrate the processed features, performing high-level reasoning to facilitate the final decision-making or classification task. A key issue with CNNs lies in their high computational and power demands for convolutional operations, which often consume up to 90% of the network’s execution time 8 . This has led to a focus on optimizing these operations in various hardware architectures, including traditional graphics processing units (GPUs) 9 , memristors 10 , and lately photonics 11 . These technologies primarily aim to enhance the efficiency of digital matrix-to-vector multiplications (MVMs) which are necessitated to execute the convolutional operations in CNNs. In this context, integrated photonic solutions such as micro-ring resonator (MRR) banks based on the broadcast-and-weight protocol 12 , coherent MRR networks 13 , photonic tensor cores 14 , 15 , cascaded fixed Mach-Zehnder interferometers (MZI) 16 , and combined MRR-MZI networks 17 risen as low-power MVM accelerators able to reach more than 1.27 tera-operations per joule. The adaptation of the MVM-based approach into the photonic domain introduces considerable challenges. One primary issue is the necessity for extensive offline numerical simulations during high-speed weight adaptation in training 13 , 17 . These simulations are meant to emulate the behavior of photonic hardware but often lead to power inefficiency and make photonic MVMs prone to errors in weights and digital-to-analogue converter (DAC) components’ resolution. These errors arise from discrepancies between the actual physical parameters of the chip and those assumed in the simulations. Additionally, current scalability constraints of photonic platforms limit the number of total MVMs that can be executed in a single pass, increasing the total execution time of these systems. Therefore, as the resolution and/or size of visual data increases, the number of processing passes on the same integrated photonic chip might also increase the total processing time and energy consumption. Recent attempts in the electronic domain to address these constraints have utilized phase-change components for in-memory computing 18 , enabling precise control over a substantial number of weights. However, even in these optimized cases, such approaches still demand extensive off-chip training and additional circuitry. In 19 , the operational principles of an unconventional integrated photonic accelerator were presented. The proposed system, namely OSS-CNN, is comprised of two main components: an analog pre-processor, which utilizes high-speed passive photonic and opto-electronic devices for feature extraction, nonlinear transformation and dimensionality reduction and a digital post-processor, in which a simple feedforward neural network is employed to associate the analog outputs with their corresponding class labels. The uniqueness of this acceleration approach stems from the adoption of an optical spectrum slicing (OSS) technique, which employs parallel passive optical filter nodes with distinct characteristics (bandwidth, central frequency) to perform analog-domain convolutions with the optical signal. Although the use of analog photonics as a preprocessor in convolutional tasks is not new, previous architectures that exploited this approach followed the MVM approach for the convolutional stage 12 , 17 , 20 . In contrast, OSS-CNN uses each filter node as an analog convolutional operator, interacting with the modulated signal through a continuous-valued transfer function or impulse response in the spectral or temporal domain to extract different spatial features from the input tensors. Unlike MVM-based approaches, the number of optical nodes in OSS-CNN is determined by the number of distinct characteristic nodes required to efficiently decompose the signal, rather than the size of the visual dataset. Furthermore, the use of a single photoreceiver after each node to appropriately average the output time-traces before digitization reduces the number of data and the overall power consumption of the system. This method results in a simple and minimal design in terms of employed photonic and electronic components, and the associated power footprint. The combination with a straightforward time-multiplexed encoding of visual data into the optical signal allows convolution operations of any size to be performed through a single pass, freeing the photonic hardware from the need to physically replicate the required MVMs. Compared to other advanced photonic CNN architectures evaluated on the MNIST handwritten digits task 21 , the OSS accelerator has theoretically shown equivalent state-of-the-art precision, comparable computational density and increased efficiency in terms of power consumption 19 . In this work, the transition of the OSS acceleration approach into the domain of reprogrammable photonics is explored. The photonic front-end of the OSS-CNN system is experimentally validated using a reconfigurable silicon photonic chip (SmartLight) 6 , 22 , with the OSS nodes now implemented using reconfigurable integrated filters. In our previous work, we demonstrated the OSS-NN concept using only numerically simulated ideal, lossless Butterworth filters 19 . Moreover, in this study, the OSS-CNN is benchmarked against a conventional digital CNN architecture, evaluating not only classification accuracy but also end-to-end power consumption for the same task, thereby highlighting the energy efficiency advantages of this approach. In specific, the first section of this document briefly analyzes the operational characteristics and performance of an ideal OSS-CNN system, characterized by no losses and the use of ideal Butterworth filters, and moves onto the analysis of an OSS-CNN system implemented within a reconfigurable photonic chip. Numerical simulations conducted using the physically accurate iPronics’ SmartLight emulation platform indicate that the proposed OSS accelerator even when implemented on a reconfigurable photonic platform notably can improve performance on benchmark datasets such as MNIST by at least 5.3% in accuracy, when utilized prior to a basic feedforward neural network. The subsequent section documents the experimental validation of the OSS accelerator on a reconfigurable silicon photonic chip, achieving a state-of-the-art 97.7% accuracy with only 3 OSS nodes while maintaining the same number of data samples at its output as the input visual data. The final section compares the reconfigurable OSS-CNN directly to a digital equivalent single-layer CNN implemented on a state-of-the-art GPU. The exploration of performance and total power consumption for both architectures is conducted in relation to the depth of the feedforward neural network employed at their back-end. Even though the OSS accelerator exhibits slightly inferior precision compared to the single-layer CNN, it achieves a substantial reduction in total power consumption during the training stage by 30%."
} | 2,851 |
31730850 | PMC6912165 | pmc | 7,722 | {
"abstract": "Summary Ocean microbial communities strongly influence the biogeochemistry, food webs, and climate of our planet. Despite recent advances in understanding their taxonomic and genomic compositions, little is known about how their transcriptomes vary globally. Here, we present a dataset of 187 metatranscriptomes and 370 metagenomes from 126 globally distributed sampling stations and establish a resource of 47 million genes to study community-level transcriptomes across depth layers from pole-to-pole. We examine gene expression changes and community turnover as the underlying mechanisms shaping community transcriptomes along these axes of environmental variation and show how their individual contributions differ for multiple biogeochemically relevant processes. Furthermore, we find the relative contribution of gene expression changes to be significantly lower in polar than in non-polar waters and hypothesize that in polar regions, alterations in community activity in response to ocean warming will be driven more strongly by changes in organismal composition than by gene regulatory mechanisms. Video Abstract",
"conclusion": "Conclusions Large-scale oceanographic sampling expeditions, such as the World Ocean Circulation Experiment (WOCE) or GEOTRACES ( Anderson et al., 2014 , Koltermann et al., 2011 , Woods, 1985 ) have been extremely valuable in building our understanding of the ocean circulation, and the distribution of major nutrients and elements including trace metals, as well as their contribution to the climate system. However, our geochemical and physical knowledge of the ocean remains incomplete without incorporating the processes that regulate biogeochemical cycles at planetary scale ( Falkowski et al., 2008 ). Analyzing the repertoire of genes and transcripts from environmental samples can inform us about the potential and activity of microbial communities that drive these cycles at global scale and thus help us to understand the intertwined processes that shape the physico-chemical state of the ocean through biological activity. In this study, we describe global biogeographical patterns of microbial community transcriptome compositions and demonstrate how changes in these compositions can be attributed to community turnover and/or gene expression changes as the underlying mechanisms. Assessing the mechanisms that underlie such compositional differences, as demonstrated here, can help us to determine whether changes in the molecular activities of microbial communities are regulated by gene expression changes or by a turnover of organisms containing genomic modifications that arose over evolutionary time. In addition, an improved understanding of the ecological factors that drive community compositional and diversity changes can help us to better predict how ocean microbial communities will respond to environmental changes. For example, the consistent identification of temperature as a major explanatory factor for global-scale community-level differences in genomic ( Sunagawa et al., 2015 ) and transcriptomic (this study) composition, as well as taxonomic diversity ( Gregory et al., 2019 , Ibarbalz et al., 2019 ), has wide-ranging implications, in particular for the Arctic Ocean, given the current projections of disproportionately high warming rates in this region ( Alexander et al., 2018 , IPCC, 2014 ). Notably, the analyses of this study were enabled by a systematic, highly contextualized, pan-oceanic set of metagenomic and metatranscriptomic data that, along with the OM-RGC.v2, complements other large-scale datasets that have been developed for eukaryotes ( Carradec et al., 2018 , Ibarbalz et al., 2019 ), prokaryotes ( Biller et al., 2018 ), and viruses ( Gregory et al., 2019 ). Together, these will pave the way for an eco-systems level understanding of ocean plankton diversity, function, and activity across boundaries of organismal size ranges. To reach this goal, it will be important to integrate temporal meta-omics data, ideally from global observations, to account for seasonal variations and other concomitant environmental changes, such as increased stratification, acidification, nutrient availability, and deoxygenation of the oceans ( Bopp et al., 2013 , Schmittner et al., 2008 ). Such concerted efforts are required to further refine gene-to-ecosystem models ( Coles et al., 2017 , Garza et al., 2018 , Guidi et al., 2016 ) and to inform environmental and climate policies ( Le Quéré et al., 2018 ), which must consider not only how microorganisms are impacted by but also how they may affect anthropogenic climate change ( Cavicchioli et al., 2019 ).",
"introduction": "Introduction Microorganisms perform ecological functions and drive biogeochemical cycles that transform matter and energy on a global scale ( Falkowski et al., 2008 ). Recent advances in sequencing technology and the analysis of DNA extracted from environmental samples (metagenomics) have made it possible to systematically characterize the taxonomic and genomic composition of microbial communities in diverse biomes ( Fierer et al., 2012 , Human Microbiome Project Consortium, 2012 , Sunagawa et al., 2015 ). In the ocean, such biodiversity surveys have been conducted on local ( Karl and Church, 2014 , Venter et al., 2004 ), as well as regional and global scales ( Biller et al., 2018 , Kent et al., 2016 , Rusch et al., 2007 , Sunagawa et al., 2015 ). These and similar efforts ( Delmont et al., 2018 , Duarte, 2015 , Kopf et al., 2015 , Tully et al., 2018 ) have provided valuable baseline data that reveal the biodiversity of ocean microbial taxa, the repertoire of genes and genomes in the ocean, and the ecological factors that structure ocean microbial communities. Despite the rich information that can be obtained about the gene-encoded functional potential in an environment, metagenomics alone cannot predict which, and in what amount, specific functions contribute to the molecular activity of microbial communities in situ , because genes may be variably expressed or not expressed at all. In contrast, metatranscriptomics enables the analysis of the pool of transcripts from genes that are actually expressed in an environmental sample ( Helbling et al., 2012 , Moran et al., 2013 , Poretsky et al., 2005 ) and therefore provides a more accurate depiction of ecologically relevant processes that are occurring (e.g., in response to diurnal or other variations in environmental conditions) ( Ottesen et al., 2014 , Poretsky et al., 2009 ). In addition, the integration of metagenomic and metatranscriptomic data to quantify levels of gene expression, that is, the relative amount of expressed transcripts per gene, has revealed a number of important insights. For example, the ecological importance of photosynthesis, carbon fixation, and ammonium uptake has been highlighted in Prochlorococcus , which is abundant in oligotrophic waters of the tropical and subtropical ocean, because genes encoding these functions were among the most highly expressed genes in their genomes ( Frias-Lopez et al., 2008 ). Picocyanobacteria, in general, have been found to contribute more to the community pool of transcripts than expected by abundances inferred from metagenomics, whereas the opposite has been shown for some heterotrophic bacteria, including those from the highly abundant SAR11 clade ( Dupont et al., 2015 , Frias-Lopez et al., 2008 , Shi et al., 2011 ). In contrast to studying differences between gene and transcript abundances within samples, understanding why a pool of community transcripts (metatranscriptome) changes from one sample to another has received much less attention. Notably, changes in metatranscriptomes can result from alterations in the relative abundance of organisms and their associated genes (community turnover) and/or by changes in the expression of genes encoded among the community members ( Satinsky et al., 2014 ) ( Figure S1 ). For microbial communities in the Amazon River Plume, it has been shown, for example, that higher transcript levels for some functions (e.g., acquisition of phosphorous) could be explained by increased gene abundances in free-living communities whereas for other functions (e.g., sulfur cycling, vitamin biosynthesis, and aromatic compound degradation) higher transcript levels were attributed to increased gene expression levels in particle-attached communities ( Satinsky et al., 2014 ). However, global-scale biogeographic patterns of community turnover versus gene expression-driven changes in metatranscriptomes, and the ecological determinants of the relative contribution driving these two mechanisms, have not yet been studied for marine or any other environmental microbial communities. Figure S1 Transcript Abundance Profile as a Function of Community Composition and Gene Expression, Related to STAR Methods Cartoon exemplifying how an initial community with a given expression profile may result insimilar transcript abundance profiles through two different mechanisms: (i) changes in the community composition (upper arrow), represented by three different species (green, red, and blue), or (ii) changes in gene expression (lower arrow), represented by two different genes (purple and orange, with low and high expression levels, respectively). Here, in order to better understand the basis of metatranscriptomic differences across environmental gradients (e.g., latitude and depth) in the ocean, we leveraged efforts from the Tara Oceans (2009–2013) expeditions ( Karsenti et al., 2011 ) and analyzed an environmentally contextualized dataset ( Pesant et al., 2015 ) of metatranscriptomes and metagenomes, which includes a circumpolar representation of the climate change-impacted Arctic Ocean ( Hoegh-Guldberg and Bruno, 2010 , Overland et al., 2018 ). To capture the abundances of genes and transcripts from ocean microbial communities at the species level, we established a reference catalog of non-redundant protein-coding sequences (hereafter, genes). Using this integrated information, we determined for a number of biogeochemical processes involved in photosynthesis, as well as in the cycling of carbon, nitrogen, and sulfur, varying contributions of community turnover, and gene expression changes to metatranscriptome differences across latitude and depth. We further compared, as a function of temperature, the relative contributions of these mechanisms and hypothesize how they will differ between polar and non-polar regions in response to ocean warming.",
"discussion": "Results and Discussion A New Meta-omics Resource for Global Ocean Microbiome Research The dataset for this study consists of metatranscriptomic (n = 187) and metagenomic (n = 370) samples collected at 126 globally distributed sampling stations across a latitudinal range of 142° ( Figure 1 ; https://doi.org/10.5281/zenodo.3473199 ). The samples originate from the light-penetrated, epipelagic waters from the surface (SRF), deep chlorophyll maximum (DCM), and mixed water layer, and dark waters from the mesopelagic (MES) layer, from 5 m to 1,000 m in depth (median depths of 5 m, 50 m, and 550 m for SRF, DCM, and MES, respectively). The 187 prokaryote-enriched metatranscriptomic libraries were generated and sequenced to an average depth of 28 Gbp per sample ( https://doi.org/10.5281/zenodo.3473199 ), after protocol optimization for low-input RNA samples ( Alberti et al., 2014 ) ( STAR Methods ). These data were analyzed in conjunction with a set of 131 virus-, 59 giant virus-, and 180 prokaryote-enriched metagenomes ( https://doi.org/10.5281/zenodo.3473199 ), which include prior sequencing efforts of Tara Oceans ( Sunagawa et al., 2015 ), virus-enriched metagenomes from polar (n = 44) and non-polar (n = 42) regions ( Gregory et al., 2019 , Roux et al., 2016 ) (see STAR Methods for definitions), and 41 prokaryote-enriched metagenomes from the Arctic Ocean (new to this study). Figure 1 Geographic Coverage of the Meta-omics Dataset Analyzed in This Study Geographic distribution of the sampling stations of the Tara Oceans (2009–2013) expeditions ( Pesant et al., 2015 ). Several size-fractionated samples were collected from different depth layers at each station for a total of 557 samples (370 metagenomes and 187 metatranscriptomes). Stations numbered 155 and above represent the Tara Oceans Polar Circle campaign undertaken between June and October 2013. Colors indicate the type of samples collected for the prokaryote-enriched fractions at each station: metagenome only (orange, 18 stations); metatranscriptome only (blue, 40 stations); metagenome and metatranscriptome for at least one of the depth layers (green, 68 stations). We aimed to capture whole community-level variations in community turnover and gene expression changes and to place these data into the context of geographic and environmental gradients at a global scale. Notably, the applicability of this approach critically depends on the evolutionary distances between the organisms present in the environment and those represented in genomic sequence databases ( Nayfach et al., 2016 ). Ideally, genome sequences would be available for all organisms that comprise the communities of interest, thus facilitating the integration of gene abundance and gene expression data to assess whole-community compositions. Such analyses appear to be within reach for the human gut microbiome, for which appropriate genomic resources have recently become available ( Almeida et al., 2019 , Nayfach et al., 2019 , Pasolli et al., 2019 ). However, for ocean microbiome samples, less than 10% of metatranscriptomic, and less than 5% of metagenomic data, can currently be resolved at the species-level using available marine genomic sequence databases ( Figure 2 A). Figure 2 Gene Detection Rates and Annotation of the OM-RGC.v2 (A) Percentage of reads from 180 prokaryote-enriched metagenomes (orange) and 187 prokaryote-enriched metatranscriptomes (blue) aligned with a 95% identity cutoff to: the MarRef database v3, updated 2019/01/19 ( Klemetsen et al., 2018 ), a collection of metagenome-assembled genomes (MAGs) reconstructed from Tara Oceans samples ( Delmont et al., 2018 ), and the OM-RGC.v2 (this study). To fairly compare the alignments to the MarRef database or MAGs and the catalog, we corrected for the gene coding density in prokaryotic genomes ( STAR Methods ). Boxplots show the median values as horizontal lines, interquartile ranges as boxes with whiskers that extend up to 1.5 times the interquartile range, and outliers as individual data points. (B) The accumulation of OM-RGC.v2 genes detected in 180 prokaryote-enriched samples. The dashed line separates the prokaryote-enriched non-Arctic metagenomes (n = 139) ( Sunagawa et al., 2015 ) from the Arctic metagenomes (n = 41). The increase in slope reflects an increase in the rate of detection of new genes in the Arctic Ocean. The non-prokaryote-enriched metagenomes (n = 190) and the metatranscriptomes (n = 187) are excluded from this analysis. (C) The taxonomic annotation of genes at the domain level (and viruses; LUCA, last universal common ancestor) and the breakdown of gene functional annotations into ∼9 k KEGG and ∼76 k eggNOG orthologous groups (KOs and OGs, respectively). The remaining fraction of unannotated genes was used to generate de novo gene clusters (GCs) for further functional characterization of the catalog. To overcome this limitation, we generated an updated version of the Ocean Microbial Reference Gene Catalog (OM-RGC.v2; original version in Sunagawa et al., 2015 ) based on 370 metagenomes with extended geographic coverage, particularly for the Arctic Ocean ( Figure 1 ). Among the 47 million non-redundant genes, 24.5% were reconstructed, although partially detected elsewhere ( Figure 2 ), in the Arctic Ocean samples alone, highlighting the added value of sampling genomically underexplored environments. Using this reference, nearly 70% of the genes could be taxonomically annotated, and 61% showed homology to known (i.e., existing) orthologous groups (OGs) in the database used for gene functional annotation (eggNOG version 4.5) ( Huerta-Cepas et al., 2016 ) ( STAR Methods ). We further grouped the remaining 39% of the genes in the OM-RGC.v2 that represent unknown genes (i.e., genes of unknown function without detectable homology to known sequences), into ∼250,000 gene clusters (GCs) based on shared sequence similarity ( Figure 2 C; STAR Methods ). We identified significant differences when comparing transcript abundances between depth layers (for 5,439 GCs) or between polar and non-polar regions (for 31,339 GCs), or correlations with environmental parameters (for 21,648 GCs) ( Figure S2 ). These findings suggest ecologically relevant yet unknown functions of these genes in response to environmental variation. A benchmarked analysis of conserved co-expression as a method for identifying functionally related genes ( Stuart et al., 2003 ) suggests that some of the GCs are likely to represent unidentified players in signal transduction, transcriptional regulation, and energy production/conversion ( Figure S3 ; Table S1 ). Figure S2 Prevalence and Statistical Associations to the Environment of OGs and GCs, Related to STAR Methods Gene abundance-based prevalence versus transcript abundance-based prevalence (i.e., number of samples in which detected) for (A) eggNOG-based orthologous groups (OGs) and (B) de novo gene clusters (GCs) based on the 122 paired metagenomes and metatranscriptomes. Prevalence distributions are shown in the side and upper panels. The numbers of OGs and GCs with significant associations of transcript abundances to depth layers (C) and polar/non-polar regions and (D) to environmental variables are shown. Associations were detected as statistically significant differences in transcript abundance by Wilcoxon tests for depth layers and polar/non-polar regions (p < 0.05, after Holm correction for multiple comparisons) and as significant Pearson correlations for environmental variables (|r| > 0.6 and p < 0.05, after Holm correction for multiple comparisons). In both cases only the OGs and GCs with a transcript abundance-based prevalence higher than 10% were considered in order to avoid spurious associations. Figure S3 Rationale for the Use of Co-expression Data to Associate Groups with Unknown Functions to Known Functional Groups, Related to STAR Methods Evaluation of model performance for the link between OGs based on co-variation analysis. (A) Receiver operating characteristic (ROC) curves for all models. Variation in (B) false positive rate and (C) sensitivity with increasing Pearson correlation values used as a cut-off for classification ( r min ). The r min is a value to be optimized corresponding to the minimum Pearson r that provides sufficient predictive power (false positive rate < 5%). A total of nine models are represented, which used co-abundance, co-transcription, and co-expression for the prediction of shared KEGG reactions, modules, and pathways, respectively, between pairs of OGs (see details in STAR Methods ). In contrast to existing ocean genomic reference databases, we found the OM-RGC.v2 to capture the majority of gene-encoding metagenomic and metatranscriptomic data (70% and 51%, respectively) ( Figure 2 A) used in this study, making it a suitable resource to address our aim of analyzing whole-community metatranscriptomic compositions. All gene sequences can be queried online for their abundance, expression, and geographic distribution ( Villar et al., 2018 ), and they are linked to contextual environmental parameters ( Pesant et al., 2015 ) facilitating additional gene-centric explorations in the future. Variation of Meta-omic Compositions across Latitude and Depth Having established resources to quantify whole-community taxonomic, genomic, and transcriptomic compositions, we next sought to identify patterns and drivers of compositional structure across major axes of environmental variation in the ocean biome at a global scale. Numerous studies have revealed that microbial communities are vertically stratified in the ocean, with a striking boundary between epipelagic and mesopelagic zones ( DeLong et al., 2006 , Giovannoni and Stingl, 2005 , Sunagawa et al., 2015 ). Polar and non-polar communities have also been shown to separate into distinct groups with different species-level taxonomic compositions ( Ghiglione et al., 2012 , Gregory et al., 2019 ). Critically, however, the shared gene content between different strains of the same species may be as low as 40%, as has been shown, for example, in Escherichia coli ( Mira et al., 2010 ). Furthermore, gene functional redundancy in microbial communities (i.e., when the same gene functions are encoded by different taxa) may help to maintain important community functions in cases of biodiversity loss ( Bell et al., 2005 ). Thus, it is difficult to predict whether gene functional compositions and gene expression-regulated transcriptomic repertoires would follow the same patterns of taxonomic composition changes. To address this question, we first aimed to locate the boundaries of differentiation ( Ludwig and Cornelius, 1987 ) in epipelagic waters (SRF and DCM) along the latitudinal gradient for different community-compositional measures derived from the prokaryote-enriched metatranscriptomes and metagenomes ( STAR Methods ). From the equator northward, no significant differentiation was identified in epipelagic waters until a latitude of 40°N. At this point, the degree of differentiation increased significantly for all community-compositional measures and peaked at around 60°N. A similar trend was also observed for the southern hemisphere ( Figure 3 ) and is consistent with the taxonomic compositional differences observed between polar and non-polar waters for bacterial ( Ghiglione et al., 2012 , Gregory et al., 2019 ) and viral communities ( Ghiglione et al., 2012 , Gregory et al., 2019 ). We further found that the differentiation is reflected by significant enrichments of operational taxonomic units (OTUs) from the order Flavobacteriales (e.g., Formosa , Polaribacter , NS5, NS7, and NS9 marine groups), the class Gammaproteobacteria (OM182 clade and Piscirickettsiaceae), and eukaryotes (e.g., Phaeocystis ), as well as by depletions of Prochlorococcus spp., members of the Rhodospirillaceae family, and members of the SAR11 and SAR406 clades toward higher latitudes ( Figure S4 ). Here, the congruent patterns observed for both metagenomic and metatranscriptomic differentiation—measured as changes in the relative abundance of gene and transcript copies at the level of OGs—indicate that on a global scale, taxonomic composition largely shapes the composition of gene functional content. Organismal composition also dominates over gene regulatory variations in shaping community-level transcriptomic compositions across ecological boundaries. Figure 3 Latitudinal Partitioning of Global Ocean Microbiome Compositions The schematic on the left illustrates the underlying concept of the split moving-window analysis of ecological differentiation ( Ludwig and Cornelius, 1987 ). It consists of a comparison of the pairwise distances between communities on opposite sides of a putative boundary with the pairwise distances between communities on the same side. A high differentiation value captures an increase in the distance between the two sides of the boundary compared with the distances within each side. The analysis was conducted with a window width of 10 samples and shows an ecological boundary centered around 60°N based on the taxonomic composition (gray, relative abundance of OTUs), metagenomic composition (orange, per-cell abundance of genes), and metatranscriptomic composition (blue, relative per-cell abundance of transcripts) of prokaryote-enriched samples from surface (SRF) and deep chlorophyll maximum (DCM) waters (both belonging to the epipelagic layer). A similar pattern is evident for the southern hemisphere; however, the limited number of samples precluded detection of an ecological boundary. Significance was determined using 99% confidence intervals computed with 10,000 random permutations of the latitude values. Vertical lines represent the window of the latitudinal range of significant values. The insufficient number of samples and latitudinal coverage prevented us to perform this analysis for the mesopelagic layer. See also Figure S4 . Figure S4 Differential Abundance of the Dominant OTUs along the Latitudinal Gradient, Related to Figure 3 Latitudinal niche value (i.e., the abundance-weighted mean absolute latitude) for the 60 most abundant OTUs in the epipelagic subset of samples. Latitudinal niche values significantly higher and lower than the value expected from a random distribution of abundances (represented by the horizontal bold lines; see STAR Methods ) are color coded. The dot size is proportional to the mean relative abundance of each OTU. Indeed, we found that all community-compositional measures were highly correlated ( Figure S5 ), and their variability in the epipelagic ocean was, among a set of 27 environmental parameters, best explained by seawater temperature ( Figure 4 A). This result complements earlier reports of temperature as an important factor driving the taxonomic composition of ocean microbial communities ( Fuhrman et al., 2006 ), which was corroborated by a later analysis of a globally distributed set of samples that accounted for geographic effects and disentangled temperature from other environmental parameters to confirm that it acts as a key driver of taxonomic and gene functional compositions in epipelagic, non-polar open ocean waters ( Sunagawa et al., 2015 ). In fact, the identification of an ecological boundary starting at 40°N and peaking at 60°N coincides with a steep temperature decrease between the North Atlantic and Arctic waters that were sampled ( Figure S6 ) and relates to additional oceanographic features. At ∼40°N/S, the 15°C annual-mean isotherm effectively delineates the permanently stratified ocean from the subpolar and polar regions ( Behrenfeld et al., 2006 ), while winter mixing in the North Atlantic is the strongest (deepest mixed layer depth) at ∼60°N ( Montégut et al., 2004 ). The ecological boundary we describe here for microbial community compositions could thus be due to physico-chemical changes driven by the variability in the vertical mixing of oceanic water masses, which is linked to differences in sea surface temperature. Figure S5 Correlations between the Taxonomic, Metagenomic, and Metatranscriptomic Composition, Related to Figure 4 All pairwise correlations between the Euclidean distance of the (log 2 -transformed) taxonomic, metagenomic, and metatranscriptomic profiles were computed for 122 samples for which all three profiles were available. The correlation strength and significance were assessed using Mantel tests with 10,000 permutations. Figure 4 Patterns and Drivers of Global Ocean Microbiome Compositions across Depth Layers and between Polar and Non-polar Regions (A) Taxonomic, metagenomic, and metatranscriptomic composition of epipelagic samples (based on mi tags, and the normalized abundances of eggNOG-derived OGs from metagenomic and metatranscriptomic data, respectively) were related to each of 27 environmental factors using partial (geographic distance-corrected) Mantel tests with 10,000 permutations and Bonferroni correction. Pairwise comparisons of environmental factors are shown below, with a color gradient denoting Spearman’s correlation coefficients. Temperature is the best explanatory variable for all of the profiles in the epipelagic ocean (taxonomic profile: Pearson’s r = 0.75; metagenomic profile: Pearson’s r = 0.69; metatranscriptomic profile: Pearson’s r = 0.64; all p < 0.05), followed by oxygen concentration, which is highly correlated to temperature (Pearson’s r = −0.72). A more detailed description of the variables is available in https://doi.org/10.5281/zenodo.3473199 . (B) Compositional richness of polar and non-polar microbiomes across three depth layers. Taxonomic and functional metagenomic richness (numbers of OTUs and OGs, respectively) increases with depth, although the richness is consistently lower in polar samples than in non-polar samples (two-way ANOVA: p < 0.05 for depth layers and polar/non-polar, for both taxonomic and metagenomic functional richness). By contrast, there was no significant difference in functional metatranscriptomic richness (number of OGs), either across depths or between polar and non-polar samples (two-way ANOVA: p > 0.05 for depth layers and polar/non-polar). Violin plots represent the (mirrored) density distribution of the data with the median shown as a horizontal line. (C) Correlations among species richness (number of OTUs), functional metagenomic (metaG) richness and metatranscriptomic (metaT) richness (number of OGs). Data were rarefied before richness computation ( STAR Methods ). Pearson’s correlation was used for all comparisons (OTU-metaG; r = 0.78, p < 0.001; OTU-metaT: r = 0.16, p = 0.06; metaG-metaT: r = 0.39, p < 0.05). The solid line corresponds to the best linear fit. N.S., not significant (p > 0.05). See also Figures S5 and S6 . Figure S6 Latitudinal Distribution of Seawater Temperature in the Epipelagic, Related to Figure 4 Seawater temperature (°C) measurements (n = 528) at the surface (SRF) and the deep chlorophyll maximum (DCM) along the Tara Oceans course in relation to (A) raw latitude values and (B) bins of the absolute latitude. Data are available at https://doi.org/10.1594/PANGAEA.875576 . We next quantified metatranscriptomic richness (i.e., the unique number of OGs detected by cDNA sequencing), as a proxy for the diversity of transcribed gene functions, and compared this to taxonomic and metagenomic richness (i.e., the unique number of detected OTUs and OGs, respectively, detected by DNA sequencing). As measures of diversity, the latter two provide information about the stability ( McCann, 2000 ), functionality ( Cardinale et al., 2006 ), and possibly productivity ( Tilman, 1995 , Vallina et al., 2014 ) of ecological communities. In addition, we sought to quantify the fraction of the gene-encoded functional potential in a given community that is actually transcribed at a given time by comparing metatranscriptomic and metagenomic richness. Taxonomic and metagenomic richness were highly correlated, without showing signs of saturation, supporting the previous observation that functional redundancy in the marine ecosystem is rather low ( Fierer et al., 2013 , Galand et al., 2018 ), and both were found to be significantly lower in polar than in non-polar communities at all tested depth layers ( Figure 4 B). These data are congruent with studies suggesting a decrease in the taxonomic diversity of communities with increasing latitude ( Fuhrman et al., 2008 , Gregory et al., 2019 , Ibarbalz et al., 2019 , Sul et al., 2013 ) and an associated decrease in gene functional diversity, although other studies have also proposed alternative patterns of latitudinal diversity gradients ( Ghiglione et al., 2012 , Ladau et al., 2013 , Raes et al., 2018 ). In contrast, metatranscriptomic richness was not correlated with taxonomic richness and only poorly correlated with metagenomic richness, and no significant difference was found between polar and non-polar microbiomes or between any depth layers ( Figure 4 B). This unexpected disparity between metagenomic and metatranscriptomic richness patterns suggests that the non-transcribed proportion of a given metagenome is higher in mesopelagic waters and non-polar regions relative to epipelagic waters and polar regions. This could be due to a higher proportion of dormant or dead, and passively sinking, microbes in the mesopelagic compared to the epipelagic ocean. Alternatively, these observations may reflect the prevalence of genome streamlining in surface ocean waters ( Swan et al., 2013 ), where per genome, the number of genes is expected to be lower ( Mende et al., 2017 ). The proportion of transcribed genes is thus expected to be higher than in mesopelagic waters. Future studies will be required to determine whether the apparent saturation of simultaneously transcribed gene functions, despite increasing numbers of encoded gene functions, is a feature that is also common in microbial communities from other biomes. Differential Abundance and Expression of Biogeochemical Cycling Genes The pool of microbial community transcripts may vary along environmental gradients as a function of community turnover and/or changes in gene expression ( Figures S1 and S7 ; STAR Methods ). To disentangle the individual contributions of these mechanisms across environmental gradients for genes that are involved in ecologically relevant processes, we integrated 122 prokaryote-enriched, matched metatranscriptomes and metagenomes and quantified the differential abundances and expression levels for a set of biogeochemical marker genes across depth layers and between polar and non-polar waters ( Figure 5 ). Figure S7 Derivation of the Decomposition of a Metatranscriptome, Related to STAR Methods Mathematical basis for (A and B) the within-sample decomposition of metatranscriptomes (transcript copies / cell) into abundance (gene copies / cell) and expression (transcript copies / gene copy) components, and for (C) the between-sample decomposition of the Euclidean distance between metatranscriptomes (transcript abundance differences) into the abundance component (gene abundance differences), the expression component (expression differences), and an interaction term (abundance - expression covariation). See details in STAR Methods . Figure 5 Differences in Gene Abundance and Expression Determine Differential Transcript Abundances of Metabolic Marker Genes across Depth Layers and between Polar and Non-polar Regions (A and B) Differences in the abundance of genes and transcripts, and the gene expression level of metabolic marker genes (KOs) were determined (A) between epipelagic and mesopelagic layers and (B) between polar and non-polar regions. The data points show the differences in the mean transcript abundances, mean gene abundances, and mean gene expression (i.e., transcript abundance normalized by gene abundance) of KOs. Differences were computed using log 2 -transformed values ( STAR Methods ) and tested for significance by Mann-Whitney tests. Differences were considered significant if p values after Holm correction were smaller than 0.05. Only epipelagic samples were used for the data shown in (B). See also Figures S8 , S9 , S10 , and S11 . As a first step, we sought to validate both data quality and our analytical approach by testing whether patterns for genes involved in well-studied processes, including carbon fixation, photosynthesis, and nitrogen cycling could be observed. As expected, we found that the most differentially abundant transcripts between epipelagic and mesopelagic layers included those from the photosynthesis marker genes, psaA and psbA , and genes encoding the subunits of RuBisCO ( rbcL and rbcS ), the key enzyme required for carbon fixation ( Figure 5 A). Moreover, we observed that abundances of the rbcL and rbcS transcripts were highly correlated with those of psaA and psbB , which is consistent with the expectation that carbon fixation is primarily driven by photoautotrophs rather than chemoautotrophs ( Raven, 2009 , Shively et al., 1998 , Swan et al., 2011 ). This is further supported by the observation of low RuBisCO gene expression levels in mesopelagic waters, despite the presence of chemoautotrophs ( Figure S8 ). In addition to psbA , the abundances of other photosynthetic marker genes, including markers for the photosynthetic reaction center ( petC , petE , and petH ) and the cyanobacteria-specific antenna proteins ( apcA , apcF , cpcA , cpeA , and cpeT ), were lower in polar than in non-polar waters ( Figure 5 B). This result likely reflects the depletion of cyanobacteria in colder environments ( Marchant et al., 1987 ) ( Figure S4 ) and an underrepresentation of eukaryotic phototrophs in the prokaryote-enriched samples we analyzed here. Figure S8 Gene and Transcript Abundance of RuBisCO Subunits and PSI and PSII Marker Genes, Related to Figure 5 Distribution of whole-community (log 2 -transformed) (A) gene and (B) transcript abundances of the RuBisCO subunits ( rbcS and rbcL ) and the marker genes for photosystem I ( psaA ) and II ( psbA ) in the epipelagic and mesopelagic depth layers. Pairwise correlations based on the (C) gene and (D) transcript abundances of the four genes are shown below. All comparisons, except the ones denoted with N.S. in (A) and (B) were significant (p < 0.05 using Wilcoxon test and Holm correction for multiple comparisons). All Pearson correlations in (B) and (C) were significant (p < 0.05). With respect to nitrogen cycling, we detected both gene and transcript abundances for denitrification marker genes (napA, nirS, norB, and nosZ) to be enriched in mesopelagic versus epipelagic waters ( Figure 5 A). As expected for this predominantly anaerobic process ( Zehr and Ward, 2002 ), transcript abundances were particularly high in oxygen-depleted waters, although interestingly, similar transcript levels were also observed in some well-oxygenated Arctic water samples ( Figure S9 ). Transcripts of nitrogen fixation marker genes (nifK, nifH, and nifD) were more abundant in non-polar than in polar regions, with the highest abundances detected in waters between 20° and 35° (absolute latitude) with low nitrate and nitrite concentrations ( Figure S10 ). These data generally agree with the long-standing expectations that nitrogen fixation activity is higher under conditions of nitrogen limitation and is primarily driven by cyanobacteria in tropical and subtropical regions ( Dixon and Kahn, 2004 , Stal, 2009 ). However, more recent studies have provided additional evidence for an extended geographic and depth range ( Blais et al., 2012 , Harding et al., 2018 , Moisander et al., 2017 ) and for a wider taxonomic breadth of nitrogen fixing organisms including non-cyanobacterial heterotrophic diazotrophs ( Bombar et al., 2016 , Delmont et al., 2018 ). Given these findings, we further investigated the biogeography of the nifH gene in more detail and determined which organisms not only encode this gene, but also express it. Specifically, we analyzed the distribution of nifH gene and transcript abundances among 24 nifH-encoding “species” that were detected in the 122 matched metagenomes and metatranscriptomes. From this analysis, we found that a number of Gamma- and Deltaproteobacteria, for which genomes have recently been reconstructed ( Delmont et al., 2018 ), were not only abundant, but also among the top contributors to the nifH transcript pool in the studied samples ( Figure 6 ). Additionally, for the first time, to our knowledge, we detected nifH gene expression in mesopelagic Arctic waters and reconstructed the nif operon-containing genome of its carrier ( http://doi.org/10.5281/zenodo.3352180 ; STAR Methods ), a candidate heterotrophic Deltaproteobacterium or a member of the Myxococcota phylum according to a recent proposal for a standardized bacterial taxonomy ( Parks et al., 2018 ), that awaits further characterization. Figure S9 Transcript Abundance of Denitrification Marker Genes along the Oxygen Gradient, Related to Figure 5 The log 2- transformed transcript abundances of nirS , norZ , nosB, and napA in relation to the oxygen concentration at the sampling location, showing a high transcript abundance in samples taken from anoxic waters (< 100 μM) and interestingly, from oxygenated waters at stations 206, 208, and 210. The depth layer (EPI or MES) and polar/non-polar nature of the sample are coded as the symbol type and color, respectively. The dot size is proportional to the concentration of NO 2 and NO 3 (μM) when available. Figure S10 Expression and Transcript Abundance of the nifH , nifD , and nifK Genes in Relation to Nitrate and Nitrite Concentration, Related to Figure 5 Gene expression and transcript abundance of the nifH, nifD, and nifK genes in relation to the total nitrate plus nitrite concentration (μM), showing a fast decay of gene expression and transcript abundance with increased in nitrate/nitrite concentrations from 0 to 0.2 μM at absolute latitudes between 20° and 35°. Solid lines correspond to the result of local regression. Figure 6 Relative Gene and Transcript Abundance of 24 Nitrogenase Genes ( nifH ) Representing nifH -Encoding “Species” (A–D) Relative gene (orange) and transcript (light blue) abundance distributions of the 24 nifH genes from the OM-RGC.v2 that were detected in 122 matched metagenomes and metatranscriptomes (A) are shown and broken down by latitude (B) and by depth (C) of the sample origin. Genes (IDs in the bottom panel) were annotated using a nifH -specific database (see STAR Methods ). Boxplots in (A–C) show the median values as horizontal lines, interquartile ranges as boxes with whiskers extending up to 1.5 times the interquartile range, and all values overlaid as individual data points. Colors denote phylum-level taxonomic annotations, naming corresponds to finer grain taxonomy or database-specific identifiers (D), and stars indicate genes that were previously identified in MAGs of heterotrophic bacterial diazotrophs (HBDs) ( Delmont et al., 2018 ). The genome containing a nifH gene for which transcripts were detected in the mesopelagic layer in the Arctic (OM-RGC.v2.019519152, bold) was reconstructed (see STAR Methods and http://doi.org/10.5281/zenodo.3352180 ). Horizontal dashed lines denote the latitude and depth that were used to define polar and non-polar (B) and epipelagic and mesopelagic waters (C), respectively. In spite of the potential biases inherent to our approach that are related to the collection of spatially discrete data over a period of more than 3 years and to the sampling process itself (e.g., unaccounted effect of seasonality or potential changes in transcript abundances during the sampling process), we were able to corroborate expected patterns of metabolic processes using metatranscriptomic data at global scale. In addition to validating our methods, we demonstrated how our community-centric approach for analyzing metatranscriptomes can be used in conjunction with metagenomic data, and furthermore, bridge to new genome-resolved insights. Building on the robustness of our analysis, we next focused on disentangling the mechanisms that underpin the differences in community transcriptomes across depth and latitude. Notably, we observed cases in which transcript abundance changes could be mainly attributed either to differences in gene abundance or gene expression or a combination of these mechanisms. As described above, the enrichment of transcripts from denitrification marker genes in mesopelagic versus epipelagic waters are mainly driven by changes in gene abundance ( Figure 5 A). In this case, gene abundance changes, due to environmental filtering of organismal community composition in response to higher nitrate and nitrite concentrations in mesopelagic waters, dominate the observed community transcriptomic differences. Conversely, a higher transcript abundance of marker genes for anaerobic dissimilatory sulfate reduction ( aprA and aprB ) in epipelagic waters is driven by an increased expression of these genes, despite no significant differences in the abundance of these genes between depth layers ( Figure 5 A). A taxonomic breakdown shows that 39% and 59% of aprA and aprB genes were encoded by Proteobacteria, and only 2% of each gene could be assigned to taxa containing known sulfate reducers (Archaea, Firmicutes, Nitrospirae, and Deltaproteobacteria) ( Muyzer and Stams, 2008 ). These results suggest that the significance of alternative uses for aprA and aprB in oxic waters, namely to detoxify cells by catalyzing the oxidation of sulfite accumulated in the cytoplasm, as described for clades such as SAR11 and SAR116 ( Meyer and Kuever, 2007 , Smith et al., 2016 ), may be of global relevance. A more complex scenario for observing differences in transcript pools is exemplified by a number of marker genes for assimilatory sulfate reduction ( cysD , cysH , cysI , cysJ , and cysN ), for which the observed differences across the latitudinal gradient (i.e., higher transcript abundances in non-polar versus polar regions) result from a combination of community turnover and gene expression changes. In this case, the increased transcript abundance in non-polar waters results from higher expression levels, despite a lower abundance of genes. Interestingly, we found the transcript abundance of these marker genes to be anticorrelated with that of dmdA ( Figure S11 ), the key gene for the demethylation of dimethylsulfoniopropionate (DMSP) ( Howard et al., 2006 ), which results in incorporation of carbon and sulfur into bacterial biomass ( Kiene et al., 1999 ). Based on these data, we hypothesize that the global-scale expression of the assimilatory sulfate reduction pathway may be downregulated in response to the availability of DMSP, which is used by prokaryotes as an alternative source for sulfur assimilation ( Kiene et al., 2000 ). Notably, if turnover and differential gene expression are both operative, relying on gene abundance alone may lead to false predictions including patterns that would suggest the opposite of what is manifested at the transcript level (e.g., non-photosynthetic carbon pathways with higher epipelagic expression levels but higher mesopelagic gene abundances of mct and abfD ). Figure S11 Correlation between Assimilatory Sulfate Reduction Marker Genes and the dmdA Gene, Related to Figure 5 Transcript abundance and expression of the genes involved in the assimilatory sulfate reduction pathway in relation to the transcript abundance of the dmdA gene involved in the dimethylsulfoniopropionate (DMSP) demethylation pathway. Pearson correlation was used to test for significance of the correlation. Pearson r values and significance are shown on the plot. Log 2 -transformed data were used in all cases. The correlation with the transcript abundance was significant for all genes and was especially high (−0.73) for cysD and cysN, the genes encoding the initial step of the pathway (i.e., the reduction of sulfate). Turnover Dominates over Gene Expression Differences in Polar Water Communities In light of global climate change, a better understanding of how ocean microbial communities will respond to ongoing changes is urgently needed ( Cavicchioli et al., 2019 , Overland et al., 2018 ). In particular, the Arctic region has experienced some of the highest ocean surface water temperature anomalies recorded to date ( Hoegh-Guldberg and Bruno, 2010 ). Ocean warming models (scenario RCP 8.5, business as usual) predict that mean surface water temperatures will increase by 2°C to 5°C in the Arctic by the end of the century ( Alexander et al., 2018 ), highlighting a critical need to better understand how these changes will impact microbial communities in this region. Given that these projections focus on surface temperature changes and due to their major contribution to biogeochemical cycles ( Field et al., 1998 ), we sought to assess the response of epipelagic communities to environmental variation, as reflected by measurable differences in their metatranscriptomic composition, and subsequently to use these spatially discrete data to hypothesize on future projections. Specifically, we aimed to disentangle ( Figure S7 ; STAR Methods ) whether differences in microbial community transcriptomes are impacted more strongly by community turnover and/or by gene expression changes along the temperature gradient at their sampling locations. To this end, we divided all samples into groups of 15 samples (bins) using a sliding window along the temperature gradient, so that each group reflected the range of ocean warming expected before the end of the century (median temperature difference within each bin: 1.6°C; Figure S12 A). We then quantified the different mechanisms of metatranscriptome changes within each bin ( Figure 7 ; STAR Methods ) and found that in warmer epipelagic waters, the relative contribution of community turnover to metatranscriptomic compositional dissimilarities is significantly lower than that of gene expression changes. In contrast, the effect of community turnover in colder (predominantly Arctic) waters is higher or in the same range as gene expression changes ( Figure 7 A). Overall, community turnover was found to be significantly higher in polar communities than in non-polar communities (p < 0.001), whereas gene expression changes displayed the opposite pattern (p < 0.001) ( Figure 7 B). Interestingly, the shift in the relative contributions of the different mechanisms of metatranscriptome changes occurs at ∼15°C and therefore coincides with the ecological boundary previously identified, which, as such, not only delineates communities differing in their composition but also in the mechanism shaping their transcript pool. We further found that the effect of temperature was greater than that of other environmental variables, such as nitrate/nitrite concentrations and salinity ( Figure S12 ), suggesting a higher acclimatory capacity of microbial communities in warm than in cold epipelagic waters in response to temperature variations. Figure S12 Temperature Dominates over Other Environmental Variables in Structuring the Relative Contribution of Community Turnover and Gene Expression Changes to Metatranscriptomic Differences between Epipelagic Communities, Related to Figure 7 Panel (A) mirrors the data in Figure 7 A, so that it represents the groups of 15 samples (bins) along the temperature gradient on the x axis. The y axis, however, captures the distribution of the temperature differences within each bin. Notably, the distributions of these differences are highly similar in polar and non-polar waters. This indicates that the higher relative contribution of turnover in polar waters and gene expression changes in non-polar waters occurs for a similar range of temperature differences. (B) The distribution of the interaction component (see Equation 1 in STAR Methods ) for all the polar-to-polar and non-polar-to-non-polar comparisons across the bins are not significantly different from each other (Wilcoxon test), which indicates that the absolute values of turnover and gene expression changes are comparable between polar and non-polar communities ( Figure 7 B). Panel (C) is based on Figure 7 A and serves as an explanatory schematic for panel (D). To evaluate the influence of an environmental parameter on the relative contribution of community turnover and gene expression changes, a similar analysis to the one in Figure 7 A was performed. A score was attributed to each parameter as the sum of the deviation of each bin from 1 (where the effect of both mechanisms is identical). The deviation of each individual bin is visualized as a gray line. The results are summarized in panel (D) for the environmental parameters that were tested. The vertical lines indicate the distribution of this score for 100 random binnings (solid line denotes the median value and dashed lines represent the 95% interval of the distribution). As a result, we identify that daylength, temperature and chlorophyll concentrations have significant effects on the relative contributions. We further investigated these parameters, by assessing the distribution of environmental variation for polar and non-polar regions across the bins [panels (E), (G), and (I)], and the relationship between the relative contributions (of community turnover and gene expression changes) and the variation in the environmental parameter across the whole (unbinned) dataset [panels (F), (H), and (J)]. The left-side [(E), (G), and (I)] aims at answering whether the difference in regimes that are observed between polar and non-polar regions may simply be due to a different range of environmental variation. The distributions display little differences in the case of temperature, while they are strongly contrasted for daylength and chlorophyll concentrations. Furthermore, (F), (H), and (J) provide a direct estimation of the relationship of the relative contributions of community turnover and gene expression changes with the environmental distance. Based on linear models, temperature differences capture most of the variance, both in polar and non-polar regions. In contrast, daylength and chlorophyll concentrations show a weaker or no trend, especially in polar regions (despite a wide range of variation). Overall, this confirms that among the parameters tested, temperature is the best explanatory variable for the difference in the relative contribution of community turnover and gene expression changes observed between polar and non-polar epipelagic communities. Figure 7 Relative Contributions of Community Turnover and Gene Expression Changes to Variations in Metatranscriptome Composition Determination of the relative contributions of community turnover and gene expression changes to variations in the metatranscriptome composition requires the decomposition of metatranscriptomic distances between communities ( Figure S7 ; STAR Methods ). Specifically, the relative contribution is determined as the ratio of the gene abundance-based distance (community turnover) and the gene expression-based distance (gene expression changes) between two metatranscriptomes. (A) The relationship of the ratio with temperature was analyzed by dividing the epipelagic samples into groups (bins) of 15 samples each using a sliding window along the temperature gradient. For each bin, we report the median ratio (among all the pairwise comparisons within each bin) as a function of the median temperature of the samples present in the bin. The significance is determined by a Wilcoxon test comparing the within-bin distribution of the ratios to 1 (in which case the relative contributions of community turnover and gene expression changes are the same). The Holm correction was used to adjust for multiple testing. The ratio was considered to be significantly different from 1 if p < 0.05. (B) The inner panel represents the difference for community turnover and gene expression changes between polar and non-polar regions. The distributions capture the distances of each component for all pairwise comparisons of polar and non-polar epipelagic samples. Violin plots represent the (mirrored) density distribution of the data with the median shown as horizontal line. Significance was tested by the Wilcoxon test; ∗∗∗ p < 0.001. See also Figure S12 . Finally, by extrapolating our results from spatially discrete data to potential consequences of climate change ( Blois et al., 2013 ), we hypothesize that the relative impact of organismal composition changes on microbial community transcriptomes will be greater in polar than in non-polar waters. This extrapolation, however, needs to be interpreted within the limitations of the data analyzed here, namely that it cannot account for the evolutionary adaptation of microbial communities to gradual changes with time. As such, further studies resolving long-term temporal dynamics of metatranscriptome changes are required to improve our understanding of the contributions of community turnover and gene expression changes in the context of environmental changes. Notwithstanding, the present results provide a first global-scale evaluation of the mechanisms underpinning the changes in community transcriptomes as well as a framework for future work. Conclusions Large-scale oceanographic sampling expeditions, such as the World Ocean Circulation Experiment (WOCE) or GEOTRACES ( Anderson et al., 2014 , Koltermann et al., 2011 , Woods, 1985 ) have been extremely valuable in building our understanding of the ocean circulation, and the distribution of major nutrients and elements including trace metals, as well as their contribution to the climate system. However, our geochemical and physical knowledge of the ocean remains incomplete without incorporating the processes that regulate biogeochemical cycles at planetary scale ( Falkowski et al., 2008 ). Analyzing the repertoire of genes and transcripts from environmental samples can inform us about the potential and activity of microbial communities that drive these cycles at global scale and thus help us to understand the intertwined processes that shape the physico-chemical state of the ocean through biological activity. In this study, we describe global biogeographical patterns of microbial community transcriptome compositions and demonstrate how changes in these compositions can be attributed to community turnover and/or gene expression changes as the underlying mechanisms. Assessing the mechanisms that underlie such compositional differences, as demonstrated here, can help us to determine whether changes in the molecular activities of microbial communities are regulated by gene expression changes or by a turnover of organisms containing genomic modifications that arose over evolutionary time. In addition, an improved understanding of the ecological factors that drive community compositional and diversity changes can help us to better predict how ocean microbial communities will respond to environmental changes. For example, the consistent identification of temperature as a major explanatory factor for global-scale community-level differences in genomic ( Sunagawa et al., 2015 ) and transcriptomic (this study) composition, as well as taxonomic diversity ( Gregory et al., 2019 , Ibarbalz et al., 2019 ), has wide-ranging implications, in particular for the Arctic Ocean, given the current projections of disproportionately high warming rates in this region ( Alexander et al., 2018 , IPCC, 2014 ). Notably, the analyses of this study were enabled by a systematic, highly contextualized, pan-oceanic set of metagenomic and metatranscriptomic data that, along with the OM-RGC.v2, complements other large-scale datasets that have been developed for eukaryotes ( Carradec et al., 2018 , Ibarbalz et al., 2019 ), prokaryotes ( Biller et al., 2018 ), and viruses ( Gregory et al., 2019 ). Together, these will pave the way for an eco-systems level understanding of ocean plankton diversity, function, and activity across boundaries of organismal size ranges. To reach this goal, it will be important to integrate temporal meta-omics data, ideally from global observations, to account for seasonal variations and other concomitant environmental changes, such as increased stratification, acidification, nutrient availability, and deoxygenation of the oceans ( Bopp et al., 2013 , Schmittner et al., 2008 ). Such concerted efforts are required to further refine gene-to-ecosystem models ( Coles et al., 2017 , Garza et al., 2018 , Guidi et al., 2016 ) and to inform environmental and climate policies ( Le Quéré et al., 2018 ), which must consider not only how microorganisms are impacted by but also how they may affect anthropogenic climate change ( Cavicchioli et al., 2019 )."
} | 14,889 |
26454280 | PMC4734042 | pmc | 7,723 | {
"abstract": "Summary: Analyzing the functional profile of a microbial community from unannotated shotgun sequencing reads is one of the important goals in metagenomics. Functional profiling has valuable applications in biological research because it identifies the abundances of the functional genes of the organisms present in the original sample, answering the question what they can do. Currently, available tools do not scale well with increasing data volumes, which is important because both the number and lengths of the reads produced by sequencing platforms keep increasing. Here, we introduce SUPER-FOCUS, SUbsystems Profile by databasE Reduction using FOCUS, an agile homology-based approach using a reduced reference database to report the subsystems present in metagenomic datasets and profile their abundances. SUPER-FOCUS was tested with over 70 real metagenomes, the results showing that it accurately predicts the subsystems present in the profiled microbial communities, and is up to 1000 times faster than other tools. Availability and implementation: SUPER-FOCUS was implemented in Python, and its source code and the tool website are freely available at https://edwards.sdsu.edu/SUPERFOCUS . Contact: \n redwards@mail.sdsu.edu Supplementary information: \n Supplementary data are available at Bioinformatics online.",
"conclusion": "4 Conclusions Here, we present SUPER-FOCUS, an agile solution to identify the subsystems present in metagenomic samples that first determines the taxonomic composition of the entire metagenome by using FOCUS, and uses this knowledge to create, on the fly, a reduced database only containing the subsystems present in the organisms found. This makes SUPER-FOCUS a faster and still accurate tool to profile the functional subsystems in metagenomes. SUPER-FOCUS reports very similar results to currently available tools, but does so faster and using less memory.",
"introduction": "1 Introduction Prokaryotes and the viruses that infect them are the most abundant organisms on earth ( Whitman et al. , 1998 ), and it is important to understand both who they are and what they are doing. In many environments, the majority of the microbes cannot be cultured by using standard laboratory techniques, and metagenomics is the preferred way to study them as a whole community ( Handelsman, 2004 ). Next-generation DNA sequencing (NGS) technologies have sped up the sequencing process, reduced the cost, increased the sampling resolution and opened new horizons in the biological sciences ( Zhang et al. , 2011 ). Understanding microbial communities is important in many areas of biology. For example, metagenomes can distinguish taxonomic and functional signatures of microbes associated with humans ( Consortium, 2012 ), sponges ( Trindade-Silva et al. , 2012 , 2013 ), red seaweed ( Oliveira et al. , 2012 ) and diseased and healthy states of corals ( Garcia et al. , 2013 ). Functional annotation of metagenomic reads normally requires the alignment of sequences to a large database of annotated sequences to identify homologs ( Mendoza et al. , 2015 ). There are many databases for annotations at the functional system or pathway level, including the SEED ( Overbeek et al. , 2005 ) which contains subsystems (sets of protein families with a similar function), and the large metabolic pathway databases KEGG ( Kanehisa and Goto, 2000 ) and MetaCyc ( Caspi et al. , 2010 ). Many of the tools for functional profiling are slow ( Lindgreen et al. , 2015 ), suggesting that there is an opportunity for improvements of these tools. Currently, available tools generally use either homology, i.e. by aligning the metagenomic sequencing reads against an annotated reference database, or use exact matches (k-mers) to link metagenomic sequencing reads to the annotated sequences. Homology-based metagenome annotation tools programs are still frequently based on pre-NGS algorithms such as BLAST ( Altschul et al. , 1997 ) or BLAT ( Kent, 2002 ) to identify the best hit in a large database, although new homology search algorithms such RAPSearch2 ( Zhao et al. , 2012 ) and DIAMOND ( Buchfink et al. , 2015 ) have recently been developed to reduce the run time. MG-RAST ( Meyer et al. , 2008 ) and MEGAN 5 ( Mitra et al. , 2011 ) both align sequences to a reference database to profile the metagenomic sample. MG-RAST first predicts the open reading frames (ORFs) on the metagenomic sequence using FragGeneScan ( Rho et al. , 2010 ) and then aligns the translated amino acid sequences to the M5NR database via BLAT. MEGAN accepts as input the tabular results files created from programs such as blastx/blastp, DIAMOND, or RAPSearch2 to the NR database, and creates taxonomic and/or functional profiles based on the search output. Exact match or k-mer based approaches use oligonucleotides to identify the hits in a metagenome. For example, real time metagenomics (RTMg) ( Edwards et al. , 2012 ) identifies all words of length k (where k is typically between 7 and 12 amino acids) that are a unique signature for a set of functionally related proteins, and uses them to profile the functions present in the metagenomic sample. This approach has also been successfully applied to assign taxonomic labels to metagenomic sequences using a length of k of 31 nt ( Wood and Salzberg, 2014 ; Ounit et al. , 2015 ). We developed a novel approach, named SUPER-FOCUS, which classifies each sequence in the metagenome into a subsystem. SUPER-FOCUS aligns all the input data against a reduced database with contains only the subsystems present in the organisms in that metagenome. The speed up derives from three improvements compared with the standard metagenome annotation pipelines. First, the SEED database was clustered using CD-HIT ( Huang et al. , 2010 ) using a similar approach as previously discussed and applied to the Genbank NR database ( Li et al. , 2012 ); second, the metagenomic query sequences are profiled using FOCUS ( Silva et al. , 2014 ), an ultra-fast tool that identifies the organisms in the metagenome; and finally, comparisons are performed using RAPSearch2 which is ∼2–3 times faster than BLAT and 100 times faster than blastx, but has no reduction in sensitivity or specificity when compared with BLAST ( Berendzen et al. , 2012 ). We compare the performance of SUPER-FOCUS to RTMg, MEGAN and MG-RAST using different reduced databases and over 70 metagenomic datasets of different sizes and from different environments. The results shows that SUPER-FOCUS speeds up the process of sequence functional annotation 37, 60 and 1000 times faster than RTMg, MEGAN and MG-RAST, respectively. We apply SUPER-FOCUS to a novel dataset from a coral reef environment and show that the taxonomy is conserved across islands while functions adapt to local conditions.",
"discussion": "3 Results and discussion 3.1 Validation of clustered database and SUPER-FOCUS evaluation Prior to testing SUPER-FOCUS we independently validated our database construction and size reduction. The HMP and viromes testing set metagenomes were aligned against the SUPER-FOCUS database DB_100 using blastx, as described in the methods, and each query sequence was assigned to a subsystem using the SUPER-FOCUS best-hit method. Next, DB_100 blastx's assignments were assumed to be the right answer, and same testing sets were aligned against DB_100, DB_98, DB_95 and DB_90, but now using RAPSearch2 as the aligner with the same parameters previously used with blastx. RAPSearch2 was run with 24 threads in the sensitive and fast modes using the SUPER-FOCUS workflow. RAPSearch2 without the SUPER-FOCUS workflow using database DB_100 represents how a regular user would profile a metagenomic dataset only using the complete SEED and a fast aligner. This analysis was added to show that part of the loss of sensitivity and precision from SUPER-FOCUS profiling comes from RAPSearch2. For the 50 HMP metagenomes with short reads, we measured both the sensitivity ( Fig. 3 a for sensitivity using subsystem level 1 classification, Supplementary Fig. S2 for level 2 and 3 classifications) and precision ( Fig. 3 b) for precision using subsystem level 1 classifications, Supplementary Fig. S3 for level 2 and 3 classifications.\n Fig. 3. Percent classification sensitivity ( A ) and precision ( B ) of level 1 subsystems and speed of RAPSearch2 and SUPER-FOCUS using different databases and parameter modes. This analysis was based on a comparison of 50 HMP metagenomes, where blastx assignments using DB_100 were considered to be the true answer We used blastx as our notion of ‘truth’ as it is the most widely used algorithm in metagenomics analysis. For the 50 HMP metagenomes the average processing time (sequences/minute) is compared with the sensitivity at each of the three subsystem levels. The more sensitive the profile, the longer it is going to take, and the more broader the categories (subsystems level 1), the more sensitive the results are. Figure 3 also shows that the SUPER-FOCUS approach was slightly faster, but less sensitive when compared with RAPSearch2 without the SUPER-FOCUS workflow alignments; it was faster because on average 88.53 ± 5% of the 1290 subsystems in the SEED database were used to profile all the metagenomes in the testing set ( Fig. 4 ) in step 2 of the SUPER-FOCUS pipeline, and the misclassified sequences explain the loss of sensitivity as described later.\n Fig. 4. Percentage of level 3 subsystems present in all the testing set metagenomes predicted by SUPER-FOCUS Overall, there was <1% loss of sensitivity ( Supplementary Fig. S4a ) and precision ( Supplementary Fig. S4b ) between all the levels, which shows the efficiency of using the clustered database presented in this paper and the SUPER-FOCUS approach compared with blastx searches. 3.2 Understanding SUPER-FOCUS misclassifications In order to understand the SUPER-FOCUS misclassifications when compared with RAPSearch2 without the SUPER-FOCUS workflow, two confusion matrices were generated to compare predicted and real sequence annotations: Figure 5 a presents the RAPSearch2-Sensitive results using the DB_100, and the only significant loss of sensitivity was 6.45% of sequences that were supposed to be classified into the ‘Plant cell walls and outer surfaces’ subsystems but were not classified; this loss of sensitivity is explained because RAPSearch2 is based on a reduced amino acid alphabet of 10 symbols, which makes the tools less sensitive, while blastx uses a complete amino acid alphabet. Figure 5 b shows the results for SUPER-FOCUS-Sensitive using DB_100, and now 67.74% of the sequences that were supposed to be classified into the ‘Plant cell walls and outer surfaces’ subsystems and 6.03% of the sequences that were supposed to be classified into the ‘Photosynthesis’ subsystems were not classified.\n Fig. 5. Confusion matrix displaying the percentage of correct assignments in each level 1 subsystem for the 50 HMP metagenomes. ( a ) Shows the RAPSearch2 assignments in the sensitive mode to DB_100. ( b ) Shows the SUPER-FOCUS assignments in the sensitive mode to DB_100 To investigate the SUPER-FOCUS misclassifications, those sequences were aligned against the NR-database using blastx and an E-value cutoff of 1 e − 5 . The best-hit was selected for each sequence, and the taxonomic classification for each hit was recovered using Biopython ( Cock et al. , 2009 ). The results show that ∼25% of the wrong assignments were from the Eukaryota and ∼1.3% of the wrong assignments were derived from viruses. The rest of the misclassified sequences were associated with Bacteria. Ninety-nine percentage of those were not identified by FOCUS and 78% were not identified by MetaPhlAn because neither FOCUS nor MetaPhlAn include those bacteria in their databases. Of the 22% that were identified by MetaPhlAn (used to analyzed the data in the HMP paper) ( Segata et al. , 2012 ), just over half of those (57%) had <2% relative abundance, suggesting that both FOCUS and MetaPhlAn are missing the rare species in the environment ( Supplementary Table S6 ). SUPER-FOCUS uses FOCUS in its pipeline, a tool that was developed to taxonomically profile microbial data. The FOCUS database only contains bacterial and archaeal genomes, which explains the misclassification of metagenomic reads from other microbial clades. We showed that the real metagenomes used herein contained viral and eukaryotic sequences, even after the sequences were filtered. For example, while human contamination from the HMP data was already removed using BMTagger ( Rotmistrovsky and Agarwala, 2011 ), we were still able to identify human reads with the DeconSeq tool ( Schmieder and Edwards, 2011 ). These contaminations affect biological conclusions ( Weiss et al. , 2014 ) and lead to increased computing time. SUPER-FOCUS was designed to only classify microbial data. The SUPER-FOCUS pipeline guarantees a more accurate microbial functional analysis and does not classify eukaryotic or viral sequences. Thus, if hits to the Eukaryotic and Viral Kingdoms are ignored, the SUPER-FOCUS approach would present a better profile than the one present in Fig. 3 a as shown in Fig. 6 for sensitivity at level 1, Supplementary Fig. S5a for level 2 and 4 (b) for level 3.\n Fig. 6. Classification sensitivity using level 1 classifications and speed comparison of 50 HMP metagenomes using RAPSearch2 and SUPER-FOCUS using different databases and modes, but removing Eurkaryota and viral assignments. blastx assignments using DB_100 were considered to be the true answer 3.3 Comparison of SUPER-FOCUS with other tools SUPER-FOCUS was compared with RTMg, MEGAN and MG-RAST, and all the four tools were tested using default parameters. MEGAN 5.10 and SUPER-FOCUS were run using default parameters and their default reference database, either online (MG-RAST and RTMg) or using one core on a server with 24 processors × 6 cores Intel(R) Xeon(R) CPU X5650 @ 2.67 GHz and 189 GB RAM. MEGAN uses the NR database, and it was downloaded on January 22, 2015. MG-RAST uses the M5NR as database, and its last update is not known because MG-RAST does not provide that information. RTMg is a web server which uses the SEED as database and was updated on November 8, 2013. For the three viromes that were sequenced using 454 technology and thus had longer reads, SUPER-FOCUS sensitivity and precision were evaluated against the annotations based on the blastx searches against DB_100 as the true assignments, and its runtime was compared with MEGAN, MG-RAST and RTMg. Here, RAPSearch2’s performance was tested using different numbers of threads (24, 18, 12 and 6) ( Supplementary Fig. S6 ), blastx was also used to align the sequences because RAPSearch2 is known to be less sensitive for 454 data ( Buchfink et al. , 2015 ), and because DIAMOND was designed for large datasets, and it is slower than blastx for small datasets. RAPSearch2 is less sensitive for long sequences as shown in Fig. 7 a. SUPER-FOCUS had a high precision, using all the databases, of ∼98.4 ± 1% using blastx with level 1 classifications ( Fig. 7 b). Subsystems sensitivity and precision measurements for level 2 and 3 subsystem classifications are shown, respectively, in Supplementary Figs S7 and S8 . BLAST is slow, thus the blastx results for Fig. 7 was generated in a cluster at the San Diego State University facilities, and the run time was approximated based on the knowledge that RAPSearch2 is ∼100 times faster than blastx.\n Fig. 7. Classification sensitivity ( a ) and precision ( b ) percent using level 1 and speed comparison of three viromes using RAPSearch2 and SUPER-FOCUS using different databases and modes. blastx assignments using DB_100 were considered to be the true answer In the timing tests, SUPER-FOCUS had the fastest profiling using RAPSearch2 in its fast mode. RTMg was faster than RAPSearch2 sensitive ( Fig. 8 ). MEGAN can be enhanced by using RAPSearch2, and the underlying framework hinders MG-RAST.\n Fig. 8. Run time comparison for the three marine viromes using SUPER-FOCUS, RTMg, MEGAN and MG-RAST Interestingly, despite the absence of phages in the FOCUS database, SUPER-FOCUS was able to predict correctly the subsystems in the viromes. We hypothesize that FOCUS predicted the microbial host for the phages, and from those genera that carry the subsystems present in the viral metagenomes. Phages are very diverse which means that the subsystems associated with phages (‘Phages, Prophages, Transposable elements, Plasmids’, ‘Virulence’ and ‘Virulence, Disease and Defense’), did not cluster well in the SUPER-FOCUS database creation. This lack of clustering explains why the sensitivity across different databases does not change much in Fig. 7 . For the one big data metagenome analyzed here, the runtime of SUPER-FOCUS was only compared with MEGAN, MG-RAST and RTMg, and due to the large number of sequences (30 917 457 reads), DIAMOND was used as default aligner for the SUPER-FOCUS and MEGAN tools. As shown in Fig. 9 , SUPER-FOCUS was the fastest tool followed by RTMg, MEGAN, and then MG-RAST. It is important to point out that SUPER-FOCUS (and MEGAN) used 24 threads, and even if the program had been set to use less threads, it would be still expected to be faster than RTMg because as shown in Fig. 8 SUPER-FOCUS is ∼4.4 (most sensitive) and 37.7 (fastest) times faster than RTMg. The SUPER-FOCUS profiling was compared with MEGAN, RTMg and MG-RAST as shown in Fig. 10 , and the results show that the three tools are comparable to each other, except to MG-RAST which apparently over predicted hits to the ‘Clustering-based subsystems’ and RTMg which did not report any hits this subsystem as they are ignored.\n Fig. 9. Run time comparison for the one big data metagenome using SUPER-FOCUS, RTMg, MEGAN and MG-RAST \n Fig. 10. Comparison of level 1 subsystems profile of one big data metagenome using SUPER-FOCUS, RTMg, MEGAN, MG-RAST and blastx that are considered to be the true answer 3.4 Coral metagenomes functional profiling Twenty coral metagenomes from four sites were also analysed to test the robustness of using SUPER-FOCUS in the marine environment. First, the coral sequences were profiled using FOCUS, which is part of SUPER-FOCUS, to understand the taxonomic profile among the sites; and second, the same sequences were classified into subsystems using SUPER-FOCUS with RAPSearch2 and DB_98 as database which we have already shown to be sensitive and precise. In addition to RAPSearch2, blastx was also added to the analysis because the sequenced reads average ∼218 ± 64.4 bp and RAPSearch2 is known to be less sensitive for reads longer than 100 bp ( Buchfink et al. , 2015 ). The difference in sensitivity for the long reads coral data between blastx and RAPSearch2 is shown in Fig. 11 a. The median RAPSearch2 sensitivity was ∼83%, although it was still precise (∼99%; Fig. 11 b); in comparison, the median blastx sensitivity was (∼99.5%) and precision was (∼99.5%) as shown in Fig. 9 c and d.\n Fig. 11. Box plots displaying the percent sensitivity ( A and C ) and precision ( B and D ) of RAPSearch2 ( A and B ), blastx ( C and D ) annotation of the 20 coral metagenomes. RAPsearch2 was tested in the fast and sensitive modes For the taxonomic profiling, Fig. 12 a was generated using SciPy ( Jones et al. , 2001 ) and matplotlib ( Hunter, 2007 ) (both open source python programming language libraries), and it shows the hierarchical clustering based on the pairwise Euclidean distances of the relative genus abundance of the each coral metagenome. The clustering shows that the samples from the same site cluster together, which suggest they have a similar microbial community.\n Fig. 12. Hierarchical clustering of the taxonomic ( A ) and functional ( B ) annotations of 20 coral metagenomes. Genus level taxonomic annotation was performed using FOCUS. Functional annotation of level 3 subsystems was performed using SUPER-FOCUS using blastx and DB_98 For the functional profiling, the same hierarchical clustering approach, using the distances between the relative abundances of the level 1, 2 and 3 subsystem classification were used. Figure 12 b also shows that samples from the same sites had similar subsystem level 3 profiles ( Supplementary Fig. S9 for levels 1 and 2). Both figures show that there is a microbial core that is preserved on some of the reefs; however, the nature of that core differs between taxonomy and function, as we have shown before ( Dinsdale et al. , 2008 ). This trend suggests that the organisms are quite stable at different sites while the functions change to allow adaptation to local environments. 3.5 Final considerations FOCUS is integrated with a SUPER-FOCUS pipeline which permits the tool to also report the taxonomic profile for a given metagenomic dataset. Both functional and taxonomic profiles are also provided by MEGAN and MG-RAST; RTMg only reports the functional assignments. SUPER-FOCUS provides three advances in the functional profiling compared with other tools: (i) It uses a fast aligner; (ii) it uses clustered databases in order to obtain a fast profile with little loss of sensitivity and (iii) focuses on the microbes present in the input data to attain a more microbial profile. As a default the tool, uses three advances; however, the user can select any combination of options."
} | 5,341 |
34208985 | PMC8307034 | pmc | 7,724 | {
"abstract": "Protective textiles used for military applications must fulfill a variety of functional requirements, including durability, resistance to environmental conditions and ballistic threats, all while being comfortable and lightweight. In addition, these textiles must provide camouflage and concealment under various environmental conditions and, thus, a range of wavelengths on the electromagnetic spectrum. Similar requirements may exist for other applications, for instance hunting. With improvements in infrared sensing technology, the focus of protective textile research and development has shifted solely from providing visible camouflage to providing camouflage in the infrared (IR) region. Smart textiles, which can monitor and react to the textile wearer or environmental stimuli, have been applied to protective textiles to improve camouflage in the IR spectral range. This study presents a review of current smart textile technologies for visible and IR signature control of protective textiles, including coloration techniques, chromic materials, conductive polymers, and phase change materials. We propose novel fabrication technology combinations using various microfabrication techniques (e.g., three-dimensional (3D) printing; microfluidics; machine learning) to improve the visible and IR signature management of protective textiles and discuss possible challenges in terms of compatibility with the different textile performance requirements.",
"conclusion": "5. Conclusions Smart textiles offer great potential for the next generation of camouflage products. While preserving the comfort and wearability of traditional textiles, they can take advantage of various existing technologies to provide the adaptive visible and IR signature management that is critically needed due to the advances in infrared sensing. These technologies, which can be combined to work in synergy, include surface coloring and pigmentation, embedded additives, chromic materials, low emissivity coatings, phase change materials, shape memory materials, and different thermal and mechanical actuation strategies. Combining these technologies can overcome some of the limitations outlined in Table 1 . Current work on adaptive camouflage has looked at nature as a source of inspiration, with embedded microfluidics and microcapsules. In terms of active thermal camouflage, the strategies used generally rely on IR emissivity/reflectivity control or thermal radiation/heat conduction guiding. However, many challenges remain on the path towards adaptive camouflage for clothing applications. The garment must be comfortable, durable, and easy to maintain. The manufacturing process must be compatible with industry capabilities and cost effective. Therefore, solutions proposed must consider the trade-offs between function, comfort, and cost to produce camouflage smart textile systems that can be manufactured at scale. In addition, the development of appropriate test methods for quality control is critically needed. A response to these different requirements may be found by taking advantage of the opportunities offered by 3D printing to produce fabric structures combining metamaterials and microfluidics. Above all, an interdisciplinary, holistic approach that considers the needs of the users and the capability of the manufacturing industry at the initial step of the development is key to solving this multifaceted problem involving dynamic interactions between humans and their changing environment.",
"introduction": "1. Introduction Today’s protective textiles must defend against a multitude of threats and fulfill a variety of functional requirements. In particular, textiles for military applications require durability, resistance to ballistic threats and environmental conditions (e.g., ultraviolet (UV) light, moisture, fire, heat, and wind), all while being comfortable and lightweight [ 1 ]. In addition, these textiles must provide camouflage and concealment under various environmental conditions and, thus, a range of wavelengths on the electromagnetic spectrum, especially the visible region (400–800 nm), near-infrared region (NIR) (750–1200 nm), and thermal or far-infrared region (FIR) (3–5 and 8–14 μm) [ 2 ]. Textiles must also adhere to military textile standard specifications/requirements such as colorfastness to light, washing, and perspiration to ensure changes to visible or infrared concealment are not compromised in use or as a result of cleaning [ 3 ]. Historically, military textiles for close-range battle were colorful and bright, intended for enemy intimidation and regimental identification [ 2 ]. With advancements in long-range weaponry and visual detection equipment at the beginning of the 20th century, the purpose of military uniforms was now to blend into the soldier’s background [ 2 ]. Although the desired colors may differ based on the wearer’s environment, modern camouflage textiles are typically olive, green, khaki, brown and black [ 3 ]. Camouflage properties for military textiles can come in the form of clothing, light flexible nets, garnishing and covers [ 2 ]. Textiles are flexible, three-dimensional materials made from fibers that can be spun into yarns and woven (interlaced) or knitted (interlooped) into a fabric. Alternatively, non-woven textiles can be made by bonding fibrous webs through mechanical entanglement, using resin, or thermally or chemically fusing the fibers together [ 4 ]. Embroidery is also used for the creation of technical textiles where yarns or threads are stitched on to the surface of a ground material [ 4 ]. Additional functional and aesthetic properties can be imparted into a textile by the application of surface finishes, coatings, or lamination techniques. Military textiles have traditionally been made from woven cotton fabrics, eventually blending with synthetic fibers such as nylon to be more lightweight and to dry more quickly [ 5 ]. As fiber manufacturing technology improved, the addition of high-performance fibers such as para-aramids and ultra-high molecular weight polyethylene (UHMWPE) have been integrated into uniforms for ballistic protection [ 6 ]. Protective fabrics are continuously being improved as new threats arise and technologies to address these threats are integrated into textile systems. The diverse requirements of military fabrics have been a motivator for the rapid development of smart textiles [ 7 , 8 ]. Put simply, smart or intelligent textiles are textiles that can automatically sense, react, and adapt to environmental stimuli [ 7 ]. There are varying degrees of “smartness” within a textile system [ 9 ] (p. 109). At a passive level, textiles will only sense an external stimulus; active smart textiles have an added sensor and actuator or display function; and responsive or ultra-intelligent textiles sense, react, and adapt to stimuli [ 9 , 10 ]. A simple smart textile can automatically sense and react to a stimulus as with responsive phase change, chromatic or shape memory materials or an active sensor can be added to the material to detect external stimuli [ 9 , 11 ]. This sensor may respond to the stimuli in a manner that is directly visible (e.g., changes to physical properties such as color, shape, or size), through an indirect response at the molecular level, or by electric or magnetic mechanisms [ 9 ]. These responses may be undetectable to the naked eye, but still trigger a controlled reaction [ 9 ]. Smart technologies can be integrated into fabrics to protect against various threats including biological, mechanical, or chemical hazards, or to improve the functionality of fabrics such as changing appearance or through enhanced thermal regulation. These technologies are applied through different techniques and mediums such as nanoparticles, microencapsulation, lamination, woven or knitted into a textile, or added at the polymer stage prior to fiber extrusion. There are three generations of smart textiles, which are differentiated based on the stage of the manufacturing process they are added [ 10 , 12 ]. First generation smart textiles are added to a fully assembled garment; for second generation textiles, the active component is incorporated during textile manufacturing such as weaving or knitting; lastly, active components are integrated within the fiber or yarns of a textile for third generation smart textiles [ 10 , 12 ]. While the possibilities for smart textile applications are diverse, the functionality of smart textiles must be compatible with how these textiles are used and maintained in its real-world application. Therefore, it is important to consider how smart solutions may interfere with textile performance. In this article, we review various smart textile mechanisms from simple to increasing complexity and how they have been applied for the purpose of achieving visible and infrared (IR) camouflage. Section 2 describes these existing technologies applicable to visible and IR camouflage for textiles. Following this, Section 3 reports on current progress in terms of adaptive camouflage. Lastly, Section 4 discusses the requirements and potential strategies for improved visible and IR signature management using microfabrication techniques.",
"discussion": "4. Discussion and Path Forward The opportunities for incorporating microstructures into textiles for the purpose of visible and infrared camouflage are promising. The above review demonstrates how research is already combining various technologies into single smart textile systems to achieve more effective camouflage properties, for instance at the fiber stage (e.g., embedded additives), in woven or knitted structures, and as surface finishes, coatings, or laminations. However, it is important to keep in mind the challenges in balancing the addition of smart microfabrication with the functional requirements of textiles. Several factors must be considered when it comes to incorporating smart technologies into garments to be produced at industry scale and in a cost-effective manner. These considerations are outlined in this section along with possible strategies for improved visible and IR signature management. 4.1. Functional Requirements of Smart Textile Systems for Camouflage Applications Synthetic fibers (e.g., polypropylene, polyethylene, polyester, nylon) are well suited for smart textile manufacturing as their properties can be modified during the production process [ 4 ]. Such modifications include changing fiber size and shape, the fiber’s molecular structure, embedding additives into a polymer, or co-extruding multiple polymers at once [ 4 ]. While durable and having high tensile strength, synthetic fibers can be less comfortable as they have decreased moisture absorbency compared to natural fibers [ 4 ]. Absorbency and air permeability can be further diminished with added finishes, coatings, or lamination [ 4 ]. Therefore, systems must not add to the thermal physiological stress of the wearer, especially if the wearer is already carrying heavy gear and performing physical tasks, as is the case with many military applications [ 2 ]. The work of many researchers demonstrates the importance of wearer comfort, resistance to laundering, and maintaining the functionality of the garment over time [ 14 , 16 , 23 , 31 , 34 , 39 , 40 , 44 ]. There are benefits in avoiding the need for embedded electronics as this adds complexity in terms of washing and powering of devices, especially since energy harvesting capabilities are not ready for use in textile systems yet. Similarly, the compatibility, modularity, interoperability, and ergonomics of the smart textile system with multiple components should be factored in through intuitive design [ 8 ], user-centered design, wear trials, and testing to adhere to existing textile standard specifications. Furthermore, smart textile systems must consider the current manufacturing capabilities in the textile industry as some camouflage methods are expensive (e.g., the use of expensive materials or processes) and not suitable for industrial manufacturing (e.g., relying on processes not easily scalable or the use of toxic compounds) [ 33 ]. The integration of metal fibers and wires or metallic and galvanic coatings must be carefully considered as these can cause damage to textile machinery, add weight to a textile, or be limited due to adhesion difficulties and corrosion resistance [ 9 ]. Technologies can certainly be combined to develop minimum viable products, but consideration of who will pay for the products must be taken into account. Smart microfabrication should take advantage of existing infrastructure and manufacturing processes to minimize costs and consider the needs and usefulness (e.g., comfort, protection) of these products for industry and end-users. Smart textiles combine materials from different sectors such as textiles, electronics, and chemicals, making standardized testing a particular challenge [ 8 ]. Smart textiles for military applications must also consider adherence to military textile standards. As of December 2020, only eighteen standards exist or are in development for smart textiles, indicating a current lack of standardization, comparability, and quality-control of smart textile products [ 67 ]. As smart textiles are worn close to the body, health and safety risks, durability, and compatibility with other materials should be considered. Shuvo et al. suggests a tri-factor framework for assessing smart textile performance which includes the durability, safety, and efficiency of the product, as well as the product features and longevity, user experience, and the cost versus benefits of the product [ 67 ]. As military personnel face a multitude of threats, compromises between protection, comfort, and the ability to complete tasks while wearing protective clothing must be made [ 6 ]. However, this protection may only be required for a short amount of time during the overall wear period [ 8 ]. Solutions that are adaptive or only “on” when needed can help to balance protection and comfort [ 8 ], thus minimizing these trade-offs. 4.2. Three-Dimensional Printing with Metamaterials and Microfluidics A response to these different requirements may be found by taking advantage of the opportunities offered by three-dimensional (3D) printing to produce fabric structures combining metamaterials and microfluidics. For instance, polymer-based nanophotonic was used to produce a hybrid metamaterial radiative cooling textile [ 68 ]. An electrospun layer of Si 3 N 4 nanoparticle/poly(vinylidene fluoride) nanocomposite nanofibrous mat was sandwiched between a nanoporous polyethylene layer on the outer side and a dopamine-modified nanoporous polyethylene layer on the inner side. The resulting fabric combines high spectral selectivity, IR absorbance/emittance, and sunlight reflectance. It is also water-tight and water vapor permeable. Another example is a 3D metamaterial absorber textile structure that was proposed for radar stealth application [ 69 ]. It combines a copper yarn weft-knitted fabric as the periodic resonator on one side, a conductive plain weave fabric on the other side, and a silicone dielectric layer in the middle. Using the Computer Simulation Technology software program, the authors calculated an absorption between 81 and 95% in the 8 to 12 GHz frequency range. The use of hierarchical metamaterials allows designing solutions for multispectral camouflage, i.e., covering both infrared and microwaves [ 70 ]. This was achieved by integrating an IR selective emitter with a microwave selective absorber. The authors managed to reduce the signature levels of 8–12 μm IR waves by up to 95% and 2.5–3.8 cm microwaves by up to 99%. Three-dimensional printing has opened up new perspectives for manufacturing metamaterials. In particular, it has been shown to allow the production of microstructures capable of experiencing large deformations at the microscale using embedded soft pivots [ 71 ]. For instance, 3D printed pantographic sheets comprised of straight and parabolic fibers connected by soft pivots were subjected to bias extension up to rupture. The stress–strain curves were successfully described using continuum bidimensional models. Other deformable metamaterial complex structures produced by 3D printing include stretchable circuits manufactured by coextrusion of liquid metal within thermoplastic filaments [ 72 ] and a robotic gripper combining soft and hard thermoplastics [ 73 ]. Recent progress in 3D printing includes the use of a polycarbonate (PC)/acrylonitrile butadiene styrene (ABS) core/sheet filament as feedstock for fused filament fabrication 3D printing, yielding a ductile and tough composite ABS/PC meso-structured part after annealing at a temperature between the glass transition temperatures of ABS and PC [ 74 ]. Such techniques could be adapted for more than just mechanical improvements and open up opportunities for complex internal structures of fibers to be manufactured with future printing and thermal drawing steps. In another area, droplet microfluidics can be used to form specialized microparticles with well controlled and tunable size, structure, and composition [ 75 ]. After the droplets are formed one by one using breakout methods such as dripping, jetting, or squeezing, they can be converted into solid microparticles by polymerization, temperature-induced gelation, ionic crosslinking, or solvent evaporation for instance. The microfluidic devices most commonly used to generate and manipulate droplets are glass capillary microfluidics and poly(dimethylsiloxane) (PDMS) devices produced by lithography or molding processes [ 75 ]. Recently, 3D printing has been explored with some success for the fabrication of microfluidic devices. However, some challenges remain, including the needed resolution, suitable materials, and surface modification techniques. If these precisely engineered microparticles have found numerous applications in medicine, for example for drug delivery and cell encapsulation [ 75 ], they can also be used to provide solutions for adaptive camouflage by combining chromatophores with hyperelastic matrices and varying their in-plane surface area by compressing them in the other direction. Figure 11 demonstrates the fabrication of one such microfluidic device developed by the authors using silicone-based chromatophore spheres. In this process, Ecoflex gel was mixed with silicone fluid (ratio of 1:2) and various pigments to aid in visual identification. After degassing, a disposable syringe was used to collect the silicone material and inject it while being stirred in warm (~70 °C) soapy water which ensured the emulsion remained stable long enough for the silicone to cure (~5 minutes). The size of spheres could be controlled by the stirring speed and the gauge size of the dispensing needle tip, with different sieves used for collection producing spheres in the range of 0.2–2 mm. The silicone spheres were then placed in an IR transparent polyethylene pouch with a metalized background. By applying vacuum to the pouch, the spheres were deformed and covered the low emissivity surface. Placing greater numbers of spheres in controlled spots of the pouch has the potential to change the entire surface of the device from IR transparent to IR opaque. Through applying nearly 20 years of research in droplet microfluidic technology towards the manipulation of thermoplastics, liquid metals, gels and curable polymers, entirely new fabrics made of microfluidic channels/capillaries may be manufactured for variable thermal and visible properties for a variety of applications. Applying microfluidic techniques to the production of smart fabrics is very early in its development, however, and challenges in the adoption of these technologies by large manufacturers to make economically viable products, as well as investigations into durability, adaptability and comfort as fabrics, remain an open research area for these new smart materials. 4.3. Machine Learning Machine learning for control over manufacturing and process stability of complex, multi-material extrusion and thermal drawing may also help alleviate the current challenges experienced when attempting to apply current technologies for the development of smart visible and IR camouflage solutions, especially when considering the complexity of combining several techniques in a single device. The complexity of controlling non-equilibrium processes with highly variable viscosity, viscoelasticity, shear thinning behavior and surface tension-induced droplet breakups balancing with multi-material extrusion leads to extreme difficulties in predicting all final part behavior with simple input conditions. Machine learning has been successfully used to improve the performance of textiles, for instance with the optimization of the elastic modulus of woven fabrics based on the weave factor, warp yarn count and pick density using artificial neural network (ARN) and random forest regression (RFR) approaches [ 76 ]. In an attempt to solve the challenges encountered with asymmetric interactions in thermal metamaterials, an autoencoder was also trained to optimize thermal transparency using two types of particles disposed in a periodic manner in a mechanism named periodic interparticle interaction by the authors [ 77 ]. It allowed for using a more complex lattice and varying the relative positioning of the two types of particles. A vast array of data on multi-material extrusions would need to be collected to train appropriate machine learning algorithms, but once completed the improved predictive performance of new material combinations, processing conditions and post-processing treatments for smart fibers/fabrics could unlock new capabilities for intelligent and responsive fabrics. 4.4. Interdisciplinary Approaches Finally, the most critical aspect towards achieving comfortable and durable smart camouflage textiles that can be manufactured in a cost-effective manner is a holistic approach involving the different relevant disciplines: textiles and clothing, materials, design, engineering, and manufacturing. An interdisciplinary perspective that considers the needs of the users and the capability of the manufacturing industry at the initial step of the development is key to solving this multifaceted problem involving dynamic interactions between humans and their changing environment. Researchers must work closely with textile manufacturers to ensure that the developed technologies will be feasible and scalable, applying cutting edge approaches from non-traditional disciplines to textile design, but also understanding early in the process the commercial and industrial realities of large-scale manufacturing systems through frequent dialogue. In this context, an innovation-oriented industry aiming at high value-added niche products and capable of production flexibility and fast adaptation such as the textile industry is an ideal partner for such endeavor. Figure 12 summarizes existing technologies used for smart textile systems for visible and IR camouflage and this proposed path forward. Smart textiles for camouflage application must consider various textile properties that will influence performance while also fulfilling functional requirements such as comfort, appearance, durability, and manufacturing scalability. As demonstrated in this review, existing manufacturing technologies with increasing smart capabilities have achieved visible and IR camouflage in textiles, but still have limitations in terms of speed of response, weight, washability, and wearability. An envisioned path forward for the field takes an interdisciplinary approach that utilizes existing manufacturing techniques and considers the diverse functional requirements of wearable textiles while integrating microfabrication techniques such as 3D printing and machine learning to develop smart textile camouflage systems."
} | 5,996 |
32053106 | PMC7082127 | pmc | 7,725 | {
"abstract": "Many aspects of the brain’s design can be understood as the result of evolutionary drive toward metabolic efficiency. In addition to the energetic costs of neural computation and transmission, experimental evidence indicates that synaptic plasticity is metabolically demanding as well. As synaptic plasticity is crucial for learning, we examine how these metabolic costs enter in learning. We find that when synaptic plasticity rules are naively implemented, training neural networks requires extremely large amounts of energy when storing many patterns. We propose that this is avoided by precisely balancing labile forms of synaptic plasticity with more stable forms. This algorithm, termed synaptic caching, boosts energy efficiency manifold and can be used with any plasticity rule, including back-propagation. Our results yield a novel interpretation of the multiple forms of neural synaptic plasticity observed experimentally, including synaptic tagging and capture phenomena. Furthermore, our results are relevant for energy efficient neuromorphic designs.",
"introduction": "Introduction The human brain only weighs 2% of the total body mass but is responsible for 20% of resting metabolism ( Attwell and Laughlin, 2001 ; Harris et al., 2012 ). The brain’s energy need is believed to have shaped many aspects of its design, such as its sparse coding strategy ( Levy and Baxter, 1996 ; Lennie, 2003 ), the biophysics of the mammalian action potential ( Alle et al., 2009 ; Fohlmeister, 2009 ), and synaptic failure ( Levy and Baxter, 2002 ; Harris et al., 2012 ). As the connections in the brain are adaptive, one can design synaptic plasticity rules that further reduce the energy required for information transmission, for instance by sparsifying connectivity ( Sacramento et al., 2015 ). But in addition to the costs associated to neural information processing, experimental evidence suggests that memory formation, presumably corresponding to synaptic plasticity, is itself an energetically expensive process as well ( Mery and Kawecki, 2005 ; Plaçais and Preat, 2013 ; Jaumann et al., 2013 ; Plaçais et al., 2017 ). To estimate the amount of energy required for plasticity, Mery and Kawecki (2005) subjected fruit flies to associative conditioning spaced out in time, resulting in long-term memory formation. After training, the fly’s food supply was cut off. Flies exposed to the conditioning died some 20% quicker than control flies, presumably due to the metabolic cost of plasticity. Likewise, fruit flies doubled their sucrose consumption during the formation of aversive long-term memory ( Plaçais et al., 2017 ), while forcing starving fruit flies to form such memories reduced lifespan by 30% ( Plaçais and Preat, 2013 ). A massed learning protocol, where pairings are presented rapidly after one another, leads to less permanent forms of learning that don’t require protein synthesis. Notably this form of learning is energetically less costly ( Mery and Kawecki, 2005 ; Plaçais and Preat, 2013 ). In rats ( Gold, 1986 ) and humans ( Hall et al., 1989 , but see Azari, 1991 ) beneficial effects of glucose on memory have been reported, although the intricate regulation of energy complicates interpretation of such experiments ( Craft et al., 1994 ). Motivated by the experimental results, we analyze the metabolic energy required to form associative memories in neuronal networks. We demonstrate that traditional learning algorithms are metabolically highly inefficient. Therefore, we introduce a synaptic caching algorithm that is consistent with synaptic consolidation experiments, and distributes learning over transient and persistent synaptic changes. This algorithm increases efficiency manifold. Synaptic caching yields a novel interpretation to various aspects of synaptic physiology, and suggests more energy efficient neuromorphic designs.",
"discussion": "Discussion Experiments on formation of a long-term memory of a single association suggest that synaptic plasticity is an energetically expensive process. We have shown that energy requirements rise steeply as memory load or designated accuracy level increase. This indicates trade-offs between energy consumption, and network capacity and performance. To improve efficiency, we have proposed an algorithm named synaptic caching that temporarily stores changes in the synaptic strength in transient forms of plasticity, and only occasionally consolidates into the persistent forms. Depending on the characteristics (decay and maintenance cost) of transient plasticity, this can lead to large energy savings in the energy required for synaptic plasticity. We stress that from an algorithmic point of view, synaptic caching can be applied to any synaptic learning algorithm (unsupervised, reinforcement, supervised) and does not have specific requirements. Further savings might be possible by adjusting the consolidation threshold as learning progresses and by being pathway-specific ( Leibold and Monsalve-Mercado, 2016 ). The implementation of a consolidation threshold is similar to what has been observed in physiology, in particular in the synaptic tagging and capture literature ( Redondo and Morris, 2011 ). Our results thus give a novel interpretation of those findings. Synaptic consolidation is known to be affected by reward, novelty and punishment ( Redondo and Morris, 2011 ), which is compatible with a metabolic perspective as energy is expended only when the stimulus is worth remembering. In addition, our results for instance explain why consolidation is competitive, but transient plasticity is less so ( Sajikumar et al., 2014 ), namely the formation of long-term memory is precious. Consistent with this, there is evidence that encouraging consolidation increases energy consumption ( Plaçais et al., 2017 ). We also predict that the transient weight changes act as an accumulative threshold for consolidation. That is, sufficient transient plasticity should trigger consolidation, even in the absence of other consolidation triggers. Future characterization of the energy budget of synaptic plasticity should allow more precise predictions of our theory. Combining persistent and transient storage mechanisms is a strategy well known in traditional computer systems to provide a faster and often energetically cheaper access to memory. In computer systems, permanent storage of memories typically requires transmission of all information across multiple transient cache systems until reaching a long-term storage device. The transfer of information is often a bottleneck in computer architectures and consumes considerable power in modern computers ( Kestor et al., 2013 ). However, in the nervous system transient and persistent synapses appear to exist next to each other. Local consolidation in a synapse does not require moving information. Using this setup, biology appears to have found a more efficient way to store information. Memory stability has long fascinated researchers ( Richards and Frankland, 2017 ), and in some cases forgetting can be beneficial ( Brea et al., 2014 ). Splitting plasticity into transient and persistent forms might prevent catastrophic forgetting in networks ( Leimer et al., 2019 ). Here, we argue that the main benefit of more transient forms of plasticity is to permit the network to explore the weight space to find a desirable weight configuration using less energy. While this work focuses solely on the metabolic cost of synaptic plasticity, the brain also expends significant amounts of energy on spiking, synaptic transmission, and maintaining resting potential. Learning rules can be designed to reduce costs associated to computation once learning has finished ( Sacramento et al., 2015 ). It would be of interest to next understand the precise interaction of computation and plasticity cost during and after learning."
} | 1,960 |
34402639 | PMC8407213 | pmc | 7,726 | {
"abstract": "ABSTRACT A central paradigm in microbiome data analysis, which we term the genome-centric paradigm, is that a linear (non-branching) DNA sequence is the ideal representation of a microbial genome. This representation is natural, as microbes indeed have non-branching genomes. Tremendous discoveries in microbiology were made under this paradigm, but is it always optimal for microbiome research? In this Commentary, we claim that the realization of this paradigm in metagenomic assembly, a fundamental step in the “metagenomics analysis pipeline,” suboptimally models the extensive genomic variability present in the microbiome. We outline our efforts to address these issues with a “genome-free” approach that eschews linear genomic representations in favor of a pan-metagenomic graph."
} | 196 |
24877729 | null | s2 | 7,727 | {
"abstract": "A neural network with symmetric reciprocal connections always admits a Lyapunov function, whose minima correspond to the memory states stored in the network. Networks with suitable asymmetric connections can store and retrieve a sequence of memory patterns, but the dynamics of these networks cannot be characterized as readily as that of the symmetric networks due to the lack of established general methods. Here, a reduction method is developed for a class of asymmetric attractor networks that store sequences of activity patterns as associative memories, as in a Hopfield network. The method projects the original activity pattern of the network to a low-dimensional space such that sequential memory retrievals in the original network correspond to periodic oscillations in the reduced system. The reduced system is self-contained and provides quantitative information about the stability and speed of sequential memory retrieval in the original network. The time evolution of the overlaps between the network state and the stored memory patterns can also be determined from extended reduced systems. The reduction procedure can be summarized by a few reduction rules, which are applied to several network models, including coupled networks and networks with time-delayed connections, and the analytical solutions of the reduced systems are confirmed by numerical simulations of the original networks. Finally, a local learning rule that provides an approximation to the connection weights involving the pseudoinverse is also presented."
} | 385 |
37711644 | PMC10497401 | pmc | 7,729 | {
"abstract": "Abstract High cellular pigment levels in dense microalgal cultures contribute to excess light absorption. To improve photosynthetic yields in the marine microalga Picochlorum celeri , CAS9 gene editing was used to target the molecular chaperone cpSRP43. Depigmented strains (>50% lower chlorophyll) were generated, with proteomics showing attenuated levels of most light harvesting complex (LHC) proteins. Gene editing generated two types of cpSRP43 transformants with distinct lower pigment phenotypes: (i) a transformant (Δ srp43 ) with both cp SRP43 diploid alleles modified to encode non‐functional polypeptides and (ii) a transformant (STR30309) with a 3 nt in‐frame insertion in one allele at the CAS9 cut site (non‐functional second allele), leading to expression of a modified cpSRP43. STR30309 has more chlorophyll than Δ srp43 but substantially less than wild type. To further decrease light absorption by photosystem I in STR30309, CAS9 editing was used to stack in disruptions of both LHCA6 and LHCA7 to generate STR30843, which has higher (5–24%) productivities relative to wild type in solar‐simulating bioreactors. Maximal productivities required frequent partial harvests throughout the day. For STR30843, exemplary diel bioreactor yields of ~50 g m −2 day −1 were attained. Our results demonstrate diel productivity gains in P. celeri by lowering pigment levels.",
"introduction": "1 INTRODUCTION Microalgae are able to efficiently convert sunlight into biomass (stored chemical energy), which can be used in a variety of sustainable biotechnological applications including food security, sustainable aviation fuels, nutraceuticals, and renewable biopolymers (Davis et al., 2011 ; Gimpel et al., 2013 ; LaPanse et al., 2021 ; Schenk et al., 2008 ; Sheehan et al., 1998 ). Importantly, some microalgae thrive in saline water and can be grown using marginal lands/marshes/offshore sites, which make these saltwater photoautotrophs attractive complements, and potential alternatives, to land‐based feedstocks that have substantial freshwater demands (Benedetti et al., 2018 ; Dismukes et al., 2008 ). The full realization of the potential of microalgae is currently limited by production costs, which will benefit from biomass productivity improvements. While theoretical solar‐to‐biomass conversion efficiency estimates of 8–10% (18–22% PAR➔ biomass) are calculated to yield 77–100 g m −2 day −1 , the best sustained productivities of outdoor dense algal cultures are in the range of 20–35 g m −2 day −1 translating to an ~3% conversion efficiency (Grobbelaar, 2000 ; Lee, 2001 ; Melis, 2009 ; Moheimani & Borowitzka, 2007 ; Sheehan et al., 1998 ; Weissman & Nielsen, 2016 ). The “low‐light” acclimated state of microalgae with high pigment content in dense‐mass cultures is a primary contributor to this inefficiency (Grobbelaar, 2010 ; Grobbelaar et al., 1995 ; Kirst et al., 2017 ; Melis, 2009 ; Weissman, 1978 ; Weissman & Nielsen, 2016 ). Here, individual cells compete for light by maximizing their peripheral light harvesting antenna to increase their ability to capture photons in light‐limited environments. High optical cross sections lead to saturation of the photosynthetic apparatus with the consequence that many absorbed photons are dissipated by non‐photochemical routes leading to reduced photosynthetic efficiency. One proposed solution is to develop strains with lower pigment levels (Cazzaniga et al., 2014 ; Dabes et al., 1970 ; Formighieri et al., 2012 ; Jeong et al., 2017 ; Kirst, Garcia‐Cerdan, Zurbriggen, Ruehle & Melis, 2012 ; Kirst et al., 2014 ; Kwon et al., 2013 ; Melis et al., 1998 ; Mussgnug et al., 2007 ; Nakajima & Ueda, 2000 ; Negi et al., 2020 ; Oey et al., 2013 ; Perin et al., 2015 ; Perrine et al., 2012 ; Polle et al., 2003 ; Radmer & Kok, 1977 ). In principle, depigmentation will reduce light absorption by individual cells enabling more cells to photosynthesize, contributing towards culture productivity. Multiple strategies have been used to manipulate light harvesting antenna sizes and generate depigmented algae including but not limited to Chlamydomonas, Dunaliella, Nannochloropsis, Chlorella, Phaeodactylum, and cyanobacteria. Almost all had lower chlorophyll content, smaller peripheral antenna, and used high light more effectively (Huesemann et al., 2009 ; Jeong et al., 2017 ; Kirst, García‐Cerdán, Zurbriggen & Melis, 2012 ; Kirst, Garcia‐Cerdan, Zurbriggen, Ruehle & Melis, 2012 ; Mussgnug et al., 2007 ; Nymark et al., 2019 ; Perrine et al., 2012 ; Polle et al., 2000 , 2003 ) (Baek et al., 2016 ; Beckmann et al., 2009 ; Cazzaniga et al., 2014 ; Kirst et al., 2014 ; Nakajima et al., 2001 ; Negi et al., 2020 ; Perin et al., 2015 ). However, only a few of them demonstrated successful improvements in productivity/growth in mass culture (Cazzaniga et al., 2014 ; Nakajima et al., 2001 ; Negi et al., 2020 ; Perin et al., 2015 ). Here, we describe depigmented strains of a remarkably fast‐growing, high‐light tolerant microalga, Picochlorum celeri . The Picochlorum genus (Henley et al., 2004 ) has become the focus of several recent studies because of their high stress tolerances, as well as their small, compact genomes (Barten et al., 2020 ; Dahlin et al., 2019 ; Foflonker et al., 2016 ; Gonzalez‐Esquer et al., 2019 ; Manjre et al., 2022 ; Weissman et al., 2018 ). P. celeri demonstrates broad halotolerance, an exceptional ability to thrive at high light intensities (>2000 μmol photons m −2 s −1 ), temperatures (>35°C), and salinities (2–3× seawater), and can attain rapid doubling times of ~2 h under optimal conditions (Krishnan et al., 2021 ). Recently, P. celeri demonstrated among the highest monthly productivity (~36 g m −2 day −1 ) reported to‐date for a marine alga in an outdoor field trial at the Arizona Center for Algae Technology and Innovation testbed (Krishnan et al., 2021 ). \n P. celeri's chlorophyll content is very responsive to irradiance, becoming relatively low (~1.2% of ash‐free‐dry‐weight [AFDW]) at high irradiance (~2000‐μmol photons m −2 s −1 ) and almost tripling at low light (60 μmol photons m −2 s −1 ) and diel culturing (~4.5% of AFDW) (Weissman et al., 2018 ). Because of the ability of P. celeri to develop a “low‐light acclimated” physiology with high chlorophyll content, it is likely necessary to engineer a strain that remains in a more “high‐light acclimated” state with less photosynthetic pigment content even in dense culture to improve solar‐to‐biomass conversion efficiencies. Photosynthetic machinery in oxygenic phototrophs is composed of (i) two coupled, light‐driven reaction enters, photosystem II (PSII) and photosystem I (PSI), involved in water splitting and the generation of low‐potential electrons and (ii) peripheral light harvesting pigment‐protein antenna complexes (LHCs), associated with each photosystem that are predominantly involved in the absorption of photons and transfer of energy to the respective reaction centers. In higher plants and microalgae, there are distinct LHC proteins, called light harvesting proteins of photosystem I (LHCA) and light harvesting proteins of photosystem II (LHCB), associated with PSI and PSII, respectively (Jansson et al., 1992 ; Grossman et al., 1995 ). All the LHCs are nuclear encoded and post‐translationally imported to the chloroplast envelope where they are guided to the thylakoid membrane using the chloroplast signal recognition particle 43 (cpSRP) pathway. cpSRP43 and cpSRP54 form a soluble transit complex by binding to the hydrophobic LHCPs that traverses the stroma and docks to the thylakoid membrane by the interaction with cpFTSY and then ALB3 insertase (Kirst et al., 2014 ; Ziehe et al., 2018 ). In both microalgae and higher plants, cpSRP43, a molecular chaperone, plays a central role in LHCP recognition and in preventing misfolding of the incoming LHCPs (Falk & Sinning, 2010 ). Δ srp43 mutants have been shown to display considerable decreases in the LHC pigment‐protein complexes and have therefore been used as targets for engineering “truncated peripheral‐antenna” mutants (Amin et al., 1999 ; Kirst, Garcia‐Cerdan, Zurbriggen, Ruehle & Melis, 2012 ; Kirst et al., 2014 ; Kirst et al., 2018 ; Klimyuk et al., 1999 ). Here, we apply CAS9 technology to edit cpSRP43 to lower pigment levels in the biotechnologically relevant alga P. celeri and further stack two LHCI polypeptide deletions into an edited cpSRP43 strain to better balance light capture between PSI and PSII and enable increased photosynthetic efficiencies. Lastly, we explore the effects of harvesting frequency in bioreactors to modulate biomass and pigment levels to improve biomass productivities.",
"discussion": "3 DISCUSSION Light saturation of photosynthesis, which is exacerbated in lower‐light acclimated states, is a major problem that limits algal productivity in mass cultures (Grobbelaar, 2010 ; Grobbelaar et al., 1995 ; Kirst et al., 2017 ; Melis, 2009 ; Weissman, 1978 ; Weissman & Nielsen, 2016 ). This has prompted research into developing strains with permanently truncated peripheral antenna (and decrease in the number of reaction centers if necessary) to minimize pigment content so as to increase the irradiance at which photosynthesis saturates. P. celeri is currently among the most promising outdoor biomass production strains due to its high productivity and robustness in cultivation (46), and we have initiated a series of pigment lowering efforts to further improve biomass productivities. By targeting the molecular chaperone cpSRP43 using CAS9, we were able to simultaneously attenuate protein levels of both PSII and PSI peripheral antenna subunits generating mutants with >50% loss in pigmentation and no significant changes in the remaining photosynthetic machinery under our assay conditions (Table 1 , Figures 1 , 3 , and 4 ). cpSRP43 knock‐outs have been previously used to generate truncated antenna mutants in other photosynthetic organisms including Arabidopsis thaliana , Nicotiana tabaccum , and Chlamydomonas reinhardtii (Amin et al., 1999 ; Kirst, Garcia‐Cerdan, Zurbriggen, Ruehle & Melis, 2012 ; Kirst et al., 2018 ; Klimyuk et al., 1999 ). Interestingly, our CAS9 strategy yielded two phenotypically distinct depigmented mutants resulting from the variations in editing: (i) a complete homozygous knock‐out of cpSRP43 (Δ srp43 ) and (ii) a modified version of cpSRP43 that had a single amino acid inserted at the cut site (STR30309) in one allele; whereas the second allele was undetected by PCR (Figure 1 ). While Δ srp43 had a severe >75% lower pigment levels, the modified version of cpSRP43 (STR30309) demonstrated a smaller decrease in total pigments (~65% decrease) and had a lower Chl a/b ratio (Figure 3 ). Differences in Chl a/b ratios and the abundances of LHC proteins between Δ srp43 and STR30309 indicate the ability of the modified SRP43 protein to partially rescue protein function (Figures 1 and 3 , Table 1 ). Even with the complete loss of the cpSRP43, Δ srp43 accumulated all the LHC proteins, albeit at lower concentrations, suggesting the presence of an additional SRP43‐independent pathway for LHC transport in P. celeri . Existence of a such pathway has been alluded to in Chlamydomonas though a mechanism that remains elusive (Bujaldon et al., 2020 ; Kirst & Melis, 2014 ). Strains with one WT copy of the protein and the other allele eliminated exhibited normal pigmentation, indicating that the single cpSRP43 amino acid insertion in STR30843 (Figure S2 ) is the causative agent of the intermediate phenotype. Variations in chaperoning ability of cpSRP43 caused by point mutations have been demonstrated in vitro (McAvoy et al., 2018 ). Shin et al. ( 2016 ) found a point mutation in cpSRP43 protein of a depigmented Chlorella strain (~50% lower pigments) despite this strain having high levels of cpSRP43 expression. Identification and use of modulated function mutants that generate phenotypes intermediate of a complete knockout and WT may be a useful strategy to fine tune pigment levels in photosynthetic organisms and identify improved biomass strains. Both STR30309 and STR30843 had higher chlorophyll normalized α (initial slope of the P‐I curve), higher E k , and no significant effect on the P max per AFDW (Figure 2 ) suggesting that at least the photosynthetic efficiency as well as the coupling of the photosystems to CO 2 fixation is not significantly affected in these strains. This is consistent with the proteomics data that show only minor changes in the levels of proteins in the photosynthetic electron transport chain and Calvin Benson Bassham cycle. Genotypically, the difference between STR30309 and STRP30843 was the further elimination of LHCA7 and LHCA6 from the STR30309 strain by CAS9. No significant differences in the reaction center proteins were observed for the two strains (Figure 3a ), and the 77‐K fluorescence profile suggested changes in the relative PSI‐LHCI levels in STR30843 (Figure 2b ). This indicates that the increase in productivity (PAR‐to‐biomass conversion efficiency) observed in STR30843 as compared with STR30309 (Table 1 ) is most likely a direct outcome of better balance between LHCI‐PSI and LHCII‐PSII light absorption. These results reinforce the idea that depigmentation is necessary, but not a sufficient condition for higher productivity, and needs to be complemented with the balanced absorption of light between the two photosystems. Currently, we are continuing to target additional LHCAs in Δ srp43 and STR30843 to determine if photosystem excitations can be balanced further. It was recognized in early work (Cook, 1951 ; Dabes et al., 1970 ; Eppley & Dyer, 1965 ; Myers & Graham, 1959 ; Weissman & Benemann, 1979 ) that when biomass productivity is measured as a function of specific growth rate (which is equal to dilution rate in continuously operated algal cultures), there is usually an optimum. This optimum may be narrow (Cook, 1951 ; Eppley & Dyer, 1965 ). Presumably at low dilution rates (high biomass densities), productivity declines due to high maintenance burden. At higher dilution rates, productivity may also decrease due to photoinhibition and eventually because as dilution rate approaches the maximum specific growth rate of the organism under the growth conditions, not all of the incident light is absorbed by the culture. Myers and Graham ( 1959 ) recognized that the cells may also acclimate to the different average light levels in the reactor at different dilution rates, broadening out the productivity optimum. They pointed out that algal cells acclimated to higher light, having lower pigment/AFDW ratios, have a higher saturating irradiance and hence higher photosynthetic efficiency leading to this broader optimum. None of these earlier studies imposed a diel light regime, but the underlying principles remain similar. The data in Figure 4c show that both the WT and STR30843 have decreasing chlorophyll/AFDW ratios with increasing dilution rate; that is, the chlorophyll content of the cells is acclimating to the different light levels. However, these changes in chlorophyll content are mitigated by pigment packaging effects. Lower chlorophyll/AFDW increases the optical cross section per unit chlorophyll, which diminishes the decreases in this cross section based on biomass density. Increases in the ratio of other pigments to chlorophyll also increase the optical cross sections. The ratio of lutein and β‐carotene to chlorophyll increases in both the WT and 30843 with dilution rate (Figure S4C ). The consequence is that the light acclimation observed may broaden the optimum of biomass productivity versus specific growth rate but may not increase the actual productivity of these higher light acclimation states by very much, if at all. All these factors led to a very flat response of biomass productivity with dilution rate, over a large range of the latter, for both strains studied here. Within these broad optima, increases in biomass density were nearly proportional to decreases in specific growth rate. In fact, there was very little change in biomass productivity for dilution rates above 40% day −1 with either the WT or 30843, like an earlier study with a Chaetoceros species under diel conditions (Weissman & Nielsen, 2016 ). There may have been such a decrease for STR30843 at high light (high μ), in terms of total chemical energy (maintenance plus net productivity, Figure 4h ). Depigmented cells with a “truncated antenna” phenotype absorb fewer photons per cell, leading to less loss due to light saturation. Inter strain comparisons of the productivity of a WT with a depigmented strain can be done in several ways. One way is to compare them for cases for which the light profile in the reactor runs is similar; that is, the extinction coefficient for light attenuation (the absorption cross section times the biomass density) is the same. The specific growth rate in mass culture, or any column containing algae, is governed by the light profile within the reactor and the physiological responses to this light profile of the alga under study including its maintenance burden, sensitivity to high irradiance, and factors that affect the minimum quantum requirement. Of interest here is to try to isolate the potential increases in biomass productivity of the depigmented strain due to lowered light saturation losses, from the other factors affecting productivity (maintenance burdens and photoinhibition). A comparison of the productivity at moderate specific growth rates (dilution rates) does this. The WT and STR30843 are compared at a dilution rate of 80% day −1 in Case 1 of Table 2 . The light profiles, that is, extinction coefficients, were similar. Strain 30843 produced 24% more biomass and had a total chemical energy (maintenance plus biomass) also 24% higher. When the strains were compared at 60% day −1 dilution rates, the results were similar (23% higher for STR30843, data not shown in the table). This increase may be taken to represent the impact of differential light saturation on increasing productivity with about 55% lower chlorophyll content. TABLE 2 Selected comparisons of WT versus STR30843 reactor trials. Case 1 Case 2 Case 3 WT 80 30843 80 WT 40 30843 60 WT 130 30843 130 Chl density, mg L −1 \n 33.1 19.2 60.4 25.2 17.3 8.4 \n \n a \n K, m −1 \n 150 131 ~300 ~170 96 58 Chl, %AFDW 4.3 2.0 5.2 2.1 3.3 1.6 AFDW, g L −1 \n .76 .94 1.18 1.22 .54 .53 m 2 (gChl) −1 \n 5.25 6.8 ~5.0 ~6.7 5.53 6.94 m 2 (gAFDW) −1 \n .196 .136 ~.24 ~.15 .177 .116 Biomass productivity, g m −2 day −1 \n 40.8 50.7 31.5 49.1 46.7 45.9 m, g m −2 day −1 \n 12.9 16.3 19.9 21.0 9.1 9.1 Total energy, g m −2 day −1 \n 53.7 66.8 51.4 70.1 55.7 55.0 Ratio, productivity \n b \n \n 1.24 1.56 .98 Ratio, total energy \n b \n \n 1.24 1.36 .98 Abbreviations: AFDW, ash‐free‐dry‐weight; WT, wild type. \n a \n K calculated by multiplying chlorophyll density times optical cross section. \n b \n For each case, the ratio was calculated by dividing the 30843 values by WT value. The two strains are compared on a different basis: equal but relatively high AFDW densities in Case 2. The WT can attain an average specific growth rate of 40% day −1 , whereas the de‐pigmented 30843 attains 60% day −1 . Thus, the biomass productivity of the latter is about 50% higher, and the total chemical energy is 36% higher. In this situation, the maintenance burdens are about the same because they are based on biomass density. The larger increase in productivity than in Case 1 may be attributed to not only the depigmentation of 30843 but also to the even lower state of light acclimation of the WT in this case than in Case 1. The former leads to less loss due to light saturation of the depigmented strain and the latter to increased losses of the WT. In Case 3, the comparison is again based on equal AFDW densities but much lower ones. Again, the maintenance burdens are the same for the strains, but there is almost no increase in biomass productivity nor total chemical energy conversion from strain 30843 despite the lower pigmentation (by a factor of two) and lower cross section/AFDW (by almost one‐third). The extinction coefficient for 30843 is much lower, which is expected due to the substantially lower chlorophyll density. But the average specific growth rates attained are the same despite the large difference in average irradiance in the reactor (extinction coefficient). The mismatch between extinction coefficients and average specific growth rates reflects a very different physiological response to the light field. Here, it may be that strain 30843 is experiencing more photoinhibition than the WT. Except for a couple of studies (Cazzaniga et al., 2014 ; Huesemann et al., 2009 ; Negi et al., 2020 ), most of the research reported for depigmented strains uses constant light levels. Our bioreactor results clearly show that the depigmented strains display higher productivity than WT under dense culture‐diel light conditions (Table 1 , Figure 4 ) and underscore the need for optimizing dilution regimes for each type of strain for maximizing productivity. In summary, through systematic engineering, we were able to generate a highly depigmented strain of P. celeri that under dense diel light conditions, outperformed the already high wild type (WT) with biomass productivity gains of 5–24% and achieving bioreactor yields of 50 g m −2 day −1 . The highest productivities for STR30843 were achieved at moderate dilution rates, that is, denser cultures. At dilutions of 100% day −1 or lower, STR30843 consistently maintained higher biomass density (14–24%) and productivity (14–24%) and demonstrated better light utilization as compared to the WT at equivalent dilution rates (Table 1 , Figure 4 ). E k values remained consistently higher than the WT. The advantages of depigmentation might be partially negated by other factors such as enhanced carotenoid fraction relative to chlorophyll (Figures 4 and 5 ) or enhanced photoinhibition and needs to be probed further. While STR30843 did not have the dramatic alterations in thylakoid membrane architecture upon the loss/decrease of LHCs as observed in prior work (Asakura et al., 2004 ; Friedland et al., 2019 ; Mitra et al., 2012 ; Mussgnug et al., 2005 , 2007 ; Negi et al., 2020 ), the thylakoid branching was affected, and additional research is required to probe how changes in the architecture of the thylakoids may affect cell physiology, connectivity of photosystems, and electron conduction (Figure 3b–e ). The genetic alterations did not change the maintenance burden of the STR30843 compared with the WT, and the maintenance burden was low for both. This minimizes any penalty for operating the modified strain at higher biomass density than the WT. Thus, from a practical standpoint, productivity gains by the altered strain are attainable, whereas achieving the maximum productivity of the WT requires more dilution than is practical. The >50% decrease in pigments coupled with higher productivity without significant effect on respiration relative to WT under dense cultivation makes STR30843 a useful chassis strain to further pursue higher productivities in depigmented strains with balanced photosystem activities."
} | 5,850 |
27914060 | null | s2 | 7,730 | {
"abstract": "In recent years there has been increasing interest in nanostructure design based on the self-assembly properties of proteins and polymers. Nanodesign requires the ability to predictably manipulate the properties of the self-assembly of autonomous building blocks, which can fold or aggregate into preferred conformational states. The design includes functional synthetic materials and biological macromolecules. Autonomous biological building blocks with available 3D structures provide an extremely rich and useful resource. Structural databases contain large libraries of protein molecules and their building blocks with a range of sizes, shapes, surfaces, and chemical properties. The introduction of engineered synthetic residues or short peptides into these building blocks can greatly expand the available chemical space and enhance the desired properties. Herein, we summarize a protocol for designing nanostructures consisting of self-assembling building blocks, based on our recent works. We focus on the principles of nanostructure design with naturally occurring proteins and synthetic amino acids, as well as hybrid materials made of amyloids and synthetic polymers."
} | 294 |
21334091 | null | s2 | 7,733 | {
"abstract": "Horizontal gene transfer is increasingly described between bacteria and animals. Such transfers that are vertically inherited have the potential to influence the evolution of animals. One classic example is the transfer of DNA from mitochondria and chloroplasts to the nucleus after the acquisition of these organelles by eukaryotes. Even today, many of the described instances of bacteria-to-animal transfer occur as part of intimate relationships such as those of endosymbionts and their invertebrate hosts, particularly insects and nematodes, while numerous transfers are also found in asexual animals. Both of these observations are consistent with modern evolutionary theory, in particular the serial endosymbiotic theory and Muller's ratchet. Although it is tempting to suggest that these particular lifestyles promote horizontal gene transfer, it is difficult to ascertain given the nonrandom sampling of animal genome sequencing projects and the lack of a systematic analysis of animal genomes for such transfers."
} | 255 |
39948085 | PMC11825717 | pmc | 7,734 | {
"abstract": "Salinity is a major challenge for plant growth, but Populus euphratica , a species native to desert regions, has a remarkable ability to tolerate salt stress. This study aimed to explore how salinity affects the rhizosphere microbiome of P. euphratica , focusing on diversity patterns, assembly mechanisms, network characterization, and the functional roles of specialists and generalists under salt stress conditions. The findings revealed that increased salinity enhances the complexity of the rhizosphere microbial network and the diversity of bacterial specialists. Specialists demonstrated a wider range of environmental adaptation and played a pivotal role in species interactions within the microbial network. Notably, salinity stress altered the structure and assembly of plant rhizosphere specialists, facilitating functional compensation and potentially augmenting the health of P. euphratica . This research offers critical insights into the microbiome dynamics of P. euphratica under salinity stress, advancing the understanding of specialists and generalists in the rhizosphere.",
"introduction": "Introduction Soil salinization significantly hampers plant growth and crop productivity worldwide 1 . The rhizosphere microbiome, pivotal in aiding plant tolerance to salt stress, is often considered as the plant’s ‘second genome’ 2 , 3 . Numerous studies have shown when faced with high salinity conditions, plants can attract specific beneficial soil bacteria to their rhizosphere, fostering growth 3 – 8 . Populus euphratica , renowned for its ability to thrive in saline environments, possesses a distinct soil microbiome that may underpin its stress resistance 9 . Despite the critical interplay between P. euphratica and its rhizosphere microorganisms, research in this area remains sparse. Traditional studies have generally classified microorganisms as either abundant or rare taxa, neglecting their ecological niche. In response to diverse environmental conditions, species are often identified as specialists, generalists, or neutral taxa based on their niche breadth, which reflects the range of resources, habitats, or environments a species utilized 10 . This approach offers a novel perspective in microbial classification. Previous research indicated that microbial generalists had a stronger ability to evolve toward specialists 11 , although the underlying reasons for this evolutionary process remain unclear. Xu et al. 12 observed that the impact of generalists and specialists on soil microbial diversity in farmland varies, depending on network perspective, community assembly, and biogeographic patterns 12 . Specialists are more governed by deterministic processes, whereas generalists are swayed by stochastic processes. Liao et al. 13 found that while stochastic processes predominantly influenced the distribution of generalists in plateau lakes of China, deterministic processes played a more significant role in the assembly of specialists 13 . However, a study on Tibetan lake sediment microorganisms found that stochastic processes significantly affected both generalists and specialists 14 , with specialists maintaining robust connections within the network and exhibiting high modularity. Nevertheless, there is a lack of research on the assembly mechanisms and networks of specialists and generalists, underscoring the need for more exploration into their assembly and network characterization in the rhizosphere under salt stress. This will help uncover the various adaptive dynamics and evolutionary reasons behind the existence of specialists and generalists. Plants can modify the composition of the rhizosphere microbiome to better adapt to various soil conditions. There is evidence that plants may recruit beneficial microorganisms to aid growth under stressful and nutrient-limited environments 15 . Ren et al. 16 explored rhizosphere function across different soil environments, uncovering an effect for functional compensation 16 . The availability of nutrients likely influenced plant rhizosphere microbial communities, triggering functional compensation to boost host fitness. For instance, in nutrient-rich soil, nutrient cycling functions in the rhizosphere bacterial community might be downregulated, whereas nutrient cycling might become more crucial in nutrient-poor soil 16 . The question of whether functional compensation is a widespread phenomenon remains open for exploration, particularly regarding the roles of microbial generalists and specialists. This study aimed to (i) examine the composition and characteristics of specialist and generalist microbiomes in the rhizosphere of P. euphratica ; (ii) understand the community assembly mechanisms and network characterization of specialists and generalists under salt stress; (iii) decipher the functional potential of rhizosphere microorganisms in P. euphratica under salt stress. By elucidating the assembly patterns, network characterization, and functional roles of specialists and generalists in P. euphratica , we aspire to deepen our understanding of the ecological processes of rhizosphere microorganisms under salt stress. This knowledge may pave the way for novel strategies to manipulate microorganisms and enhance ecosystem functions in the P. euphratica rhizosphere community.",
"discussion": "Discussion Specific root-associated bacteria can be recruited by plants when confronted with salinity stress 4 . This adaptive strategy is notably evident in halophytes, which leverage root-associated microorganisms to enhance their resilience to salt stress 7 . Our research indicated that the rhizosphere microbiome of P. euphratica demonstrated functional compensation, with specialists in the rhizosphere adapting and performing necessary functions to aid the plant in surviving in salty conditions. We postulated that the harsh saline environment might have driven the evolution of these specialists, uniquely adapted to their specific ecological niche 11 , 18 . Conversely, it might also be interpreted as a strategic response by the plant, a “cry for help” to attract these specialist microorganisms, thereby fortifying its defense against the challenges posed by salinity 15 , 19 . Our findings revealed a positive correlation between soil salinity levels and the α diversity of bacterial specialists within the rhizosphere. The influence of salinity on the differentiation between specialist and generalist bacteria is notable. This distinction underscores the critical role of salt resistance among bacterial specialists. Our results showed that Bacteria had a stronger reaction to changes in salinity during the salt stress period than fungi. Bacterial Specialists demonstrated a positive response to salt stress, resulting in increased diversity. Although the community structure of fungi changed, there was no notable increase in fungal diversity. It has been reported that bacterial generalists are vital for maintaining community and functional stability in dynamic environments due to their broad ecological resistance and diversification 12 . However, in more static environments, specialists are key contributors to community diversity and function. Our research supports this paradigm within the relatively stable rhizosphere of P. euphratica . Further supporting our findings, we observed an increased abundance of bacterial specialists, particularly from the genera Planocucus and Planomicronium , in environments with elevated salinity. This aligns with existing research indicating these microorganisms’ potential for salt tolerance 20 , 21 . For instance, Planococcus rifietoensis , known for its moderate halotolerance 20 , has been shown to facilitate wheat growth under salinity conditions by converting ammonia into nitrogen, thereby enhancing soil fertilization. Additionally, this bacterium’s ability to metabolize potassium contributes to maintaining ion balance within plant cells 21 . Similarly, the discovery of Planomicrobium iranicum sp. nov . highlights the emergence of slightly halophilic bacteria adapted to saline environments 22 . Moreover, Li et al. suggest that the broader capability of soil bacteria to mitigate salt stress in plants, extending beyond the microorganisms’ own salinity tolerance levels 4 . This trend suggests that these salt-tolerant species may colonize the rhizosphere to compensate for functional deficits induced by salt stress. We found that specialists exhibit a more deterministic assembly process, this is consistent with the results of previous studies 12 , 13 , 18 . Heterogeneous selection was found to play a more substantial role in the assembly of specialists compared to generalists 23 . Bacterial specialists and generalists exhibit greater certainty in high salinity stress and extremely low salinity conditions. A previous report had shown that environmental filters are more pronounced under extreme, such as highly variable soil pH 24 , indicating that challenging environmental conditions may amplify the deterministic processes governing microbial community assembly mechanisms. With increasing salinity, especially bacterial specialists experienced enhanced diffusion limitations. This finding suggests that salinity may act as a deterministic force influencing the diffusion process of microorganisms. This is consistent with the results of previous studies, stochastic processes, particularly dispersal limitation, played critical roles even under high-stress conditions 25 , 26 . Previous results have also found that bacteria are more affected by diffusion limitations than fungi 27 . However, the stronger stochastic exhibited by fungi may be due to the better stability of fungal communities under stress 23 , 28 . The assembly mechanism of fungi seemed to be more influenced by randomness, indicating that salinity factors had a smaller impact on fungi in comparison to bacteria. In summary, within the rhizosphere of P. euphratica , specialists play a pivotal role in shaping community diversity through deterministic processes and dispersal limitation. Our study revealed that specialist organisms in the rhizosphere of P. euphratica exhibited a greater environmental threshold compared to their generalist counterparts. Despite their narrower ecological niche, specialists demonstrated an ability to thrive across a broader spectrum of environmental factors within specific habitats. This finding aligned with the ecological principles of categorizing species as specialists or generalists based on Levins’ niche breadth. Generalists, despite their adaptability to a wide range of environments, are at a disadvantage in specialized habitats. On the other hand, specialists, with their narrow ecological niche, demonstrate superior competitiveness and adaptability in specialized environments when compared to generalists. The rhizosphere of P. euphratica was characterized by reduced fluctuation, providing specialists with a survival advantage. While generalists were capable of adapting to diverse ecological settings, they tended to be outcompeted by specialists within certain niche ranges. It has been noticed that in stable environments, specialists are more likely to contribute to enhancing community diversity than generalists 12 . Furthermore, an increase in salinity had been observed to complicate the network dynamics within rhizosphere communities, which exhibited distinct network interactions in response to external disturbances 29 . Despite these complexities, specialists constituted a significant portion of the network across various salinity levels. Our analysis of the microbial co-occurrence network indicated a closer association between bacterial specialists and neutral groups, with specialists and neutral groups dominating in eight observed cases. Network eigenvalues revealed that specialists generally had higher values than generalists, except betweenness centrality and eccentricity. This could be attributed to the prominent centrality and eccentricity of specific ASV mediators among generalists, suggesting that specialists occupied central roles within the network and maintained strong connections with neutral groups, thereby exhibiting high modularity 14 . The composition of most modules primarily included specialists, neutral taxa, and a few generalists, apart from one module where generalists predominated, implying a significant interconnection and functional exchange among specialists and neutral tax 12 . Keystone species predominantly consisted of bacterial specialists and neutral taxa. Fungi had only two keystone species. This underscored the pivotal role of bacterial specialists over generalists within the network. In a previous study, researchers identified the influential microbial players in a network, using IVI and some other centrality measures 30 . In our research, we also used the IVI to identify key microbes. We obtained a significant positive correlation between the key microbes and the network complexity and salinity. Many of these key microbes are bacterial specialists and neutral groups, highlighting the significance of bacterial specialists in high salt stress conditions rather than fungi. These microorganisms play a crucial role in regulating microbial interactions under high salt stress, potentially aiding in functional compensation. Salt stress can significantly alter the metabolic and ecological functions of root-associated bacteria 4 . Rhizosphere microbes play a crucial role in enhancing plant salt stress tolerance by re-establishing ion and osmotic homeostasis, thereby preventing damage to plant cells and facilitating the resumption of plant growth under salt stress conditions 2 . Our research demonstrated that in high salinity environments, the functions of the rhizosphere microorganisms, particularly those that bolster plant tolerance to abiotic and biotic stress-are increasingly valued, leading to functional compensation in the P. euphratica rhizosphere. Specifically, metabolic pathways such as “Ascorbate and aldarate metabolism”, “Arachidonic acid metabolism”, “Ubiquinone and other terpenoid−quinone biosynthesis”, and “Terpenoid backbone biosynthesis”, “glycan biosynthesis, are generally involved in enhancing plant stress tolerance 16 . For instance, l -ascorbic acid (AsA) was a plentiful metabolite in plants, playing crucial roles in stress physiology as well as growth and development 31 . Similarly, arachidonic acid has been identified as a signaling molecule that can attract beneficial microbiota to the rhizosphere, thus promoting plant growth and facilitating nutrient turnover in the soil 32 . Additionally, the triterpenoid compound cucurbitacin has been found to improve plant disease resistance by regulating the rhizosphere flora 33 . However, there are numerous metabolic functions related to stress resistance that remain unexplored. Our observations indicated that while many of these functions were marginalized in low salt conditions, they gained prominence in high salt soil environments. Further investigation into the roles of these metabolisms revealed significant implications for plant health. These findings highlight the multifaceted roles of bacterial specialists in supporting plant resilience and health in saline conditions. In summary, our study elucidates the significant impact of salinity on the formation and function of specialist versus generalist bacteria within the rhizosphere, highlighting the adaptive strategies that enable certain bacteria to thrive under saline stress. Our findings also confirmed that the P. euphratica rhizosphere microbiome also employed a functional compensation in response to salt stress, highlighting the pivotal role of bacterial specialists in this process. This adaptive response may be attributed to the recruitment of more salt-resistant and microorganism specialists by P. euphratica as salinity levels increase. Previous studies have also found that generalist-to-specialist transformations occur three times more frequently than the reverse transformations 11 . It is hypothesized that in the rhizosphere of P. euphratica , the increase in salinity may trigger functional compensation, leading to the shift from generalists to specialists. This insight not only advances our understanding of microbial ecology in saline environments but also points to potential avenues for leveraging these microbial adaptations under salinity stress aimed at exploring the formation causes of specialists and generalists and enhancing crop resilience to salinity stress. Moving forward, we aim to identify and further investigate salt-tolerant strains among these specialists, exploring their function and the potential for creating synthetic microbial communities (SynCom). The development of artificially selected microbiomes that confer salt tolerance represents a promising strategy to enhance agricultural productivity 34 . The engineering of the desert microbiome into SynCom capable of protecting plants in natural soils from abiotic stress opens new avenues for agricultural innovation 35 . Quite a few studies have demonstrated that root endophytes also have the ability to help plants withstand stress tolerance 36 – 39 . By integrating the findings on endophytes with the importance of rhizosphere microorganisms in salt tolerance, we anticipate uncovering novel and intriguing insights in further studies. Our findings not only shed light on the dynamics of the P. euphratica rhizosphere microbiome under salt stress but also provide a valuable framework for the selection of salt-resistant strains. This research lays the foundation for future studies on the interplay between specialists' and generalists' microorganisms in the P. euphratica rhizosphere, offering insights that could lead to the development of resilient agricultural systems in arid and saline environments. Research on the rhizosphere microorganisms of P. euphratica has revealed that an increase in salinity will lead to an increase in the α diversity of bacterial specialists and alterations of structure. Changes in salinity levels have an effect on the assembly of bacterial specialists and generalists, with the former being more characterized by deterministic processes and exhibiting wider adaptation. Furthermore, bacterial specialists are found to play a more significant role in the microbial community. The relationship between key microbes, particularly bacterial specialists, and network complexity is strongly positive. As salinity levels increase, the metabolic function of microorganisms becomes more crucial, shaping the assembly of plant rhizosphere microbial communities under stress. This stress prompts a functional compensation that enhances plant health, as P. euphratica recruits specialized rhizosphere microorganisms. This research highlights the importance of the plant-microbe interaction in promoting resilience and adaptability in the face of environmental challenges, shedding light on the diversity, assembly, network characterization, and functions of bacterial specialists and generalists in the rhizosphere of P. euphratica ."
} | 4,786 |
31263456 | PMC6584816 | pmc | 7,735 | {
"abstract": "An intriguing aspect in microbial communities is that pairwise interactions can be influenced by neighboring species. This creates context dependencies for microbial interactions that are based on the functional composition of the community. Context dependent interactions are ecologically important and clearly present in nature, yet firmly established theoretical methods are lacking from many modern computational investigations. Here, we propose a novel network inference method that enables predictions for interspecies interactions affected by shifts in community composition and species populations. Our approach first identifies interspecies interactions in binary communities, which is subsequently used as a basis to infer modulation in more complex multi-species communities based on the assumption that microbes minimize adjustments of pairwise interactions in response to neighbor species. We termed this rule-based inference minimal interspecies interaction adjustment (MIIA). Our critical assessment of MIIA has produced reliable predictions of shifting interspecies interactions that are dependent on the functional role of neighbor organisms. We also show how MIIA has been applied to a microbial community composed of competing soil bacteria to elucidate a new finding that – in many cases – adding fewer competitors could impose more significant impact on binary interactions. The ability to predict membership-dependent community behavior is expected to help deepen our understanding of how microbiomes are organized in nature and how they may be designed and/or controlled in the future.",
"introduction": "Introduction Microbial communities develop emergent properties through complex networks of interspecies interactions, which cannot be understood by simply examining the behavior of all members individually. Microbial interaction dynamics underpin community functions associated with various critical processes for society, such as environmental sustainability ( Gougoulias et al., 2014 ; Rousk and Bengtson, 2014 ), human health ( Cho and Blaser, 2012 ; Young, 2017 ), and bio-manufacturing industry ( Gustavsson and Lee, 2016 ; Schmidt-Dannert and Lopez-Gallego, 2016 ; Valdivia et al., 2016 ). Predicting – and ultimately engineering – stable and productive functioning of microbial communities requires a fundamental understanding of how microbes influence each other and self-organize into complex interaction networks in response to environmental changes and compositional shifts ( Lindemann et al., 2016 ; Song, 2018 ; Song et al., 2018 ). Microbe-microbe interactions can be either bidirectional or unidirectional and hence can be systematically classified as pairs: positive-positive (mutualism and synergism), negative-negative (competition), positive-negative (antagonism such as predation and parasitism), positive-neutral (commensalism), negative-neutral (amensalism), and neutral-neutral (no significant interaction). Types and strength of interspecies interactions can be computationally inferred from experimental data with respect to these categories. These data are typically observed as population abundances and growth kinetics, because comprehensive molecular-level multi-omics data that relate to microbial interactions are rare. Despite an increasing number of inference methods reported hitherto ( Faust and Raes, 2012 ; Song et al., 2014 ; Faust et al., 2015 ), reliable, comprehensive prediction still faces challenges. Difficulties often arise from insufficient resolution of input data (species population or abundance), typically associated with data sparsity and noise, as well as the uncertainty in differentiating direct ecological interactions from indirect relationships ( Li et al., 2016 ). However, in our view, more fundamental challenges lie in the lack of our understanding of how the microbial interactions in a community are organized as an interaction network and how they affect complex ecological dynamics ( Konopka et al., 2015 ). Researchers have sought to develop new tools to manage these challenges from both experimental and modeling perspectives. One such tool has been the use of combinatorial co-cultivation, which provides insight into core pairwise interactions between cultivable microbes ( Wintermute and Silver, 2010 ; Bernstein and Carlson, 2012 ; Grosskopf and Soyer, 2014 ; Khan et al., 2018 ). Behavioral changes in single species that are treated with specific binary partnerships can be synergistically examined through experiments ( Bernstein et al., 2012 , 2017 ) and modeling analyses ( Song et al., 2014 ; Henry et al., 2016 ). In many cases, however, knowledge from simple binary cultures is not directly translatable to complex, multi-species communities due to the effect of additional members. That is, interaction coefficients between species identified in simple and complex communities often show quantitative and even qualitative differences, which is defined here as neighbor dependence of interactions . In community ecology, this aspect of interactions has been studied as part of a broader concept termed context dependence, which includes the impact of spatial and abiotic gradients, as well as biotic factors ( Chamberlain et al., 2014 ). A prime example of neighbor-dependent interaction in the murine and human gut microbiome showed that, after antibiotic treatment, a pathogen Clostridium difficile in the digestive tract suppressed the growth of residing intestinal bacteria by producing cytotoxin ( Buffie et al., 2015 ). Intestinal bacteria developed resistance to C. difficile , however, by the addition of another species, Clostridium scindens , which produces secondary bile acids that can impair the growth of the pathogen. Microbial interactions modulated by additional members are often implied by population shifts. In colonic fermentation of carbohydrates not accessible to degradation by human enzymes, population sizes of lactobacilli incapable of directly degrading long-chain inulin were modulated by organisms possessing enzymes to degrade these polysaccharides ( Rossi et al., 2005 ). Furthermore, differential metabolism of members of the altered Schaedler flora, a well-studied model microbiome of mice, was observed when members grown in fresh media vs. spent media from other members of the community ( Biggs et al., 2017 ). These effects are not specific to the gut microbiota; synergistic growth responses were observed for ternary communities of environmental microbes in fermentations of cellulose and xylan ( Deng and Wang, 2016 ). The above examples indicate that interaction coefficients between two species are not constant, but can vary as a function of community membership and species population density. Changes in pairwise interactions by a third-party organism can be caused in two ways ( Wootton, 1993 ; Momeni et al., 2017 ): (1) interaction modification and (2) interaction chain. In the first type, a third-party organism affects interaction by forming a multi-way relationship, which is thus known as higher-order interaction ( Kato et al., 2008 ; Voit, 2013 ; Zelezniak et al., 2015 ; Bairey et al., 2016 ; Grilli et al., 2017 ; Levine et al., 2017 ). In the second type, a third-party organism does not directly participate in interaction, but affects the density of existing species (thus known as density-mediated indirect interaction). The values of pairwise interaction coefficients in this case can also be modified when the pairwise interaction is driven by a non-linear functional response, while they may remain constant if not. As such, the changes in interactions can be caused by either higher-order interactions or non-linearity. Regardless of what the main source is, understanding such membership-dependent interactions is an ecologically important goal, which requires advanced concepts beyond the scope of traditional theoretical approaches that assume fixed interactions. The analysis of population data alone does not allow to identify the true source of modification in pairwise interactions. Current practices of ecological community modeling tend to account for this effect by adding higher-order interaction terms. However, that approach substantially increases the number of interaction parameters, making it challenging to reliably infer microbial interactions, particularly for compositionally complex natural communities, which are often composed of hundreds of different taxa or more. In this article, we propose a novel concept of a rule-based network inference method that enables predicting neighbor-dependent interactions without assuming any functional forms of neighbor dependence of interactions and therefore is scalable to complex communities. Our method predicts the modulation of interaction in complex communities from the knowledge of pairwise interactions derived from binary growth dynamics based upon the assumption that the presence of neighbor species will perturb these intrinsic interactions but only to a minimal degree. We have named this concept the minimal interspecies interaction adjustment (MIIA), the utility of which was demonstrated through comprehensive in silico experiments. Specifically, the MIIA hypothesis led to reliable predictions of the change in interaction coefficients in a complex community in an either positive or negative direction. In the analysis of experimental data derived from a microbial community of eight competing soil bacteria ( Friedman et al., 2017 ), we obtained an interesting finding that interspecies relationships can be significantly changed when perturbed by fewer competitors. The modulation of interspecies interactions diminished as competitive neighbors increase. This is important knowledge for the effective control and design of microbial communities. The unique concept of MIIA may contribute to revealing many other intriguing aspects of interspecies interactions in ecological communities beyond microbial systems.",
"discussion": "Discussion We present and critically evaluate MIIA a new concept of network inference that addresses the following, related ecological questions: (1) how are interspecies interactions modulated by the shifts in community composition and species populations? and (2) to what extent can interspecies relationships observed in simple cultures be translated into complex communities? Our results show that pairwise interactions in binary communities can serve as meaningful references for predicting interactions in more complex communities, e.g., when translated through the minimal adjustment assumption. It needs to be noted that the term “minimal” does not imply little changes because as shown in our examples, this minimal assumption enables predicting dramatic changes of interactions, encouragingly from the analysis of population changes in communities, even without fundamental knowledge of underlying interaction mechanisms. Beyond a conceptual basis for understanding context-dependent microbial interactions, our work also provides critical information practically useful for the control and design of microbial interactions and community function. First, we demonstrated that our framework can provide reliable prediction of the direction of interaction modulation, i.e., how pairwise interactions are weakened or strengthened under the influence of third-party organisms. Second, through the case study of competing bacteria, we also found that the competitive relationship can be significantly modulated when perturbed by a small number of species, but the level of modulation diminishes as the number of new competing members increases. These results together can serve as useful lessons for a rational engineering of microbial community functioning through compositional and population changes. How can the minimal adjustment be rationalized from an ecological perspective? While this is an open question, we conjecture that the answer would be associated with stability characteristics and energetics of microbial interaction networks. MIIA implicitly assumes that pairwise interactions observed from a given binary environment are unique and stable – otherwise, taking them as references may not be reasonable. When perturbed by a new member, the community’s stability landscape will be altered and the previous interaction network may be unstable or less stable, leading to the reorganization of interaction network toward a more stable state. In this reorganization, species may need to change their enzyme settings to adapt to a new environment (created by new members) at the expense of energy. The solution that MIIA chose, minimizes the adjustment in pairwise interactions, implying that network reorganization will occur toward spending minimal energy, the amount of which is assumed to be proportional to the level of modulation. In a different area, the minimal adjustment concept has been employed for predicting the metabolic response of individual microorganisms to gene knockouts ( Segre et al., 2002 ; Shlomi et al., 2005 ). However, our approach is fundamentally different in that we apply this idea to microbes that (can) change their function based on the context, whereas genes generally do not. Higher-order interactions are a major factor that exerts a profound impact on pairwise interactions. Ideally, evaluation of this effect would require the use of simulated data generated from properly formulated higher-order interaction models. However, identification of appropriate values of interaction parameters for ensuring stable population profiles was a taxing task, particularly for complex communities composed of a large number of member species. In this work, we instead used random variations of interaction parameters to realize neighbor-dependent interactions, including the cases where modulated interaction coefficients show significant deviations from binary parameters (as controlled by the parameter β up to 1), which therefore provided unbiased datasets for assessing the validity of our method. The resulting datasets generated as such could be broadly useful because our focus is not necessarily limited to higher-order interactions but includes non-linear interactions. The understanding of context dependence of interactions is a key component for making microbial ecology as a more predictive discipline. Toward this end, the generalizability of the proposed concept will be tested through further case studies in diverse contexts and applications including natural microbial communities (such as in human body, soil, and marine environment), as well as engineered synthetic consortia. Modern meta-omics analyses and other advanced experimental techniques offer powerful tools for quantifying interspecies interactions in microbial communities, the resulting data of which will serve as a valuable basis for informing and validating computational network inference approaches such as the one we proposed in this work."
} | 3,737 |
26499760 | PMC4620479 | pmc | 7,736 | {
"abstract": "Modern microbial mats can provide key insights into early Earth ecosystems, and Shark Bay, Australia, holds one of the best examples of these systems. Identifying the spatial distribution of microorganisms with mat depth facilitates a greater understanding of specific niches and potentially novel microbial interactions. High throughput sequencing coupled with elemental analyses and biogeochemical measurements of two distinct mat types (smooth and pustular) at a millimeter scale were undertaken in the present study. A total of 8,263,982 16S rRNA gene sequences were obtained, which were affiliated to 58 bacterial and candidate phyla. The surface of both mats were dominated by Cyanobacteria, accompanied with known or putative members of Alphaproteobacteria and Bacteroidetes. The deeper anoxic layers of smooth mats were dominated by Chloroflexi, while Alphaproteobacteria dominated the lower layers of pustular mats. In situ microelectrode measurements revealed smooth mats have a steeper profile of O 2 and H 2 S concentrations, as well as higher oxygen production, consumption, and sulfate reduction rates. Specific elements (Mo, Mg, Mn, Fe, V, P) could be correlated with specific mat types and putative phylogenetic groups. Models are proposed for these systems suggesting putative surface anoxic niches, differential nitrogen fixing niches, and those coupled with methane metabolism.",
"conclusion": "Conclusions This is the first study to characterise bacterial distribution at a millimeter scale in Shark Bay microbial mats, employing high throughput sequencing platforms. Elemental analyses have provided insights into potential links with the prevailing biota, although further work is needed to definitively conclude whether a given element is indeed being sequestered by a particular biological process, or whether geochemical processes are influencing their enrichment. It is postulated that niche differentiation may occur and can help our understanding of how microbial communities can survive and adapt in environments of high salinity, UV radiation, desiccation and diel fluctuation of biogeochemical gradients. Although this study has provided key insights into the spatial distribution of bacteria with potential niches postulated, as alluded earlier delineating the depth distribution of archaea in particular will be important to fully characterise these communities. Seasonal analyses may also facilitate building potential models of microbial succession in these systems. Another caveat is that 16S rDNA data can be limited in facilitating exact links with proposed functions, and thus recent metagenomic studies 23 , as well as future work examining defined gene expression in specific organismal groups in Shark Bay, will allow for robust conclusions to be drawn on the complex interactions proposed here. Nonetheless this study has enabled a holistic view of these ecosystems, and facilitated a greater understanding of the complex network of putative niches and microbial interactions in modern microbial mats.",
"discussion": "Discussion This study delineates for the first time the spatial distribution of bacteria with depth in distinct Shark Bay microbial mats, complemented by biogeochemical measurements and elemental analyses. One significant finding of this study was the large proportion of novel bacterial phyla found in these mats. All of these new phyla were not identified in previous studies on these microbial ecosystems except TM6 8 ( Supplementary Table 2 ). The new sequences substantially increase the representation of these novel candidate bacteria in Shark Bay, and illustrate the power of next generation sequencing to improve our understanding of microbial diversity in these systems. The novel bacteria comprise up to 10–15% of the microbial population, potentially representing (a range of) novel metabolic pathways. Among the novel bacterial phyla, Caldithrix, OP3, OP8 and GN01 were found exclusively in smooth mats. All of these sequences were found in the anoxic zone, demonstrating the potential preference towards a strict anaerobic lifestyle. It has been suggested that OP8 possesses versatile metabolic capabilities 36 37 , and some of its members have branched out to form the novel phylum Nitrospirae 17 . Candidate phylum GN01 has also been found in hypersaline mats in Guerrero Negro, Mexico, suggesting a link between microbial mats in disparate geographic locations 22 . It is hypothesized that different biogeochemical environments (i.e., shallow oxygen penetration and higher sulfide concentrations, higher rates of sulfate reduction and O 2 production and consumption) observed in smooth mats in the present study may contribute to the prevalence of these relatively unique groups. However exactly why these candidate phyla were only found in smooth mats, and their ecological role(s), remains unknown. Previous studies suggest that metabolite exchange occurs at a micrometer scale in microbial mats, posing a steep biochemical gradient 9 38 . Niche differentiation is a concept that describes the tendency for co-existing species to differ in their adaptations to the habitat, and is believed to occur among this gradient to sustain the vast biological diversity of microbial mats 39 . The high microbial diversity observed in the present study is likely to be a result of niche differentiation in the Shark Bay systems. Furthermore, the diel fluctuations of O 2 and H 2 S, along with the phylogenetically stratified nature of the microbial mats, suggest metabolic cooperation, which leads to niche differentiation. Putative phototrophic consortium The surface layer of smooth mats is dominated by Cyanobacteria, which comprise 40% of sequences. Cyanobacteria are generally considered the main producers of photosynthates, including exopolymeric substances (mucilage) that fuel and protect the mats 9 40 , and the high abundance of Cyanobacterial rRNA genes in the surface layer of the mat is consistent with that role. Other known phototrophic or putative phototrophic members of the Alphaproteobacteria (e.g. Rhodobacterales) and Gammaproteobacteria (e.g. Chromatiales) were also localized at the surface, with both representatives containing bacteriochlorophylls and some that are photoheterotrophic, (photolithotrophic or chemolithotrophic) in light 41 . STAMP analysis quantitatively demonstrated that a significantly abundant cluster of Cyanobacteria, Bacteroidetes, Alpha- and Gammaproteobacteria was localized at the surface layer of smooth mats, while pustular mat surfaces were characterised by various Cyanobacteria and Bacteroidetes classes ( Supplementary Figs S4a and S5a ). Furthermore, network correlation revealed an interconnected cluster consisting of Alphaproteobacteria (Rhodobacterales), Gammaproteobacteria, Cyanobacteria and Bacteroidetes (Cytophagia and Saprospirae) in smooth mats ( Supplementary Fig. S6 ), while similar patterns were observed at the surface of pustular mats, except that Gammaproteobacteria was negatively correlated to the Bacteroidetes-Alphaproteobacteria-Cyanobacteria cluster (Supplementary Fig. S7). These significant co-occurrences and associations between the bacterial taxa and elements may be used as an indicator of potential niche preferences or synergetic relationships 30 . Empirical evidence from a global ocean metagenomics expedition has shown that Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes are major carriers of proteorhodopsin 42 , a novel photo-enhanced energy-harvesting protein. Proteorhodopsin-based phototrophs represent a potentially unique mode of energy acquisition and carbon assimilation in marine environments 43 44 . It is proposed that in this putative phototrophic niche at the mat surface, cyanobacteria and putative proteorhodopsin-carrying Proteobacteria and Bacteroidetes produce organic matter that maintains the heterotroph biomass. Bacteroidetes may also have a putative ecological role of breaking down high molecular weight (MW) macromolecules, including EPS, and cyanobacterial derived metabolites, making organic matter more accessible to the microbial community 23 45 . Bacteroidetes are known to have a preference towards metabolising high MW organic matter and complex polysaccharides 46 . It is suggested that Bacteroidetes are associated with Cyanobacteria in close proximity in order to access the high molecular weight organic exudates 47 , as smooth mats may have a surplus of organic carbon on the surface due to the high O 2 production and consumption rates observed. Therefore the microbial community underneath the surface of mats could gain access to the degraded organic matter, and it is this exploitation of energetic opportunities that likely supports the high microbial diversity in Shark Bay. Putative anoxic niche at mat surfaces As light has limited penetration through microbial mats 9 48 , it was not surprising that both mats had enriched phototrophs at the surface, which decreased along with depth (Supplementary Fig. S8). However, there was also an abundance of potential anaerobic fermenters residing at the surface in both mat types. Pustular mats also had a lower abundance of cyanobacteria at the surface compared to smooth mats, and unexpectedly anaerobes were more abundant than phototrophs even at the surface layer in pustular mats. Aerobic heterotrophs and sulfate-reducing bacteria (SRB) were well distributed throughout smooth mats, however the obligately anaerobic SRB were found even at the oxic surface layers (10% of OTUs in smooth mats, 5% of OTUs in pustular mats). The majority of the detected representatives of the Deltaproteobacteria in Shark Bay were Desulfococcus and Desulfovibrio , suggesting sulfate reduction even at the oxic surface layer. The presence of SRB has been reported for several mats with smooth surfaces and high metabolic activity of the entire community 18 19 49 . The finding of SRB in the oxic zone of the mat was supported by data in the present study on sulfate-reduction in the mats ( Fig. 5 ), in addition to a recent study whereby sulfate-reducing activity was also observed at the oxic surface layer in smooth mats in Shark Bay 15 . It is speculated that the anaerobic fermenters at the surface may have a role in providing low molecular weight organic carbon and H 2 to the SRB. Thus it is suggested that an anoxic niche exists in the surface mat matrix to protect SRB from oxidative stress. The distribution of SRB in microbial mats forming marine stromatolites in the Bahamas was mapped using Geographical Information Systems and concluded clusters of SRB were common in lithifying mats 50 . Furthermore, our data indicating a large portion of other anaerobic members at the surface oxygenated layer (e.g. Anaerolineae, Spirochaetes, Planctomycetes, Deltaproteobacteria) provides further support for the existence of an anoxic niche in the photic-oxic zone of Shark Bay microbial mats, or alternatively, some possess oxygen-insensitive metabolisms or survival strategies 51 . As alluded to earlier, one potential ecological role of the surface anoxic niche might be mat lithification. Sulfate reduction had also been found in the oxygenated zones in a diverse range of microbial mats 19 , such as hypersaline mats in Guerrero Negro, Solar Lake, Kiritimati Atoll, and Bahamas 11 18 52 53 . SRB that are localised in the photic-oxic zone have been linked to carbonate precipitation at the mat surface 24 28 54 . The coupling of SRB activities and cyanobacteria may be linked to mat lithification (net carbonate precipitation) 3 . The recent identification of specific niches of sulfide production coinciding with cyanobacterial photosynthesis 15 , further supports the proposal here of metabolic cooperation in modern microbialites, especially as it relates to the higher lithification potential proposed for smooth mats than for pustular mats 27 . Putative spectral photoheterotrophic niche under photic zones In hypersaline environments, energy-expensive osmotic regulation is required for microbial survival, thus anaerobic microorganisms that have low energy-yielding metabolisms should theoretically be eliminated due to bioenergetics constraints 55 . Although some studies have demonstrated that hypersalinity can limit microbial diversity 56 57 , our study indicates a highly diverse microbiota. Given both the high diversity and abundance of putative anaerobic heterotrophs in both mat types (Supplementary Fig. S8), it suggests an ecological model in Shark Bay where niche differentiation may facilitate tightly coupled microbial interactions. Interestingly, as vibrations of water molecules and the surface coarse sediment can scatter light, a series of distinct spectral niches can be generated along the light spectrum even under photic zones 58 59 . As both mats examined in the present study are situated in the intertidal zone in Shark Bay, daily disturbance of tidal waves may potentially contribute to creating distinct spectral niches, thus facilitating a significantly different microbial structure in smooth and pustular mats. Indeed, conceptually it seems unlikely for phototrophs to occur deep in the mat, however an unexpected trend of increasing cyanobacteria (e.g. Microcoleus sp.) was observed below the oxic zone at 5–20 mm depth in smooth mats in the present study, suggesting the potential for a distinct spectral niche deep in the mat. There has been the report of a cyanobacterium isolated from Shark Bay stromatolites containing a novel chlorophyll f , which can absorb light at the infrared part of the spectrum 60 . Given the fact that infrared light can penetrate deeper into microbial mats 48 , it is possible that cyanobacteria deep in the smooth mats potentially possess these (or other) novel chlorophylls, where the red-shifted pigment could extend the phototrophic light spectrum. Microcoleus has also been shown to switch from an oxygenic to anoxygenic phototrophic lifestyle under anoxia with high levels of H 2 S 61 62 . However it should be noted that Microcoleus in particular has been shown to migrate vertically in a diel cycle 63 64 65 , and also demonstrates alternate metabolisms such as fermentation of endogenous carbohydrates under dark, anoxic conditions 66 67 . A recent study has found phototrophic Gemmatimonadetes in geographically dispersed environments containing carotenoids as the phototrophic pigment 68 , suggesting this poorly described phylum as photoheterotrophic. Thus Gemmatimonadetes found in pustular mats may potentially be putative phototrophic and form a distinct spectral phototrophic niche in the anaerobic zone. The potential phototrophic niche below the photic zone may have an ecological role for “filling the niches”, increasing the efficiency of energy production and nutrient cycling, allowing microbial co-existence and diversity through niche creation 59 69 . A H 2 S-molybdenum paradox? Interestingly, using the methods employed in the current study, molybdenum (Mo) was only detected in smooth mats. The predominant form of Mo in typical marine habitats is the molybdate salt 70 . Studies have demonstrated that the presence of molybdate inhibits sulfate respiration, hence restricting H 2 S concentrations in microbial mats 19 71 72 73 . However, in situ microelectrode measurements undertaken in the present investigation indicated that smooth mats have more than twice the concentration of sulfide than pustular mats over depths of 4–8 mm ( Fig. 3 ), and higher sulfate reduction rates ( Fig. 5 ). Thus the co-existence of a H 2 S-rich zone and Mo leads to a potential sulfide-Mo paradox in smooth mats. It is possible that Mo may not manifest itself in the form of molybdate in the Shark Bay mats, or there is a specific niche protecting SRB from molybdate inhibition. Furthermore, Mo detected in smooth mats might be in the form of molybdoenzymes (e.g. nitrogenase, sulfite oxidase) 74 75 , and certain species of SRB can produce MoSO 2 76 which potentially contributes to the higher detectable levels of both Mo and sulfide observed in smooth mats. However, further work is needed to clarify which Mo species is present and whether Mo has any defining roles in the Shark Bay microbial mat systems. Putative nitrogen fixation niches in smooth mats Although cyanobacteria are thought to be the primary nitrogen fixers in microbial mats, research has shown direct coupling of nitrogen fixation with sulfate reduction in microbial mats and marine sediments 77 78 79 . In our study, STAMP analyses revealed that bottom mat layers are characterised by Deltaproteobacteria, and studies in deep-sea systems have shown that nitrogen-fixing Deltaproteobacteria are often coupled with methane-oxidising archaea in a syntrophic relationship 80 . Furthermore, horizontal gene transfer of the nitrogenase gene has been observed globally, whereby nitrogenase gene sequences were found in Clostridium Firmicutes, Delta- and Gammaproteobacteria 81 . The sharp peak of Firmicutes at the bottom layer in smooth mats (Supplementary Fig. S2g), along with the increasing trend of Microcoleus and the abundance of Deltaproteobacteria, suggests a potential nitrogen-fixing niche in the lower depths of smooth mats. Consortial N 2 fixation is the suggested symbiotic strategy in smooth mats where microorganisms cooperate to enhance nitrogen cycling 82 83 . Mo is a co-factor of nitrogenase, which may potentially facilitate the formation of a nitrogen fixation niche in smooth mats 74 . Of further relevance to nitrogen cycling, network correlation analyses indicated that Mo was only correlated to the phylum Caldithrix ( Fig. 8d ), and this bacterial phylum was only found in smooth mats. This novel phylum is known to be involved in the nitrogen cycle by performing nitrate reduction 84 , and may play such a role in the smooth mats. Interestingly, typical nitrifying bacteria (i.e. Nitrosomonas, Nitrobacter ) were not detected in both mats in the present study. This is consistent with a recent metagenomic study, where it was suggested an alternative pathway of nitrification may be present in these systems 23 , potentially filled by ammonia-oxidising archaea. Further work on archaeal diversity and distribution in the Shark Bay mats is required to ascertain the exact role(s) of this domain in nitrogen cycling. Putative nitrogen fixation and methanogenic/methanotrophic niches in pustular mats Network correlation and elemental analyses in pustular mats demonstrated that vanadium is positively correlated to surface Bacteroidetes and Verrumicrobia ( Fig. 9b ). In addition, elemental analyses revealed that pustular mats have a higher concentration of vanadium and the levels peak at the surface. Although a recent study demonstrated that some Bacteroidetes are able to carry out nitrogen fixation 85 , it is not known if this is the case in Shark Bay. However, as Mo was not detected in pustular mats, nitrogen fixation in this system could potentially be performed in part by Mo-independent nitrogenase, such as iron or vanadium nitrogenase 86 87 . Taken together, this suggests a putative nitrogen fixation niche at the surface in pustular mats, although there is the potential for vanadium to be associated with other metabolisms. Interestingly, it has been shown that Mo-independent nitrogen fixation is coupled with a significant amount of H 2 as a by product 88 . The large quantity of H 2 produced is suggested to maintain a population of methanogenic archaea in pustular mats 89 . Although SRB also reside at the mat surface and it has been shown that SRB thermodynamically outcompete methanogens 90 91 , the use of non-competitive substrates 21 or spatial compartmentalisation may allow the coexistence of SRB and methanogens. Furthermore, the excess amount of photosynthetically originated organic carbon derived from the steep O 2 gradients observed in the present study may support methanogenesis at the surface 21 . Verrucomicrobia are more abundant in pustular mat surfaces and known to exhibit aerobic methanotrophy 92 93 , which could potentially metabolise methane produced by the putative archaeal community in pustular mats. Therefore it is suggested that there is a putative nitrogen fixation niche at the pustular mat surface, coupled with a putative methanogenic and methanotrophic niche. Potential evolutionary significance of smooth mats The presence of Mo in smooth mats may have potential evolutionary implications. Marine Mo concentrations were less than one-tenth of modern values in the Precambrian, where Mo-nitrogen co-limitation may have influenced the early evolution of cyanobacteria 94 . Thus, there was likely strong selection for bacteria with strategies for Mo acquisition and retention during the early development of microbial mats. Furthermore, enrichment of Mo is also observed in Guerrero Negro microbial mats 95 , and it is proposed that elevated levels of Mo can be used as a geochemical tool to indicate conditions associated with the presence of microbial mats in ancient hypersaline environments 73 95 . Potential ecological significance of niche differentiation in Shark Bay mats The dominance of the Chloroflexi class Anaerolineae in smooth mats but not in pustular mats, along with the novel candidate bacterial phyla found exclusively in smooth mats may be a result of niche differentiation. Metabolic specialisation of a range of microorganisms in different putative niches might have contributed to the dominance of Anaerolineae in smooth mats, and shaped the assembly of distinct microbial communities 96 . Furthermore, pustular mats are located closer to the shore along a tidal transect compared to smooth mats 27 , where potentially higher desiccation, temperature, and salinity stress could occur. Thus as pustular mats are exposed to these extremes for longer periods, this may have contributed to the different geochemical gradients and hence microbial differentiation observed in the present study. In addition to some expected overlap, metabolic versatility may also allow microbes to be present in different mat layers, contributing to the heterogeneous nature of the mats. In contrast, metabolic specialisation may allow switching metabolic pathways in specific depths. For example, Microcoleus observed in the present study may switch from oxygenic phototrophy to anoxygenic phototrophy in deeper anoxic layers. Metabolic speciation and synergies may also be potentially moderated by quorum sensing in Shark Bay mats, as suggested in Bahamian microbialites 97 . Shark Bay mats may represent a unique ecological model containing a wide spectrum of autotroph-heterotroph interactions ranging from mutualistic nitrogen fixation niches and phototrophic niches, to opportunistic niches that exploit every energetic opportunity to maintain the microbial diversity observed."
} | 5,716 |
26841977 | PMC4594333 | pmc | 7,738 | {
"abstract": "Soybean is an important crop, with processed soybeans being the second largest source of vegetable oil and the largest source of animal protein feed in the world. Nodules on soybean roots are responsible for symbiotic nitrogen fixation, enabling soybean plants to obtain sufficient nitrogen for growth and seed production. Because nitrogen is an essential, but often limiting, nutrient for plant growth, improvements in nitrogen fixation are highly required in agriculture. We recently reported a comprehensive analysis of rhizosphere bacterial communities during soybean growth in a field in Kyoto prefecture, Japan. The bacterial communities of the rhizosphere changed significantly during growth, with potential plant growth-promoting rhizobacteria, including Bacillus , Bradyrhizobium , and Rhizobium , increasing in a stage-specific manner. In this addendum, we focus on changes in Bradyrhizobium during soybean growth, suggesting that soybean plants select for symbiotic partners."
} | 247 |
33915126 | PMC8163978 | pmc | 7,741 | {
"abstract": "Geobacter bacteria are able to transfer electrons to the exterior of the cell and reduce extracellular electron acceptors including toxic/radioactive metals and electrode surfaces, with potential applications in bioremediation or electricity harvesting. The triheme c -type cytochrome PpcA from Geobacter metallireducens plays a crucial role in bridging the electron transfer from the inner to the outer membrane, ensuring an effective extracellular electron transfer. This cytochrome shares 80% identity with PpcA from Geobacter sulfurreducens , but their redox properties are markedly different, thus determining the distinctive working redox potential ranges in the two bacteria. PpcA from G. metallireducens possesses two extra aromatic amino acids (Phe-6 and Trp-45) in its hydrophobic heme core, whereas PpcA from G. sulfurreducens has a leucine and a methionine in the equivalent positions. Given the different nature of these residues in the two cytochromes, we have hypothesized that the extra aromatic amino acids could be partially responsible for the observed functional differences. In this work, we have replaced Phe-6 and Trp-45 residues by their nonaromatic counterparts in PpcA from G. sulfurreducens . Using redox titrations followed by UV–visible and NMR spectroscopy we observed that residue Trp-45 shifted the redox potential range 33% toward that of PpcA from G. sulfurreducens , whereas Phe-6 produced a negligible effect. For the first time, it is shown that the inclusion of an aromatic residue at the heme core can modulate the working redox range in abundant periplasmic proteins, paving the way to engineer bacterial strains for optimal microbial bioelectrochemical applications.",
"conclusion": "Conclusions and implications Electrochemical measurements on naturally grown biofilms of G. sulfurreducens ’ cells on electrode surfaces showed an optimal electron transfer to the electrode at −0.15 V ( 25 ). This potential was correlated with the working redox potential range of the abundant periplasmic PpcA cytochrome ( 26 ), establishing this cytochrome as one of the main targets for the control of the working redox potential range in Geobacter cells. This is particularly useful as one of the means for improving Geobacter -based microbial bioelectrochemical applications in which a fine tuning of the redox properties of the electron transfer proteins can be explored. The markedly different functional properties of PpcA from G. metallireducens and PpcA from G. sulfurreducens are striking when considering that these proteins only differ in 13 amino acids, constituting a useful model for testing the impact of specific residues in their overall redox behavior. We investigated the role of Phe-6 and Trp-45 in the modulation of the functional properties of PpcA from G. metallireducens by replacing these amino acids by their counterparts in PpcA from G. sulfurreducens —Leu-6 and Met-45. We constructed two single mutants—PpcAF6L and PpcAW45M—and a double mutant—PpcAF6LW45M—and confirmed by NMR that the mutations did not introduce significant structural modifications. The impact of the mutations in the macroscopic redox behavior of the protein was probed by redox titrations followed by UV–visible spectroscopy, showing that only PpcAW45M produced a marked alteration in the redox profile of the protein to more negative heme redox potentials in comparison with the wildtype protein. A full thermodynamic characterization of the PpcAW45M mutant showed that Trp-45 is determinant in the modulation of the redox properties of heme III and, in less extent, heme I. This study provides a better understanding on the fundamental factors that modulate the redox properties of cytochromes with a pivotal role in Geobacter ’s electron transfer chain, setting the stage for the rational engineering of cytochromes with specific working redox potential windows. Consequently, the design of periplasmic proteins with enhanced electron transfer driving force from upstream or downstream partners will contribute to the creation of Geobacter cells with improved electron transfer capabilities. Overall, this work highlights the key role played by a specific aromatic residue in the modulation of the redox properties of highly abundant and homolog periplasmic cytochromes, offering new directions for the rational modulation and tuning of the electron transfer flow in Geobacter .",
"introduction": "The introduction of mutations maintains the global fold of the mutant proteins In the present work, the aromatic residues at position 6 (F6) and 45 (W45) in PpcA from G. metallireducens were replaced by the counterparts in PpcA from G. sulfurreducens (L6 and M45, respectively—see Fig. 2 A ). NMR spectroscopy is a sensitive technique to probe conformational changes caused by the replacement of residues in proteins. This is particularly notorious in the case of small low-spin triheme cytochromes, as it is the case of PpcA from G. metallireducens . In fact, the relatively small number of amino acid residues per heme group yielded well-resolved NMR spectra, which permitted us to evaluate the impact of the replacements in the cytochrome's heme core and polypeptide chain. For all mutants, the introduction of the mutation produced an alteration in the backbone signals of the amino acids located in the vicinity of the mutations ( Fig. 3 ). In the case of PpcAF6LW45M, the most affected backbone signals are a combination of the most affected ones of each of the two single mutants. Concerning the heme core, the good correlation obtained for the chemical shifts of heme protons in the mutant and wildtype cytochromes ( Fig. 4 ) indicates that the heme cores were unaffected by the mutations. In fact, the signals are only marginally affected in the three mutants. The rmsd value of PpcAF6LW45M (0.10 ppm) is slightly higher in comparison with the other values of the mutants since there was substitution of two aromatic rings and elimination of their ring-current effect contribution to the observed chemical shifts. In all mutants, the most affected signals correspond to the substituents facing the hydrophobic core of the protein, which have NOE connectivities with the mutated residues in the wildtype protein ( Fig. 4 , B and C ). It is worth noting that, for both single mutants, the differences between the wildtype and the mutants' chemical shifts are of the same magnitude (with exception of 3 2 CH 3 III substituent in W45M), which indicates that the mutations have a comparable impact in the heme substituent chemical shifts. In the double mutant, the differences between the wildtype and the mutant's chemical shifts are slightly higher because of the simultaneous removal of two aromatic amino acids.",
"discussion": "Discussion The introduction of mutations maintains the global fold of the mutant proteins In the present work, the aromatic residues at position 6 (F6) and 45 (W45) in PpcA from G. metallireducens were replaced by the counterparts in PpcA from G. sulfurreducens (L6 and M45, respectively—see Fig. 2 A ). NMR spectroscopy is a sensitive technique to probe conformational changes caused by the replacement of residues in proteins. This is particularly notorious in the case of small low-spin triheme cytochromes, as it is the case of PpcA from G. metallireducens . In fact, the relatively small number of amino acid residues per heme group yielded well-resolved NMR spectra, which permitted us to evaluate the impact of the replacements in the cytochrome's heme core and polypeptide chain. For all mutants, the introduction of the mutation produced an alteration in the backbone signals of the amino acids located in the vicinity of the mutations ( Fig. 3 ). In the case of PpcAF6LW45M, the most affected backbone signals are a combination of the most affected ones of each of the two single mutants. Concerning the heme core, the good correlation obtained for the chemical shifts of heme protons in the mutant and wildtype cytochromes ( Fig. 4 ) indicates that the heme cores were unaffected by the mutations. In fact, the signals are only marginally affected in the three mutants. The rmsd value of PpcAF6LW45M (0.10 ppm) is slightly higher in comparison with the other values of the mutants since there was substitution of two aromatic rings and elimination of their ring-current effect contribution to the observed chemical shifts. In all mutants, the most affected signals correspond to the substituents facing the hydrophobic core of the protein, which have NOE connectivities with the mutated residues in the wildtype protein ( Fig. 4 , B and C ). It is worth noting that, for both single mutants, the differences between the wildtype and the mutants' chemical shifts are of the same magnitude (with exception of 3 2 CH 3 III substituent in W45M), which indicates that the mutations have a comparable impact in the heme substituent chemical shifts. In the double mutant, the differences between the wildtype and the mutant's chemical shifts are slightly higher because of the simultaneous removal of two aromatic amino acids. Trp-45 is a key regulator of the redox properties of PpcA The initial screening of the impact of the mutations on the overall or macroscopic redox behavior of each mutant indicated that residues 6 and 45 have opposite effects in the modulation of the redox properties of PpcA from G. metallireducens and that, in the case of a double mutation, their effect is additive ( Table 1 and Fig. 5 ). In the case of PpcAF6L, the redox curve is similar to the one obtained for the wildtype cytochrome, though shifted to slightly higher redox potential values. On the contrary, PpcAW45M and PpcAF6LW45M mutants have their redox curves shifted to more negative values. In addition, the mutants' curves are steeper compared with the wildtype, particularly in the last third of oxidation, at higher redox potential values, a region that is essentially dominated by the oxidation of the heme with the highest redox potential value (see the arrow in Fig. 5 ). Therefore, from the analysis of the macroscopic redox curves, it can be predicted that the redox potential values of the hemes are more negative but less separated compared with the wildtype cytochrome. Since the more pronounced effects were observed for PpcAW45M, this mutant was selected to test these hypotheses. To attain this, a thermodynamic characterization at the microscopic level of the PpcAW45M mutant was pursued, including the determination of the reduction potential values of the individual hemes and their redox and redox–Bohr interactions. The thermodynamic parameters obtained for PpcAW45M are listed in Table 2 and show that in the fully reduced and protonated form, just like in the wildtype protein, the redox potentials of the hemes in PpcAW45M are all negative and different from each other. In addition, the redox interactions between each pair of hemes are positive, indicating that the oxidation of a particular heme stabilizes the reduced form of its neighbor. The strongest redox interactions are observed between the hemes that are closer in proximity: I–III and III–IV. However, and despite this apparent conservation of the redox properties in the mutant and wildtype cytochromes, the values clearly show that the redox potential of heme III, followed by that of heme I, are considerably more negative in the mutant, whereas that of heme IV is essentially unaltered. These results are in line with the spatial location of Trp-45 near hemes I and III. The redox–Bohr interactions between the hemes and the redox–Bohr center are negative, which means the oxidation of the hemes facilitates the deprotonation of the acid–base center and vice versa. The highest redox–Bohr interaction value is observed for heme IV, similarly to the wildtype protein, indicating that the mutation does not impact the redox–Bohr center and its interactions' network. This is also corroborated by the p K a values determined for the redox–Bohr center in the different oxidation stages ( Table 3 ). In fact, the p K a values cover a similar region, and the total redox–Bohr effect is essentially the same (1.5 and 1.6 pH units for PpcAW45M and wildtype protein, respectively). Effect of the mutation on the cytochrome's functional mechanism at physiological pH From the thermodynamic values indicated in Table 2 , it is now possible to evaluate the functional mechanism of PpcAW45M at physiological pH, including (i) the profile of the oxidation curve for each individual heme, which directly provides their redox potential values and order of oxidation as well as (ii) the relative contribution of each microstate during the oxidation ( Fig. 8 ). It is important to stress that, because the individual heme redox potential values are modulated both by redox interactions and redox–Bohr interactions, the potentials at pH 7 differ from those observed for the fully reduced and protonated protein (reported in Table 2 ). The data obtained showed that the redox potential values of the hemes cover a smaller range (−123 to −67 mV) compared with the wildtype cytochrome (−121 to −18 mV) ( Fig. 8 , upper panels ). A comparison of the microscopic reduction potential values ( e app ) of PpcAW45M and the wildtype protein shows that the reduction potential of heme IV is essentially unaltered (−123 versus −121 mV in the mutant and wildtype, respectively), whereas the reduction potential values of hemes I and III are decreased by 11 and 49 mV, respectively ( Fig. 8 , upper panel ). The narrow range observed for the mutant is then explained by the significant decrease of the redox potential value of heme III (−67 versus −18 mV in the wildtype). Despite these significant changes in the overall oxidation profiles, the order of oxidation of the hemes is conserved: IV–I–III ( Fig. 8 ). However, this does not imply that the functional mechanism of the protein is also conserved, since important changes were observed in the redox properties of hemes I and III. The functional mechanism of the mutant can be evaluated by determining the relative contribution of each of the 16 possible microstates along the redox cycle of the protein ( Fig. S1 ). A well-defined electron pathway is favored when one microstate clearly contributes (highest molar fraction) over that of another microstate within the same oxidation stage, thus favoring the directionality of electron transfer. In the wildtype cytochrome ( Fig. 8 ), due to the considerable separation of the heme redox potential values, the microstate with heme IV oxidized ( P 4H ) will dominate the first oxidation stage ( S 1 ), followed by the microstate with both hemes IV and I oxidized ( P 14 ) in the second oxidation stage ( S 2 ) and finally by the oxidation of heme III ( P 134 ). Thus, a well-defined preferential electron transfer pathway is established: P 0H → P 4H → P 14 → P 134 ( Fig. 8 , lower panel ). This pathway is no longer observed in PpcAW45M ( Fig. 8 , lower panel ). In fact, because of the decrease of the redox potential values of hemes III and I, the contribution of the microstates with these hemes oxidized in the oxidation stage 1 is higher compared with the wildtype protein. Consequently, the fractional contribution of the dominant microstate P 4H is much lower in the mutant and equals that of the microstate P 4 ( Fig. 8 , lower panel ). Thus, the preferred pathway observed for the wildtype protein is no longer observed in the mutated protein. Thus, the data obtained clearly indicate that the tryptophan residue at position 45 plays a crucial role in the regulation of the working redox potential range of PpcA and in the maintenance of a well-defined electron transfer pathway. Figure 8 Effect of the Trp-45 mutation in the microscopic properties of PpcA at pH 7. The upper panel shows the individual heme oxidation profiles for both the PpcAW45M mutant and wildtype cytochromes. The green , orange , and blue curves correspond, respectively, to hemes I, III, and IV. The curves were calculated as a function of the solution potential using the parameters indicated in Table 2 . The midpoint reduction potentials ( e app ) of the individual hemes are also indicated. The lower panel shows the molar fractions of the 16 individual microstates of PpcAW45M mutant and wildtype cytochromes. The curves were also calculated as a function of the solution potential using the parameters indicated in Table 2 . The protonated and deprotonated microstates are depicted in solid and dashed lines , respectively ( Fig. S1 ). To not overcrowd the figure, only the relevant microstates are labeled."
} | 4,164 |
39555911 | PMC11727244 | pmc | 7,742 | {
"abstract": "Abstract Mechanism‐based metamaterials, comprising rigid elements interconnected by flexible hinges, possess the potential to develop intelligent micromachines with programmable motility and morphology. However, the absence of efficient microactuators has constrained the ability to achieve multimodal locomotion and active shape‐morphing behaviors at the micro and nanoscale. In this study, inspiration from the flight mechanisms of tiny insects is drawn to develop a biomimetic microhinged actuator by integrating compliant mechanisms with soft hydrogel muscle. A Pseudo‐Rigid‐Body mechanical model is introduced to analyze structural deformation, demonstrating that this hydrogel‐based microactuator can undergo significant folding while maintaining high structural stiffness. Furthermore, multiple microhinged actuators are combined to facilitate folding in multiple degrees of freedom and arbitrary directions. Fabricated by a multi‐step four‐dimensional (4D) direct laser writing technique, the microhinged actuators are integrated into 2D and 3D metamaterials enabling programable shape morphing. Additionally, micro‐kirigami with photonic structures is demonstrated to show the pattern transforming actuated by the microhinges. This bioinspired design approach opens new avenues for the development of active mechanism‐based metamaterials capable of intricate shape‐morphing behaviors.",
"introduction": "1 Introduction Mechanism‐based metamaterials are a distinct class of shape‐morphing metamaterials found in complex mechanical systems. [ \n \n 1 \n , \n 2 \n , \n 3 \n \n ] These metamaterials can transform into various structural forms by employing rigid skeleton frameworks and flexible articulations, resembling systems such as the natural musculoskeletal system or reconfigurable tensegrity system. By programming the functional units within their systems, these rigid‐flexible coupled mechanical metamaterials can switch between shapes with unconventional physical properties, such as negative thermal expansion, [ \n \n 4 \n \n ] negative refractive index, [ \n \n 5 \n \n ] and invisibility cloaking. [ \n \n 6 \n \n ] Furthermore, the stimuli‐responsive properties of smart materials, such as hydrogels, [ \n \n 7 \n , \n 8 \n , \n 9 \n , \n 10 \n , \n 11 \n \n ] liquid crystal elastomers (LCEs), [ \n \n 12 \n , \n 13 \n , \n 14 \n , \n 15 \n \n ] and shape memory polymers (SMPs), [ \n \n 16 \n , \n 17 \n , \n 18 \n \n ] enable these metamaterials to adapt to environmental conditions, facilitating active shape transformations across multiple configurations. These active shape‐morphing metamaterials have attracted considerable attention for applications in flexible electronics, [ \n \n 19 \n , \n 20 \n , \n 21 \n \n ] biomedical devices, [ \n \n 22 \n \n ] autonomous robotics, [ \n \n 23 \n , \n 24 \n \n ] deployable antenna, [ \n \n 25 \n \n ] and optical microdevices. [ \n \n 26 \n \n ] \n The development of active mechanism‐based metamaterials necessitates the design of flexible microhinges within constrained spatial dimensions to efficiently manipulate rigid skeleton structures. [ \n \n 27 \n \n ] One of the main challenges is to produce flexible microhinged actuators that can exert substantial force and displacement outputs. Unlike macroscale machines, microactuators require high integration of sensing, actuation, and mechanical systems into compact bodies. Additionally, at the microscale, issues such as multiple component assembly, limited fabrication technologies, and increased surface forces, pose difficulties in microhinged actuator design. Typically, shape‐morphing behavior in metamaterials is achieved by precisely tuning the strain mismatch between multiple layers of films or beams. [ \n \n 28 \n , \n 29 \n , \n 30 \n , \n 31 \n \n ] However, this method poses limitations such as asynchronous responsiveness of different materials, mismatching in thickness and stiffness between layers, and failure in interface bonding, all of which constrain the achievable deformations. Four‐dimensional (4D) direct laser writing based on two‐photon polymerization has been developed to address these limitations. [ \n \n 32 \n \n ] Using this advanced microprinting technique with submicron precision, mechanical advantage is introduced to significantly enhance shape‐morphing capabilities by printing microscale passive hinges with active heterogeneous bi‐layered beams. However, increasing deformation by reducing the layer thickness produces relatively low hinge torque which hinders the morphological transformation of large‐scale metamaterials. [ \n \n 33 \n \n ] Additionally, this one‐step fabrication strategy also prints bi‐layered microactuators using a single photopolymerized material with nonlinear crosslinking gradients via an adjustable dose laser. The adjustable range of heterogeneity between different layers is narrow, and the stringent material selectivity imposes significant restrictions on potential applications. To break through the single material constraint, diverse materials need to be precisely programmed into rigid‐flexible coupled structures of active metamaterials. Drawing inspiration from nature, the musculoskeletal system exemplifies a rigid‐flexible coupled morphable structure that facilitates complex 3D motion. Even tiny flying insects, such as Drosophila, which measure as small as 400 µm, [ \n \n 34 \n \n ] have evolved efficient wing hinge mechanisms within their thorax. These mechanisms involve specialized thoracic skeletal structures and interconnected flight muscles that induce thoracic vibrations. [ \n \n 35 \n \n ] Amplified by the skeletal mechanisms, minute muscle contractions result in significant and powerful wing folding, allowing for remarkable stability and nimble maneuverability at high speeds of flight. This compact musculoskeletal design serves as a model for developing bioinspired microscale actuators for active mechanism‐based metamaterials. In this paper, we introduce a multi‐material microhinged actuator inspired by insect wing hinges. Employing the multi‐step direct laser writing, a compliant skeleton mechanism with high stiffness and soft hydrogel muscles with environmental responsiveness is integrated into the mechanism‐based metamaterial. Through structural analysis and parameter optimization, compliant mechanisms are designed to enable microhinged actuators with substantial folding deformation and high structural stiffness. These initial microhinges are then modularly assembled to create advanced microhinges with multi‐orientation folding and multi‐degree‐of‐freedom (DOF) folding deformations. Further, these microhinges are integrated into kirigami and network metamaterials. This deformation strategy enables the realization of micro‐kirigami with 2D in‐plane and out‐of‐plane morphing, as well as micro‐networks with 2D and 3D anisotropic shape transformations. Finally, to demonstrate the potential applicability of this strategy, photonic structures are programmed onto the micro‐kirigami to achieve active pattern transformation. The resulting shape‐morphing metamaterials, incorporating both soft and rigid components, offer significant potential for integrating functional materials and advancing active mechanism‐based metamaterials.",
"discussion": "3 Discussion This study presents microhinged actuators, inspired by insect wing hinges, designed to facilitate active shape‐morphing in mechanism‐based metamaterials. These hydrogel‐based microhinged actuators enable a wide range of adjustable folding deformations while maintaining high structural stiffness. Additionally, we have developed microhinges that support multi‐orientation and multi‐DOF folding deformations. Leveraging the versatility and efficiency of these microhinges, we fabricated programmable shape‐morphing 2D and 3D kirigami and network micro‐metamaterials. The resultant micro‐kirigami have been used as carriers of photonic structures to achieve environmentally responsive pattern transformations. These microscale mechanism‐based metamaterials exhibit significant shape‐morphing capabilities despite being constructed primarily from highly rigid materials. Our proposed actuation methodology introduces a new approach to multi‐material active shape‐morphing devices. The rigid parts, made of IP‐L 780 commercial photoresist, provide a platform for integrating functional components such as microcircuits and on‐board sensors. Metamaterials activated by microhinges can serve as responsive substrates for electrical and magnetic systems, with potential applications in wearable devices, wireless microrobotics, and tissue engineering. By incorporating different stimulus‐responsive materials within the microhinges, multi‐stimulus‐responsive metamaterials can be achieved. Furthermore, the independence of the microhinge deformations facilitates the individual control of each actuator within the metamaterial. This approach enables the realization of active shape‐morphing devices with multiple functions and wide‐ranging applicability."
} | 2,237 |
28450886 | PMC5402650 | pmc | 7,744 | {
"abstract": "Background Biofuels obtained from first-generation (1G) sugars-starch streams have been proven unsustainable as their constant consumption is not only significantly costly for commercial scale production systems, but it could potentially lead to problems associated with extortionate food items for human usage. In this regard, biofuels’ production in alkali-thermophilic environs from second-generation (2G) bio-waste would not only be markedly feasible, but these extreme conditions might be able to sustain aseptic fermentations without spending much for sterilization. Results Present investigation deals with the valuation of ethanologenic potential of locally isolated moderate alkali-thermophilic fermentative bacterium, Bacillus licheniformis KU886221 employing sugarcane cane bagasse (SCB) as substrate. A standard 2-factor central composite response surface design was used to estimate the optimized cellulolytic and hemicellulolytic enzymatic hydrolysis of SCB into maximum fermentable sugars. After elucidation of optimized levels of fermentation factors affecting ethanol fermentation using Taguchi OA L27 (3^13) experimental design, free cell batch culture was carried out in bench-scale stirred-tank bioreactor for ethanol fermentation. Succeeding fermentation modifications included subsequent substrate addition, immobilized cells fibrous-bed bioreactor (FBB) incorporation to the basic setup, and performance of in situ gas stripping for attaining improved ethanol yield. Highest ethanol yield of 1.1406 mol ethanol/mol of equivalent sugars consumed was obtained when gas stripping was performed during fed-batch fermentation involving FBB under aseptic conditions. Despite the fact that under non-aseptic conditions, 30.5% lesser ethanol was formed, still, reduced yield might be considered influential as it saved the cost of sterilization for ethanol production. Conclusion Effectual utilization of low-priced abundantly available lignocellulosic waste sugarcane bagasse under non-aseptic moderate alkali-thermophilic fermentation conditions as directed in this study has appeared very promising for large-scale cost-effective bioethanol generation processes. Electronic supplementary material The online version of this article (doi:10.1186/s13068-017-0785-1) contains supplementary material, which is available to authorized users.",
"conclusion": "Conclusion Purpose of the present investigation was to exploit the locally available lignocellulosic waste for the production of clean and environment friendly energy fuel bioethanol employing extremophiles. In this regard, the bioethanol production by alkali-thermophilic fermentative microbes from bio-waste such as sugarcane bagasse would not only be sustainable, but the extreme conditions will employ a bioprocess for high selection pressure for reducing the likelihood of microbial contaminations. Thus, it will be possible to maintain aseptic environment without spending much for sterilization. Successful exploitation of low-cost substrate for biofuels’ production under moderate alkali-thermophilic conditions appeared promising for the development of large-scale bio-fermentation processes. For this purpose, it is extremely important to explore the locally available fermentative ethanologenic thermophiles and understand their metabolic requirements while utilizing lignocellulosic biomass. It is foreseeable that understanding of non-aseptic extremophilic fermentations utilizing agro-industrial waste as necessitated in the present investigation for development of large-scale cost-effective, eco-friendly biofuel generation processes will pave the way to achieve one of the greatest benefits of mankind.",
"discussion": "Discussion Principal aim of this study was to extract the energy from renewable waste sugarcane bagasse under non-aseptic extreme conditions in the form of bioethanol. For this purpose, alkali-thermophilic ethanologens were isolated from the soil sampled from the vicinity of hot water effluent near Balkasar oil refinery, Chakwal, Pakistan. The mud temperature was 50–55 °C with pH ranging 9–10 at the sampling time. Such habitats are favorable conditions for isolation of alkali-thermophiles [ 26 ]. Further valuation of cellulolytic/ethanologenic potential of the isolates was done by replacing glucose with sugarcane bagasse as substrate in the fermentation medium. On the basis of higher ethanol production, isolate ML-07 was selected, pure-cultured, and identified through 16S rRNA gene sequencing, and was recognized as a cellulolytic and mild alkali-thermophilic strain of B. licheniformis. Consequently, it was allotted the accession no. KU886221. It was previously reported that some strains of B. licheniformis were found to produce ethanol in fermentation medium [ 27 ]. As mentioned before, SCB was selected as substrate for ethanol fermentation. Composition of sugarcane bagasse was 31.168% cellulose, 21.048% xylan, and 23.633% lignin. Many thermophilic bacteria capable of utilizing hexoses/pentoses from cellulose/hemicellulose, respectively, directly/indirectly are considered promising for ethanol fermentation [ 16 , 28 – 30 ]. For that reason, it can be speculated that both cellulose and hemicellulose are valuable carbon substrates for thermophilic ethanol fermentation. Cellulose is generally recalcitrant to hydrolysis; therefore, ethanol production from lignocellulosic biomass requires separation of its fractions through pretreatment methods [ 31 ]. The chemicals used in pretreatment expose the crystalline cellulose and enhance its porosity by breaking down lignin protection around it, which, in turn, facilitates polysaccharide hydrolysis [ 32 , 33 ]. For this purpose, optimum pretreatment concentrations of H 2 SO 4 , H 3 PO 4 , HCl, and NaOH for SCB hydrolysis were explored. The outcomes indicated that the alkaline pretreatment with 1 N sodium hydroxide of SCB significantly reduced lignin contents and produced cumulatively greater sugar contents (glucose and xylose) after enzymatic hydrolysis as presented in Additional file 2 . This is analogous to the findings of previously reported research indicating improved performance in sugar yield after enzymatic hydrolysis of alkaline pretreated SCB, delignification, and biomass utilization, compared to acid pretreatment [ 33 , 34 ]. When utilization potential of differently pretreated SCB for ethanol production was evaluated, it appeared that ethanol production from B. licheniformis KU886221 was also greater for alkaline pretreated SCB possibly due to less inhibitory compounds’ production compared to acid pretreatments that produce more inhibitory compounds for subsequent fermentation processes as reported previously [ 35 ]. This is parallel to the previous finding (see Additional file 2 ) presenting that microbe Saccharomyces cerevisiae was able to yield significantly greater ethanol (0.39 g/g sugar) employing timothy grass hydrolysate containing 57.4 g/L sugar compared to the ethanol yield (0.36 g/g sugar) employing pine wood hydrolysate containing 68.5 g/L sugar and wheat straw hydrolysate’s ethanol yield (0.35 g/g hexose) from 63.6 g/L sugar [ 36 ]. In subsequent experiment, ethanol production was evaluated employing pretreated SCB following enzymatic hydrolysis (SCBH). It was found that ethanol production increased 32.89% for SCBH. Furthermore, when the bacterium was analyzed to ferment glucose, xylose, and cellulose separately in fermentation medium FM4, it was revealed that after 72 h of incubation, glucose was the preferred sugar substrate compared to cellulose. It also consumed significant amounts of xylose and mannose. Approximately 60% reduced ethanol was produced using cellulose directly as substrate in comparison with glucose. Thence, alkaline pretreated SCB hydrolysate was selected as the carbon source in fermentation medium for further investigation. RSM is an assemblage of statistical performances for scheming experimentations, constructing models, valuing the special effects of experimental factors, and scrutinizing for the optimum conditions facilitating the whole phenomenon. Enzymatic hydrolysis was performed by a combination of two enzymes, i.e., cellulase (Cellic Ctec-2 Novozyme) and hemicellulase (Htec Novozymes) for breaking down biomass to maximum amounts of fermentable sugars. The effect of Ctec and Htec enzymes’ quantities on consequent alkaline pretreated SCB hydrolysis for attaining simple fermentable sugars was studied by central composite design. A quadratic (second-order) polynomial equation was proposed including all interaction terms relating independent and response variables: \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{Y }} = \\, \\beta_{0} + \\beta_{1} {{X}}_{1}^{2} + \\beta_{2} {{X}}_{2}^{2} + \\beta_{3} {{X}}_{1} + \\beta_{4} {{X}}_{2} + {{X}}_{1} {{AX}}_{2} + \\varepsilon.$$\\end{document} Y = β 0 + β 1 X 1 2 + β 2 X 2 2 + β 3 X 1 + β 4 X 2 + X 1 A X 2 + ε . \n The results of the second-order response surface models for the sugar hydrolysis in the form of regression and analysis of variance (ANOVA) were studied under 95% of confidence interval ( P ≤ 0.05) and pure error. Desirability profiling with the help of Statistica software enabled to predict enzymes’ optimum quantities. After validation of the experiment, enzymes optimum dose was determined as 1.024 g Cellic Ctec Novozymes + 0.468 g Htec Novozymes/100 g of alkaline pretreated SCB that was found to be compliant to the previously investigated [ 37 ]. The purpose of the present investigation was to exploit the low-cost substrate for the production of clean and environmentally friendly biofuels. In this regard, corn steep liquor (CSL), a chief by-product of corn starch processing, was used as low-cost nitrogen source. It contains significant amount of proteins, amino acids, vitamins, minerals, and trace elements, and can suitably replace yeast extract and peptone in alcoholic fermentations [ 38 – 42 ]. Specifically, CSL has been reported effective for ethanol fermentations previously [ 39 , 40 , 43 – 45 ]. In advance experiment, potential of CSL as complex nitrogen source for ethanol production substituting peptone and yeast extract in the fermentation medium was evaluated. It was found that 3.24% more ethanol was produced when CSL was used in comparison with peptone + yeast extract in the fermentation medium. This increased ethanol production was possibly due to the complex nutritious nature of CSL [ 38 , 40 ]. Concentration of alkaline pretreated SCB hydrolysate, CSL and NaCl, incubation temperature, pH, and fermentation period were optimized for ethanol fermentation using B. licheniformis KU886221. Experimentation assisted by Taguchi OA experimental design using the Design Expert 8 software enabled us to infer that changing of incubation temperature (optimum: 45 °C) and alkaline pH (optimum: 9) are most significant factors affecting ethanol fermentation. Furthermore, a batch culture was carried out under optimized conditions of ethanol fermentation in bench-scale stirred-tank bioreactor. Ethanol yield of 0.909 mol of ethanol/mol of sugars consumed was obtained after 84 h of incubation, corresponding to 0.279 g ethanol/g of sugars consumed. Though reduced ethanol yield in comparison with the yield by mesophilic yeast, S. cerevisiae (as shown in Additional file 2 ) was obtained, but the use of thermophilic fermentative bacteria is considered more advantageous where LCB is used as fuels’ feed as unlike most mesophilic yeasts, the thermophilic fermentative bacteria are able to ferment hexoses as well as pentoses to ethanol [ 36 , 46 ]. The percent reduction of substrate was 98.5 showing significant conversion of substrate into product with ethanol titer of 11.301 g/L. In terms of biomass, 0.114 g ethanol/g of alkaline pretreated SCB was yielded. It is noticeable that the slope for ethanol production regarding sugar consumption was nearly constant, signifying that this integrated fermentation was very steady and reliable with ethanol as the primarily chief product. Next, the fed-batch fermentation was performed to understand the influence of substrate addition on batch fermentation. The overall ethanol titer was 16.896 g/L associated with 78.94% substrate consumption. In terms of biomass, 0.123 g ethanol/g of alkaline pretreated SCB was obtained at the end of fed-batch fermentation experiment. The molar ethanol yield was 0.973 mol ethanol/mol sugars consumed. This lower substrate consumption was possibly due to accumulation of inhibitory fermentation metabolites including ethanol. To lessen the toxic effect of higher density of cells on their further growth/fermentative potential, further adaptation in fed-batch bioreactor was applied by attaching fibrous-bed bioreactor (FBB) to the fermentor [ 22 , 47 ]. It is also very innovative fermentation approach to improve solvent tolerance facilitating ethanol fermentation. Increased ethanol production with 85.031% substrate utilization was observed with 1.076 mol ethanol/mol of sugars consumed. In terms of biomass, 0.131 g ethanol/g of alkaline pretreated SCB was achieved at the end of fermentation. This revealed that immobilized-cell fermentation might be capable of enhancing ethanol production by trapping the cells in fibrous-bed, thus facilitating increased cell tolerance to toxic fermentation metabolites and increase cell viability inside main bioreactor. Ethanol titer was significantly increased from 16.896 to 19.39 g/L with the attachment of FBB. To investigate the ethanol-induced inhibition, further study was done by partially removing ethanol by in situ gas stripping from the fermentation medium. Gas stripping is an established product recovery technique for ABE separation [ 23 , 48 – 51 ]. It was confirmed that ethanol intolerance was one of the reasons of growth inhibition. Ethanol titer of 21.637 g/L (1.1406 mol ethanol/mol of sugars consumed) was observed with 94.295% substrate consumption. Results are compliant with the previous findings that removal of inhibitory compounds leads to a better performance in sugar conversion to products [ 48 , 52 ]. Thus, gas stripping is proved to be an effective separation technique used to apprehend synchronized ethanol recovery and overcome product repression due to ethanol-induced inhibition [ 52 , 53 ]. In terms of biomass, 0.135 g ethanol/g of alkaline pretreated SCB was obtained at the end of fed-batch fermentation involving FBB and gas stripping. Finally, fed-batch fermentation involving FBB was studied under non-aseptic conditions. The results indicated that 13.466 g/L ethanol corresponding 0.570 mol ethanol/mol of sugars consumed was produced at the end of fermentation (at 120 h). In terms of biomass, 0.089 g ethanol/g of alkaline pretreated SCB was obtained. Comparing aseptic conditions, 30.5% reduced ethanol yield was obtained under non-aseptic conditions showing the likelihood of some contaminant in the fermentation culture. However, cost of sterilization is one of the major obstacles for developing lignocellulosic bio-refineries. Therefore, even a reduced product yield might be considered appealing while saving the cost of sterilization. In the present study, besides ABE, significant amounts of acetic acid and some butyric acid were also produced during fermentation. ABE were also formed in the condensate after gas stripping. Realizing separation of ABE along with acetic acid and butyric acid would make the fermentation methodology more efficient as these by-products are very profitable."
} | 3,941 |
36080594 | PMC9460880 | pmc | 7,745 | {
"abstract": "Self-healing materials, especially self-healing polyurea/polyurethane, to replace traditional coating has been of increasing interest in the past decade. The frequency of regular maintenance work can also be reduced as the coating is capable of forming bonds at ruptured sites. This reduces the cost of maintenance and the risk involved in workers engaging in maintenance work. The extremely short curing time of polyurea coating could potentially outweigh the cost due to its short down time. With a high self-healing efficiency, self-healing polyurea could be the ultimate choice of protective coating. This report aims to find the optimum formulation for fabrication of polyurea with a high self-healing efficiency. This is conducted by changing the composition of the components chosen for formulation of polyurea. The choice of isocyanate and amine is varied to explore its impact on chain mobility and microphase separation, which are important factors affecting self-healing efficiency. A series of characterizations, including ATR-FTIR, DSC, optical microscope and mechanical tester, is used to analyze the factors affecting the self-healing efficiency of fabricated polyurea and to eventually determine the best formulation. The ideal formulation of toluene 2,4 diisocyanate-amine (TDI-P1000) polyurea managed to achieve a self-healing of 42%. Further studies could be done to include multiple healing mechanisms after different area of polyurea to boost its self-healing efficiency after repeated healing.",
"conclusion": "4. Conclusions and Position Self-healing capability and efficiency is highly dependent on the components included during fabrication of polyurea. Important factors that affect self-healing of polyurea include chain mobility, microphase separation, and time before surfaces are in contact. However, there is a trade-off between achieving good chain mobility and adequate microphase separation (phase-mixed). From this experiment, it can be concluded that polyurea with self-healing capability has to be fabricated with isocyanate (hard segment) and long chain amine (soft segment) to achieve the factors mentioned. The choice of isocyanate has more impact on chain mobility than the choice of amine. On the other hand, the choice of amine has more impact on the phase separation than the choice of isocyanate. To achieve the best of both, TDI-P1000 has been tested to portray the best self-healing capability and efficiency of up to 42% at room temperature and elevated temperature even after repetitive healing cycles. From the recent available research, self-healing polyurea has mostly been fabricated with D2000. From this experiment, it has been observed that P1000 has a lesser reactivity with air, and it could be a better choice for more commercialized self-healing polyurea.",
"introduction": "1. Introduction In the midst of the rising global cost of corrosion, solutions such as application of a protective coating and employing regular maintenance work have been put in place. All of these efforts have been made primarily for the lifetime extension of what is being protected [ 1 ]. To further reduce need for regular maintenance, the area of self-healing materials is explored. Polyurethane has always been the more popular coating among other polymers [ 2 ] because of its lower cost of raw materials, easy fabrication, and fast curing time [ 3 , 4 ]. Although polyurea is known to cost more, it has an even shorter curing time [ 5 ]. The extremely short curing time of polyurea coating could potentially outweigh the cost due to its short down time [ 6 ]. With a high self-healing efficiency, self-healing polyurea could be the ultimate choice of protective coating [ 7 ]. In this series study, a protective layer of polyurea coating is proposed, which can be prepared quickly and used for a long time with minimum downtime [ 8 ]. Quick preparation can ensure fast coating of substrate, while minimum downtime is to maintain constant protection of substrate. Besides that, if microcracks are not healed in time, the healing components on the fracture surface may undergo unwanted side reactions, such as reduction of disulfide bonds or saturation of hydrogen bonds [ 8 ]. These side reactions hinder further self-healing processes. Hence, it is important to accelerate the self-healing process and enhance self-healing efficiency. Therefore, the ability of polyurea to self-heal within a short period of time with an acceptable recovery of mechanical strength is important to reduce the extent of corrosion on substrate [ 9 ]. This study aims to find the optimum formulation for polyurea to self-heal in a fixed duration of time, even after multiple cuts at the same location regarding its performance in the aspects healing efficiency of strength. This is conducted by exploring a combination of components. The choice of isocyanate is varied to explore how chain mobility affects the self-healing efficiency of polyurea [ 10 ]. The choice of amine (in hard segment or soft segment) is varied to explore how microphase separation can affect the self-healing efficiency of polyurea [ 10 , 11 ]. The samples fabricated will undergo a series of characterizations before being cut, which includes ATR-FTIR, DSC, optical analysis, and tensile test. After being cut, the samples will undergo another round of tensile testing to determine self-healing efficiency. The sample with the highest self-healing efficiency after multiple cuts will be deemed the optimum formulation.",
"discussion": "3. Result and Discussion 3.1. Fabrication Features and Results Fabrication of samples labelled from Sample 3 to Sample 6 were successful. Solid samples of polyurea were obtained. However, Samples 1 to 2 did not form a solid polyurea. Prior to this method of fabrication, two other methods have also been tested to produce a sample based on the initial plan ( Figure S1a ). The main difference in the fabrication methods is the sequence of components being added. The initial method was to add D2000 dropwise into a respective amount of diisocyanate to form a prepolymer. Then, the chain extender was added to the prepolymer to form polyurea ( Figure S2 ). However, the prepolymer solidified immediately. Addition of a chain extender resulted in a layer of liquid above the initial solid prepolymer. This fabrication failure could be due to an immediate reaction between diamine and isocyanate. The fast reaction means that the chain extender could not be added into the backbone of the polyurea chain [ 14 ]. This could be the reason why the red solution of the chain extender remained as a liquid above the solid prepolymer. An improved method is planned to include the chain extender into the backbone of polyurea before solidification. D2000 is added to the chain extender and mixed well. This mixture is then added dropwise to a respective amount of diisocyanate. A liquid mixture is obtained. The mixture is then placed in an oven at 80 °C to remove residual solvent. However, the mixture changed to a dark red color and a solid formed in the mixture ( Figure S1b ). This method of fabrication could not guarantee that the chain would turn out similar to that described in Figure S1b as the chain extender and D2000 are both diamines. Both components could react with diisocyanate and lead to formation of the extender-diisocyanate-chain extender polymer chain or D2000-diisocyanate-D2000 [ 15 , 16 ]. The dark red color could be due to oxidation of the mixture at an elevated temperature. The solid formed could be the reaction between the diisocyanate with either D2000 or a chain extender to establish urea bond. At this point, it seems impossible to fabricate a polyurea with all three components: diisocyanate, D2000, and chain extender. Hence, the samples are fabricated without chain extenders [ 17 ]. A conclusion that can be drawn from all of the fabrication methods is that D2000 may have reacted with air, leading to it not reacting with polyurea. This can be further supported by sample MDI-D2000-PPD from the first method, a layer of solid uneven solid is obtained when the liquid is removed from the mold. This could possibly be due to D2000 reacting with moisture in the air. In the region underneath, the D2000 has yet to react with air and hence is able to react with diisocyanate. On the other hand, the other long chain diamine, P1000, does not react with air as rapidly and allows reaction with diisocyanate. However, it also cures quickly, which prevents the amendment of the polyurea backbone. In order to fabricate polyurea using D2000, a nitrogen atmosphere is needed. Hence, the formulation is changed to those illustrated in Table 1 . 3.2. ATR-FTIR Analysis FTIR is capable of providing both qualitative and quantitative analysis of the samples. In terms of a qualitative analysis, FTIR analysis will show the presence of certain functional groups represented by respective wavenumbers. In terms of a quantitative analysis, the concentration of the functional groups can be determined by the peak areas. FTIR analysis has been used in the characterization of polyurea samples to verify that the fabrication has successfully produced polyurea and also to find the concentration of different types of hydrogen bonds present in the samples. To verify that the samples are all polyurea, the peak representing the isocyanate function group at 2250 cm − 1 should not be present as the urea bond is formed during the reaction of isocyanate and amine functional groups [ 18 ]. Besides that, there should also be peaks at approximately 1620–1720 cm − 1 due to the presence of the stretching vibration of C=O in the urea linkage. There should also be peaks present at approximately 3320–3450 cm − 1 due to the presence of the stretching vibration of N-H in the urea linkage [ 19 ]. However, the absorption peak of N-H groups is in the same range as the overtone of carbonyl groups [ 20 ]. The signals for N-H groups are also weak, hence it is not conducive for quantitative analysis to be conducted. Figure 1 shows the FTIR spectra for the samples. All of the samples have been successfully fabricated to consist of urea linkages. The purpose of finding the concentration of different types of hydrogen bonds present in the samples is to analyze the packing of the hard segment [ 21 ]. The packing of the hard segment plays an important role in self-healing efficiency. With the use of ATR-FTIR analysis, the extent of hard segmental packing can be determined and the formulation’s impact on microphase separation within polyurea can be examined [ 22 ]. The peaks at different wavenumbers from 1620 cm − 1 –1760 cm − 1 represent different types of hydrogen bonds. A compilation of the wavenumber and representative stretching of the carbonyl functional group is shown in Table 2 [ 23 ]. The presence of “ordered” C=O is representative of a regularly arranged hard segment. A denser packing of the hard segment also means a higher amount of “ordered” C=O is present in the sample [ 24 ]. On the other hand, the presence of “disordered” and “free” C=O is representative of an irregularly arranged hard segment [ 25 ]. The looser the packing of the hard segment, the higher the amount of “free” and “disordered” C=O. To find the amount of each type of bond, OriginPro software was used to apply a Gaussian function to find the peak area under the graph. Followed by that, a simple law of proportion can be applied to calculate the proportion of each bond and to analyze the overall packing of the hard segment within the function [ 26 ].\n (1) X f = A f A f + A d + A o \n (2) X d = A d A f + A d + A o \n (3) X o = A o A f + A d + A o X f , X d , X o are proportion carbonyl groups with of no hydrogen bond, monodentate hydrogen bonds and bidentate hydrogen bond respectively. A f , A d , A o are peak areas on the graphs representing carbonyl groups with no hydrogen bond, monodentate hydrogen bonds, and bidentate hydrogen bond, respectively [ 26 , 27 ]. Neither MDI-DDS nor MDI-PPD is considered as an ideal formulation group due to its poor transparency and fabrication defects, as indicated in Figure S2 . The same FITR characterization has been performed, however, the low absorbance of MDI-DDS and MDI-PPD in the range of 1620 to 1760 cm −1 indicates only a low percentage of hydrogen bonding are present in these samples [ 27 ]. Self-healing capability in MDI-DDS and MDI-PPD is very limited and hence excluded from the following quantitative analysis. Figure 2 shows the application of a Gaussian function onto the graph using OriginPro software for MDI-P1000 and TDI-P1000 [ 28 ]. By using Equations (1)–(3), the proportion of carbonyl groups with no hydrogen bond, monodentate hydrogen bonds, and bidentate hydrogen bond are compiled in Table 3 . The peak area for no hydrogen and monodentate hydrogen bonds cannot be defined distinctively, hence it is taken as one peak. Similarly, the proportion of carbonyl groups for TDI-P1000 with no hydrogen bond, monodentate hydrogen bonds, and bidentate hydrogen bond are also compiled in the right column. Comparison of different types of hydrogen bonds are represented in a bar graph ( Figure 3 ). Comparing the composition of different types of hydrogen bonds in MDI-P1000 and TDI-P1000 ( Figure 3 ), MDI-P1000 has a higher proportion of bidentate hydrogen bonds than TDI-P1000. Having a higher proportion of bidentate hydrogen bonds means that there is a hard segment of MDI-P1000 that is more tightly packed. This means that the microphase separation between the hard segment and the soft segment is more distinctive as there is a smaller extent of penetration of the soft segment into the hard segment [ 22 , 29 ]. As a result, MDI-P1000 is more phase-separated [ 30 ]. On the other hand, TDI-P1000 has a higher proportion of monodentate hydrogen bonds and no hydrogen bond. The hard segment of TDI-P1000 is more loosely packed, which allows for a larger extent of the soft segment to penetrate into it [ 20 , 31 ]. TDI-P1000 is more phase-mixed than phase-separated. The presence of phase-mixed morphology in TDI-P1000 is further verified by a higher absorbance detected at peaks representing hydrogen bonds in TDI-P1000 than MDI-P1000, signifying more dynamic interactions (H-bonding). From FTIR analysis, it is predicted that TDI-P1000 will show better self-healing capability and efficiency than MDI-P1000 [ 32 ]. 3.3. DSC Analysis Chain mobility is one of the factors that affects the self-healing efficiency of polyurea. Upon reaching the glass transition temperature, the polyurea change from a solid to a rubbery state [ 33 , 34 ]. The soft segment of polyurea gain a higher chain mobility and hence a higher capability to diffuse to the damaged location, forming new bonds with the broken bonds for self-healing to occur. In order to analyze the chain mobility of the soft segment of polyurea at a different temperature, DSC analysis has been conducted to determine the samples’ glass transition temperatures [ 35 ]. Glass transition is an endothermic process that does not occur at one temperature but over a range of temperatures. The temperature in the middle of the region is taken as the glass transition temperature [ 36 ]. Since the DSC graph was plotted in an “exo up” setting, the glass transition temperature is expected to be represented in the middle of a decline in heat flow, as shown in Figure S3 . From the DSC result of MDI-P1000 ( Figure 4 ), it can be seen that the glass transition temperature is not very distinctive. Upon further rescaling of the graph, the glass transition temperature of MDI-P1000 can be determined as 51.56 °C. The reason for the sample having a non-distinctive wide range of glass transition temperatures could be due to the arrangement of soft and hard segments in the polyurea [ 37 ]. According to Lefebvre et al. [ 38 ], a broad glass transition temperature is expected for gradient copolymers ( Figure S4 ), where its homopolymers have very different glass transition temperatures. A gradient copolymer is defined as a lamellar with a composition that is smooth, varying such that the domains are never made of purely one homopolymer [ 39 ]. In the case of MDI-P1000, the soft segment is comprised of P1000, which has molecular weight of 1000 g/mol. The hard segment is MDI, which has a molecular weight that is approximately five times lower than the soft segment. The significant difference in molecular weight could be representative of a distinct difference in glass transition temperatures between the homopolymers (hard segment and soft segment) [ 40 , 41 ]. From the DSC result of TDI-P1000 ( Figure 4 b), it can be seen that the glass transition temperature is also not distinctive. Similar to MDI-P1000, a molecular weight of TDI is also much lower than that of P1000. Hence, the sample could also be a gradient copolymer. Upon further zooming in at the locations selected, there are two transition temperatures found in the result [ 42 ]. One glass transition temperature could be determined as −36.57 °C and the other as 64.66 °C. The reason for the appearance of two glass transition temperatures could be due to the presence of two separated phases in the sample. The fabricated polyurea should be a copolymer but not a homopolymer. However, due to the extremely short curing time of polyurea, the hard and soft segments in the sample could not mix well and were unable to form a homogenous solution (copolymer) [ 43 ]. Hence, this leads to the observation of two glass transition temperatures, where both copolymer and homopolymer are in the sample. Since polyurea that shows capability of self-healing at room temperature have a sub-zero glass transition temperature, the glass transition temperature of copolymer of TDI-P1000 will be taken to −36.75 °C [ 44 ]. This temperature will be further verified at a later part of the report, when self-healing capability is tested [ 45 ]. From the DSC result of MDI-DDS in Figure 4 c, the glass transition temperature can be determined as 54.66 °C, whereas for the DSC result of MDI-PPD, the glass transition temperature can be determined as 54.71 °C ( Figure 4 d) [ 46 ]. From the analysis of DSC results, it can be concluded that the glass transition temperature is more dependent on the diisocyanate (MDI/TDI) used than the amine (P1000/DDS/PPD). All the samples with MDI have glass transition temperatures in the 50 °C range, regardless of the choice of amine. On the other hand, TDI-P1000 has a lower glass transition temperature than samples with MDI. This could mean that, in this case, the chain mobility is more affected by the bulkiness of structure than the molecular weight [ 47 ]. MDI-P1000 has a higher glass transition temperature than TDI-P1000. This means that the soft segment in MDI-P1000 has a lower chain mobility than the soft segment in TDI-P1000. In another word, a soft segment in MDI-P1000 is more hindered than that of TDI-P1000 [ 48 ]. This corroborates with the bulkiness of MDI and TDI structures. Since MDI has a bulkier structure than TDI, the mobility of P1000 in MDI-P1000 is hindered to a greater extent [ 49 ]. Hence, a higher energy is needed to allow the polymer chain to move and MDI-P1000 has a higher glass transition temperature. With the information of the glass transition temperature of the samples, if self-healing can take place in these polyurea, it is expected that TDI-P1000 can self-heal at room temperature, while MDI-P1000, MDI-DDS and MDI-PPD will need heating before any potential self-healing can occur [ 50 ]. 3.4. Optical Analysis The time before surfaces are in contact is one of the factors that affects the self-healing efficiency of polyurea. Once the polyurea has been cut, there has to be contact between two interfaces for self-healing to occur. If no contact is made, there is no chance of self-healing taking place [ 51 ]. The waiting time also plays a role in ensuring the occurrence of self-healing instead of self-adhesion. An optical microscope was used to get a magnified view of the cut made on the samples by the razor and to check the self-healing capability of the sample [ 52 ]. Self-healing capability of the sample is defined as the ability of the sample for self-healing to take place. With the use of microscope, only the capability and speed of self-healing will be analyzed [ 53 ]. The efficiency of self-healing in terms of retention of mechanical property will be analyzed in another characterization. For each sample, 3 cm × 1.2 cm rectangular pieces were cut out. On the rectangular pieces, two different types of cuts were made: scratch and clean cut. MDI-DDS and MDI-PPD were too brittle to be scratched or cut by razor. Upon exertion of force, the samples broke into multiple pieces. Hence, MDI-DDS and MDI-PPD were exempted from the scratch test [ 54 ]. 3.4.1. Optical Analysis for Scratching A single scratch was made on the samples. The samples were then left at room temperature. The samples were also checked with the naked eye for visible changes at 5 min, 1 h, 3 h, and 6 h. Within the first 6 h, no observable change was seen with naked eyes for all samples. The samples were then placed in an oven at 80 °C. The samples were also checked with the naked eye for visible changes at 5 min, 1 h, 3 h, and 6 h. No observable change was seen with the naked eye for all samples. The samples were then sent under the microscope. From the images obtained from the microscope ( Figure 5 ), the razor cut had a width of approximately 50 μm. In both samples, no healing took place. Upon fine tuning, both scratches were still observable on the surfaces of the samples. Hence, MDI-P1000 and TDI-P1000 showed no self-healing capability for scratching. 3.4.2. Optical Analysis for Cutting The second type of optical analysis is on a rectangular shaped sample cut, which is cut cleanly into two. For this cut, three sets of broken pieces were prepared. The first set (self-healing after 10 min) was placed in contact with each other only after 10 min. For the second (immediate self-healing at room temperature) and the third sets (immediate self-healing at 80 °C), the broken pieces were immediately placed in contact with each other but in different temperature environments. Figure 6 indicates the self-healing features after 6 h of different samples. The difference in curvature does not reflect the degree of softness of the materials, whereas it serves as a demonstration of the cross-section recovery performance of the self-healing progress [ 55 , 56 ]. Self-healing after 10 min After cutting the samples into two, the broken pieces were left separately at room temperature for 10 min. After 10 min, the broken pieces were forced together and left at room temperature. Due to MDI-DDS and MDI-PPD being too brittle, it was not possible to properly cut the samples into two. Unevenly shaped and broken pieces of MDI-DDS and MDI-PPD were instead tested [ 57 , 58 ]. Broken pieces of MDI-P1000, MDI-DDS, and MDI-PPD were unable to be mended after being left separately for 10 min. The broken pieces of MDI-P1000, MDI-DDS and MDI-PPD remained as separate pieces even after 6 h [ 59 ]. On the other hand, TDI-P1000’s broken pieces were able to be attached to each other and be lifted up without falling apart. This means that TDI-P1000 showed self-healing capability [ 60 ]. From the results of this set of optical observation, it seems the self-healing results were not coherent with the self-healing capability of samples seen in scratching test. TDI-P1000 showed no self-healing capability in the scratching, but it was shown in such cutting experiments. The reason could be due to the gap between the interfaces. The scratch made in scratching left a 50 μm gap between the two interfaces. However, the broken pieces were forced together in cutting which leaves almost no gap between the interfaces [ 61 ]. This shows that the 50 micrometer is beyond the diffusion limit of the samples at room temperature and at 80 °C. MDI-P1000 showed self-healing capability, while MDI-DDS and MDI-PPD do not indicates that the presence of long chain diamine is critical in achieving self-healing ability. The presence of long chain diamine allows for the chain to be more flexible than the sample with just a short chain diamine [ 62 ]. This explains the brittleness of MDI-DDS and MDI-PPD. The long chain amine could have more donors and acceptors available to form hydrogen bonding, which heals the sample. Immediate self-healing at room temperature For immediate self-healing at room temperature, the broken pieces were immediately forced together and left at room temperature. At 5 min, only TDI-P1000′s broken pieces were connected. After 6 h, the broken pieces of MDI-P1000 and TDI-P1000 managed to form a bridge across the interfaces and stay connected ( Figure 6 a,b). Nevertheless, the cut at the interfaces were still present for both samples [ 63 ]. The gap between the interfaces can still be felt by touch. Slight bending of the MDI-P1000 could cause a fracture along the cut but not on TDI-P1000 ( Figure 6 c). After 6 h, MDI-DDS and MDI-PPD remained as two separate pieces. The result from this set shows some similarity with the results obtained from the first set (self-healing after 10 min): MDI-DDS and MDI-PPD do not heal, while TDI-P1000 has self-healing capability. However, the sets differ in whether MDI-P1000 can self-heal. In this set of testing, MDI-P1000 shows self-healing capability. This could be due to the waiting time before the broken pieces were put together [ 64 ]. During this waiting time, self-adhesion could not take place because the concentration of open stickers may have dropped further as the open stickers formed looped with dangling chains, and eventually they do not have sufficient open stickers available to form bonds across the interfaces for self-adhesion or self-healing to take place. Putting the broken pieces of MDI-P1000 immediately allow it to form a piece of sample could be due to the process of self-adhesion or self-healing. Considering the low fracture toughness along the cut of the newly combined MDI-P1000 broken pieces, the attachment of broken pieces could be due to self-adhesion or poor self-healing capability. Immediate self-healing at 80 °C For this set of cutting and self-healing experiments, the broken pieces were immediately forced together and left in the oven at 80 °C. The results obtained were similar to those obtained at room temperature. As usual, MDI-DDS and MDI-PPD did not heal. TDI-P1000 managed to heal within 5 min. Although the scar is still visible, the cut could not be felt by touch. The surface of TDI-P1000 after 6 h is smoother than at 5 min. The broken pieces of MDI-P1000 were able to form a single piece within 5 min and slight bending did not result in fracture ( Figure 6 d,e). Nevertheless, a scar is still visible. In this case, it can be concluded that self-healing occurs for MDI-P1000 instead of self-adhesion. At an elevated temperature of 80 °C, the broken pieces are able to form a stronger bridge across the interface due to the increase in chain mobility [ 65 ]. This means that at room temperature, bad self-healing may have taken place instead of self-adhesion as the bridges formed across the interface would not be stronger at a higher temperature of self-adhesion since it is dependent on concentration of open stickers [ 66 ]. 3.4.3. Coherency with FTIR Analysis From FTIR results, the concentration of hydrogen bonds present in the samples decreased from TDI-P1000, MDI-P1000, MDI-PPD to MDI-DDS. This is coherent with the optical analysis, which showed that at room and elevated temperatures, TDI-P1000 had better self-healing capability than MDI-P1000, while MDI-PPD and MDI-DDS showed no capability. 3.4.4. Coherency with DSC Analysis The presence of self-healing capability of TDI-P1000 verified the glass transition temperature of −36.75 °C in DSC analysis. The results obtained from all the sets tested for healing under room temperature are consistent with the conclusion drawn from DSC analysis. In the case where healing is possible, TDI-P1000 with a glass transition temperature lower than room temperature was expected to possess self-healing capability, while the other samples with a glass transition temperature higher than room temperature were not expected to heal at room temperature. 3.5. Tensile Analysis Tensile analysis is used to compare the self-healing efficiency of the samples. Self-healing efficiency is defined as the ratio of the recovered ultimate tensile strength to its initial [ 67 ]. It is used to determine the retention of mechanical properties after repetitive healing processes, which is representative of the sample’s self-healing efficiency. To compare the self-healing efficiency, only TDI-P1000 and MDI-P1000 were tested as self-healing capability was seen from optical analysis. From the optical analysis, although TDI-P1000 has a smooth surface when healed at 80 °C, the crack can still be seen on the subsurface which indicates that the self-healing may not be 100% complete. During the tensile test, the samples tend to fail along the healing site ( Figure 7 ), which further supported that the samples were not healed fully. The sample were first tested for tensile strength by pulling till fail. On another sample of the same dimension, a cut that separates the sample into two is made. The second sample is left in the oven at 80 °C to heal for 6 h. The second sample is then tested for tensile strength. The same sample is healed in the oven at 80 °C for 6 h. The sample that is healed for the second time is then tested for tensile strength. Only two stress–strain curves were obtained as MDI-P1000 could not be healed for the second time. MDI-P1000 has a healing efficiency of 39%, whereas TDI-P1000 has a healing efficiency of 42% from the first heal from the horizontal comparison between Figure 7 a,b. It has a healing efficiency of 22% from second heal. From the results obtained from tensile tests, it can be seen that tensile strength decreases with every repetition of breaking and healing. TDI-P1000 has a higher retention of mechanical strength after healing as compared to MDI-P1000. This is coherent with the conclusion derived from FTIR, where a higher number of dynamic interactions (H-bonds) in TDI-P1000 means a higher self-healing efficiency ( Figure 7 c). This could be the reason for MDI-P1000 not being able to heal the second time, as the self-healing efficiency had degraded [ 68 ]."
} | 7,641 |
31814613 | null | s2 | 7,747 | {
"abstract": "Microfluidic devices allow for the manipulation of fluids, particles, cells, micro-sized organs or organisms in channels ranging from the nano to submillimeter scales. A rapid increase in the use of this technology in the biological sciences has prompted a need for methods that are accessible to a wide range of research groups. Current fabrication standards, such as PDMS bonding, require expensive and time consuming lithographic and bonding techniques. A viable alternative is the use of equipment and materials that are easily affordable, require minimal expertise and allow for the rapid iteration of designs. In this work we describe a protocol for designing and producing PET-laminates (PETLs), microfluidic devices that are inexpensive, easy to fabricate, and consume significantly less time to generate than other approaches to microfluidics technology. They consist of thermally bonded film sheets, in which channels and other features are defined using a craft cutter. PETLs solve field-specific technical challenges while dramatically reducing obstacles to adoption. This approach facilitates the accessibility of microfluidics devices in both research and educational settings, providing a reliable platform for new methods of inquiry."
} | 312 |
36215489 | PMC9586279 | pmc | 7,748 | {
"abstract": "Significance Acetate-producing microorganisms (acetogens) have long been hypothesized as among the earliest evolving organisms. Concomitantly, serpentinite formations have been widely considered the most likely habitat to have supported early life. Yet, little is known of acetogens in contemporary serpentinites. Here, evolutionary and genome-guided metabolic reconstructions provide insights into the characteristics of two deeply branching, uncultured acetogens that are found in serpentinite-hosted waters. The two types are differentially distributed among serpentinite fluids and exhibit physiological traits unique to their respective environments, with the “type II” acetogens specialized for growth on substrates generated by water–rock interaction. Collectively, these data help bridge a knowledge gap between hypotheses of early acetogenic life on Earth and contemporary acetogen characteristics in early Earth analog environments.",
"conclusion": "Conclusions Type II Acetothermia from the SO exhibit physiological/metabolic traits that allow them to conserve energy in contemporary serpentinization-influenced environments. Phylogenetic data indicate that many of the traits, including modifications to the WL pathway, originated early during the evolution of this lineage, suggesting selective pressures on this cell type have been fairly constant since the metabolism first originated in the lineage. Type II Acetothermia are characteristic inhabitants of highly serpentinized waters in the SO and are likely key contributors to ecosystem productivity in these environments. Further, these organisms encode proteins involved in key energy conservation pathways that are early branching within the Bacteria or derive from deep within the archaeal domain (e.g., ATP synthases and carbon monoxide dehydrogenases). The phylogenetic and physiological characterization of contemporary type II Acetothermia helps bridge theoretical inferences regarding the role of serpentinization in supporting the earliest forms of life on Earth and reveals the key physiological characteristics that could allow this metabolism to operate at the thermodynamic limits of life ( 76 ) by persisting with substrates derived from water–rock interaction such as H 2 , HCOO − , and CO. Key questions remain regarding type II Acetothermia metabolism, including how their cellular energetics are balanced under high H 2 partial pressures and low DIC conditions relative to more-DIC-replete conditions used to cultivate canonical acetogens. Further, the form of carbon (e.g., CO 2 /HCO 3 − /CO 3 2− , HCOO − , or CO) that supports acetogenesis and carbon fixation by the WL pathway remains to be elucidated and will require cultures to definitively prove.",
"discussion": "Results and Discussion Description of Fracture Fluids Sampled from the SO. The eight fracture waters sampled from the SO in 2015 ranged, in pH, from 8.2 to 11.4, while the eight sampled in 2017 ranged, in pH, from 8.3 to 11.3 ( SI Appendix , Tables S1 and S2 ). The four waters sampled from 2020 ranged, in pH, from 7.5 to 11.4. The lithological setting of each well varied and included gabbro, peridotite, and “contact” areas between peridotite and gabbro settings that represented mixtures of alkaline and hyperalkaline fluid types ( 14 ). Fracture waters were previously classified based on their geochemical properties and extent of transformation into hyperalkaline fluids via serpentinization reactions ( 14 , 34 , 63 ). Briefly, less-reacted waters are termed “type I” and tend to have slightly alkaline pH, lower conductivity, higher oxidation reduction potentials (ORPs), and generally exhibit higher oxidant availability along with lower concentrations of reduced aqueous-phase gases like CH 4 and H 2 ( 14 , 15 , 34 , 63 ). Conversely, waters that have been subjected to increased serpentinization are termed “type II” and tend to have hyperalkaline pH, higher conductivity, lower ORPs, minimal oxidant availability, low DIC availability, and much higher reduced aqueous-phase gas concentrations compared to type I waters ( 14 , 15 , 34 , 63 ). Fracture water communities that were analyzed herein spanned a range of type I and type II waters obtained from multiple subsurface drilled water wells. Recovery of Acetothermia MAGs from the SO. Metagenomic characterization of DNA recovered from filtered biomass collected from eight fracture waters in eight wells in 2015 ( SI Appendix , Table S1 ), another eight fracture waters from seven wells in 2017 ( SI Appendix , Table S2 ), and four fracture waters from three wells in 2020 yielded 14 moderate- to high-quality MAGs affiliated with Acetothermia (two from 2015 samples, five from 2017 samples, and seven from the 2020 samples; estimated completeness: 75 to 90%; Table 1 ). The metagenomes from the 2015 samples were sequenced at a lower read depth than those from 2017 and 2020, likely resulting in recovery of fewer MAGs. Phylogenomic analysis indicated the presence of two phylogenetically distinct populations within the Acetothermia phylum (hereafter referred to as type I and type II Acetothermia), with high within-group genomic identity in the two groups (∼98 to 100% AAI within type I and type II groups; Fig. 1 A ). In addition, the MAGs were nearly identical (∼100% AAI) to samples collected from the same wells across years. Such AAI values are well above commonly used thresholds to delineate species (e.g., 95% AAI) ( 51 ), and indicate essentially homogenous population-level genomic diversity across fracture waters and years for the two phylotypes. Fig. 1. Phylogenomic analysis and ecological distribution of Acetothermia MAGs recovered from SO subsurface fracture waters. ( A ) ML phylogenetic reconstruction of MAGs recovered in this study in comparison with those recovered from previous studies, with type I MAGs from the SO highlighted in orange and type II MAGs highlighted in blue. The year of the sample is shown, followed by the well designation from which the MAGs were recovered. Biomass from fracture waters was collected from 50-m (NSHQ14B) and 85-m (NSHQ14C) depths from NSHQ14 in 2017. The three proposed orders for the Acetothermia/Bipolaricualaeota are shown on the far right. Branch length is scaled to that shown in the lower left for expected substitutions per site. Bootstrap values >90% (out of 1,000 bootstraps) are indicated by black circles. Thermotogae representatives were used as the outgroup including Petrotoga mobilis SJ95, Thermotoga petrophila RKU-1, and Mesotoga inferna. ( B ) The abundances of types I and II populations are shown as the percentage of total metagenome reads mapped to either MAG type for each 2017 metagenome. The 2015 and 2020 metagenomes were not used for abundance calculations, given their relatively low sequencing depth and limited sampling scope, respectively. Metagenomes are arranged by the fluid type designation of their well waters and then ordered by ascending pH values for well waters, as indicated in parentheses next to each metagenome. Comparison of type I and type II MAGs revealed overall genomic differences (∼48% AAI ± ∼10%) consistent with class- or order-level taxonomic differences ( 51 ). The type I SO clade formed a group with the previously described Candidatus Acetothermum autotrophicum MAG ( 35 ) and others recovered from subsurface environments ( Fig. 1 A ). The type II clade formed a lineage along with a MAG generated from a Lost City hydrothermal vent fluid metagenome (UBA7950). The Lost City MAG was generated from genome mining of publicly available metagenome data ( 64 ) and was not characterized in that study. The Lost City hydrothermal system is a well-characterized marine serpentinite system ( 21 , 22 , 65 ) and features characteristics similar to those of highly reacted type II SO waters, including hyperalkaline pH, high concentrations of H 2 , and limited DIC ( 65 ). Thus, the close phylogenetic similarity between the MAGs recovered from hyperalkaline waters in the SO and the Lost City (and their phylogenetic distinction from other Acetothermia) suggests that the clade comprises serpentinite-adapted Acetothermia from globally distributed serpentinite-hosted ecosystems. The type II MAGs exhibited smaller estimated genome sizes, lower guanine and cytosine (GC) content, and fewer protein coding genes relative to the type I MAGs ( Table 1 ), consistent with genome streamlining to minimize energy demands associated with inhabiting hyperalkaline fluids, as previously observed among genomes in serpentinization-impacted environments ( 15 , 18 ). The two populations exhibited nearly mutually exclusive distributions across fracture fluids, with the type I population more prevalent in mixed waters from “contact” wells (i.e., at gabbro/peridotite interfaces), while the type II populations were abundant in waters with hyperalkaline pH ( Fig. 1 B and Table 1 ). A notable exception was observed in waters from well NSHQ14, where the distribution of the two populations overlapped in samples taken from nearer to the surface in 2017 (50 m; NSHQ14B) and 2020, which was packed off at a depth of ∼20 m and thus only encompassed shallow waters. The type II population dominated the community in deeper waters from 2017 (85 m; NSHQ14C) and was also abundant in waters nearer to the surface in both 2017 and 2020 ( Table 1 ), while the type I populations comprised smaller fractions of the nearer to the surface communities of 2017 and 2020 waters ( Fig. 1 B and Table 1 ). Near-surface mixing of water types closer to the surface in NSHQ14 is likely, as suggested by down-borehole geochemical data from the immediately adjacent BA3A well suggesting that the deeper waters are considerably more reducing than the surface waters ( 66 ). Thus, the near-surface waters of NSHQ14 may represent a mixture of communities typically associated with type I (or mixed) and type II waters. Adaptations to Acetogenic Metabolism in Type I and II Acetothermia Populations. The metabolic potentials of types I and II MAGs were investigated to evaluate whether they exhibited different adaptations to serpentinization-influenced geochemical gradients that characterize type I and II waters, respectively ( SI Appendix , Tables S1 and S2 ). Metabolic reconstructions indicated that both type I and II populations are likely capable of autotrophic acetogenesis via the WL pathway, albeit through slight modifications that were unique to their respective environments ( Fig. 2 ). Fig. 2. Metabolic reconstructions for Acetothermia type I and II MAGs recovered from the SO subsurface fracture waters. A composite metabolic reconstruction map is shown for type I and II MAGs, with encoded proteins/pathways in orange denoting type I MAGs and those in blue denoting the type II MAGs. Energy conservation pathways highlighted in the text are grouped in gray boxes, with the rest of the pathways representing central carbon metabolism. Question marks show areas of uncertainty in reconstructing aspects of metabolism. Subunits or proteins highlighted in black were not observed in the corresponding MAGs. Abbreviations are as follows: anaerobic carbon monoxide dehydrogenase (Coo), pyruvate Fd oxidoreductase (Pfor), ADP-forming acetyl-CoA synthetase (ACD), AMP-forming acetyl-CoA synthase (ACS), citrate synthase (GltA), aconitate hydratase (Aco), isocitrate dehydrogenase (Idh), 2-oxoglutarate Fd oxidoreductase (Kor), succinyl-CoA synthetase (Suc), succinate dehydrogenase (Sdh), fumarate hydratase (FumC), malate dehydrogenase (Mdh), pyruvate orthophosphate dikanase (PpdK), pyruvate-water dikinase (PpsA), enolase (Eno), phosphoglycerate kinase (PgK), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), triosephosphate isomerase (TPI), fructose-1,6-bisphosphatase I (FBP), glucose/mannose-6-phosphate isomerase (Pgi-Pmi), glucose-6-phosphate isomerase (Pgil), Fd, reduced Fd (Fd 2− ), nitrate reductase (Nar), cytochrome c nitrite reductase (Nrf), heme-Cu oxidase (HCO), methyl-viologen reducing [NiFe]-hydrogenase (Mvh), and heterodisulfide reductase (Hdr). The first step in autotrophic acetogenesis is the reduction of CO 2 to CO in the carbonyl branch of the WL pathway. The type I MAGs encoded bacterial-like carbon monoxide dehydrogenase/acetyl-CoA synthase protein complex (CdhACDE/AcsE; nomenclature following ref. 53 ), typical of other Acetothermia ( 53 ). In contrast, type II MAGs encoded an archaeal-like carbon monoxide dehydrogenase/acetyl-CoA synthase protein complex (CdhABCDE), with an additional bacterial-like AcsE subunit ( SI Appendix , Figs. S1 and S2 and Dataset S2 ). The two complexes are assumed to be functionally equivalent ( 53 ), although their distinct distributions between type I and II populations suggest advantages of each system in their respective environmental regimes. Moreover, the presence of both an archaeal-like CdhA colocalized with other Cdh subunits in the type II genomes, along with the existence of a separate bacterial-like (but non-Acetothermia-like) CdhA within these genomes, suggests the potential for plasticity in their use ( Dataset S2 ). Genes encoding an ABC-type membrane transport system with components homologous to the bicarbonate transport system of Cyanobacteria were colocalized with the bacterial-like CdhA subunit in type II genomes, along with sodium bicarbonate transport protein encoding genes ( sbtA ) encoded by the MAGs ( Dataset S2 ). This transport system is critical for cyanobacterial Na + -dependent bicarbonate (HCO 3 − ) uptake under low CO 2 conditions ( 67 ). Thus, it is plausible that the additional bacterial-like CdhA of type II populations is associated with DIC (HCO 3 − ) uptake, potentially pointing to plasticity in their use of the archaeal- or bacterial-like CdhA, depending on DIC availability. Notably, no components of this transport system were identified in the type I MAGs ( Dataset S2 ), consistent with greater availability of DIC in type I waters. An equally important step in acetogenesis is the reduction of CO 2 to HCOO − in the methyl branch of the WL pathway. Type I MAGs encoded formate dehydrogenase complexes (Fdo) that can reduce CO 2 to HCOO − , whereas type II MAGs encoded homologs of the reversible NADH- and ferredoxin (Fd)-dependent, electron-bifurcating formate dehydrogenase (HylCBA-FdhF2) identified in Clostridium acidurici ( 68 ). In type II MAGs, the hylCBA-fdhF2 genes were colocalized together with genes coding for Fhs ( SI Appendix , Fig. S3 and Dataset S2 ), a key enzyme for carbon fixation in the WL pathway. This suggests that the electron bifurcating FdhF2-Hyl complex may provide formate for the WL pathway in the type II population, and/or also contribute to the production of CO 2 from formate for use in the carbonyl branch of the WL pathway to generate CO. A critical challenge faced by the type II Acetothermia population is the availability of CO 2 for reduction to CO and HCOO − in the initial steps of the WL pathway. CO 2 is not readily available in SO fluids with hyperalkaline pH (pH >10) ( 69 ) due to conversion to HCO 3 − (pK a = 6.4) and, ultimately, CO 3 2− (pK a = 10.2). Moreover, carbonate mineral solubility is inversely correlated to pH, resulting in substantial carbonate precipitation and DIC removal from hyperalkaline waters containing abundant Ca + ( 70 ). However, serpentinization reactions resulting in abundant H 2 can drive the generation of CO and HCOO − through the abiotic reduction of DIC ( 3 , 4 , 71 ). Notably, acetogens can use CO and HCOO − as initial, or sole, compounds for entry into the WL pathway ( 72 , 73 ). Thus, it is plausible, if not likely, that the type II Acetothermia preferentially use CO and HCOO − , when available, rather than expending reducing equivalents from NAD(P)H or reduced Fd to reduce CO 2 . Similar adaptations were observed in Methanobacterium genomes and transcriptomes recovered from hyperalkaline waters in the SO ( 20 , 27 ), wherein Methanobacterium were inferred to be dependent on HCOO − to overcome DIC limitation via cytoplasmic HCOO − oxidation, rather than H 2 , as is otherwise conserved in this methanogenic genus ( 27 ). HCOO − concentrations in the well waters hosting both Acetothermia types range from ∼1 μM to 2 μM ( 14 , 27 ), suggesting that HCOO − is bioavailable in these environments. Moreover, laboratory experiments of serpentinization reactions in SO waters have demonstrated considerable HCOO − production (up to 98 μM) coinciding with H 2 and CO 2 drawdown ( 71 ), suggesting that abiotic HCOO − generation could support HCOO − -dependent populations in these waters. Nevertheless, the potential use of two different enzyme systems to possibly catalyze the reduction of CO 2 to HCOO − represents an intriguing difference between the two Acetothermia types. Reduction of CO 2 to HCOO − ( E°’ = −420 mV) by Fdo in acetogens involves NAD(P)H ( E°’ = −320 mV) as an electron donor, and this reaction proceeds in vivo (despite standard state energetics) if the ratio of reduced to oxidized NAD(P)H is high. However, the reduction potential of CO 2 (HCO 3 − or CO 3 2− ) to HCOO − becomes increasingly negative at increasingly alkaline pH ( 75 ) and becomes even less thermodynamically favorable to reduce with NAD(P)H. This reaction barrier is due to the −60 mV/per pH unit shift from standard state for a 2 e − /2 H + charge transfer during the reduction of HCO 3 − to HCOO − when compared to only a −30 mV per pH unit shift from standard state due to a 2 e − /1 H + charge transfer for NAD(P)H ( 75 ). It is possible, then, that electron bifurcation/confurcation enables type II Acetothermia cells to overcome the thermodynamics associated with CO 2 (HCO 3 − or CO 3 2− ) reduction ( E ≈ −630 mV at pH 11) by coupling the reaction with the exergonic oxidation of Fd ( E ≈ −650 mV at pH 11) ( 8 ) and the endergonic oxidation of NAD(P)H ( E ≈ −470 mV at pH 11), so long as the ratios of reduced to oxidized Fd and NAD(P)H in cells are high. The other requisite proteins involved in the WL pathway (Fhs, FolD, and MetVF) were encoded in both type I and II MAGs. Given the minimal energy conserved during acetogenesis, phosphorylation of acetyl-CoA produced from the WL pathway by phosphotransacetylase (Pta) and its subsequent dephosphorylation to produce one ATP and acetate by acetate kinase (AckA) has been traditionally considered necessary for acetogens to balance the consumption of one ATP molecule from HCOO − fixation in the methyl branch of the WL pathway, thereby rendering the WL pathway ATP consumption neutral ( 76 ). However, as observed elsewhere for other Acetothermia ( 36 ), genes encoding AckA or Pta were not identified in the genomes of either type I or type II Acetothermia. Rather, a homolog of an archaeal-like ADP-forming acetyl-CoA synthetase (ACD) ( 77 ) was identified in the type I MAGs but not in the type II MAGs ( Fig. 2 ). Further, both the type I and type II MAGs encoded homologs of the non-ATP generating acetyl-CoA synthase [Acs, also known as the AMP-forming Acs; abbreviated here as ACS to differentiate from the Acs discussed above ( 78 )] that is generally considered an assimilatory, but potentially reversible, acetate utilization mechanism ( 78 ). Consequently, the potential to leverage acetate production from acetyl-CoA for ATP production in the type II MAGs is unclear, raising questions of how the cellular bioenergetics of the type II population might operate. The minimal energy conserved during acetogenesis leads to acetogens operating at the near limit of possible bioenergetics when growing autotrophically ( 76 ). Importantly, the overall energetics of autotrophic, hydrogenotrophic acetogenesis in the model acetogens Clostridium ljungdahlii and Acetobacterium woodii grown under CO 2 -replete conditions are directly associated with H 2 partial pressures, the ratio of reduced/oxidized cofactors (e.g., Fd) within cells, and small but consequential alterations to acetogenesis pathways ( 76 ). In particular, hydrogenotrophic acetogenesis is most favorable at elevated H 2 partial pressures and high ratios of reduced to oxidized Fd in cells ( 76 ). Further, acetogenesis in C. ljungdahlii and A. woodii is principally energy conserving due to the activity of the multisubunit Fd-NAD + oxidoreductase (Rnf) complex that couples the oxidation of Fd to the reduction of NAD + , with the free energy of the reaction used to generate a transmembrane ion (sodium or proton) gradient that can drive ATP synthesis ( 76 ). Rnf complexes were encoded in type II MAGs, but not type I MAGs ( Fig. 2 ). Low-potential reduced Fd needed to drive Rnf in type II cells could be generated by tetrameric (group 3) electron bifurcating [FeFe]-hydrogenase homologs ( 79 , 80 ), of which two loci were identified in the type II MAGs ( Fig. 2 and Dataset S2 ), also as present in A. woodii and C. ljungdahlii ( 76 ). The functionality of the [FeFe]-hydrogenase was supported by colocalized genes encoding the four subunits of tetrameric bifurcating hydrogenases (HydABCD) ( 79 ) ( Dataset S2 ) in addition to phylogenetic comparison of the catalytic subunit (HydA) to classified [FeFe]-hydrogenases in the HydDB database (group A classification; E value < 2 × 10 −129 ) ( 81 ). These enzyme complexes are predicted to reversibly bifurcate two electrons from H 2 and couple the endergonic single-electron reduction of Fd to the exergonic single-electron reduction of NAD + ( 80 ). Notably, the well waters contain substantial concentrations of H 2 [e.g., ∼2.9 mM in NSHQ14 ( 14 )] that would increase the overall bioenergetics of hydrogenotrophic acetogenesis. Sodium or hydrogen ions pumped from Rnf could then be harnessed via F-type (bacterial-like) ATP synthases ( SI Appendix , Fig. S4 ) that were also detected in type II, but not type I MAGs, with the latter encoding V-type (archaeal-like) ATPases ( Fig. 2 and SI Appendix , Fig. S4 and Supporting Text ). In addition to the critical WL steps described above, the reduction of methylenetetrahydrofolate (CH 2 -THF) to methyl tetrahydrofolate (CH 3 -THF) is catalyzed by MetVF, thereby generating the methyl group of the methyl branch of the WL pathway ( Fig. 2 ) ( 82 , 83 ). The type I Acetothermia encode MetVF, and these genes are colocalized in the genome with those encoding HdrA and MvhD ( SI Appendix , Fig. S5 and Dataset S2 ), while genes encoding HdrBC and MvhAG were, together, colocalized but encoded elsewhere in the genomes. HdrABC are commonly involved in bifurcating electrons in methanogens ( 84 , 85 ) and were recently proposed to be involved in bifurcating electrons from NADH ( E°’ = −280 mV) for the exergonic reduction of CH 2 -THF ( E°’ = −200 mV) and the endergonic reduction of a yet to be identified second electron donor in the model acetogen Moorella thermoacetica ( 83 ). Genes encoding MetVF are colocalized in the genome in M. thermoacetica , and these proteins have been copurified as a complex with HdrABC and the [NiFe]-hydrogenase protein MvhD, suggesting that the second electron acceptor may be a proton that is reduced to H 2 ( E°’ = −414 mV). This is further supported by data indicating that CH 2 -THF reduction activity is localized to the membrane ( 86 ), where an energy-converting hydrogenase (Ech) could function in the reduction of protons to H 2 . The colocalization of metVF , hdrA , and mvhD in type I Acetothermia MAGs ( SI Appendix , Fig. S5 ) suggests the presence of a similar bifurcating mechanism in these cells, albeit through a cytoplasmic MvhAGD complex (Group 3c complex, confirmed by phylogenetic comparison of the MvhA to the HydDB server; E value = 0) ( Fig. 2 ) since these cells do not encode a separate Ech, rendering the fate of intracellular H 2 produced by this reaction unclear. In contrast, neither HdrABC nor MvhD are encoded by type II Acetothermia MAGs. Rather, several homologs of RnfC, which codes for NADH dehydrogenase functionality, are found in these genomes. As such, CH 2 -THF reduction in type II Acetothermia may be directly coupled with NADH oxidation, as in the acetogen A. woodii ( 87 ). It is possible that reduction of CH 2 -THF in type II Acetothermia occurs via NADH oxidation instead of the putative bifurcating, H 2 -dependent reaction of type I Acetothermia due to unfavorable thermodynamics of the bifurcating reaction due to product (H 2 ) inhibition in the H 2 -rich environments inhabited by type II Acetothermia ( 14 ). Type I and II populations both exhibited the capacity to generate chemiosmotic potentials to drive ATP synthesis, albeit through distinct pathways and ATPases that may reflect the geochemical characteristics of the environments that they differentially inhabit ( SI Appendix , Fig. S4 and Supporting Text ). In contrast to the Rnf complexes encoded by type II MAGs, the type I MAGs encoded a nearly complete NADH dehydrogenase (Nuo) complex that allows entry of electrons in the form of NADH into canonical electron transport respiration chains of Archaea, Bacteria, and Eukarya ( 88 ). Moreover, the type I population encoded proteins allowing for the respiration of NO 3 − and NO 2 − to N 2 (or NH 4 + ) via Nar and Nrf complexes, respectively ( Fig. 2 ), and encoded cytochrome c oxidases necessary for aerobic respiration ( Fig. 2 and Dataset S2 ). This expanded respiratory capacity is also coincident with expected higher oxidant concentrations (e.g., NO 3 − and O 2 ) in type I waters ( 14 ). The capacity for NO 3 − and O 2 reduction was also observed in the first described Acetothermia MAG, Candidatus Acetothermum autotrophicum ( 35 ), as well as in other related MAGs ( 36 ). The Type II Acetothermia Are Dominant Autotrophs in Serpentinizing and Hyperalkaline Subsurface Fluids. Both type I and II Acetothermia MAGs encoded glycolytic/gluconeogenic pathways that would allow for the conversion of acetyl-CoA to six-carbon sugars that can be used in biosynthesis, albeit with partially different complements of enzymes to complete the pathways ( SI Appendix , Supporting Text ) ( 89 ). Substrates capable of supporting autotrophic metabolism of type I and II Acetothermia appear to be limited to HCOO − , CO, H 2 , and, possibly, CO 2 /HCO 3 − . In addition, the two Acetothermia types exhibited varying capacities to incorporate exogenous organic carbon compounds, with the type I MAGs encoding a greater capacity for facultatively heterotrophic metabolism ( SI Appendix , Supporting Text ). The distributions of type I and II Acetothermia were mutually exclusive among the well waters sampled in 2017 ( Figs. 1 B and 3 ), with the exception of the NSHQ14B sample taken nearer to the surface (50 m), while only the type II population was present in the deeper (85 m) waters (NSHQ14C). Type II Acetothermia were the dominant population (22%, 20%, and 35% relative abundance, respectively) in the most hyperalkaline waters analyzed in 2017 (WAB71, NSHQ14B, and NSHQ14C) and were also the dominant putative autotroph in these waters. The other putatively autotrophic populations present in high relative abundance (i.e., >5%) in the three most hyperalkaline samples collected in 2017 ( Fig. 3 ) included those related to Methanobacterium ( 20 , 27 ) (0%, 18%, and 17%, respectively), Thermodesulfovibrionales (5%, 10%, and 17%, respectively) ( 66 ), and uncharacterized Dehalococcoidia (0%, 5%, and 9%, respectively). The relative abundances of dominant taxa based on metagenomic analyses were broadly consistent with inferences from previous 16S rRNA gene analysis of these well communities from previous years ( 14 ), suggesting that they were representative of the native well communities. Consequently, the type II Acetothermia are inferred to represent key primary producers in communities inhabiting hyperalkaline SO waters and likely substantially contribute to ecosystem productivity. Consistently, these populations were also among the most abundant in the 2015 and 2020 metagenomes where they were identified ( Table 1 ). The only autotrophic pathway identified among community members inhabiting highly serpentinized waters (pH >11.0) was the WL pathway, unlike communities inhabiting more circumneutral waters where multiple autotrophic pathways (WL, Calvin Cycle, and rTCA cycle) were identified ( Fig. 3 ). In contrast, only minor populations of type I Acetothermia were present in lower pH wells (or the near-surface waters of NSHQ14B or NSHQ14 in 2020) ( Fig. 3 ). In summary, both the type I and type II Acetothermia populations exhibit evidence for autotrophy that may be supplemented by organic carbon sources. However, the type II populations appear to be more dependent on substrates produced by water–rock interaction such as HCOO − , CO, or H 2 for autotrophic metabolism ( Fig. 3 ). This observation is consistent with their general dominance in environments where these substrates are more readily available ( 14 , 69 ). Fig. 3. Putative autotrophs in subsurface fracture water communities from the SO. Each column shows the relative abundances of reconstructed MAGs (>5% relative abundance) within each well water community, with the corresponding taxonomic classification of MAGs given on the left. The circles represent individual MAGs and are scaled based on the relative abundance of that MAG among others within that community (based on the legend at the top left of the plot). The circles are colored according to the autotrophic carbon fixation pathway that is inferred to be encoded within the MAG, as based on the legend at the top right of the plot. Well names are followed by the pH of the sampled fracture waters in parentheses. Taxonomic classifications are given at the lowest characterized taxonomic designation, followed by either the specific uncultured group it belongs to (alphanumeric designations) or are otherwise followed by “unclassified” if the MAG is not related to previously characterized genomes within the Genome Toolkit Database. The type II Methanobacterium is the same as described in Fones et al. ( 27 ), and, while it encodes the WL pathway, it is dependent on formate as an electron donor and carbon source. The 2015 and 2020 metagenomes were not used for abundance calculations, given their relatively low sequencing depth and limited sampling scope, respectively. The Type II Acetothermia Are Analogs of Early-Evolving Acetogens. The prevalence of type II Acetothermia as putative autotrophs operating the WL pathway in hyperalkaline SO waters is consistent with previous suggestions of an ancient ancestry of the WL pathway that has its roots in primitive methanogens or acetogens that inhabited environments rich in iron sulfide minerals and that were actively undergoing serpentinization ( 2 , 90 ). Arguments favoring the origin of the WL pathway in serpentinizing environments include that 1) it is a simple linear pathway that can be exergonic ( 91 ); 2) it involves reactants, intermediates, and products also involved in serpentinization ( 92 ); and 3) many of the key reactions of the WL pathway are catalyzed at iron–sulfur or nickel–iron–sulfide centers that are reminiscent of mineral catalysts that perform similar chemistry ( 5 ). These observations motivated further phylogenetic analysis of type II Acetothermia and components of their WL pathways. CdhABC house the nickel–iron–sulfide active site center of carbon monoxide dehydrogenase that catalyzes the initial reduction of CO 2 to CO. Phylogenetic analysis of the type II archaeal-like CdhABC homologs indicated that they comprised a monophyletic group with CdhABC from other bacterial and archaeal MAGs largely from serpentinite-hosted environments ( Fig. 4 A ). These included the putatively acetogenic Actinobacteria-related Candidatus Hakubanella thermoalkaliphilus ( 30 ), the putatively acetogenic NPL-UPA2 organism ( 29 ), an unclassified anaerobic methanotrophic euryarchaeota (ANME) MAG from the Lost City system, and a deltaproteobacterial Candidatus Desulforudis audaxviator MAG, which are commonly observed in SO fluid communities ( Fig. 3 ) and other serpentinization-influenced communities ( 93 ). In addition, the type II CdhABC homologs were monophyletic with one another, closely related, and followed the same branching pattern as the whole genome Acetothermia phylogeny ( Fig. 1 A ). Additional phylogenetic analysis of CdhA from the type II Acetothermia along with 500 of the most closely related homologs in the NCBI database further supported the branching of Acetothermia CdhA and those from organisms identified in serpentinite environments in association with the CdhA of Archaea that formed a group distinct to the homologous CdhA/AcsA of Bacteria ( SI Appendix , Fig. S2 ). Together, these observations suggest that 1) Cdh were vertically inherited among the type II SO Acetothermia and have differentiated with minor taxonomic divergence of type II populations; 2) type II Acetothermia from the SO and putative acetogens from other serpentinite-hosted environments harbor Cdh that share a common evolutionary ancestor; and 3) the Cdh arose from a possible transfer (or shared evolutionary origin) between Archaea and Bacteria primarily from serpentinite-hosted environments ( SI Appendix , Fig. S2 ), and this shared evolutionary history appears to be deeply rooted within Archaea. The latter assertion is evinced by the well-supported monophyly of bacterial Cdh from serpentinite environments (in addition to those from other bacteria including Chloroflexi/Deltaproteobacteria) with early-evolving methanogens (i.e., class I methanogens: Methanopyrales, Methanobacteriales, and Methanococcales), to the exclusion of other Archaea, including more recently evolved methanogens (i.e., Methanosarcinales and Methanomicrobiales) and recently identified methanotrophs/alkanotrophs like the Bathyarchaeota and Syntropharchaeales ( Fig. 4 A ). This distinction is particularly striking given that Cdh complexes have been largely vertically inherited among Archaea ( 53 ), and is consistent with previous analyses hypothesizing a transfer of archaeal Cdh to Bacteria via an unidentified Euryarchaeota ( 29 , 53 ). The type II MAGs also encoded bacterial-like CdhA homologs not related to the bacterial-like CdhA in the type I MAGs ( SI Appendix , Fig. S6 ). Rather, the bacterial-like CdhA of type II MAGs were similar to those in a deep-branching group primarily comprising ANME and Firmicutes ( SI Appendix , Fig. S6 ). As suggested above, the presence of two evolutionarily distinct bacterial- and archaeal-like CdhA within the type II MAGs may indicate physiological plasticity in the WL pathway, perhaps related to the availability of specific carbon substrates. Importantly, while it is not currently possible to determine whether the catalytic Cdh subunits share an origin with deeply diverging Archaea or whether they were laterally transferred from or into the type II lineage, the deep node connecting the primarily serpentinite-hosted bacterial Cdh lineage and that of all Archaea nevertheless suggests an early origin of the Cdh within the type II Acetothermia lineage. Fig. 4. Phylogenetic placement of key proteins involved in acetogenesis and Acetothermia lineages recovered from the SO subsurface waters among other bacterial lineages. ( A ) ML phylogeny of the oxidoreductase subunits of the carbon monoxide/acetyl-CoA synthase (CODH/ACS) complex (CdhABC) encoded by the type II Acetothermia in context of other archaeal-like CdhABC (alignment length of 1,908 amino acid positions). Each subunit was individually aligned, and a concatenation of the three was subjected to ML analysis. The type II Acetothermia CdhABC are highlighted in bolded blue text. CdhABC from MAGs recovered from serpentinite-influenced environments are indicated on the right. Black circles show >90% bootstrap support (out of 1,000 bootstraps). Branch length is scaled based on the expected number of substitutions per site legend at the bottom left. CdhABC clades are collapsed as triangles, with the taxonomic groups they correspond to indicated next to the triangles. The monophyletic bacterial CdhABC clade is shown by an arrow. The tree is shown with a midpoint-rooted visualization. ( B ) Unrooted ML phylogeny of concatenated protein alignments from 30 housekeeping single-copy genes within 722 bacterial genomes representative of 128 phyla in the Genome Taxonomy Database (filtered alignment length of 9,162 amino acid positions; see Materials and Methods for additional analytical details). Phylum-level groupings are provided next to the shaded regions, with multiple phyla delineated with slashes. Genome information for the 722 entries is provided in Dataset S1 . Black circles show >90% bootstrap support (out of 1,000 bootstraps). Bootstraps are only shown for phylum or higher-level groupings for clarity in interpreting the tree. Branch length is scaled based on the expected number of substitutions per site, as indicated on the left. Type I and II Acetothermia lineages are indicated with orange and blue stars, respectively. The placement of recently hypothesized root positions for the bacterial domain is shown with black ( 44 ) or white ( 94 ) stars. To further assess the phylogenetic placement of the SO Acetothermia among other taxonomic divisions, a comprehensive phylogenomic analysis was conducted with 722 genomes representing 128 bacterial phyla in the Genome Taxonomy Database ( Fig. 4 B ). Phylogenomic analysis of 30 universally conserved phylogenetic marker genes including RNA polymerase subunits and ribosomal proteins (see Materials and Methods for additional details) suggested that the Acetothermia/Bipolaricaulota lineage comprised a highly supported outgroup to the Thermotogota that, together, were related to the Deinococcota and others ( Fig. 4 B ), consistent with previous phylogenomic analyses ( 35 ). Recent attempts to identify the root of the bacterial domain using comprehensive phylogenomic analyses have placed the bacterial root between the Thermotogota and all remaining bacterial phyla ( 44 ) or with the Thermotogota, Deinococcota, and Synergistota as a group closest to the optimal rooting position within several possible rooting scenarios ( 94 ) ( Fig. 4 B ). These results are consistent with an early phylogenomic analysis placing the first recovered Acetothermia/Bipolaricaulota genome from Candidatus Acetothermum autotrophicum near the root of the bacterial tree along with the Thermotogae (now Thermotogota) and the Deinococcus-Thermus (now Deinococcota) ( 35 ). The precise position of the root of the bacterial domain remains controversial and is dependent on the taxa included, genes considered in the analysis, and phylogenetic methods that are used. Nevertheless, the placement of the Acetothermia/Bipolaricaulota as a highly supported sister clade to the Thermotogota and recent analyses placing the Thermotogota near the root of the bacterial domain ( 44 , 94 ) suggest that the Acetothermia/Bipolaricaulota are also among the earliest evolving bacterial lineages. Together, these observations are consistent with the hypothesis that ancestors of type II Acetothermia inhabited environments undergoing active serpentinization and add further support to the supposition that serpentinization played a key role in supporting the earliest acetogens."
} | 9,870 |
37988619 | null | s2 | 7,749 | {
"abstract": "The soil environment adjacent to plant roots, termed the rhizosphere, is home to a wide variety of microorganisms that can significantly affect the physiology of nearby plants. Microbes in the rhizosphere can provide nutrients, secrete signaling compounds, and inhibit pathogens. These processes could be manipulated with synthetic biology to enhance the agricultural performance of crops grown for food, energy, or environmental remediation, if methods can be implemented in these nonmodel microbes. A common first step for domesticating nonmodel organisms is the development of a set of genetic engineering tools, termed a synthetic biology toolbox. A toolbox comprises transformation protocols, replicating vectors, genome engineering (e.g., CRISPR/Cas9), constitutive and inducible promoter systems, and other gene expression control elements. This work validated synthetic biology toolboxes in three nitrogen-fixing soil bacteria: "
} | 234 |
35492934 | PMC9050384 | pmc | 7,750 | {
"abstract": "Functionalization of synthetic suede materials with excellent superhydrophobicity can expand their application ranges. Superhydrophobic synthetic suede was obtained by coating with polydimethylsiloxane (PDMS) and octadecyltrichlorosilane (OTS). Utilizing the synthetic suede effect of the fibrous rough structures in combination with the low surface energy micro–nano rough structure on fibers resulting from PDMS and OTS, the surface was easily turned superhydrophobic with self-cleaning properties. Abrasion tests showed that the superhydrophobic synthetic suede has excellent superhydrophobic performance after more than 2000 severe abrasion tests. This research provides a facile strategy for the preparation of practical superhydrophobic synthetic suede materials.",
"conclusion": "Conclusions Using PDMS and OTS as low surface energy substances, superhydrophobic synthetic suede was prepared by step-by-step impregnation. The contact angle (CA) and sliding angle (SA) of the as-prepared synthetic suede were 162.8° and 3.4°, respectively. In addition, the as-fabricated synthetic suede can maintain its superhydrophobicity after 2000 times abrasion, showing a strong mechanical resistance. Importantly, the superhydrophobic synthetic suede has good self-cleaning and chemical resistant properties. This present fabrication method is facile and can be used for large area production of superhydrophobic synthetic suede materials.",
"introduction": "Introduction High water repellency can be observed in nature, for example, on lotus leaves, rose petals and animal feathers. 1–5 Those materials are referred to as superhydrophobic surfaces with a water contact angle (CA) greater than 150° and water sliding angle (SA) typically lower than 10°. 6–10 They can be used in potential applications including oil–water separation, 18–20 self-cleaning, 21–23 anti-corrosion, 24–26 anti-fog 27 and others 28–30 due to their water-repellency 11–14 and self-cleaning properties. 15–17 Up to now, quite an army of superhydrophobic surfaces have been prepared on various substrates such as glass, metal, wood, sponges, and textiles. 31–34 However, little attention has been paid to the construction of superhydrophobic synthetic suede which is a classical flexible substrate widely used in the wearable field. 35–38 Synthetic suede is a kind of synthetic leather material with hydrophilicity because of the superfine nylon fibers in the substrate, which have an abundance of hydrophilic groups, 39,40 such as –OH, –CO–, and –NH– of the macromolecules in nylon, which can form hydrogen bonds with water molecules. Since there is a sea of fine fluff on the surface, the synthetic suede products are vulnerable to liquid stain or easily contaminated by blotting. Making the synthetic suede superhydrophobic and giving it self-cleaning properties could be a good way to solve these problems. Additionally, turning daily materials superhydrophobic can make life much more convenient, and prolong the life span of the materials. Since the surface of such substrates inevitably experiences all kinds of friction in practical applications, so much attention has been paid to promote the wear resistance of superhydrophobic flexible materials in recent years. 41–49 One approach to create a durability of superhydrophobic surfaces is to endow materials with the ability to regenerate the surface roughness or restore the hydrophobic components. Zhang et al. 50 reported a long lasting superhydrophobic surfaces by spraying polystyrene/SiO 2 core/shell particles as a coating skeleton and polydimethylsiloxane (PDMS) as hydrophobic interconnection. The coating exposed new roughening structures during the rubbing process, thus maintaining a suitable hierarchical roughness, favouring the superhydrophobic property of the surface. Also, the superhydrophobicity of the damaged surface could be automatically restored in 12 h at room temperature. Another approach to increase the life time of superhydrophobic materials is to use mechanically durable hydrophobic materials. Zhao et al. 51 coated three kinds of nano-silica with different particle sizes by PDMS and sprayed them on the glass substrate to prepare superhydrophobic surfaces with a water contact angle of 169.8° and rolling angle of less than 4°. The persistent superhydrophobic properties were maintained within 6 months. In addition, the coating also has good mechanical stability and remarkable self-cleaning property. Wang et al. 52 prepared transparent superhydrophobic coating by a simple method, in which PDMS was used as the intermediate layer to coat the glass, and silica nanoparticles with fluoroalkyl silanes were embedded on the surface of PDMS. The obtained coating maintained superhydrophobicity and self-cleaning property. Inspired by these works, and based on utilizing the characteristics of the fine roughening structures of suede, 53,54 here we report a facile method to fabricate wear resistant superhydrophobic synthetic suede. As shown in Fig. 1 , pristine synthetic suede was soaked in PDMS then in OTS solution to obtain superhydrophobic synthetic suede after drying. Abrasion tests showed that the superhydrophobic synthetic suede was robust to machine wearing, maintaining its superhydrophobicity even after 2000 abrasion cycles. Fig. 1 Schematic illustration of the fabrication of superhydrophobic synthetic suede.",
"discussion": "Results and discussion Morphology and composition of synthetic suede SEM images of the pristine synthetic suede, Synthetic suede–PDMS, Synthetic suede–OTS and Synthetic suede–PDMS–OTS were shown in Fig. 2(a–d) . It was found that the fiber surfaces of the pristine synthetic suede are quite smooth with a diameter of about 2.86 μm ( Fig. 2(a) ). Coating of PDMS did not cause obvious change in the morphology of the fiber surface of the synthetic suede while the diameter of the fiber increased to about 3.19 μm, which becomes thicker, as shown in Fig. 2(b) . When the sample was only treated with OTS, the surfaces of fibers became rougher than ever with some bulges ( Fig. 2(c) ). When the samples were treated with PDMS followed by OTS, the surface of the fiber became covered by much more micro–nano bulges and spiny structure, the fiber increased to about 3.50 μm as shown in Fig. 2(d) , which may be caused by the self-dehydrating and condensation of OTS on the Synthetic suede–PDMS. It should be noticed that when PDMS or OTS was used alone to modify the synthetic suede, the value of CA can reach more than 150°, the value of SA can reach more than 8°. This is because OTS and PDMS can dramatically reduce the low surface energy of the suede surface, making the microscale roughening suede with fluffy microstructure superhydrophobic. However, due to insufficient surface roughness in nanoscale, the rolling angle is relatively large, and water droplets are not easy to roll down from the surface, thus affecting the self-cleaning effect of the surface. When the samples were treated with PDMS and OTS together, the value of CA can reach 163.5° and the value of SA can reach less than 5° due to decoration of these low surface energy nanoscale spiny structures on the fiber surfaces, which not only hydrophobized the surface but also provided a higher roughening structure, enhancing the superhydrophobicity on the synthetic suede. Fig. 2 SEM images of (a) pristine synthetic suede, (b) Synthetic suede–PDMS and (c) Synthetic suede–OTS, (d) Synthetic suede–PDMS–OTS. The element distribution of the Synthetic suede–PDMS and Synthetic suede–PDMS–OTS were illustrated by EDS, as shown in Fig. 3(a–d) . The dominant elements (C, O and Si) were uniformly distributed on the Synthetic suede–PDMS and Synthetic suede–PDMS–OTS, demonstrating PDMS was eventually coated on the synthetic suede surface. There was no Cl in Synthetic suede–PDMS, while the content of Cl in Synthetic suede–PDMS–OTS was 4.20% ( Fig. 3(d) ). This might be caused by the OTS accumulation 56 on the surface of PDMS coating through self-dehydrated condensation, forming OTS condensed polyoctadecylsiloxane attached to PDMS by van der Waals forces. Due to the incomplete reaction of the self-dehydration condensation and possible residuals of chlorine from OTS, the content of the chlorine element appeared in the sample of Synthetic suede–PDMS–OTS, implying the successful coating of OTS on the synthetic suede. Fig. 3 Element mapping (a) and EDS spectrum (b) of Synthetic suede–PDMS; element mapping (c) and EDS spectrum (d) of Synthetic suede–PDMS–OTS. As shown in Fig. 4 , FTIR spectra showed the changes of the functional groups of the synthetic suede. The peak at 3304 and 1543 cm −1 and a strong absorption band at around 1642 cm −1 was related to the –NH 2 and CONH groups, which are characteristic peaks of the pristine synthetic suede 57 with 2920 cm −1 and 2853 cm −1 related to the –CH antisymmetric and symmetric stretching. After PDMS coating, a broad band centred at 1080 cm −1 and 804 cm −1 appeared, which were associated with the Si–O–Si and Si–CH 3 asymmetric bond stretching vibration peaks derived from PDMS. 58 By comparing Synthetic suede–PDMS–OTS with Synthetic suede–PDMS, it can be found that the peak appears at 1465 cm −1 related to –CH 2 bending (scissors) vibration. These results further confirmed the coating of PDMS followed by condensed OTS on the fibers. Fig. 4 FTIR spectra of (a) pristine synthetic suede, (b) Synthetic suede–PDMS, and (c) Synthetic suede–PDMS–OTS. The dosage of OTS played a crucial role in determining the surface morphology and wetting behaviour. From the SEM images in Fig. 5(a–e) , it was found that, with increasing the OTS concentration, the fiber surface became roughened and nanoscale thorn structures appeared when the OTS concentration reached 1.5% as show in Fig. 5(c) . The nanoscale thorn structures on the fiber surface in combination with the microscale woven structures as well as the fluffy characteristics of the suede made the suede have typical micro/nano structures, enhancing the superhydrophobicity of the suede material. However, Fig. 5(d and e) show that further increase of OTS concentration did not contribute much to the roughening of the fiber surface as well as the CA value ( Fig. 5(f) ). Fig. 5 SEM images of the Synthetic suede–PDMS–OTS with different mass fractions of OTS. (a) 0.5%, (b) 1.0%, (c) 1.5%, (d) 2.0% (e) 2.5%; (f) CA changes of Synthetic suede–PDMS–OTS with different mass fractions of OTS. Surface wettability As shown in Fig. 6(a) , the pristine synthetic suede was easily wetted by water drops with different colour due to the capillary effect and the abundant hydrophilic hydroxyl groups of the synthetic suede. However, the water droplets were spherical on the sample of Synthetic suede–PDMS–OTS, exhibiting remarkable superhydrophobicity. In addition, contact angle measurement showed the value of CA is 162.8° and SA is 3.4° on the as-prepared superhydrophobic synthetic suede. To further demonstrate the water-repellent property of the superhydrophobic synthetic suede, the pristine synthetic suede and Synthetic suede–PDMS–OTS samples were stuck on a glass plate and immersed into water. It was found mirror-like phenomenon was displayed obviously on the surface of the Synthetic suede–PDMS–OTS, as shown in Fig. 6(b) . This is because the low surface energy material loaded on the synthetic suede in combination with its rough structure made the material superhydrophobic and easy to trap air at the interface to prevent the material from being soaked with water. Fig. 6 The pictures of (a) water drops on pristine synthetic suede and Synthetic suede–PDMS–OTS and (b) immersion of pristine synthetic suede and Synthetic suede–PDMS–OTS stuck on glass. Durability of the superhydrophobic synthetic suede The wear of the synthetic suede usually comes from the contact friction and washing friction with each other, which belong to physical damage and will affect the hydrophobic effect of the surface. Therefore, we tested the wear resistance of the synthetic suede with a friction meter. As shown in Fig. 7(a) , the droplets can still maintain the spherical shape on the synthetic suede after 2000 cycle abrasion. Fig. 7(c) showed that the CA slightly decreased with the increase of wear cycles. The first 500 abrasion cycles only made the CA decreased from 162.8° to 159°, and the CA of the sample was still higher than 155° after 800 cycles of abrasion. Further abrasion showed that the superhydrophobicity of the synthetic suede was very robust against 1600 abrasion cycles, with CA over 150°. After severe abrasion, the CA on the surface decreased to 150° with SA increased to about ±15°. The picture and SEM image of the Synthetic suede–PDMS–OTS after 2000 cycles of abrasion were displayed in Fig. 7(a) and (b) . It was found that although the synthetic suede was abraded flat and got loss of some roughness of its surface, the sample is still not broken with repellency to water droplet. The durability of the superhydrophobicity of the suede might come from the combination of PDMS and OTS. PDMS as an elastomer has not only good wear resistance but also strong adhesive force to synthetic leather fibers after curing. 59 And OTS as additional low surface energy substance to PDMS, can be dehydrated and condensed to form a long-chain polymer, which can provide surface roughness and hydrophobization. 60 As the number of friction cycles increases, the surface condensed polymer from OTS is first worn away, and then the underlying PDMS. A composite coating formed by PDMS and OTS resulted in improvement of the superhydrophobicity and abrasion resistance of the suede surface. This demonstrated that the obtained Synthetic suede–PDMS–OTS has excellent mechanically resistant superhydrophobicity. Fig. 7 Pictures of the Synthetic suede–PDMS–OTS after 2000 abrasions (a) and SEM images of (b); (c) CA and SA changes of Synthetic suede–PDMS–OTS with abrasion cycles. Chemical resistance of the superhydrophobic suede material was also evaluated. The Synthetic suede–PDMS–OTS was immersed into different pH (pH = 1–13) solutions for 72 h, then rinsed by water and dried. As shown in Fig. 8 , although a slight decrease of CA after immersion in a solution at pH 13 was observed, which might be caused by some hydrolysis of the ester groups formed by dehydration condensation of OTS under strong alkaline conditions, the CAs still remained above 150°. This might be because the air layer between the solution and the superhydrophobic surface prevented the synthetic suede from corrosion by acid or alkali solution, demonstrating that the Synthetic suede–PDMS–OTS is highly stable against chemical corrosion. Fig. 8 CAs of the Synthetic suede–PDMS–OTS after immersion in different pH solutions for 3 days. Self-cleaning performance was inspired by the phenomenon of lotus leaf in nature. Water droplets can keep the spherical shape and easily take the dust away due to the superhydrophobicity. Self-cleaning properties of the pristine and modified synthetic suede were tested here. Fine gravel dyed by methyl orange was used as a marker on the pristine synthetic suede and Synthetic suede–PDMS–OTS. It can be obviously seen from Fig. 9(a–c) that when rinsing with water droplets, the dye made the pristine synthetic suede totally wetted and polluted by the dyed fine gravels. However, the dyed fine gravel on the surface of Synthetic suede–PDMS–OTS was easily removed away by droplet without dyeing of the surface, as shown in Fig. 9(d–f) . The result confirmed that the Synthetic suede–PDMS–OTS showed excellent self-cleaning property. Fig. 9 Self-cleaning test of (a–c) pristine synthetic suede, and (d–f) Synthetic suede–PDMS–OTS."
} | 3,929 |
24781324 | null | s2 | 7,751 | {
"abstract": "Cells navigate environments, communicate and build complex patterns by initiating gene expression in response to specific signals. Engineers seek to harness this capability to program cells to perform tasks or create chemicals and materials that match the complexity seen in nature. This Review describes new tools that aid the construction of genetic circuits. Circuit dynamics can be influenced by the choice of regulators and changed with expression 'tuning knobs'. We collate the failure modes encountered when assembling circuits, quantify their impact on performance and review mitigation efforts. Finally, we discuss the constraints that arise from circuits having to operate within a living cell. Collectively, better tools, well-characterized parts and a comprehensive understanding of how to compose circuits are leading to a breakthrough in the ability to program living cells for advanced applications, from living therapeutics to the atomic manufacturing of functional materials."
} | 248 |
32596434 | null | s2 | 7,752 | {
"abstract": "Soft robotics is an emerging field enabled by advances in the development of soft materials with properties commensurate to their biological counterparts, for the purpose of reproducing locomotion and other distinctive capabilities of active biological organisms. The development of soft actuators is fundamental to the advancement of soft robots and bio-inspired machines. Among the different material systems incorporated in the fabrication of soft devices, ionic hydrogel-elastomer hybrids have recently attracted vast attention due to their favorable characteristics, including their analogy with human skin. Here, we demonstrate that this hybrid material system can be 3D printed as a soft dielectric elastomer actuator (DEA) with a unimorph configuration that is capable of generating high bending motion in response to an applied electrical stimulus. We characterized the device actuation performance via applied (i) ramp-up electrical input, (ii) cyclic electrical loading, and (iii) payload masses. A maximum vertical tip displacement of 9.78 ± 2.52 mm at 5.44 kV was achieved from the tested 3D printed DEAs. Furthermore, the nonlinear actuation behavior of the unimorph DEA was successfully modeled using analytical energetic formulation and a finite element method (FEM)."
} | 320 |
36684723 | PMC9849897 | pmc | 7,753 | {
"abstract": "The use of arbuscular mycorrhizal (AM) fungi has great potential, being used as biostimulants, biofertilizers and bioprotection agents in agricultural and natural ecosystems. However, the application of AM fungal inoculants is still challenging due to the variability of results when applied in production systems. This variability is partly due to differences in symbiosis establishment. Reducing such variability and promoting symbiosis establishment is essential to improve the efficiency of the inoculants. In addition to strigolactones, flavonoids have been proposed to participate in the pre-symbiotic plant-AM fungus communication in the rhizosphere, although their role is still unclear. Here, we studied the specific function of flavonoids as signaling molecules in AM symbiosis. For that, both in vitro and in planta approaches were used to test the stimulatory effect of an array of different subclasses of flavonoids on Rhizophagus irregularis spore germination and symbiosis establishment, using physiological doses of the compounds. We show that the flavone chrysin and the flavonols quercetin and rutin were able to promote spore germination and root colonization at low doses, confirming their role as pre-symbiotic signaling molecules in AM symbiosis. The results pave the way to use these flavonoids in the formulation of AM fungal-based products to promote the symbiosis. This can improve the efficiency of commercial inoculants, and therefore, help to implement their use in sustainable agriculture.",
"introduction": "1 Introduction The growing human population requires a considerable increase in food production, leading to overexploitation of natural resources ( Godfray et al., 2010 ). Crop varieties with higher yields and greater resistance to environmental stresses and diseases are currently being developed. However, massive use of chemical fertilizers and pesticides is still required to provide essential nutrients and reduce disease damage in agricultural production systems. The use and abuse of these chemical products in agriculture have a huge environmental impact, polluting soils and aquifers and contributing to climate change, negatively affecting human health, ecosystems and species worldwide ( Tilman et al., 2002 ; Evans et al., 2019 ; Lynch et al., 2021 ). Therefore, there is an urgent need to find more sustainable and environmentally friendly alternatives to reduce the use of these harmful agrochemicals ( Geiger et al., 2010 ). One strategy that is gaining momentum is the use of beneficial microorganisms with biostimulant properties. These microorganisms can establish symbiotic associations with plants improving agroecosystems and crop production ( Tkacz and Poole, 2015 ). Among these beneficial microorganisms stand out arbuscular mycorrhizal (AM) fungi. These soil fungi belong to the phylum Glomeromycota and establish mutualistic associations with plant roots known as AM symbiosis ( Smith and Read, 2008 ). AM symbiosis is about 450 million years old, and it is established with more than 70% of land plants, including most species of agronomic and industrial interest (cereals, vegetables, fruit trees, cotton, etc.), as well as ornamental and forest species ( Barea et al., 2005 ; Brundrett and Tedersoo, 2018 ). It is characterized for the formation of specific structures within the roots of the host plant known as arbuscules ( Parniske, 2008 ). In the arbuscules takes place the nutrient exchange between the fungus and the host plant ( Bonfante and Genre, 2010 ). In addition to the arbuscules, the AM fungus develops a large network of hyphae, known as extraradical mycelium, which serves to explore larger areas of soil and constitutes the assimilative structure for mineral nutrients and water, functioning as pseudo roots ( Parniske, 2008 ). The benefits of AM symbiosis in plant nutrition and health are well known ( Barea et al., 2005 ; Wipf et al., 2019 ). However, in addition to a better nutrition, AM symbioses offer other benefits to the host plant including improved defense responses to pathogens and increased resilience to environmental stresses, such as drought and salinity ( Pozo et al., 2015 ). Despite the potential benefits of AM fungi, their application as biostimulants in agricultural settings is still challenging due to the variability of the results in production systems, which hinders their commercialization and implementation ( Tkacz and Poole, 2015 ). This variability resides mainly in three factors: a) the quality and effectiveness of the inoculants, b) the environmental conditions and c) the management techniques, especially chemical fertilization. AM fungi are obligate biotrophs, so they depend on a host plant to develop and complete their life cycle ( Parniske, 2008 ). This makes it difficult to implement the production of stable, axenic and homogeneous inoculants based on AM fungi. Spore-based inocula are available on the market, and they are easy to quantify and store, with higher homogeneity and lower risk of contamination than soil based inocula. However, spore production in vitro is costly ( Siddiqui and Kataoka, 2011 ). The establishment and functioning of AM symbiosis requires a high degree of coordination between the AM fungus and the host plant, based on precise molecular communication ( Pozo et al., 2015 ; López-Ráez et al., 2017 ). The molecular dialogue is initiated early during the pre-symbiotic phase with the production and exudation into the rhizosphere of signaling molecules by the plant, primarily strigolactones (SLs) ( López-Ráez et al., 2017 ). SLs are specifically recognized by the AM fungus present in the vicinity of the roots, stimulating spore germination, hyphal branching and exudation of fungal Myc-factors, thus facilitating the contact between the two partners and the establishment of the symbiosis ( Akiyama et al., 2005 ; Besserer et al., 2006 ; Bonfante and Genre, 2010 ). SLs are derived from carotenoids and, according to their signaling role, they are produced at very low amounts by the plant (on the order of pico- and nanomolar), according to the plant’s nutritional status ( López-Ráez et al., 2008 ; Yoneyama et al., 2012 ; Marro et al., 2022 ). In addition to signaling compounds in the rhizosphere, SLs are plant hormones regulating plant responses to nutritional stresses, especially phosphate (Pi) deficiency ( Gomez-Roldan et al., 2008 ; Umehara et al., 2008 ; Marro et al., 2022 ). In addition to SLs, other plant-derived compounds such as flavonoids have been proposed to participate in the pre-symbiotic molecular dialogue in AM symbiosis (reviewed in Hassan & Mathesius (2012) ). However, the flavonoids specific role and functioning is not clear. Flavonoids comprise a large and diverse family of ubiquitous secondary metabolites belonging to the phenylpropanoids. They play a diverse array of biological functions in plants, acting as antioxidants, pigments in flowers, fruits and vegetables, regulators of auxin transport, fertility, defense barriers against herbivores and microbial pathogens (phytoalexins), regulating root architecture and as signaling compounds in beneficial plant-microbe symbioses in the rhizosphere ( Hassan and Mathesius, 2012 ). So far, more than 10,000 different flavonoids have been characterized. According to their chemical structure, they are subcategorized into different major groups, including flavonols, anthocyanin, flavones, isoflavonoids, flavanonols, flavanones, flavanols, and chalcones ( \n Figure 1 \n ) ( Panche et al., 2016 ). Regarding their role as signaling molecules in the rhizosphere, the best-known function is associated to the Rhizobium -legume symbiosis ( Singla and Garg, 2017 ). This beneficial symbiosis is established between legumes and certain rhizobacteria, leading to the fixation of atmospheric nitrogen and providing nitrogen to the host plant under nitrogen deficiency ( Masson-Boivin and Sachs, 2018 ). The pre-symbiotic and symbiotic stages in the Rhizobium -legume symbiosis and AM symbioses are similar, and they share some of the required signaling components forming the so-called SYM pathway ( Mukherjee and Ané, 2011 ; de Bruijn, 2020 ). In the Rhizobium-legume symbiosis, the molecular dialogue during the pre-symbiotic phase is initiated with the production and exudation into the rhizosphere of certain flavonoids (isoflavonoids) by the host plant ( \n Figure 1 \n ). These isoflavonoids are involved in the recruitment of compatible rhizobia by inducing or inhibiting bacterial Nod factors ( Shaw et al., 2006 ; Mandal et al., 2010 ). Figure 1 Schematic overview of the different groups of flavonoids according to their chemical structure. The role of flavonoids in AM symbiosis is ambiguous and unclear. Initially, they were considered not important for AM establishment ( Becard et al., 1995 ). Few years later, it was shown that certain flavonoids presented activity either stimulating spore germination or root colonization ( Akiyama et al., 2002 ; Scervino et al., 2007 ; Steinkellner et al., 2007 ). However, the role of flavonoids in AM symbiosis is still controversial as positive, negative or neutral results have been described ( Vierheilig et al., 1998 ; Singla and Garg, 2017 ). This controversy may be related to the very different experimental conditions used, as they study different flavonoids, different concentrations and different fungal genotypes ( Vierheilig et al., 1998 ; Singla and Garg, 2017 ). Thus, the specific involvement and functioning of flavonoids in AM symbiosis remains unclear. We hypothesize that the exogenous application of flavonoids may enhance the effectiveness of AM inoculants by acting as signaling molecules during the pre-symbiotic phase of the AM symbiosis. Different flavonoids belonging to different subcategories and at different concentrations were tested, both in vitro and in planta , for their capacity to induce spore germination and stimulate root colonization by the AM fungus Rhizophagus irregularis (formerly Glomus intraradices ), the most widely used AM fungus in commercial products in the market. The results confirm the bioactivity of these compounds in the symbiosis and reveal that there is class specificity and their activity depends on the dose used.",
"discussion": "4 Discussion In the present study, we carried out in vitro and in planta assays to confirm their involvement in this beneficial symbiosis with the aim of testing their potential use as additives to improve commercial AM fungal-based inoculants. The in vitro assays revealed that the flavonoids chrysin, genistein, medicarpin, quercetin and rutin, belonging to different subclasses, stimulated spore germination and hyphal growth of the AM fungus R. irregularis at different concentrations ( \n Figures 2 \n , \n 3 \n ). They showed a stimulatory germination activity similar to that of the synthetic SL analogue GR24 4DO , indicating their high and specific activity. A role for the flavone chrysin in AM fungal spore germination and hyphal development was previously described, although with contradictory results. First, an inhibitory effect on Gi. margarita was reported during the pre-symbiotic phase ( Bécard et al., 1992 ; Chabot et al., 1992 ). Conversely, a stimulatory effect in the number of entry points and root colonization was later shown for Gi. margarita , Funneliformis mosseae and R. irregularis ( Scervino et al., 2007 ). Therefore, the results seem to vary depending on the fungal genotypes, experimental conditions and, probably, the concentrations used, as this is crucial when using signaling compounds. Here, a stimulatory effect of chrysin was observed at low (‘physiological’, nanomolar range) doses, suggesting that this compound can act as a plant-derived signaling molecule during AM symbiosis establishment. Our results are also consistent with the ability to stimulate spore germination and hyphal growth of the AM fungus Gi. margarita in vitro reported for certain flavonols, specially quercetin ( Bécard et al., 1992 ; Chabot et al., 1992 ; Poulin et al., 1997 ; Scervino et al., 2005b ). A role of quercetin in stimulating spore germination and hyphal growth has been reported also for other AM fungi, such as Gi. rosea ( Scervino et al., 2005b ) and Gi. gigantea ( Baptista and Siqueira, 1997 ) , F. mosseae ( Kape et al., 1993 ), Claroideoglomus etunicatum ( Tsai and Phillips, 1991 ; Bécard et al., 1992 ) , G. macrocarpum ( Tsai and Phillips, 1991 ) and R. irregularis ( Bécard et al., 1992 ; Poulin et al., 1997 ). However, these effects were always observed at high concentrations ( Vierheilig et al., 1998 ). Here, as for chrysin, we showed that quercetin is also able to stimulate fungal development at low concentrations (0.01 and 0.1 µM), supporting the role of flavonols a signaling molecules in AM symbiosis establishment. In agreement with this, a stimulatory effect in fungal development at low doses (0.01 µM) was also observed for rutin, a glycosylated derivative of quercetin. No effect in fungal development was previously described for rutin, although high concentrations of the compound were used in these experiments ( Bécard et al., 1992 ; Chabot et al., 1992 ; Scervino et al., 2007 ). Once again, the different concentrations of the flavonoids tested could explain the divergences observed, since the dose is critical when working with signaling compounds. Based on these and previous results, it is clear that certain flavonoids can stimulate AM fungal development during the pre-symbiotic phase of AM symbiosis in vitro . However, an effect in vitro does not necessarily correlate with an increased mycorrhizal colonization in planta . Remarkably, we show here the flavone chrysin, and the flavonols quercetin and rutin were also able to promote mycorrhizal colonization in tomato plants at low doses when applied in fertigation and using AM fungal spores as inoculum. This agrees with previous results in different plant species, including tomato. In tomato, the application of the flavones chrysin and luteolin, and the flavonol morin increased root colonization by different AM fungi, while other flavonols such as rutin, kaempferol and isorhamnetin showed no effect ( Scervino et al., 2007 ). Quercetin was found to be present in mycorrhizal white clover ( Trifolium repens ) roots and shown to promote mycorrhizal colonization of Gi. margarita ( Scervino et al., 2005a ). Recently, quercetin has been related with the expansion of invasive plants ( Pei et al., 2020 ; Tian et al., 2021 ; Borda et al., 2022 ). It was shown that these plants have increased levels of quercetin in their root exudates than native plants, which was associated to an enhanced mycorrhizal colonization and capacity of expansion. The authors also showed that the exogenous application of quercetin promoted AM fungal colonization of the target plants ( Pei et al., 2020 ; Tian et al., 2021 ). The results suggest that the flavonol quercetin, and probably its derivatives such as rutin, act as signaling molecules in the rhizosphere promoting the establishment of AM symbiosis, as SLs do. Likely, both SLs and flavonols might act in tandem as ‘cry for help’ host signals to attract AM fungi and prepare the plant for colonization. In agreement with this idea, Maloney et al. (2014) proposed a role of flavonols, including quercetin, in the promotion of lateral root formation, which are the preferred place for the AM fungus to colonize the host plant. The results open up the possibility of using these compounds to improve the efficiency of commercial products based on AM fungal spores. Indeed, we show here that the addition of low doses of quercetin (at nanomolar levels) promote mycorrhizal colonization by R. irregularis , the most widely AM fungus used in commercial products. Remarkably, the effect seems to be not specific, as this assay was performed using two different tomato genotypes, including a tomato variety commonly used as rootstock. Most tomato farmers can benefit of this effect since currently the vast majority of tomato production is carried out using grafted plants (Raymond, 2013). Our findings support the use of this alternative strategy in tomato production, which could be extended to other crops produced in nursery conditions. However, further assays under field conditions should be performed before its implementation in production systems. Remarkably, most mycorrhizal plants, including crops with agronomic interest, produce these flavonoids, being probably sensitive to them. Therefore, this promoting effect of AM symbiosis could be extended to other crops. Overall, we confirm here the role of flavonols in AM symbiosis and show their relevance as rhizosphere signaling molecules during the pre-symbiotic phase, promoting spore germination, hyphal development and symbiosis establishment. The increasing demand of AM fungal-based biostimulants in agriculture needs effective and efficient commercial inoculants, especially in seasonal crops. In this scenario, the addition of selected flavonoids -such as the flavone chrysin and the flavonol quercetin- at low doses has a great potential as accelerators of the pre-symbiotic phase, promoting symbiosis establishment and improving the efficiency of commercial products. The final goal of this research is the use these signaling compounds in agricultural production systems to implement the use of AMF as biostimulants, thus reducing the use of harmful agrochemicals. Remarkably, this management requires very reduced costs, which makes it achievable for most farmers. Therefore, this management has a great potential in sustainable agriculture. However, before its implementation we need first to confirm their effect in agricultural settings, as well as their effectiveness in different crops."
} | 4,464 |
35731940 | PMC9260185 | pmc | 7,754 | {
"abstract": "Abstract Microbial pangenomes vary across species; their size and structure are determined by genetic diversity within the population and by gene loss and horizontal gene transfer (HGT). Many bacteria are associated with eukaryotic hosts where the host colonization dynamics may impact bacterial genome evolution. Host-associated lifestyle has been recognized as a barrier to HGT in parentally transmitted bacteria. However, pangenome evolution of environmentally acquired symbionts remains understudied, often due to limitations in symbiont cultivation. Using high-resolution metagenomics, here we study pangenome evolution of two co-occurring endosymbionts inhabiting Bathymodiolus brooksi mussels from a single cold seep. The symbionts, sulfur-oxidizing (SOX) and methane-oxidizing (MOX) gamma-proteobacteria, are environmentally acquired at an early developmental stage and individual mussels may harbor multiple strains of each symbiont species. We found differences in the accessory gene content of both symbionts across individual mussels, which are reflected by differences in symbiont strain composition. Compared with core genes, accessory genes are enriched in genome plasticity functions. We found no evidence for recent HGT between both symbionts. A comparison between the symbiont pangenomes revealed that the MOX population is less diverged and contains fewer accessory genes, supporting that the MOX association with B. brooksi is more recent in comparison to that of SOX. Our results show that the pangenomes of both symbionts evolved mainly by vertical inheritance. We conclude that genome evolution of environmentally transmitted symbionts that associate with individual hosts over their lifetime is affected by a narrow symbiosis where the frequency of HGT is constrained.",
"introduction": "Introduction Bacterial populations can show enormous genomic diversity, which comprises nucleotide differences between homologous sequences and variation in the accessory gene content. In particular, gene content diversity is described by the species pangenome, which consists of all the genomic sequences present across individuals of a bacterial species. The core genes in a pangenome are present in each individual while the remaining genes are considered accessory ( Brockhurst et al. 2019 ). Pangenome size and structure vary across bacterial species ( Maistrenko et al. 2020 ), and pangenomic diversity is important for bacterial adaptation in environmental species, where accessory genes are often niche-specific ( Kashtan et al. 2014 ; Liao et al. 2021 ; Conrad et al. 2022 ). To understand microbial adaptation, it is thus crucial to understand the evolutionary processes that shape pangenome diversity. The main processes that give rise to microbial pangenomes are gene duplication and loss during vertical inheritance and gene acquisition via horizontal gene transfer (HGT). HGT enables the transfer of genetic material between microbial individuals that are not related by inheritance ( Hall et al. 2017 ) and is particularly relevant for the evolution of microbial pangenomes ( Treangen and Rocha 2011 ; Tria and Martin 2021 ). Some mechanisms of HGT involve the activity of mobile genetic elements (MGEs)—such as phages, plasmids, transposons, or genomic islands—for transferring genetic material between different DNA strands. Many bacterial species are known to be symbionts, that is, they are strictly or facultatively associated with eukaryotic hosts. Symbionts can have different modes of transmission; parentally (also called vertically) transmitted bacteria are transferred from adults to their progeny, while environmentally (also called horizontally) transmitted bacteria are acquired from the environment, either from a free-living population or other hosts and mixed transmission modes are also common over long evolutionary time ( Bright and Bulgheresi 2010 ; Russell 2019 ). We here prefer the terms environmental and parental symbiont transmission to clearly distinguish these processes from HGT and vertical inheritance. The host association has important implications for the adaptation of symbionts via HGT since bacterial populations that share a habitat may be able to access the habitat-specific gene pool by HGT ( Bordenstein and Reznikoff 2005 ; Newton and Bordenstein 2011 ; Polz et al. 2013 ). Indeed, several studies demonstrated that gene transfer from locally adapted populations may facilitate host colonization. For example, in plant-associated communities, MGEs enabled the adaptation of locally adapted nitrogen-fixing soil bacteria to associate with novel crops during their domestication ( Greenlon et al. 2019 ), and in sponges, diverse functions that potentially provide a selective advantage to the symbionts in that niche were acquired by HGT ( Robbins et al. 2021 ). Notably, these examples stem from environmentally transmitted symbionts that might have a wider potential for HGT compared with parentally transmitted symbionts. First, infection of a host by multiple symbionts results in a shared environment, where the chances for HGT are higher, and second, genes can potentially be acquired from environmental bacteria during the free-living stage. Indeed, very few HGT events have been reported in well-studied insect symbioses, potentially due to genetic isolation linked to the intracellular lifestyle and parental transmission ( Pinto-Carbó et al. 2016 ; López-Madrigal and Gil 2017 ; Waterworth et al. 2020 ). Previous studies showed that the evolution of endosymbiont genomes is characterized by rare HGT and fewer accessory genes compared with environmental bacteria ( Kloesges et al. 2011 ; Brockhurst et al. 2019 ). This conclusion was drawn based on a few model symbionts that have been cultivated and sequenced and that are mostly parentally transmitted. Populations of these parentally transmitted insect symbionts are characterized by a low intra-host diversity ( Guyomar et al. 2018 ). However, less is known about environmentally transmitted symbionts, where multiple strains might colonize an individual host, resulting in within-host strain diversity. Our view of microbial diversity has been revolutionized in the last 20 years by cultivation-independent approaches, such as metagenomics (e.g., Giovannoni et al. 2014 ; Castelle and Banfield 2018 ). Additionally, since metagenomics enables us to assess the variation of all organisms in a particular environment, deeply sequenced metagenomes provide adequate datasets for studying variation within microbial populations ( Denef 2019 ; Rossum et al. 2020 ). This approach revealed abundant within-host diversity of symbiont populations, e.g., in the gut microbiome of humans and bees ( Ellegaard and Engel 2019 ; Garud et al. 2019 ). The presence of strain diversity has recently been reported for environmentally transmitted symbionts that reside in Bathymodiolus mussels, where they are hosted in bacteriocytes within the gill epithelium and provide the mussel with nutrition ( Won et al. 2003 ). After an aposymbiotic larvae stage, the symbionts are acquired rapidly during the mussels’ metamorphosis from a planktonic to a benthic lifestyle, which is associated with morphological changes in the mussel’s epithelial tissue ( Franke et al. 2021 ). As adults, mussel gills constantly develop new filaments that are continuously uptaking symbionts, where older filaments of the same mussel might contribute substantially to the source of the colonization ( Wentrup et al. 2014 ; Romero Picazo et al. 2019 ). It still remains unclear if the symbionts have an active free-living stage or whether they might be dormant and only replicate within the mussel ( Ikuta et al. 2016 ; Laming et al. 2018 ). Bathymodiolus can be infected by two chemosynthetic symbiont species, sulfur-oxidizing (SOX) and methane-oxidizing (MOX) gamma-proteobacteria. Although most Bathymodiolus species harbor only a single 16S phylotype for each symbiont, metagenomic analyses of multiple Bathymodiolus species showed that different SOX and MOX strains can be present within an individual mussel ( Ansorge et al. 2019 ; Romero Picazo et al. 2019 ). An important role of MGEs and HGT in the evolution of SOX symbiont genomes from hydrothermal vents at the mid-atlantic ridge has been suggested. Compared with free-living relatives, SOX genomes were found to contain high numbers of transposases, integrases, restriction-modification systems, and toxin-related genes, where the latter are also linked to MGEs ( Sayavedra et al. 2015 ). In addition, it has been observed that co-occurring SOX strains from these sites differ in the content of genes involved in energy and nutrient utilization and viral defense mechanisms ( Ansorge et al. 2019 ). To analyze how symbiont strain diversity varies across mussels, we have analyzed closely related, nearby hosts. Nineteen Bathymodiolus brooksi mussels were sampled from a single location at a cold seep site in the northern Gulf of Mexico. Since Bathymodiolus symbionts cannot be cultured, high-resolution metagenomics data was collected by deeply sequencing homogenized gill tissue of each mussel ( Romero Picazo et al. 2019 ). We previously obtained single-sample assemblies and used a gene-based binning approach to reconstruct the core genomes of SOX and MOX ( fig. 1 ). Based on single nucleotide variants (SNVs) within the core genes, reconstruction of core-genome-wide strains revealed eleven SOX strains that group into four clades, and six MOX strains that group into two clades ( Romero Picazo et al. 2019 ). Mussel individuals may harbor one or multiple strains of each species. In particular, they can contain strains from one to three different SOX clades and from one or both MOX clades (Fig. 1 in [ Romero Picazo et al. 2019 ]). We found a high variability of the nucleotide diversity π among samples, where samples with low π tend to have a lower strain diversity as estimated using α-diversity. We furthermore investigated genetic isolation using the fixation index F ST , which revealed generally high genetic isolation between samples and also clusters of low genetic isolation where samples show highly similar SNV states and frequencies, that is, they contain similar populations for a particular symbiont. These clusters were also detected using the ecological measure β-diversity, thus, they are also similar in strain composition and frequencies, and they are not related to the genetics of these closely related hosts. Taken together, we found that the evolution of symbiont populations in individual mussels is characterized by genetic isolation, suggesting that symbionts are only taken up at an early stage in the mussel life cycle and are then confined to one mussel, resulting in geographic isolation ( Romero Picazo et al. 2019 ). Fig. 1. High-resolution metagenomic analysis workflow. Arrows represent open reading frames (ORFs) inferred from the assemblies. Orange part: Core genome analysis as described before ( Romero Picazo et al. 2019 ). The core genome analysis included the reconstruction of core genome strain sequences and estimation of population structure measured as both, β-diversity and F ST . Note that all SNVs are included in the F ST calculation, whereas β-diversity is based on the strain composition and not all SNVs can be linked to strains by DESMAN. Nonredundant gene catalog (NRGC) comprises gene cluster representatives obtained by grouping highly similar genes across samples. Red part: Pangenome analysis presented in this paper. The analysis is shown for a single species for simplicity. The network approach allows reconstructing population pangenomes. For each sample, the complete set of contigs containing genes from the species pangenome corresponds to the reconstructed MAG (metagenome-assembled genome). Using mapping, we estimated the coverages for all genes in the pangenomes in each sample, also including genes that have not been reconstructed on the contigs of that sample. The relative abundance of genes in the pangenome is then used to estimate P ST . Additionally, we reconstructed the gene content of single strains that are dominant in a sample. Here, we study the effect of the geographical isolation on SOX and MOX pangenome evolution. To this end, we analyzed the population pangenomes of the SOX and MOX strains residing in 19 mussels sampled from a single location.",
"discussion": "Discussion Here, we used high-resolution metagenomics to examine the gene content of two co-occurring symbiont species that inhabit mussels from a single geographical site. We reconstructed the population pangenomes of the sampling sites by applying a novel approach that links accessory genes to core genes across the chromosomal contigs of multiple samples. The pangenomes reconstructed here reflect the pangenomes of the population at the sampling site (i.e., not that of the entire species). The site pangenome is expected to be smaller compared to the species pangenome, since only populations from the same niche are included and no difference in niche-specific genes is expected. We thus expect that the SOX and MOX species pangenomes are larger with a larger proportion of accessory genomes than the site pangenomes reported here. The reconstruction of pangenomes from short-read metagenomic sequencing data is a challenging task. When isolate genomes are available, pangenome analysis can include the mapping of metagenomes to infer the ecology of the species ( Delmont and Eren 2018 ; Utter et al. 2020 ). For species that cannot be cultivated, however, only MAGs might be available for a pangenome analysis. Although MAGs might miss genomic regions compared with isolates ( Nelson et al. 2020 ; Meziti et al. 2021 ), pangenomics based on MAGs is a widely used approach for studying the evolution of bacteria that are difficult to cultivate; recent analyses include Wolbachia sequenced with their hosts ( Scholz et al. 2020 ), Sulfurovum from deep-sea hydrothermal vents ( Moulana et al. 2020 ), Thaumarchaeota from river sediments ( Sheridan et al. 2020 ), and Chlamydiae present in various habitats covered by the Earth Microbiome initiative ( Köstlbacher et al. 2021 ). It is particularly difficult to study genome rearrangements and to infer unlinked MGEs such as plasmids from short-read metagenomes ( Maguire et al. 2020 ; Nelson et al. 2020 ). Additional approaches such as isolation and sequencing, methylation patterns from long reads ( Beaulaurier et al. 2018 ), or linking DNA by Hi-C ( Yaffe and Relman 2020 ) would be necessary to resolve MGEs with confidence and to link them to their host. The inferred pangenomes reported here thus comprise the core and accessory genes located on the symbiont chromosomes. We use several steps to improve the accuracy of the pangenome reconstruction from metagenomes ( fig. 1 ). First, we employ co-abundance binning to reconstruct core genomes and subsequently a network approach to include all accessory genes that are linked on contigs to core genes or accessory genes from other samples. Second, we estimate the coverages for all genes in the pangenomes in each sample, also including genes that have not been reconstructed on the contigs of that sample. Third, strain content is highly variable across samples, and we observed several samples that were dominated by one strain only. These strains show very good assembly statistics (e.g., SOX N50 above 30,000, supplementary table S2, Supplementary Material online) and have been used to infer the strains’ gene contents. We thus conclude that the reconstructed MAGs and the reconstructed pangenomes are of high quality. Finally, our analysis focused especially on the genes that can be assigned to strain clades; these have been maintained over long time scales and might have an evolutionary relevance. However, we might have missed low-frequency strains in our analysis and accessory genes that are only present in samples with a high strain diversity might be missing from the assemblies. However these missing genes are not yet fixed in a strain clade and might thus not be relevant for adaptation. We thus conclude that the potentially missing genes are transient and belong to the cloud genome, that is, they are rare or nearly unique genes ( Koonin and Wolf 2008 ). Whereas the SOX pangenome has less genes than that of MOX, the former has a higher fraction of accessory genes. The large SOX accessory genome is consistent with the recent finding that gene content variation among coexisting thiotrophic bacteria is common ( Ansorge et al. 2020 ). We found that functions associated with genome plasticity are consistently present in the accessory genomes of both species. Among them are genes encoding for defense mechanisms, in particular, genes related to restriction-modification systems. In addition to defense, restriction-modification systems can also function as MGEs. They have, for example, been shown to be involved in genome rearrangements in termite gut symbionts and in gene birth and death in the human gut bacterium Helicobacter pylori ( Furuta et al. 2011 ; Zheng et al. 2016 ). We find that MGEs, such as transposons, are more prevalent in MOX, which also has a high proportion of them in the core genome. Furthermore, MOX has a higher fraction of genes related to cell motility and signal transduction, which can be found in the core and accessory genome. Notably, both these functional categories have been found to be underrepresented in intracellular compared with free-living bacteria ( Merhej et al. 2009 ; Lo et al. 2016 ); thus, they are more relevant for a free-living lifestyle. This suggests that MOX still pursues an active free-living life stage or that the association of MOX with Bathymodiolus is recent and ancient genes can still be found in the genome. Mussel phylogenies indeed support that the association with MOX is younger than that with SOX, where the clade comprising B. brooksi evolved from an ancestor with only the SOX symbiont about 10 million years ago ( Lorion et al. 2013 ). Nevertheless, co-speciation of hosts and symbionts is rare in that system ( Won et al. 2008 ) and it can thus not be ruled out that SOX and MOX symbiont populations have been replaced multiple times during mussel evolution. Differences in pangenome size can also be caused by different rates of HGT. Nevertheless, we conclude that HGT is rare between species and between strain clades of both species ( fig. 6 ). This conclusion is supported by the observations 1) that gene content differences between mussel individuals reflect the differences in strain composition, 2) that the accessory genome is less diverged than the core genome for both symbionts, and 3) that the homologs between SOX and MOX are not highly similar. Observations 1) and 2) support that accessory genes have been acquired at most once from another lineage and then evolved by descent with modification within the symbiont lineages. Gene loss or recombination within the strains might also contribute to accessory gene evolution, whereas multiple transfer from distantly related lineages or HGT between different strain clades is rare. Since additionally the reconstruction of homologs between SOX and MOX also did not reveal signals of recent gene transfer between the species (observation 3), we conclude that also HGT between the two species is rare. Notably, when comparing SOX from different Bathymodiolus host species to its closest free-living species, a higher fraction of genes that potentially originated by HGT has been previously found ( Sayavedra et al. 2015 ). It is thus tempting to speculate that HGT was very important in early SOX adaptation, where extensive acquisition by HGT happened during evolution of SOX symbionts from free-living bacteria. Our analysis is restricted to events within the SOX population at a particular site and is thus focused on recent evolution of symbiont strains that are already adapted to the symbiotic environment. Furthermore, we mainly focus on HGT within the mussel environment ( fig. 6 ), whereas HGT with other community members is an exciting direction for future research. SOX belongs to the Thioglobaceae family, which contains many endosymbiotic bacteria that are related to the free-living SUP05 ( Ansorge et al. 2020 ). MOX belongs to the Methyloprofundus clade, also containing free-living members ( Hirayama et al. 2022 ), which opens the possibility of HGT between closely related species during the free-living phase. Fig. 6. Routes of HGT studied here. Notably, our conclusion that HGT is rare for SOX and MOX contrasts what is known for most other bacterial species, where HGT is a major evolutionary driver ( Treangen and Rocha 2011 ; Brockhurst et al. 2019 ; Graña-Miraglia et al. 2017 ; Levade et al. 2017 ) and HGT events can even be used as epidemiological markers for bacteria with few differences in the core genome ( Mateo-Estrada et al. 2021 ; Castillo-Ramírez 2022 ). We here developed a population genetics approach to compare measures on genetic isolation that are based either on nucleotide diversity or on gene content diversity. In the future, such approaches might also be applicable in the pan genomic epidemiology setting to integrate all information for the analysis of pathogen transmission ( Castillo-Ramírez 2022 ). Given that both symbionts can be found in the same mussel gill bacteriocyte ( Duperron et al. 2007 ), the rarity of HGT contrasts their potential ability to access the habitat-specific gene pool as described for other species ( Bordenstein and Reznikoff 2005 ; Newton and Bordenstein 2011 ; Polz et al. 2013 ). The rarity of HGT in this environment might be explained by the absence of DNA transfer mechanisms in the symbionts or by environmental properties. Regarding the latter, the intracellular environment may interfere with mechanisms that rely on the transfer of free DNA such as natural transformation, since the DNA could be quickly degraded by the prevalent mussel digestive enzymes ( Ponnudurai et al. 2017 ). Likely HGT mechanisms in such an environment are such where the DNA is transferred in a packaged manner (such as in phages, gene transfer agents, or outer membrane vesicles) or transferred in direct contact between donor and recipient (as in conjugation). However, these mechanisms are most likely to transfer genes between symbionts within a single bacteriocyte only. We thus conclude that potential gene transfer events rarely establish in the population and that the host association results in genetically isolated subpopulations where HGT is limited. This is consistent with the observation that endosymbionts are rarely connected in gene transfer networks ( Popa et al. 2011 ). Symbioses are traditionally distinguished by being open (symbionts are environmentally acquired, resulting in frequent HGT and average or high GC content), closed (symbionts are parentally transmitted, resulting in the absence of HGT, low GC content, and genome degradation), or mixed (symbionts are mainly parentally transmitted, but occasional environmental acquisition occurs) ( Perreau and Moran 2022 ). Here, we observe that the Bathymodiolus symbionts are environmentally transmitted, do not engage frequently in HGT, and have an average GC content of 38%. Thus this symbiosis does not fit into the traditional division. Environmental transmission is frequent in marine symbioses, potentially due to the opportunity to uptake locally adapted symbiont strains ( Breusing et al. 2022 ). The uptake is often restricted to an early developmental stage ( Franke et al. 2021 ), where the innate immune system has an important role in establishing symbioses in invertebrates, resulting in highly specific symbiont acquisition by the host from the environment ( Nyholm and Graf 2012 ). Additionally, the long-term host association results in low strain diversity within one host, as also observed for the marine bacterium Vibrio fischeri colonizing the squid light organ ( Bongrand et al. 2022 ). We conclude that these symbioses are not open, but rather restricted, thus we term them narrow symbioses. In narrow symbioses, the tight host association leads to genetically isolated subpopulations with low frequencies of gene transfer within the host environment, where low rates of recombination can rescue the genomes from extensive degradation ( Russell et al. 2020 ). Pangenome evolution differs substantially between open and narrow symbioses, where gene content evolution of the latter symbionts is mainly driven by differential gene loss and HGT happens only occasionally."
} | 6,164 |
19807088 | null | s2 | 7,755 | {
"abstract": "Molecular self-assembly using DNA as a structural building block has proven to be an efficient route to the construction of nanoscale objects and arrays of increasing complexity. Using the remarkable \"scaffolded DNA origami\" strategy, Rothemund demonstrated that a long single-stranded DNA from a viral genome (M13) can be folded into a variety of custom two-dimensional (2D) shapes using hundreds of short synthetic DNA molecules as staple strands. More recently, we generalized a strategy to build custom-shaped, three-dimensional (3D) objects formed as pleated layers of helices constrained to a honeycomb lattice, with precisely controlled dimensions ranging from 10 to 100 nm. Here we describe a more compact design for 3D origami, with layers of helices packed on a square lattice, that can be folded successfully into structures of designed dimensions in a one-step annealing process, despite the increased density of DNA helices. A square lattice provides a more natural framework for designing rectangular structures, the option for a more densely packed architecture, and the ability to create surfaces that are more flat than is possible with the honeycomb lattice. Thus enabling the design and construction of custom 3D shapes from helices packed on a square lattice provides a general foundational advance for increasing the versatility and scope of DNA nanotechnology."
} | 345 |
21611170 | PMC3097188 | pmc | 7,756 | {
"abstract": "Species richness is the most commonly used but controversial biodiversity metric in studies on aspects of community stability such as structural composition or productivity. The apparent ambiguity of theoretical and experimental findings may in part be due to experimental shortcomings and/or heterogeneity of scales and methods in earlier studies. This has led to an urgent call for improved and more realistic experiments. In a series of experiments replicated at a global scale we translocated several hundred marine hard bottom communities to new environments simulating a rapid but moderate environmental change. Subsequently, we measured their rate of compositional change (re-structuring) which in the great majority of cases represented a compositional convergence towards local communities. Re-structuring is driven by mortality of community components (original species) and establishment of new species in the changed environmental context. The rate of this re-structuring was then related to various system properties. We show that availability of free substratum relates negatively while taxon richness relates positively to structural persistence (i.e., no or slow re-structuring). Thus, when faced with environmental change, taxon-rich communities retain their original composition longer than taxon-poor communities. The effect of taxon richness, however, interacts with another aspect of diversity, functional richness. Indeed, taxon richness relates positively to persistence in functionally depauperate communities, but not in functionally diverse communities. The interaction between taxonomic and functional diversity with regard to the behaviour of communities exposed to environmental stress may help understand some of the seemingly contrasting findings of past research.",
"introduction": "Introduction While the concern about the consequences of taxon loss has spurred a burst of studies on the relation between diversity, both as driver and as response, with ecosystem functioning and compositional stability (reviewed by [1] , [2] ), a general agreement on the magnitude and even the direction of this role has not yet been reached [1] , [3] – [5] . This is particularly true for marine ecology which is lagging behind terrestrial research on this issue [6] . Historically “biodiversity” (mostly understood as species richness) has been considered as favourable in some way or other to stability and functioning of ecosystems (reviewed by e.g. [1] , [3] , [7] ). However, in contrast to earlier views, under ecologically realistic conditions species richness has recently been shown to relate weakly, not at all, or negatively to community stability [8] , [9] . Several model approaches and a few experimental findings have postulated that species richness may even decrease stability regarding community composition [e.g. 10] . Since more than a decade the debate about the role of biodiversity at the ecosystem level is unresolved (e.g. [11] ). Likely causes for the often contradictory results are that the relation between diversity and ecosystem stability or function is highly contingent on the metric of diversity used [6] , [12] , [13] , the kind of system property investigated [5] , [14] , the response variables chosen [3] , [5] , the number of trophic levels considered [6] , the spatial scale employed [15] , the duration of the investigation (immediate response versus long term re-structuring) [16] – [18] , the experimental concept (small synthetic assemblages versus natural communities, field versus micro- or mesocosm studies) [17] , [19] , and the study areas investigated [17] , [19] , [20] . This realization has generated pressing calls not to reduce “biodiversity” to species richness [6] , [12] , to scale up spatially [6] , [12] , [15] , [21] , to include multiple trophic levels [6] , to allow sufficient time for population level responses [16] , [17] , to add observational field studies using natural communities and natural multivariate stress [2] , [6] , [14] , to consider multivariate responses [14] , and/or to clearly define stability [5] . In recent years efforts have been made to identify the common denominator for the diversity-stability relationship based on the recognition of causes for past divergent results. Much of the discussion on the discrepancies among these studies boils down to the question whether small, short, but well controlled in vitro experiments represent the real world where direct cause-effect relationships are difficult to establish because of co-varying environmental factors (e.g. [22] ). Since the quality of experiments has improved and their weaknesses are increasingly recognized, some authors think that an extrapolation to natural communities is possible [18] , [23] , [24] , while others contest this [25] – [27] . To resolve this issue, the call for more natural experiments (see above) and the request for a sound replication among ecosystems [28] or regions [23] became louder. In the investigation presented here we tried to realize the recommendations and avoid the shortcomings mentioned above. We investigated the relation between biodiversity (and unoccupied substratum) of benthic communities and their capacity to maintain their structure and composition when subjected to rapid environmental change. A major threat to ecological communities and their diversity is rapid environmental change as caused by, for instance, habitat degradation, species invasions, or shifts in marine current regimes [29] , [30] . A key question in times of rapid or gradual environmental change or fluctuations is how well a community resists or recovers from pulse stress or pressure stress with regard to either its functional or compositional properties where a compositional shift will often be accompanied by a shift in community processes (e.g. [31] ). Thus, studies on the diversity – stability relationship have used as stability metric the maintenance of either function (e.g. productivity) or structure (e.g. taxonomic composition). These two community properties differ markedly from each other [5] . Ecosystem functions may respond faster to stress and return more easily to pre-stress conditions as compared to changes in taxonomic composition which react with more inertia and are less easily reversible. For the present investigation we quantified the rate of re-structuring as a response variable of marine communities to environmental change. While this may not be identical to the classical concepts of community stability (but see [5] ), it is related to it by representing a quite permanent alteration of community properties and possibly entailing shifts in ecosystem services when lost species are not replaced by functionally equivalent ones. When a community structurally re-organizes under the influence of environmental change it is gradually replaced by another community composed of different species which cope better with the new conditions. This new community may be functionally equivalent or not to the original community. To avoid confusion in terminology, however, in the following we will employ the term persistence to describe the capacity to resist re-structuring under environmental change. In this sense a community is considered non-persistent (“unstable” sensu [5] , [14] , [16] ) when an environmental shift provokes a compositional re-organization by disappearance of sensitive species and establishment of new species, driven by direct and indirect environmental impacts at the species level, by invasion events and/or by shifts in biotic interactions [10] , [14] , [16] . Conversely, a persistent (as employed in this paper) community withstands an environmental shift with less or slower compositional change than a non-persistent community. System properties that have been suggested to contribute to community stability in various ways comprise unoccupied space [32] , functional richness [3] , [21] and - most prominently – taxonomic richness (e.g. [1] , [3] , [5] , [33] ). Since no general consensus has been found to date regarding their relative importance [34] , we decided to investigate the relationship between these three system properties and the capacity of communities to persist structurally when exposed to a pressure stress which consisted in a translocation between moderately different habitats. In order to improve the generality of the results and to take into account the warnings that artificially assembled communities may not be representative of the real world (e.g. [35] ), that the diversity-stability relation may be context-specific (e.g. [28] ), and that the relationship may vary among ecosystems (e.g. [23] ), we chose a novel approach of combining small scale, moderately controlled experiments on natural communities with large scale global replication. We test the hypothesis that compositional persistence is greater when there are less unutilised resources (as substrata for growth), higher taxonomic richness, and greater functional richness (i.e. the within-community diversity in body size, growth form, feeding mode, reproduction).",
"discussion": "Results and Discussion The translocation represented an environmental change, the impact of which decreased over time as more and more introduced taxa were replaced by resident taxa. As is typical for all natural communities, both the introduced and the resident communities continued to change in composition after the day of translocation as a result of succession, seasonality and/or stochastic events. However, averaged over all communities, introduced communities changed faster by 29% relative to the background dynamics assessed in the resident communities of the same provenance (t-test, n = 545, t = 8.9, p<0.0001). The accelerated re-structuring was driven by two processes: (i) mortality under the new conditions; and (ii) recruitment by taxa (“invasion”) belonging to the local species pool of the target site but previously not present in the introduced community. Mortality could have been caused by intolerance towards the new abiotic conditions, lack of conspecific recruits, or sensitivity towards new biotic threats such as parasites or consumers. Circumstantial evidence suggested that predation (mostly by fishes), at least in some regions, was heavier on introduced than on resident communities, but this difference was not rigorously quantified. At all sites, the re-structuring provoked a convergence of introduced communities towards local resident communities. Convergence rates varied among sites and regions between 0.1% and 13.5% of similarity (Bray-Curtis) increase per week. Communities in Malaysia and New Zealand changed rapidly, while those at other sites appeared more persistent ( Fig. S1 ). The mean regional convergence rates did not relate to the biodiversity of the region (assessed as sum of all taxa found on the panels at both regional sites, r 2 = 0.017, p = 0.76) or to the change imposed (expressed as initial dissimilarity between introduced and resident communities, Table 1 , r 2 = 0.0002, p = 0.98). In contrast, the speed of convergence related strongly to regional mean temperature (r 2 = 0.69, p = 0.007). This could merely have reflect the well known trend of metabolic rates of poikilotherms being faster and generation times being shorter in warmer regions (e.g. [49] ). This effect of temperature differences and other regional particularities injected a ‘regional noise’ into the relation between diversity and stability. To account for the regional variability, biogeographic regions and sites were included as random factors in our analysis for relationships between compositional stability and the three system properties taxon richness, functional richness and available substratum. Most of these relationships have been studied before, but rarely simultaneously in a single experiment and never with a similar generality for different biogeographic regions (but see [50] for taxon richness – ecosystem function relationships of artificially assembled communities in three regions). The linear mixed model analysis revealed significant effects of available substratum, taxon richness and the interaction between taxon and functional richness on community persistence ( Table 3 , Supplementary Table S2 ). The strength and sometimes even the sign of these relationships varied among regions ( Figs. S3 , S4 ) demonstrating the necessity to replicate at a large scale when trying to generalize about the relative importance of taxonomic and functional richness in contributing to community persistence. Across all sites and regions convergence rate and available substratum were not related significantly ( Fig. S2 ). At the site level, however, six out of 16 relationships between available substratum and convergence were significant ( Fig. S3 ), four of them positive (one site each in England, Japan, New Zealand and Tasmania) and two of them negative (one site each in Chile and Japan). Available substratum can be expected to suppress structural persistence, since it is a prerequisite for the recruitment of new colonizers [51] or for dominance shifts by lateral growth of residents. The inverse relationship, accelerated convergence on panels with less available substratum, cannot be explained at present. 10.1371/journal.pone.0019514.t003 Table 3 Effects of “Taxonomic Richness”, “Functional Richness” and available “Substratum”. \n Fixed effects \n Parameter Standard error DF t-value p-value Intercept 7.81 2.06 525 3.79 <0.001 Tax. Richness −0.92 0.18 525 −5 <0.001 Funct. Richness 0.19 0.28 525 0.65 0.51 Substratum 0.02 0.009 525 2.17 <0.05 Tax. Richness × Funct. Richness 0.07 0.03 525 2.55 <0.05 Diversity and substratum effects on the variation in the speed of convergence between transplanted and resident fouling communities. Results from linear mixed-effects analysis. The different levels of spatial replication, i.e. “Biogeographic Region” (n = 8), and “Experimental Site” (n = 16) nested in “Biogeographic Region”, were fitted as random effects. Species richness has been postulated to facilitate [32] , [52] or hinder [51] invasions – one of the convergence drivers in the present study. Species richness may also determine the response of communities to environmental change because the susceptibilities to stress of the different species composing a community may not co-vary [53] . As a consequence, when stress sensitivity varies among species within a given functional group, the risk of stress impact at the level of ecosystem service is reduced (e.g. [54] ) despite possible structural changes. In the present study, taxon richness related negatively to convergence rate (i.e. enhanced persistence) in six of the 16 sites (two Japanese sites, and one site each in England, Finland, New Zealand and Tasmania) whereas it related positively to convergence rate in only one New Zealand site ( Fig. S4 ). This enhancement of structural persistence by taxon richness must, however, be viewed with caution, since it interacts significantly with functional richness. Though the relevance of functional richness is attested by this interaction, its direct effect on compositional persistence, averaged across all levels of taxon richness, was not significant. The interaction between taxon richness and functional richness indicates that the impact of the former on structural persistence under stress (i.e. convergence rate) depends on the level of the latter. Indeed, the slope of the regression between convergence rate and taxon richness increased from negative (“enhancing persistence”) to positive (“reducing persistence”) with increasing functional richness of the experimental sites ( Fig. 1 , Spearman rank correlation, n = 16, r = 0.54, p = 0.03). In sites with lower functional richness (<4.5 functional groups per panel) convergence rates decreased with increasing taxon richness (significantly so as suggested by the confidence intervals in Fig. 2 ), while in sites with higher functional richness (≥4.5 functional groups per panel) convergence tended to accelerate with taxon richness. Thus, taxon richness enhanced community persistence under environmental change significantly more at low functional richness than at high functional richness ( Fig. 2 , t-test, df = 14, t = 3.2, p<0.01). Of the variance not explained by substratum or diversity effects, 53% could be attributed to the random factors “country” (illustrating the regional differences among experiments) and a further 2.2% to the random factor “sites within countries”. 10.1371/journal.pone.0019514.g001 Figure 1 Relation between taxonomic richness and re-structuring with increasing functional richness. Average slope (±95% CI) of the relation between convergence rate (CR) and taxon richness (TR) depicted against mean functional richness. For clarity, only site means without scatter bars are shown. 10.1371/journal.pone.0019514.g002 Figure 2 Mean relation between taxonomic richness and re-structuring at functionally poor sites and functionally rich sites. Average slopes of the relation between convergence rate (CR, box = SE, whiskers = 95% CI) and taxon richness (TR) stratified by sites with higher (”High\") versus sites with lower (”Low\") functional richness. The results of a pairwise t-test of the 2 samples are given. At present we can only speculate about the interaction between taxon richness and functional richness regarding the compositional stress resistance of benthic communities. It should be noted that the selection of functional traits is always based on expert guessing and it cannot be excluded for our approach that a different choice might have produced a stronger (or weaker) effect of functional diversity. Meanwhile, the interaction detected suggests that the different combinations of functional richness and taxon richness encountered in this study represent different positions on the continuum between complementarity and redundancy and offer some room for interpretation. The extremes of this continuum would be i) 1 species per functional group when functional richness is high relative to species richness and ii) many species in only 1 functional group when functional richness is minimal. Under scenario (i) all species are functionally different resulting in maximum functional complementarity. Under scenario (ii) all species are functionally similar resulting in minimal functional complementarity and maximal functional redundancy. Functional complementarity is thought to enhance resistance to invasion [e.g. 55] , [56] which was considered one of the drivers of convergence in our experiment. Functional complementarity is determined by the number of different functional groups present in a community and not by the number of taxa per functional group. This would explain why at elevated functional richness (high complementarity) the rate of convergence is not related to species richness ( Fig. 2 ). Redundancy, on the other hand, has long been recognized as an insurance against the impact of species loss from a community (e.g. [57] , [58] ). Species loss driven by the imposed environmental change can lead to the loss of functional groups when redundancy is low (i.e. only one species per functional group), and a reduction in functional diversity enhances the risk of invasions (see above). The loss of certain functions (e.g. UV shading or chemical defense against consumers [59] ) may accelerate the loss of further species. Loss of functional groups should be more severe when functional richness is already low from the start. This would explain why at low functional richness higher species richness (more redundancy) makes communities less vulnerable to environmental change. Indeed we observed that persistence of communities is strongly related to species richness at low functional richness and little related to species richness at high functional richness. The interplay between functional redundancy (reducing the consequences of species loss) and functional complementarity (reducing the risk of invasion) seem to explain the observed interactive effects of species and functional richness with regard to community level impacts of environmental change. Our initial hypothesis was partially confirmed. Available substratum in most sites destabilized communities as expected, however, the effect of functional richness is more indirect than expected, i.e. it modulates the strength of the stabilizing effect of taxonomic richness. The variation in responses between geographic locations illustrates the complexity of how the relationships between taxonomic and functional richness help communities persist. We conclude that the drivers of compositional persistence in marine fouling communities exposed to environmental change (i.e. one aspect of stability) are multivariate and interactive. Considering only single community properties in diversity-stability studies must forcibly produce variable results in different settings."
} | 5,261 |
39987219 | PMC11847368 | pmc | 7,757 | {
"abstract": "Background Lignocellulose is the most abundant renewable bioresource on earth, and its biodegradation and utilization would contribute to the sustainable development of the global environment. Ruminiclostridium papyrosolvens , an anaerobic, mesophilic, and cellulolytic bacterium, produces an enzymatic complex known as the cellulosome. As one of the most highly evolved species among Ruminiclostridium -type species, R. papyrosolvens is particularly relevant for understanding how cellulolytic clostridia modulate their biomass degradation mechanisms in response to diverse carbon sources. Results Our study investigates the transcriptional responses of Ruminiclostridium papyrosolvens to different carbon sources to understand its lignocellulose utilization. Using RNA-seq, we analyzed gene expression under glucose, cellobiose, xylan, cellulose, and corn stover, identifying distinct metabolic preferences and regulatory responses. We found significant gene expression changes under corn stover compared to other carbon sources, with enrichment in ABC transporters and cell growth pathways. CAZyme gene expression was regulated by TCSs, affecting sugar transporter systems. Metabolic profiling showed R. papyrosolvens produced more complex metabolites during corn stover fermentation, revealing its adaptability to various carbon sources and implications for metabolic engineering. Conclusion This study not only uncovers the intricate response mechanisms of R. papyrosolvens to lignocellulose and its hydrolysates, but it also outlines the strategy for using R. papyrosolvens as a cellulolytic chassis in genetic engineering. Supplementary Information The online version contains supplementary material available at 10.1186/s13068-025-02619-4.",
"conclusion": "Conclusion In conclusion, based on transcriptomic and metabolomics techniques, the differential expression of genes and metabolites in R. papyrosolvens under different substrate conditions was comprehensively analyzed. The important related functional genes of the lignocellulose degradation process were also analyzed. The key genes of R. papyrosolvens degrading lignocellulose model compounds are CAZymes, ABC transporters, and two-component systems. The pathways and metabolic profiling of lignocellulose degradation by R. papyrosolvens were determined. This study deepens our understanding of the composition of the lignocellulose-degrading enzymes and metabolic pathways, and laid the foundation for the future development of strategies to systematically regulate the lignocellulose biodegradation.",
"discussion": "Discussion The type species of Ruminiclostridium , such as Ruminiclostridium cellulolyticum , Ruminiclostridium josui , and R. papyrosolvens [ 3 , 4 , 39 ], are renowned for their secretion of a repertoire of carbohydrate-active enzymes, including cellulases, hemicellulases such as xylanases, pectinases, and chitinases. Notably, some of these enzymes form cellulosomes for the efficient degradation of lignocellulose biomass. R. papyrosolvens stands out as one of the most highly evolved species within the Ruminiclostridium genus [ 15 ]; thus, it is worth to analyze its mechanisms for the degradation of lignocellulose. In our study, we focused on 131 CAZyme genes that are specifically involved in the degradation of lignocellulose, which includes 57 cellulosomal genes and 74 free CAZyme genes. However, there were significant differences in their expression patterns. The cellulosomal genes predominantly exhibited high expression levels in cellulose and corn stover, whereas the genes encoding free CAZymes were found to be highly expressed in carbon sources that are more easily utilized, such as cellobiose and xylan. This pattern aligns with our previous findings in R. cellulolyticum [ 39 ], suggesting that cellulosomes may play a more crucial role in the degradation of lignocellulosic materials compared to free CAZymes. On the other hand, the enzymes with distinct functions were selectively induced to express in response to their specific substrates. For example, members of the GH94 family involved in intracellular phosphorolytic cleavage of cellodextrin and cellobiose were upregulated on cellobiose. Similarly, GH10 xylanases were found to be upregulated when xylan was present. High levels of GH5 and GH9 glucanases were observed on cellulose biomass. GH43 hemicellulases, which are responsible for the hydrolysis of a variety of different hemicelluloses, were highly expressed on corn stover. In addition, the expression patterns observed for the cip-cel and xyl-doc gene clusters in R. papyrosolvens are remarkably similar to those in R. cellulolyticum . This suggests that the regulatory mechanisms found in R. cellulolyticum , including carbon catabolite repression (CCR) [ 39 ], TCS [ 39 , 43 ], and selective RNA processing and stabilization (SRPS) [ 44 ], are likely to function in R. papyrosolvens as well. Anaerobic, mesophilic, and cellulolytic bacteria, such as Ruminiclostridium termitidis , and R. cellulolyticum , mainly employ ABC transporter systems to import sugars derived from lignocelluloses [ 39 , 45 , 46 ]. The expression of these ABC transporters is typically controlled by TCSs, which are encoded by genes located in proximity to the transporter genes themselves (Fig. 4 ). Canonical TCSs are composed of a histidine kinase (HKs) equipped with an extracellular sensor domain and a response regulator (RRs). However, three genomic loci of TCSs harbor an additional gene that encodes sugar-binding proteins (SBPs). These SBPs serve as signal collectors, amplifying the TCS’s response to extracellular sugars. As a result, these augmented systems can be classified as three-component systems, which provide an additional layer of regulation and control over the bacterial sugar uptake mechanisms. The three-component system is also found in R. cellulolyticum and Clostridium beijerinckii [ 43 , 47 ], indicating that it may play a broad role in these bacteria. Furthermore, our results revealed that TCSs not only control the expression of ABC transporters located adjacent to their genes but also exert control over ABC transporters that are situated at a considerable genomic distance. For example, the TCS encoded by P0092_RS13475-13485 simultaneously control the expression of the ABC transporters encoded by P0092_RS15915-15900 and P0092_RS13500-13490. It underscores the intricate and far-reaching nature of TCS-mediated gene regulation, which could have implications for understanding how bacteria adapt to varying environmental conditions and nutrient availability. It also highlights the potential for TCSs to serve as master regulatory switches for the expression of multiple transport systems, influencing the bacteria’s metabolic flexibility and ecological fitness. Bacteria, like R. papyrosolvens , exhibit adaptable metabolic machineries that are able to handle fluctuating environmental carbohydrate availability. The regulation of these processes is complex and appears to be controlled by a combination of CCR and operon-specific regulators [ 39 ]. First, R. papyrosolvens , like other cellulolytic clostridia such as R. cellulolyticum [ 48 ] and Clostridium thermocellum [ 49 ], shows a preference for cellobiose over glucose as a carbon source. Notably, R. papyrosolvens harbors an additional PTS for the transport of cellobiose, complementing the ABC transporter system that is a common ortholog among cellulolytic bacteria [ 5 ]. This preference suggests that cellobiose may act as a CCR trigger in these bacteria, contrasting with the role of glucose in Escherichia coli and Bacillus subtilis . Second, genes that encode metabolic pathways of hexoses and pentoses tend to cluster together, such as those involved in the pentose phosphate pathways. Meanwhile, a gene encoding a transcriptional regulator is typically found upstream or downstream of these gene clusters. These gene clusters were co-transcribed and exhibited substrate specificity, suggesting that the regulator acts as an operon-specific regulator to control the expression of these gene clusters. Third, R. papyrosolvens , known for their production of major metabolic pathway products of Ruminiclostridium species such as acetic acid, formate, ethanol, and lactic acid [ 6 , 45 , 50 ], also exhibits the ability to generate short-chain fatty acids like butyric acid by the butyrate kinase (P0092_RS03275). Thus, it can be considered as a probiotic for animal intestines, akin to Clostridium butyricum [ 51 ], due to its butyric acid production, which promotes intestinal health. Furthermore, R. papyrosolvens is capable of releasing valuable end-products from corn stover, including xylitol, vanillic acid, and ferulate [ 52 – 54 ]. This study has not only enhanced our understanding of the physiology and metabolism of R. papyrosolvens , but also laid a foundation for future metabolic engineering research."
} | 2,240 |
25415989 | null | s2 | 7,765 | {
"abstract": "Reliable real-time low-latency spike sorting with large data throughput is essential for studies of neural network dynamics and for brain-machine interfaces (BMIs), in which the stimulation of neural networks is based on the networks' most recent activity. However, the majority of existing multi-electrode spike-sorting algorithms are unsuited for processing high quantities of simultaneously recorded data. Recording from large neuronal networks using large high-density electrode sets (thousands of electrodes) imposes high demands on the data-processing hardware regarding computational complexity and data transmission bandwidth; this, in turn, entails demanding requirements in terms of chip area, memory resources and processing latency. This paper presents computational complexity optimization techniques, which facilitate the use of spike-sorting algorithms in large multi-electrode-based recording systems. The techniques are then applied to a previously published algorithm, on its own, unsuited for large electrode set recordings. Further, a real-time low-latency high-performance VLSI hardware architecture of the modified algorithm is presented, featuring a folded structure capable of processing the activity of hundreds of neurons simultaneously. The hardware is reconfigurable “on-the-fly” and adaptable to the nonstationarities of neuronal recordings. By transmitting exclusively spike time stamps and/or spike waveforms, its real-time processing offers the possibility of data bandwidth and data storage reduction."
} | 383 |
36687629 | PMC9846038 | pmc | 7,766 | {
"abstract": "Background and aims: Intercropping, a widely used planting pattern, could affect soil physicochemical properties, microbial community diversity, and further crop yields. However, its impacts on soil microbial diversity and ecosystem functioning and further soil sustainability are poorly understood. Methods: We conducted field experiments by intercropping maize with four important crops (i.e., sesame, peanut, soybean, and sweet potato), and examined soil microbial community diversity and ecosystem functioning such as microbial biomass and enzyme activities under monocropping and intercropping. We quantified their intercropping effects on microbial diversity and ecosystem functions with effect size metric Cohen d by comparing to the monocropping of maize. Results: We found that the four intercropping systems significantly increased soil aggregates in respective of the 2–0.25 mm grain size. Intercropping consistently elevated ecosystem functioning, such as soil enzyme activities of urease, phosphatase, and catalase, soil microbial biomass carbon and soil microbial biomass nitrogen. The Cohen d of bacterial richness also increased from 0.39 to 2.36, the latter of which was significant for maize/peanut intercropping. Notably, these ecosystem functions were strongly associated with the diversity of bacteria and fungi and the relative abundance of their ecological clusters identified with network analysis. Conclusion: Together, our findings indicate that intercropping generally affected soil physicochemical properties, ecosystem functions, and promoted microbial community diversity. More importantly, our findings highlight the important roles of microbial diversity of ecological clusters (that is, network modules) in maintaining ecosystem functioning after intercropping. These results will help to better understand the microbial diversity and ecosystem function in intercropping systems and guide agricultural practice.",
"conclusion": "Conclusion Our study described the differences in soil physicochemical properties, ecosystem functions, and microbial diversity between monocropping and intercropping systems, as well as their effects on the link between microbial diversity and ecosystem functions. Specifically, in terms of soil physicochemical properties, most intercropping systems significantly increases soil macroaggregates and beneficial for soil physical structure. In terms of ecosystem functions, most intercropping had significantly positive effects on urease activity, microbial biomass carbon and nitrogen. In terms of diversity, the maize/peanut intercropping significantly elevated the species richness of bacteria. We further found that rather than the whole microbial diversity, the microbial diversity of ecological clusters showed important roles in maintaining ecosystem functions. We thus expect that intercropping not only contributes to improve soil quality in agricultural ecosystems, but also provides further insights into the relationships between soil microbial diversity and ecosystem functions.",
"introduction": "Introduction Intercropping systems, also known as polyculture or mixed cropping, are the growing of different species of plants on the same field during specific periods ( Xu et al., 2020 ). It is a traditional model of farming that can be used to reduce relative inputs to achieve sustainable intensification, and improve the quality of agriculture by taking advantage of complementary species relationships ( Xu et al., 2020 ). Intercropping can provide greater ecosystem services, such as those related to climate change mitigation and ecosystem restoration ( Pierre et al., 2022 ). Intercropping is generally more productive than monocropping because it could protect soil fertility by better using resources, suppressing runoff and soil erosion, retaining water and nutrients, and reducing weed and pest infestation ( Wang et al., 2014 ; Nyawade et al., 2019a ). Compared to monocropping, intercropped soils have greater microbial biomass and respiration ( Chen et al., 2019 ), which may be due to the fact that intercropping can increase biomass as well as litter, and promote nitrogen production ( Cong et al., 2015 ). In addition, crop species differ in their ability to release, capture, or retain specific soil nutrients. If crops in an intercropping system could balance the ability to obtain different nutrients, the soil can benefit from multiple elements over time ( Güldner and Krausmann, 2017 ; Choudhary and Choudhury, 2018 ). However, few studies simultaneously consider both microbial diversity and ecosystem functions affected by intercropping, and further, how these two aspects are linked. Generally, intercropping could affect microbial diversity and ecosystem functions via its effect in soil properties ( Nyawade et al., 2019b ; Curtright and Tiemann, 2021 ). Intercropping can affect the content and distribution of soil aggregates to improve soil physicochemical properties ( Hu et al., 2022 ). For instance, soil macro-and microaggregates are higher in maize/faba bean or maize/pigeon pea intercropping systems than in the monocropping system of maize ( Garland et al., 2016 ; Tian et al., 2019 ), which could further increase soil stability ( Tong et al., 2020 ). Regarding ecosystem functions, intercropping can influence the cycling of key nutrients by increasing soil enzyme activity and microbial biomass ( Curtright and Tiemann, 2021 ). For instance, the invertase, urease, and microbial biomass carbon contents are notably higher in the Solanum tuberosum/Malus domestica intercropping systems in terraces ( Xiao et al., 2021 ). Dehydrogenase activity and microbial biomass carbon in smallholder farms could be elevated by potato/legume intercropping systems ( Nyawade et al., 2019b ). Microbial community diversity is also highly relevant to cropping systems ( Yang et al., 2016 ). For instance, tree-based intercropping system increases arbuscular mycorrhizal fungal richness and contains several taxa not present in the conventional monocropping system ( Bainard et al., 2012 ). Tobacco/peanut intercropping systems could affect soil bacterial community structure by increasing the proportions of Bacillus and Lactococcus ( Gao et al., 2019 ). Legume-based intercropping systems can increase symbiotic and non-symbiotic beneficial population to improve rhizobacterial community diversity ( Chamkhi et al., 2022 ). In addition, compared to the intercropping without legumes, legume-based intercropping are more likely to increase microbial diversity ( Wang et al., 2020 ). In general, the diversity of bacteria and fungi affects soil ecosystem processes and functions, and the abundant bacteria and fungi help retain most nutrients and drive nutrient recycling, maintain soil structure, and convert organic carbon ( Lange et al., 2014 ; Sugiyama et al., 2014 ; Lisuma et al., 2020 ; Ma et al., 2021 ). Although intercropping could have positive effects on soil properties, microbial diversity and functions, these influences remain elusive especially regarding the links between diversity and ecosystem functions. In this study, we performed field experiments with the monocropping of maize and its intercropping with another four combinations, that is, maize/sesame, maize/peanut, maize/soybean, and maize/sweet potato. We further examined soil physicochemical properties (e.g., aggregates and nutrients), ecosystem functions (e.g., microbial biomass and enzyme activities), and the diversity of bacteria and fungi in these monocropping and intercropping systems. We hypothesized that, compared to monocropping, intercropping could (1) improve soil physicochemical properties and ecosystem functions, (2) increase microbial diversity, and (3) influence the association between microbial diversity and ecosystem functions.",
"discussion": "Discussion Using “metafor” package to calculated effect size for the field experiments of cropping systems in maize fields, we found that the intercropping generally enhanced microbial diversity and ecosystem functions, and affected the linkages between these two components. Moreover, intercropping altered microbial composition and resulted in higher microbial richness and ecosystem functions such as soil nutrients, crop nutrients, enzyme activity, MBC, and MBN. These findings highlighted the importance of intercropping in maintaining microbial community diversity and ecosystem functions, and provided evidence to improve soil fertility by regulating cropping systems in the agroecosystem. These knowledges are important to highlight that intercropping could improve soil fertility to alleviate the on-going issue of soil deterioration. Intercropping significantly increased grain nitrogen, but not the yields of grain and straw. This result is in line with that of Fossati et al. (1993) , showing that grain yield is negatively correlated with grain nitrogen concentration, which could be explained by the fact that nitrogen use efficiency is associated with yield variation ( Triboi et al., 2006 ). Further explanation could be that the spacing of intercropping of crops affects the interspecific interaction and yield performance ( Li et al., 1999 ; Ren et al., 2016 ; Raza et al., 2020 ). For instance, when the rotation strip width and sowing width were increased to 14 and 6 cm in winter wheat/white clover intercropping, respectively, the interspecific interactions could be enhanced and lead to elevated grain yield and nitrogen uptake ( Thorsted et al., 2006 ). We also found that intercropping systems improved the physicochemical properties of the soils, especially in soil aggregates. Specifically, the grain size 2–0.25 mm of soil aggregates (that is, macroaggregates) was positively associated with intercropping, whereas the grain size 0.25–0.053 mm of soil aggregates (that is, microaggregates) was negatively associated with intercropping. This finding agrees with previous literature which shows that intercropping significantly increases macroaggregates (that is, >0.25 mm) compared to monocropping ( Tian et al., 2019 ). Such changes in soil aggregates can affect soil health by improving organic matter content and stability. For instance, organic matter of macroaggregates is more stable and shows higher concentrations than microaggregates ( Cambardella and Elliott, 1993 ). The changing proportion of macroaggregates and microaggregates in agricultural fields could alter soil biological activity and nutrient retention ( Xiao et al., 2021 ). We further found that microbial biomass carbon and nitrogen, urease activity, and most nutrients were significantly elevated after intercropping. These results are in agreement with a previous study of Zhou et al. (2011) , and could be explained by three non-exclusive explanations. First, the complex crop composition of intercropping increases residues, thereby enhancing soil microbial biomass ( Bichel et al., 2017 ; Sekaran et al., 2020 ). Second, intercropping affects nutrient mobilization such as increasing soil organic carbon in the inter-rooted soil, which could improve microbial biomass carbon and nitrogen ( Inal et al., 2007 ; Singh et al., 2021 ). Third, intercropping can improve urease activity via increasing soil microbial population and affecting soil phenolic allelopathic substances of crop root exudates ( Dai et al., 2012 ; Li et al., 2012 ; Xiao et al., 2012 ). Notably, the most significant increase, such as in microbial biomass carbon and nitrogen, happened for maize/soybean intercropping. This is understandable as the soybean is a legume crop with nitrogen-fixing nodules caused by the interaction of roots with a beneficial soil microorganism, Rhizobium, and could thereby provide the supply of nitrogen regardless of fertilization ( Rivest et al., 2010 ; Dyer et al., 2012 ; Wang et al., 2016 ). Furthermore, there could also be a significant difference in microbial diversity after intercropping ( Pang et al., 2021 ). For example, mulberry/alfalfa intercropping increases the richness and diversity of bacterial community by affecting the content of soil total carbon, available potassium, and phosphate ( Zhang et al., 2018 ). Morus alba/Lespedeza bicolor intercropping significantly elevates the evenness and diversity of fungal community by affecting the contents of soil total carbon, nitrogen, and phosphate ( Liu J. et al., 2022 ). Correspondingly, our results also observed that maize/peanut intercropping system can significantly increase bacterial richness. This phenomenon may result from the following two reasons. First, peanut has a nitrogen fixation effect, and thereby can influence the distribution of nitrogen in soils and further microbial composition ( Guo et al., 2020 ; Han et al., 2022 ). Second, maize/peanut intercropping can promote the population of microorganisms associated with nitrogen-fixing by affecting the structure and functions of microorganisms in rhizosphere soils ( Chen et al., 2018 ). Finally, compared to overall microbial diversity, we found a stronger relationship between ecosystem functions and the species richness or relative abundance of microbial ecological modules. For instance, each module showed varying effects on ecosystem functions, which collectively indicates the importance of microbial diversity of ecological modules in maintaining ecosystem functioning after intercropping. This is partly consistent with previous studies in other agriculture ecosystems. For instance, ecological modules based on soil microbial phylotypes in response to precipitation influence soil carbon or nitrogen mineralization rates under either the dry or wet conditions in a typical semi-arid steppe, thereby affecting ecosystem functions ( Wu et al., 2020 ). The application of organic fertilizers results in various ecological clusters of microbial species, which are relevant to essential ecological functions such as nutrient cycling and organic degrading ( Wang et al., 2022 ). Such results could be explained by the fact that compare to the whole community, soil functions may be conducted by specialized microbes which were expected to build ecological clusters to maintain vital ecosystem functions ( Harvey et al., 2017 ; Li et al., 2020 )."
} | 3,552 |
34539080 | null | s2 | 7,770 | {
"abstract": "In this work, certain aspects of the structure of the overlapping groups of neurons encoding specific signals are examined. Individual neurons are assumed to respond stochastically to input signal. Identification of a particular signal is assumed to result from the aggregate activity of a group of neurons, which we call information pathway. Conditions for definite response and for non-interference of pathways are derived. These conditions constrain the response properties of individual neurons and the allowed overlap among pathways. Under these constrains, and under the simplifying assumption that all pathways have similar structure, the information capacity of the system is derived. Furthermore, we show that there is a definite advantage in the information capacity if pathway neurons areinterspersed among the neuron assembly."
} | 209 |
32429161 | PMC7288057 | pmc | 7,771 | {
"abstract": "The liquid metal lyophobicity of a rough substrate was, in previous articles, found to be rather independent on the surface wettability. In this article, we scrutinize the impact of surface wettability of a structured (rough) surface on the liquid metal wettability and adhesion. As a model system, a structured diamond coating was synthesized and modified by air plasma. We show that surface wettability (surface free energy) does not play a prominent role for static contact angle measurements and for the liquid metal repelling properties of the diamond coating in droplet impact experiments. In contrast, roll off angles and repeated deposition experiments illustrate that the increased hydrophilicity impacts the long-term liquid metal repellency of our coating. Liquid metal adhered after around 50 deposition/removal cycles on the hydrophilic diamond coating, while no liquid metal adhesion was visible after 100 cycles on the hydrophobic diamond coating, illustrating the fundamental role for the adhesion of liquid metal. The effect of repeated deposition in conjunction with gentle applied force was employed for coating the liquid metal lyophobic (hydrophilic) diamond coating with a thin liquid metal layer. The observed effect may find application in flexible electronics and thermal management systems as a means to improve interfacing of the liquid metal with conductive non-metal coatings.",
"conclusion": "4. Summary and Conclusions At first glance, the ability to withstand adhesion of liquid metal (oxide) was found to be independent on the surface free energy. For example, liquid metal contact angles on the hydrophilic and hydrophobic rough diamond coating were virtually the same. Similarly, the initial height for overcoming the anti-adhesion effect of the structured diamond surface upon droplet impact was the same for both surfaces. However, in contrast to previous results, we show that liquid metal may adhere to the (hydrophilic) structured surfaces upon repeated deposition/removal cycles, which we assigned to the generation of small adhering LM (oxide) patches at the surface asperities of the substrate, enabled by stronger interaction of the liquid metal oxide and the hydrophilic substrate surface, as well as gentle force (compare forced wetting approaches). The difference in wetting/adhesion of the liquid metal on these two structured diamond surfaces upon repeated depositions cycles with gentle force was exploited to achieve a high contact area between the liquid metal (wetting) and the diamond substrate. Finally, we showed that the wetting of the sample is completely reversible, the liquid metal can be removed from the surface by addition of acid or base, and the diamond-coated samples, as well as the liquid metal, can be recycled. Therefore, these diamond coatings are an interesting approach for stable interfacing of liquid metal with solid metal, especially taking into account the effect observed in this research project (increased adhesion). The observed effect may find application in flexible electronics and thermal management systems as a means to improve interfacing of the liquid metal with non-metal coatings.",
"introduction": "1. Introduction Room temperature liquid metals, metals, and alloys liquid at or near room temperature have shown great promise as emerging materials and building blocks for a wide variety of applications, ranging from soft electronics to thermal interface materials [ 1 , 2 , 3 , 4 ]. The great impact of these room temperature liquid metals (for simplicity we call them in this article “liquid metals”) for soft electronics and thermal management systems is related to the atypical liquid state of the metals and alloys while possessing good electric and thermal conductivity [ 5 ]. Therefore, liquid metals are potent candidates for shaping reconfigurable and stretchable electronics, as well as adaptable electromagnetic components, such as polarizers [ 6 ] and antennas [ 7 ]. In the flexible electronics setting, liquid metals are encumbered with at least two issues, liquid metal corrosion, and poor adhesion/interfacing. First, to conduct electricity, the liquid metal needs to interface with the solid metal. However, liquid metals typically corrode solid metals, often via an alloying mechanism. Though this alloying (reactive wetting) can be employed in selective wetting/patterning methods [ 1 , 8 , 9 , 10 , 11 , 12 ], this alloying may lead to degradation of device performance or even to device failure [ 13 ]. To mitigate this liquid metal corrosion (liquid metal embrittlement), several approaches have been suggested. For example, coating of the liquid metal with conductive micro/nano particles was advocated as a means to avoid direct contact between liquid metal and solid metal, thus preventing liquid metal-induced corrosion of the solid metal [ 14 ]. On the other hand, conductive diffusion barriers were proposed to mitigate the corrosivity of the liquid metal toward the solid metal [ 15 , 16 , 17 , 18 , 19 , 20 ]. Second, a drawback of these approaches is that the liquid metal often shows poor wetting and adhesion towards the coating of the substrates. While this poor adhesion may be useful for reconfigurable electronics [ 21 ], such as antennas, it is detrimental to the performance of a device due to low contact area between the liquid metal and the coated substrate. In some recent publications, it was suggested that rendering a smooth surface hydrophilic increases the adhesion of the liquid metal (conductor), which improved the performance of liquid metal-based flexible electronic devices upon bending and stretching [ 22 , 23 , 24 ]. This effect was ascribed to enhanced adhesion due to lower surface roughness and hydrogen–bond interaction between the hydrophilic liquid metal oxide [ 25 ] and the (coated) substrate surface. However, transfer of this technique to rough surfaces, is not trivial, as the liquid metal features an oxide skin with high yield stress [ 26 ], which results in peculiar wetting and adhesion properties of the liquid metal. For example, liquid metal does not wet most surfaces and exhibits a high “non-equilibrium” (static) contact angle on these substrates. Notably, the wetting and adhesion of liquid metal is strongly dependent on the surface roughness of the substrate surface [ 21 , 27 ], as well as the liquid metal oxide [ 28 ]. Rough (superhydrophobic) surfaces are often found to exhibit strong liquid metal lyophobicity, enabling high mobility of the liquid metal on these surfaces [ 29 ]. Recently, a conductive composite polymer made of poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) and graphene oxide was proposed as a diffusion barrier [ 15 ], necessitating carbon nanofiber coating of the liquid metal to enable movement of the liquid metal on the conductive composite material, as this coating is rather smooth. However, polymer coatings suffer from poor mechanical (abrasion) stability, chemical stability, and low adhesion toward the solid support (solid metal). Other properties to change the adhesion of the liquid metal are application of force, typically in so-called forced wetting approaches for liquid metal printing [ 30 , 31 , 32 , 33 , 34 , 35 ], loading of liquid metal with particles, so-called greases [ 36 , 37 , 38 ], and so on, while for rough surfaces, the surface free energy was found to be of subordinate importance for wetting and adhesion of a substrate with liquid metal [ 29 ]. Herein, we investigate the effect of surface chemistry (surface free energy) of a structured liquid metal repellent surface on the wetting and adhesion properties of liquid metal (oxide) combined with the aid of slight force (forced wetting). To facilitate this, we employed a structured diamond coating, which shows good liquid metal lyophobicity in the as-prepared (hydrophobic) state, as well as excellent chemical stability towards the employed liquid metal (Galinstan) [ 39 ]. The diamond coating was also chosen due to the high chemical and mechanical resistance, as well as high tuneability in terms of electrical and thermal properties [ 40 , 41 ], which might be investigated in future. By oxidation with air plasma, the diamond coating becomes hydrophilic while maintaining morphology and topography. The wettability and adhesion of liquid metal toward the two diamond coatings with different surface free energy (water wettability) were scrutinized by contact angle measurements, both water contact angle and liquid metal contact angle, by roll off angle measurements of liquid metal droplets, droplet impact experiments, and repeated deposition and removal experiments. Upon gentle force and repeated deposition on the oxidized diamond coating, the liquid metal was found to adhere to the diamond coating, which was employed to achieve a high contact area between the liquid metal (oxide) and the diamond coating, while this is difficult to achieve for the hydrophobic (as-prepared) diamond coating.",
"discussion": "3. Results and Discussion 3.1. Synthesis of Hierarchical Rough Diamond Coating on Ti Alloy The synthesis of the structured non-conductive diamond film on Ti alloy is schematically illustrated in Figure 1 a and based on prior research. The synthesis procedure relies heavily on colloidal chemistry. Briefly, the deposition process comprised of two successive seeding and CVD growth processes. In the first process, unfavorable seeding parameters (i.e., taking into account Zeta potential of DNDs and charge of substrate surface) were employed to afford sparse seeding of the DNDs on the Ti alloy surface. Subsequently, diamond was grown via the HFCVD process, yielding hemispherical grown diamond on the alloy. Afterwards, a second seeding and growth cycle was executed. However, these time-favorable seeding parameters were employed, affording high seeding densities. After the second growth process, a dense diamond coating was deposited on the whole sample, yet the hemispheres remained on the substrate surface, yielding in hierarchical roughness (micro/nano roughness), as shown in Figure 1 b,c. The height and width of the hemispheres on the surface were 2.5 ± 0.2 µm and 3.5 ± 0.6 µm, respectively. The thickness of the continuous nanocrystalline diamond coating was around 200 nm (see also supporting information cross-sectional view of the diamond coating on Si, Figure S1 ). Successful coating of the Ti alloy was confirmed by Raman spectroscopy. Before coating with diamond, the Ti oxide (rutile phase) of the Ti alloy could be clearly detected, as Raman active modes with symmetries E g and A 1g at 439 cm −1 and 603 cm −1 , respectively, and the multi phonon process at 241 cm −1 were detected [ 42 ]. The existence of a (nanocrystalline) diamond coating on the Ti alloy was confirmed by Raman, as shown in Figure 1 d. The generation of diamond was evidenced by the peak at 1333 cm −1 . Furthermore, the peaks at 1350 cm −1 and 1560 cm −1 denote the D and G band of graphite, while the shoulder at 1148 cm −1 and the minuscule peak at 1460 cm −1 represent the ω 1 and ω 3 vibration modes of trans-polyacetylene (t-PA) [ 43 , 44 ]. t-PA is often formed as a byproduct during generation of nm-sized diamond crystals [ 45 ]. From the presence of these Raman bands, the generation of nanocrystalline diamond can be inferred [ 46 ]. 3.2. Wettability of the Diamond Coatings (Toward Liquid Metal) Due to the reductive reaction regime in the CVD chamber, as-made diamond coatings are hydrophobic (H-terminated) [ 47 , 48 ]. The structured diamond coating we produced features (super)hydrophobicity with water contact angles (WCAs) around 150° due to a Cassie state and the presence of air pockets. One feature of diamond coatings is the ability to change the surface water wettability (surface free energy) without degradation of the integrity, morphology, or topography of the coating. By exposition to air plasma, the water contact angle was lowered to ≤ 5°, as shown in Figure 2 . This change in WCA is attributed to the oxidation of the surface, yielding terminal groups such as –OH, =O, –O– and so on [ 47 ]. The change in surface termination also impacted the surface free energy (SFE) of the diamond coating, and the SFE determined via the van Oss method increased upon oxidation from around 50 mN/m to around 60 mN/m, while the contribution to this SFE changed substantially. The apolar Lifshitz-van der Waals γ s LW interactions were nearly the exclusive contributor for the SFE of hydrophobic diamond coatings, while polar Lewis acid-based γ s AB interactions can be detected for the oxidized diamond surface [ 47 ]. To evaluate the impact of surface free energy on the adhesion and wetting of liquid metal, liquid metal contact angles and roll-off angles were measured at first. The contact angles were determined to be 163.4 ± 2.0° and 162.5 ± 2.0° on the oxidized and H-terminated diamond coating, respectively. Thus, the liquid metal contact angle appeared to be independent on the surface free energy (surface termination), while it was impacted to some extend by surface roughness [ 29 ]. This might be the reason for the fact that, in the literature, the wetting and adhesion behavior of liquid metal was mostly attributed to surface roughness. The low impact of surface free energy on the LM contact angle might be attributed to the high yield stress [ 26 ], rendering deformation of the LM droplet upon changes in surface free energy challenging. Notably, deposited LM droplets could be removed swiftly by a tweezer or tilting the sample. Despite similar LM contact angles measured on both samples, the roll off angles differed significantly. For rolling off a LM droplet from the oxidized sample, a tilt angle of 11.3 ± 1.5° was necessary, while for the hydrophobic coating, a tilt angle of only 8.5 ± 1.0° was needed. This difference that can be ascribed to hydrophilic interactions between the hydrophilic oxide skin of the LM and the hydrophilic substrate was only possible for the oxidized diamond coating. Such hydrophilic interactions were already suggested to impact adhesion of LM [ 22 , 23 ], for example Gua et al. employed a coating based on a glue made of hydrophilic polymethacrylates (PMA) [ 24 ], which increased the adhesion of the LM toward various substrates. The obtained increased adhesion and wettability of the LM toward the surface was presumably due to hydrogen–bond interaction and lower roughness [ 24 ]. 3.3. Comparison of Adhesion Resistance In contrast to simple deposition techniques, such as contact angle and angle measurements, droplet impact measurements are more repeatable due to the fact that the former experiments rely heavily, for liquid metal (oxide), on the deposition procedure (care during deposition). Therefore, droplet impact experiments with varying impact heights were conducted (1–24 cm). The droplet velocity ( V ), which is dependent on the impact height (H) of the droplet and the droplet radius (R), can be estimated under the assumption that drag forces are negligible by Equation (1) [ 49 ].\n (1) V = 2 g ( H − R ) The droplet diameter (D) can be determined by the mass (m) and the density (ρ) of the LM by Equation (2).\n (2) D = ( 6 m n π ρ ) 1 / 3 The Weber number, a key dimensionless number, describes the ratio between inertial forces and stabilizing cohesive forces of a liquid. The Weber number (We) can be calculated by knowledge of the density, impact velocity, radius, and interfacial tension (liquid/gas) of the droplet, as shown in Equation (3) [ 50 ].\n (3) W e = ρ V 2 R γ Due to the presence of the oxide skin for gallium-based liquid metals and the accompanying change in surface tension (interfacial tension), the full extent of which is unknown at this point [ 51 ], we assume for the calculation of the We that the surface tension does not change, which might underestimate the Weber number. For low impact heights (<15 cm), the LM droplet could be readily removed and no residual LM could be detected with optical microscopy (see Figure 3 a), while at impact heights equal or bigger than 15 cm residual LM could be found on both the (super)hydrophobic and (super)hydrophilic samples. By employing Equations (1) and (2), the radius of the droplet was calculated to 0.013 cm, yielding an impact velocity of 1.708 m/s and a We number of 4.11 at an initial height of 15 cm. We ascribe this wetting to a high peak impact force, overcoming the anti-adhesion properties of the coating. Here, the impact force may be comparable to the effect of forced wetting, which is often employed for patterning techniques [ 30 , 31 , 32 , 33 , 34 , 35 ]. In forced wetting approaches, the LM is pressed with a force toward the surface, yielding in better wettability and stronger adhesion. The LM oxide adhering to the rough substrates was found to be preferentially located at the surface asperities, which can be explained by higher pressure acting at these asperities, allowing a conformal contact between the rough LM oxide surface and the rough diamond coating [ 28 ]. The presence of the LM at the asperities was proven by SEM micrographs and EDS mapping, as shown in Figure 3 e,f. Furthermore, we assumed that the adhesion is highly dependent on impact force, which in turn is dependent on impact velocity of the droplet and on the topography of the sample, while the surface free energy has only a minute impact on this. Up to this point, the hydrophilic and hydrophobic surfaces showed a similar performance to repel LM (oxide). Similarly, in the literature, only a few examples discuss the relevance of surface free energy on anti-adhesion performance and found that surface free energy does not have an impact, even though intuitively it should have [ 29 , 52 ]. To further validate the impact of surface free energy on anti-adhesion performance, the long-term resistance of the surface toward LM (oxide) adhesion was investigated by repeated deposition and removal experiments. LM was deposited carefully on both the hydrophilic and hydrophobic samples up to 100 times. For the hydrophobic sample, no residual LM could be observed in microscopy images and, after the 100 deposition and removal cycles, the substrate was still LM lyophobic. In contrast, already after a few deposition and removal cycles, LM residue was visible by microscopy on the oxidized diamond sample. After around 50 depositions and retraction cycles, the LM droplet stuck to the diamond coating (see Figure 3 g) and could not be removed with a tweezer, pipette, or by tilting the sample. We assumed that this effect was related to adhesion of micro/nano-patches of the LM oxide, induced by enhanced hydrophilic interactions, and evidenced by increases roll off angles (see above) between the oxidized LM and the hydrophilic substrate surface. These LM oxide patches were in contact with the rough surface and overcame the cohesion of the LM (oxide). As only a small ratio of LM oxide was in direct contact with a rough surface due to surface roughness and the previously discussed LM oxide roughness, the adhering LM patches denoted only a minute part of the total surface area. However, by repeated contact and retraction cycles, these patches grew and, at some point, dominated the wettability/adhesion of substrate. It should be noted that this adhesion upon repeated deposition might be an effect of exertion of force during the deposition and retraction cycles, though we were careful during this process. Therefore, we invite other researchers to investigate this phenomenon to ascertain its validity and extent, i.e., its effect on different coatings, substrates, and deposition forces. A comparison of the behavior of the liquid metal in contact with the hydrophilic and hydrophobic surface is given in Table 1 , showing at first glance similar behavior of the hydrophilic and hydrophobic diamond surface towards liquid metal wetting and adhesion. However, as detailed before, it shows different resistance toward liquid metal adhesion upon repeated liquid metal adhesion cycles, which will be exploited in the next section to enable enhanced interfacing of the liquid metal with the diamond surface. 3.4. Application of the Liquid Metal Adhesion to the Hydrophilic Diamond Coating LM is often employed as an electric conductor in flexible electronics and a thermal conductor. In these applications, the good electric and thermal transport properties of the LM were exploited. Yet, for a good performance, several prerequisites had to be met. First, the LM interfaced with a material with high thermal and/or electrical conductivity, which were typically solid metals. However, solid metals are subject to corrosion by the LM, resulting in liquid metal embrittlement. This can be overcome by employing diffusion barriers [ 15 , 16 , 17 , 18 ]. On most of these diffusion barriers, LM showed poor wettability and adhesion, which limited the electric and thermal performance, i.e., the stability and so on. To improve the performance, a high contact area between the LM and the solid sample had to be established. Here, the adhesion of the LM upon repeated deposition on an oxidized surface could be exploited to obtain excellent interfacing of the LM with the solid sample. In Figure 4 , the oxidized structured diamond coating on Ti is shown before and after the deposition of the LM by massaging the LM onto the substrate with a soft plastic syringe. An initially deposited LM droplet showed a high contact angle ( Figure 4 b). During massaging of liquid metal on the diamond surface, the liquid metal transitioned from a Cassie-Baxter state, stabilized due to trapped air and high yield stress of the liquid metal oxide, to a Wenzel state, where the liquid metal oxide was in contact with the cavities of the diamond coating. The final liquid metal (oxide) layer on the diamond coating appeared in the optical photograph to have some roughness, as shown in Figure 4 c. This roughness could be related to the fabrication procedure, which was executed with an excess of liquid metal during massaging and subsequent removal of bulk liquid metal with a syringe, leaving excess liquid metal oxide wrinkling on the surface. Interestingly, the LM could be facilely removed by acid (i.e., hydrochloric acid, sulfuric acid, etc., at a concentration of 0.1 or 1 mol/L) or base without destroying the sample morphology, as shown in Figure 4 , and even after adhering to the diamond coating for 4 days. Therefore, the removal process was anticipated to be independent on the duration of adhesion of the LM. The removal of the liquid metal was related to the removal of the oxide skin and the high interfacial tension of the “bare” liquid metal in these solutions, as published earlier [ 51 , 53 , 54 ]. Subsequently, the liquid metal could be removed facilely by tilting the sample and recycling it [ 54 ]. Afterwards, rare residual small beaded liquid metal located in the pitches between the surface protrusions was removed by gently rubbing and with subsequent rinsing with ethanol. Images during and after the cleaning steps are shown in Figure 4 d,e, illustrating the facile and complete removal of the liquid metal from the surface. Similarly, the optical and microscopy images in Figure 4 f,g illustrate the complete removal of the liquid metal from the surface. This cleaning step is enabled by the high chemical inertness, high mechanical robustness, and excellent adhesion of the diamond coating [ 40 , 41 ]. Moreover, in the view of the excellent tunability of diamond coatings, such as crystal size, electric conductivity, thickness of the coating, etc. [ 41 ], these coatings should have a great impact on liquid metal research in the future."
} | 5,929 |
36037308 | PMC9494938 | pmc | 7,772 | {
"abstract": "Organic hydrophobic\nlayers targeting sustained dropwise condensation\nare highly desirable but suffer from poor chemical and mechanical\nstability, combined with low thermal conductivity. The requirement\nof such layers to remain ultrathin to minimize their inherent thermal\nresistance competes against durability considerations. Here, we investigate\nthe long-term durability and enhanced heat-transfer performance of\nperfluorodecanethiol (PFDT) coatings compared to alternative organic\ncoatings, namely, perfluorodecyltriethoxysilane (PFDTS) and perfluorodecyl\nacrylate (PFDA), the latter fabricated with initiated chemical vapor\ndeposition (iCVD), in condensation heat transfer and under the challenging\noperating conditions of intense flow (up to 9 m s –1 ) of superheated steam (111 °C) at high pressures (1.42 bar).\nWe find that the thiol coating clearly outperforms the silane coating\nin terms of both heat transfer and durability. In addition, despite\nbeing only a monolayer, it clearly also outperforms the iCVD-fabricated\nPFDA coating in terms of durability. Remarkably, the thiol layer exhibited\ndropwise condensation for at least 63 h (>2× times more than\nthe PFDA coating, which survived for 30 h), without any visible deterioration,\nshowcasing its hydrolytic stability. The cost of thiol functionalization\nper area was also the lowest as compared to all of the other surface\nhydrophobic treatments used in this study, thus making it the most\nefficient option for practical applications on copper substrates.",
"conclusion": "Conclusions We have demonstrated the long-term durability\nof thiol coating\nas compared to silane and PFDA polymer-grafted coatings using iCVD.\nSilane-coated copper substrates immediately failed in the flow condensation\nsetup as soon as the steam flow is initiated. At 3 and 9 m s –1 steam velocity, the thiol-coated copper substrate exhibited enhancements\nof 6 and 5.8 times in HTC, respectively, compared to the reference\nCuO nanostructured superhydrophilic substrate. The PFDA-coated copper\nsample also showed similar heat-transfer performance. In the durability\ntest, however, the thiol-coated copper substrate maintained DWC for\nabout 63 h, whereas the PFDA polymer-grafted copper surface exhibited\nlocalized FWC already after 30 h. Further, the thiol-coated substrate\nproved to be more economical with a cost estimation per area 2.5 times\nless than that of the PFDA polymer grafting. Considering their superior\ndurability, cost-effectiveness, and competitive heat-transfer performance,\nthiol treatment can be an efficient strategy for industrial applications\nrelated to condensation heat transfer.",
"introduction": "Introduction Condensation of water vapor plays a vital\nrole in multiple energy\nconversion applications such as thermal management, power generation,\nrefrigeration, air conditioning, and water desalination. 1 − 4 Depending on the formation and growth of the condensate on the surface,\ncondensation can be distinguished into two different modes, i.e.,\nfilmwise condensation (FWC) and dropwise condensation (DWC). Due to\nits, up to by an order of magnitude, higher heat-transfer coefficient,\nDWC is highly desired in all heat-transfer applications. 5 − 9 Most of the condenser surfaces (copper, aluminum, steel, etc.) used\nin the industry are naturally hydrophilic. Hence, to achieve DWC on\nsuch surfaces, an additional hydrophobic coating is required. 10 − 17 However, the long-term durability of such coatings is a major bottleneck\nfor practical applications. Typically, durability can be enhanced\nby increasing the coating thickness. Since organic coatings have low\nthermal conductivity (e.g., 0.3 W m –1 K –1 for Teflon), 18 , 19 increasing the thickness will\nincrease thermal resistance and thus reduce the heat-transfer performance,\nan undesired counterproductive outcome. An ideal hydrophobic coating\nshould simultaneously achieve high droplet shedding efficiency, low\nthermal resistance, high mechanical durability, and low-cost scalable\nfabrication capability. It is obviously a great challenge to develop\na coating exhibiting all of these properties. Recent works studied\nthe durability of polymer coatings using initiated\nchemical vapor deposition (iCVD) 5 , 20 and chemical vapor\ndeposition (CVD) 21 in condensation environments.\nThe advantage of these techniques is that they can produce ultrathin\n(tens of nanometers) and conformal hydrophobic coatings, which are\nexpected to be considerably more robust compared to monolayers. However,\nsuch coatings have not been exposed for a prolonged time in a harsh\ncondensing environment (high-temperature shear flow), while their\ndurability has only been compared with one type of hydrophobic monolayer,\ni.e., a silane coating. 5 Comparing iCVD-fabricated\ncoating with existing monolayers is very critical to understanding\nthe true potential of this technique because the iCVD technique lacks\nin terms of scalability compared to silanization or thiolation and\nneeds special adaptation for practical applications, which can be\nexpensive. 5 , 22 , 23 Towards this,\nwe investigate the heat-transfer performance and durability of a 1H,1H,2H,2H-perfluorodecanethiol\n(PFDT) coating and compare it with those of 1H,1H,2H,2H-perfluorodecyltriethoxysilane\n(PFDTS) and 1H,1H,2H,2H-perfluorodecyl acrylate (PFDA) coatings, the\nlatter fabricated using iCVD, on flat copper substrates. The thiol\ncoating outperformed the silane coating in terms of both durability\nand enhancement in heat transfer, and unexpectedly outperformed the\nPFDA coating using iCVD in terms of durability. Furthermore, the thiol\ncoating is found to be the most economical hydrophobic surface treatment\ncosting ∼2.5 times less than the grafted PFDA polymer coating\nusing iCVD.",
"discussion": "Results and Discussion Fabrication of Hydrophobic Surfaces We have fabricated\nthree different surfaces with silane (PFDTS), thiol (PFDT), and grafted\npolymer (PFDA), the latter using the iCVD technique on flat copper\nsubstrates (RMS roughness of flat uncoated copper surface used = 22\nnm, see Methods ). Further, as a reference\nfor FWC, we used a copper oxide (CuO) nanostructured surface without\nany organic coating 8 , 9 (for details of fabrication, see Methods, Figure S1 ) and a clean flat copper surface. Figure S2a shows a scanning electron microscopy\nimage of the reference surface. We adopted a sessile drop method to\nmeasure surface wettability. The reference CuO nanostructured surface\nshowed superhydrophilic behavior with a static contact angle (SCA)\nof ∼0° (see Figure S2b ). For\nthe clean flat copper surface, the advancing contact angle (ACA) was\n38.3 ± 5.6°, and contact angle hysteresis (CAH) was 23.6\n± 5.3°. Both silane and thiol involve a simple dip coating\nstep (see Methods ). The static contact\nangle, advancing contact angle, and contact angle hysteresis were\nmeasured for all of the fabricated surfaces. Both silane-coated (SCA,\n114.4 ± 0.9°; ACA, 123.3 ± 0.6°; CAH, 21.3 ±\n1.2°) and thiol-coated (SCA, 113.6 ± 1.1°; ACA, 134\n± 2.7°; CAH, 27 ± 2.2°) substrates exhibited hydrophobic\nbehavior. The grafted polymer coating using iCVD was deposited in\ntwo steps (see Methods, Figure 1 a). In this process, we adopt a solvent-free approach to deposit\nin the gaseous phase an ultrathin film and graft it to the substrate\nto enhance durability. 22 , 24 Film deposition using iCVD is\nconformal, and the thickness of the coating can be precisely controlled.\nAfter some thickness optimization tests, an ultrathin layer (∼40\nnm) of PFDA was deposited on a flat copper substrate, leading to a\nhydrophobic coating (SCA, 119.6 ± 1.1°; ACA, 127 ±\n3°; CAH, 10.3 ± 3.4°). In Figure 1 b, we have plotted the advancing contact\nangle and contact angle hysteresis (inset) of water droplets for all\nof the hydrophobic (silane, thiol, and PFDA using iCVD) copper substrates.\nA lower CAH implies that dropwise condensation is enhanced, leading\nto a higher heat-transfer coefficient (HTC). However, CAH is not the\nonly parameter influencing the HTC; nucleation site density and population\ndistribution do affect HTC. 25 − 28 Therefore, it is challenging to predict the outcome\nof condensation purely on the basis of CAH. Figure 1 (a) Schematic of the\niCVD tool used to deposit the PFDA polymer\ncoating on copper substrates. (b) Advancing contact angle and contact\nangle hysteresis (inset) of water droplets on hydrophobic copper substrates. Microscale Condensation Behavior Since most of the\nheat transfer during DWC is attributed to droplets with diameters\n<100 μm, 29 − 31 we characterized the microscale condensation behavior\nof water droplets on all of the hydrophobic test substrates using\nenvironmental scanning electron microscopy (ESEM, see Methods, and Video S1 ). Figure 2 shows the selected snapshots captured at similar droplet\ndiameters on all of the test substrates. For silane-, thiol-, and\nPFDA-coated surfaces, the contact angles of the growing condensed\ndroplets varied between 40 and 120°, 32 and we observed distinct droplets showcasing their ability to manifest\nDWC. If we compare this result with the macroscopic contact angles,\nwe can conclude that all three surfaces show good hydrophobic performance.\nApparently, on the short term and without being challenged, all coatings\nappear to be good candidates for enhancing the condensation heat transfer. Figure 2 Water\ncondensation on test substrates observed using ESEM: (a)\nsilane-coated copper, (b) thiol-coated copper, and (c) PFDA-coated\ncopper using iCVD (scale bar, 100 μm; all images have the same\nmagnification). Heat-Transfer Measurement\nat High Pressures We evaluated\nthe performance of all of the test substrates under harsh conditions\nin a high-pressure flow chamber (pressure, P = 1.42\nbar) by exposing them to superheated steam at temperature, T = 111 ± 0.3 °C. The flow direction of the superheated\nsteam was vertically downward (in the direction of gravity, see Figure S3a ). The samples were tested at two steam\nvelocities, i.e., V = 3 m s –1 (laminar\nflow, , where ρ = steam\ndensity = 0.82 kg-m –3 , V = steam\nvelocity, D = hydraulic diameter, and μ = dynamic viscosity of steam = 12.62 μPa-s) 8 , and V = 9 m s –1 (transition\nflow, Re = 3900). This setup (see Figure S3b ) enables us to study the effect of vapor shear\non the hydrophobic coating present on the surface during condensation. Figure 3 shows the\nheat flux and heat-transfer coefficient (HTC) data for all of the\ntest substrates with respect to subcooling, Δ T = T – T s , where T is the steam temperature and T s is the surface temperature of the substrate. The silane-coated copper\nsubstrate failed right at the beginning of the experiment; as soon\nas the steam flow started, we observed FWC on the surface with a significant\ndecrease in HTC. Further, the subcooling on the silane-coated copper\nsubstrate was slightly lower than that on the reference CuO nanostructured\nsurface and the clean flat copper surface. This might be due to the\npresence of mixed DWC and FWC on the silane-coated copper substrate.\nHowever, it was difficult to optically confirm this during the experiment\ndue to initial fogging on the viewing window, which prevented capturing\na clear snapshot with the camera. Both the thiol-coated and PFDA-coated\ncopper substrates exhibited DWC in the flow chamber until the end\nof the experiment (∼3 to 4 h). For V = 3 and\n9 m s –1 , the thiol-coated copper substrate showed\nenhancements of ∼6 and ∼5.8 times in HTC, respectively,\nas compared to the reference substrates (CuO nanostructured surface\nand the cleaned flat copper surface) for FWC. Similarly, the PFDA-coated\ncopper substrate also manifested DWC with ∼5.1 and ∼6.6\ntimes enhancements in HTC at V = 3 and 9 m s –1 , respectively. In our experiments, both vapor shear\nand gravity play a role in the droplet slide-off events since both\nact in the same direction during condensation. Considering the experimental\nerror limits, both thiol and PFDA coatings using iCVD exhibited similar\nheat-transfer performance. The silane-coated copper, CuO nanostructured\nsurface, and the clean flat copper surface exhibited FWC with similar\nvalues of HTC. For DWC, the total thermal resistance between the substrate\nand steam is lower compared to FWC. 8 , 12 This causes\nthe T s value to be closer to the steam\ntemperature T in the case of DWC. For this reason,\nwe have a narrow range of Δ T for thiol- and\niCVD-coated copper substrates (which exhibited DWC) and the range\nof Δ T is larger for the CuO nanostructured,\nclean flat copper- and silane-coated copper substrates (which exhibited\nFWC). We also calculated the average droplet departure diameter of\nthe water droplets on the PFDA coating using iCVD and thiol coating.\nFor the PFDA coating using iCVD, the average droplet departure diameter\n( d avg ) values were 2 ± 0.4 mm and\n1.5 ± 0.2 for steam velocities V = 3 and 9 m\ns –1 , respectively. Similarly, for the thiol coating,\nthe d avg values were found to be 2.4 ±\n0.4 and 1.8 ± 0.2 mm for steam velocities V =\n3 and 9 m s –1 , respectively. Figure 3 Heat flux (a), (c) and\nheat-transfer coefficient (HTC) (b), (d)\nfor all of the substrates vs. subcooling at steam velocities of 3\nm s –1 (a and b) and 9 m s –1 (c\nand d) and steam saturation pressure of 1.42 bar. In contrast to low-pressure conditions typically\nused in condensers,\nhere, much harsher conditions (high pressures and temperatures) were\nused as an accelerated aging test, providing an indication for the\nprolonged performance of the substrates. 8 , 33 The steam\nconditions in our test chamber are designed to challenge the coatings\nin different ways. Compared to conditions in condensers found in typical\nsteam cycles, i.e., saturated steam at pressures on the order of tens\nof millibars and approximately room temperature, the pressure and\ntemperature in our chamber are much higher. In addition, steam flow\nin a narrow channel, where our test surfaces are installed, exposes\nthem to high shear stresses. For a steam flow of 3 m s –1 in our chamber, the maximum shear stress on the coating from the\nsteam flow is estimated to be at least 65 mPa, from our recent work. 8 As dropwise condensation proceeds, repeated droplet\ndeparture exposes the bare coating (as the condensate is removed)\nto these stresses by which it is worn over time until eventual failure.\nThe coating lifespan we provide here serves as a starting reference\nfor estimation of the actual lifespan, as these shear stresses vary\nin industrial conditions depending on specific operational requirements\nand condenser designs, resulting in different steam flow profiles.\nTherefore, this experiment confirms that both thiol- and PFDA-coated\ncopper substrates using iCVD can well withstand such harsh conditions\nwith an enhancement in HTC. To further test the robustness of these\ntwo coatings, we also performed longer durability tests (discussed\nnext) in the flow chamber until the substrates failed, i.e., the mode\nof condensation transitioned from DWC to FWC. Durability Test For the durability tests on the thiol-\nand PFDA-coated copper substrates, we maintained the same chamber\nconditions for a prolonged time, i.e., P = 1.42 bar, T = 111 ± 0.3 °C, and V = 3 m\ns –1 . The test was run initially for 30 h continuously\non a given substrate. Then, we shut down the steam flow (leaving the\nsubstrate inside the flow chamber) and restarted the test the next\nday, running it for 8 h before shutting down the steam flow again.\nThis 8 h cycle was repeated every day until the substrate fails. During\nthe tests, we used high-speed imaging to observe the temporal variation\nof the condensation events on the substrates (see Figure 4 a,b). Surprisingly, despite\nthe thiol thickness being at the molecular level (monolayer) 34 compared to the ∼40 nm thickness of the\niCVD coating, the former exhibited DWC for about 63 h, whereas the\nPFDA-coated copper substrate could sustain DWC for about 30 h. The\nimages in Figure 4 b\ncorresponding to 32 and 36 h of steam exposure, respectively, clearly\nshow localized FWC on the PFDA-coated copper substrate as depicted\nby the dashed line. Further, the thiol-coated copper substrate maintained\nhigher HTC ( Figure 4 c) throughout the test and survived without any visual degradation\nof its DWC for up to 63 h. Figure 4 Temporal evolution of condensation on (a) thiolated\nand (b) PFDA-coated\ncopper surfaces. The regions with the orange dashed line show the\nlocalized FWC on the PFDA-coated surface. (c) Heat-transfer coefficient\n(HTC) plotted vs. time during the durability experiment to observe\nthe trend. Whenever we shut down the steam flow at the end of the\nexperiment, there is an obvious decrease in HTC between shutdown and\nrestart. Regarding the wettability of both\nsubstrates, ACA and CAH were\nmeasured after the durability test (see Table S1 ). The increase in CAH proves that the coatings indeed failed\neventually, leading to localized FWC. The higher degree of robustness\nof thiol as compared to that of the PFDA coating can be explained\nin terms of the bonds it forms with the substrate. Prior to PFDA coating,\nwe treat the copper surface with a trichlorovinylsilane (TCVS) layer\n(see Methods). This provides anchoring points for the PFDA coating\nusing its vinyl groups (−CH 2 ). 20 The vinyl group in TCVS forms a strong bond with the vinyl\ngroup in PFDA. On the other hand, silane is known to form oxane (Si–O)\nbonds with the surface hydroxyl groups present on the substrate. However,\nthe surface oxide on copper is not hydrolytically stable. We believe\nthat the oxide dissolves when exposed to water 35 , 36 and the bond between silane and copper breaks. In the case of the\nPFDA coating, even a small defect in the PFDA coating can cause the\nunderlying surface to be exposed to water, which would cause the failure\nof TCVS and hence PFDA coating. In contrast, thiol directly\nforms a very strong covalent bond (M\n+ HS–(CH 2 ) n –R\n→ MS–(CH 2 ) n –R,\nM—Metal, HS–(CH 2 ) n –R—general formula of thiol where H is hydrogen and\nS is sulfur) 37 , 38 with the copper substrate and\ntherefore is hydrolytically more stable than silane. Thus, the higher\ndegree of robustness of thiol as compared to that of the PFDA coating\ncan be explained in terms of the bonds it forms with the substrate.\nOur result is in contrast with the literature where the grafted PFDA\npolymer using iCVD on aluminum has been claimed to be the most robust\nfor water condensation compared with silane. 5 In the same work, the authors demonstrated a successful prototype\nof iCVD coating on a copper substrate exhibiting DWC. As thiol does\nnot bind well with aluminum, we have used a copper substrate in this\nstudy. Thus, a simple thiol coating can be really promising for practical\napplications on copper substrates. In this work, the superiority of\nthiol coating over PFDA coating using iCVD has been demonstrated specifically\nfor superheated steam under extreme temperature (111 ± 0.3 °C)\nand flow speed (3 m s –1 ) at high pressure (1.42\nbar). Further, thiol can be coated on any metal used for a condenser,\nwith an additional copper layer using electroplating, 39 , 40 physical vapor deposition, 41 , 42 etc. Since copper has\nvery high thermal conductivity (401 W m –1 K –1 ), 8 such an additional\nlayer should hardly affect the overall heat-transfer efficiency. Cost Calculation We then calculated the costs of all\nof the hydrophobic coatings studied. The cost estimation includes\nthe cost of cleaning the surfaces, electricity used to operate the\ntools used in the process, and the cost of all of the chemicals used.\nThe costs of silane, thiol, and PFDA coatings for an area of 1 m 2 in US dollars (USD) are 320, 304, and 771, respectively (see\nSupplementary Information, Section S2 ).\nAdditionally, the equipment purchase cost required for the iCVD tool\n(∼USD 216250) is much higher than that of the minimal equipment\nthat the thiol functionalization requires. Therefore, thiol could\nbe used for practical applications for long-term DWC considering its\nrobustness, easy and scalable coating process, and low cost."
} | 5,022 |
29328062 | PMC5776472 | pmc | 7,773 | {
"abstract": "We report the isolation of a pinnacle-forming cyanobacterium isolated from a microbial mat covering the sediment surface at Little Salt Spring—a flooded sinkhole in Florida with a perennially microoxic and sulfidic water column. The draft genome of the isolate encodes all of the enzymatic machinery necessary for both oxygenic and anoxygenic photosynthesis, as well as genes for methylating hopanoids at the C-2 position. The physiological response of the isolate to H 2 S is complex: (i) no induction time is necessary for anoxygenic photosynthesis; (ii) rates of anoxygenic photosynthesis are regulated by both H 2 S and irradiance; (iii) O 2 production is inhibited by H 2 S concentrations as low as 1 μ M and the recovery rate of oxygenic photosynthesis is dependent on irradiance; (iv) under the optimal light conditions for oxygenic photosynthesis, rates of anoxygenic photosynthesis are nearly double those of oxygenic photosynthesis. We hypothesize that the specific adaptation mechanisms of the isolate to H 2 S emerged from a close spatial interaction with sulfate-reducing bacteria. The new isolate, Leptolyngbya sp. strain hensonii, is not closely related to other well-characterized Cyanobacteria that can perform anoxygenic photosynthesis, which further highlights the need to characterize the diversity and biogeography of metabolically versatile Cyanobacteria. The isolate will be an ideal model organism for exploring the adaptation of Cyanobacteria to sulfidic conditions.",
"introduction": "Introduction Cyanobacteria are the only chlorophototrophs that carry out oxygenic photosynthesis and presumably provided the first significant source of O 2 on early Earth. The evolution of oxygenic photosynthesis in ancient Cyanobacteria transformed Earth, ultimately providing conditions that ushered in complex multicellular life forms. However, even after the first global rise of atmospheric O 2 during the Great Oxidation Event, concentrations in ocean surface waters remained low and sulfidic conditions were common throughout much of the Proterozoic ( Canfield, 1998 ; Meyer and Kump, 2008 ), particularly in restricted basins and along productive continental margins ( Scott et al. , 2008 ; Lyons et al. , 2009 ; Poulton et al. , 2010 ; Poulton and Canfield, 2011 ). Multiple phylogenetic analyses suggest that the less complex, anoxygenic modes of photosynthesis evolved before oxygenic photosynthesis ( Blankenship, 2001 ; Xiong and Bauer, 2002 ; Sadekar et al. , 2006 ; Bryant and Liu, 2013 ). Anoxygenic phototrophs use one reaction center, which may be either type 1 or type 2, and do not evolve oxygen. Anoxygenic photosynthesis relies on a supply of reducing equivalents from reduced sulfur compounds, organic acids, hydrogen, nitrite, arsenite or Fe(II) to drive CO 2 reduction. Despite the ubiquity of H 2 O as an electron donor for oxygenic photosynthesis, observations of extant Cyanobacteria capable of performing anoxygenic photosynthesis have been documented in environments where sulfide is present in the photic zone and in a handful of pure cultures ( Cohen et al. , 1975a , 1975b ; Padan, 1979 ; de Wit and van Gemerden, 1987 ; Garcia-Pichel and Castenholz, 1990 ; Klatt et al. , 2015a ). The capability to perform anoxygenic photosynthesis has been considered a relic of cyanobacterial ancestors living before the evolution of oxygenic photosynthesis ( Oren et al. , 1977 ; Padan, 1979 ). An alternative hypothesis is that photosynthetic versatility in Cyanobacteria represents an intermediate state during the evolution and fine-tuning of oxygenic photosynthesis, in which Cyanobacteria used either H 2 O or H 2 S in sulfidic photic zones such as those present at continental margins throughout most of Earth’s history ( Hamilton et al. , 2016 ). In extant Cyanobacteria, the ability to perform anoxygenic photosynthesis, while not common, is widespread among phylogenetically diverse Cyanobacteria ( Miller and Bebout, 2004 ). To date, all characterized Cyanobacteria that perform anoxygenic photosynthesis encode a sulfide quinone oxidoreductase (SQR), which oxidizes sulfide to sulfur and transfers electrons to photosystem I (PSI) ( Arieli et al. , 1994 ; Theissen et al. , 2003 ). SQRs can also have other functions including sulfide detoxification. There is evidence that SQRs have been transferred horizontally ( Theissen et al. , 2003 ), and genes encoding multiple types of SQRs are often found within the same genome. For SQR sequences in general, there is only rough correlation between the topology of phylogenetic trees of 16S rRNA and SQR sequences from the same organism ( Pham et al. , 2008 ). The widespread taxonomic distribution of Cyanobacteria capable of performing anoxygenic photosynthesis in the absence of detectable heritability of this trait supports the emergence of this physiology multiple times through horizontal gene transfer. Still, these observations do not discern if SQR, or more specifically, anoxygenic photosynthesis, is an ancestral trait in Cyanobacteria. Regardless, little of the physiology and ecology of ancient Cyanobacteria can be gleaned from the fossil record. Today Cyanobacteria are key primary producers in laminated mats of varying morphologies. Sulfur cycling is crucially important in these mats, which also host sulfate-reducing organisms and other anoxygenic phototrophs. Similar communities dominated by Cyanobacteria are thought to have been present in ancient stromatolites ( Walter, 1976 ). Other fossil information including lipid, chlorophyll, and carotenoid biomarkers provide clues about ancient microbial community structure and physiology, but their interpretation is complicated by the diverse organisms that produce them. For instance, hopanoids methylated at the C-2 position were originally thought to be synthesized exclusively by Cyanobacteria ( Summons et al. , 1999 ). More recent studies indicate that other organisms also produce these lipids, including anoxygenic phototrophs in proteobacterial clades ( Rashby et al. , 2007 ; Welander et al. , 2010 ). The study of extant Cyanobacteria that make stromatolitic structures and lipid biomarkers under microoxic, sulfidic conditions may yield the insights necessary to interpret biosignatures in the rock record as well as to understand the physiology and ecology of ancient Cyanobacteria. Cyanobacterial isolates of known purity capable of anoxygenic photosynthesis are rare ( Cohen et al. , 1975a , 1975b ; de Wit and van Gemerden, 1987 ; Klatt et al. , 2015a ) and the regulation of photosynthetic modes appears to vary among them ( Cohen et al. , 1975a , 1975b ; Klatt et al. , 2015a ). Biochemical characterization so far has only identified a single enzyme needed for H 2 S-driven anoxygenic photosynthesis: a SQR, providing electrons to PSI via the plastoquinone pool (PQ). Recent observations suggest that the light-independent enzyme kinetics of SQR control the rates of anoxygenic photosynthesis in Cyanobacteria when the sulfide concentration is low, whereas at higher levels of sulfide, light intensity dictates the upper limit of anoxygenic photosynthesis rates ( Klatt et al. , 2015a , 2016a ). These observations are, however, complicated by the variability of specific adaptations to fluctuating sulfide concentrations and irradiance in the environment, particularly in microbial mats, and by our lack of understanding of the mechanism of sulfide inhibition of PSII ( Garcia-Pichel and Castenholz, 1990 ; Klatt et al. , 2015b ). The affinity of SQR to both H 2 S and PQ, for instance, varies substantially among the few studied Cyanobacteria ( de Wit and van Gemerden, 1987 ; Castenholz et al. , 1991 ; Griesbeck et al. , 2000 ; Klatt et al. , 2015a , 2016a ). Overall, there are many gaps in our understanding of how environmental factors interact to determine the balance of oxygenic and anoxygenic photosynthesis in metabolically versatile Cyanobacteria. Little Salt Spring (LSS), a flooded sinkhole in Florida, hosts a seasonal bloom of red pinnacle mats dominated by Cyanobacteria and green sulfur bacteria (GSB) ( Hamilton et al. , 2017 ). The water column is sulfidic and despite the abundance of Cyanobacteria in the mat, only a small increase in oxygen (0.2 μ M ) has been observed in the water column during midday ( de Beer et al. , 2017 ). The mat contains abundant hopanoids, including a significant fraction methylated at the C-2 position ( Hamilton et al. , 2017 ). Here, we report the isolation and draft genome of a metabolically versatile, pinnacle-forming cyanobacterium from LSS. We examined whether the isolate is closely related to other Cyanobacteria capable of both types of photosynthesis and if the isolate could be the source of 2-methyl hopanoids in the pinnacle mats. We used microsensors to determine if the isolate can perform both types of photosynthesis, and developed a model of photosynthetic electron transport to explore potential regulatory mechanisms.",
"discussion": "Results and discussion Isolation of Leptolyngbya sp. strain hensonii Samples of red pinnacle mat collected from the sediment–water interface in LSS were used for isolation by serial dilution. Multiple transfers of the red filaments making up the majority of the mat biomass resulted in an axenic culture of a cyanobacterium which contains chlorophyll a , phycocyanin, phycoerythrin and allophycocyanin. The isolate is red colored, filamentous ( Supplementary Figure S2 ), motile and forms pinnacles in pure culture. The 16S rRNA gene sequence of the isolate is identical to a sequence recovered from the red pinnacle mat in LSS (KP728185; Hamilton et al. , 2017 ). Based on BLASTN analyses against all non-redundant nucleotide sequences in the NCBI-NT database, the isolate is closely related to an uncultured clone from copper mine water (KF287742; 95% sequence identity) and copper mine tailings (JQ769661; 95% sequence identity). The 16S rRNA gene sequence of the isolate also shared 94% sequence identity with clones recovered from the benthic zone of an east Antarctic Lake (DQ181675, DQ181685). The phylogenetic position of the isolate was evaluated by comparing a 1293 bp fragment of the 16S rRNA gene sequence with closely related Cyanobacteria for which draft or full genomes are available ( Figure 2 ). In this analysis, the 16S rRNA sequence of the isolate formed a monophyletic branch (bootstrap value 0.91) with the most closely related sequence— Leptolyngbya sp. strain JSC1 within subsection III. Leptolyngbya sp. strain JSC1 was isolated from a ferrous iron-rich hot spring with circumneutral pH in Yellowstone National Park ( Brown et al. , 2010 ). The LSS isolate is also closely related to Geitlerinema sp. PCC 7407 and Oscillatoriales cyanobacterium UVFP2 (~94% sequence identity)( Figure 2 ). A highly enriched culture of Oscillatoriales cyanobacterium UVFP2 was obtained from Fuente Podrida, a sulfide-rich spring close to the Cabriel River in Valencia, Spain ( Camacho et al. , 2005 ), while the isolation source of Geitlerinema sp. PCC 7407 has not been published. The pinnacle-forming LSS cyanobacterium strain was named Leptolyngbya sp. strain hensonii, for its growth habit resembling the fur of Jim Henson’s famous puppets. We acknowledge that Leptolyngbya are polyphyletic; however, a systematic nomenclature for Cyanobacteria has not been published ( Komárek, 2016 ) and the description of Leptolyngbya is consistent with our isolate—long filaments with solitary or coiled clusters and fine mats. Genome features The draft genome of strain hensonii contains 77 contigs and 5 940 030 bp with an average GC content of 52.3% ( Table 1 ). The draft genome encodes 61 tRNAs and 5627 protein coding genes ( Table 1 ), including all of the conserved housekeeping genes ( Supplementary Table S1 ) and 96% of the phylum-specific marker genes identified with Phyla-AMPHORA. All of these metrics suggest that the genome is nearly complete. The genome encodes the enzymes necessary for aerobic photoautotrophic growth including Form I RuBisCO and a complete Calvin–Benson cycle; two high affinity terminal oxidases, a cytochrome c oxidase and a bd -type quinol oxidase; PSI and PSII; chlorophyll biosynthesis pathway enzymes; and a cytochrome b 6 f complex. The bd -type quinol oxidase and cytochrome c oxidase differ in their affinity for oxygen (0.35 vs 1.0 μ M ) ( Pils and Schmetter, 2001 ), and the former is postulated to be expressed under low oxygen conditions ( Hart et al. , 2005 ). Genes encoding a succinate dehydrogenase ( sdhABC ), an F-type ATPase and a NAD(P)H:quinone oxidoreductase (NDH) are also present. Thus, hensonii is expected to be capable of aerobic respiration under variable O 2 concentrations. In the environment hensonii is indeed exposed to fluctuating O 2 over a diel cycle ( de Beer et al. , 1997 ). O 2 is, however, not available from the water column in situ but is exclusively produced by oxygenic photosynthesis, with hensonii as a main source. Under in situ conditions in LSS the enzymatic machinery for aerobic respiration likely serves to maintain cellular redox balance, with the terminal oxidases, for instance, serving as electron valves for the photosynthetic electron transport reactions ( Supplementary Figure S3 ). This implies that terminal oxidases would thus never be used in the presence of the inhibitory H 2 S ( Beauchamp et al. , 1984 ; Cooper and Brown, 2008 ). The genome also encodes the enzymatic machinery necessary for assimilatory nitrate reduction and assimilatory sulfate reduction and nitrogen fixation via a Mo-dependent nitrogenase. Three enzymes integral to photosynthesis—coproporphyrinogen III oxidase, heme oxygenase and Mg-protoporphyrin IX monomethylester cyclase—require oxygen for activity. However, oxygen levels at the water depth hosting the red pinnacle mats in LSS reach only 0.2 μ M oxygen ( de Beer et al. , 2017 ). Genes encoding alternative forms of these enzymes have been observed in genomes of Cyanobacteria from diverse environments ( Panek and O’Brian, 2002 ) and, in the cyanobacterium Synechocystis sp. PCC 6803, the alternative forms of these enzymes are expressed under low oxygen conditions ( Aoki et al. , 2011 ). Consistent with environmental conditions in the natural habitat of the isolate, we found homologs of both the aerobic and anaerobic forms of these enzymes in the strain hensonii genome. Multiple psbA genes (which encode a subunit of PSII) have been observed in the genomes of sulfide-tolerant and/or sulfide-using Cyanobacteria ( Grim and Dick, 2016 ). These additional copies of psbA facilitate oxygenic photosynthesis under conditions of varying oxygen and light ( Mohamed et al. , 1993 ). In the strain hensonii draft genome, we observed the copies of the canonical oxygenic group 4 psbA ( Cardona et al. , 2015 ) as well as a group 3 psbA , which have been recovered from cyanobacterial genomic bins from a low-oxygen cyanobacterial mat in the Middle Island Sinkhole ( Voorhies et al. , 2012 ), and a group 2 psbA. Transcripts of group 2 psbA have been observed under microaerobic conditions in cultures of Synechocystis PCC 6803, Anabaena PCC 7120 and Thermosynechococcus elongatus ( Sicora et al. , 2009 ). The draft genome of Leptolyngbya sp. strain hensonii encodes a single SQR—specifically an SQR type F. SQR catalyzes the oxidation of sulfide to zero-valent sulfur and may have a physiological role in both energy transduction and sulfide detoxification. SQR has also been implicated in anoyxgenic photosynthetic activity in Cyanobacteria ( Shahak et al. , 1998 ) (see Figure 1 ). SQR sequences can be divided into seven classes (A, B, C, D, E, F or X) and a single genome can encode multiple SQR homologs ( Gregersen et al. , 2011 ). SQRA are typically found in Cyanobacteria, Proteobacteria and Aquificaceae. SQRD and SQRX form two paralogous clades—SQRD homologs are encoded by strains of GSB, Proteobacteria and Actinobacteria ( Gregerson et al. , 2011) , while SQRX homologs are encoded by GSB. No representative of SQRC has been demonstrated to oxidize sulfide. SQRB homologs are often recovered from eukaryotes, while SQRE catalyzes sulfide oxidation in the archaeon Acidianus ambivalens ( Brito et al. , 2009 ). SQRF homologs are commonly observed in the genomes of in GSB, Proteobacteria, Aquificaceae and Cyanobacteria ( Supplementary Figure S5 ); however, an SQRF from a Cyanobacteria has not been characterized. In the GSB Chlorobaculum tepidum , SQRF is important for growth at high sulfide concentration (≥4 m M ) ( Chan et al. , 2009 ; Holkenbrink et al. , 2011 ). The hensonii SQRF is only distantly related (25% sequence identity) to the Type A SQR sequences of Aphanothece halophytica and Geitlerinema sp. PCC 9228 (formerly Oscillatoria limnetica ), which have been implicated in anoxygenic photosynthesis in these isolates ( Cohen et al. , 1986 ). Red pinnacle mats collected from LSS contain elevated concentrations of hopanoids, including those methylated at the C-2 position ( Hamilton et al. , 2017 ). The isolate genome encodes homologs of two enzymes that are presumably necessary for the biosynthesis of 2-methylhopanoids—a squalene–hopene cyclase and a radical SAM methylase (HpnP) ( Welander et al. , 2010 ). The translated HpnP homolog branches with other cyanobacterial HpnP sequences ( Supplementary Figure S6 ) and is identical to the hpnP transcript recovered from LSS ( Hamilton et al. , 2017 ). These results suggest that the isolate is a source of 2-methyl hopanoids in LSS. Several lines of evidence suggest that anoxic conditions favor the production of 2-methyl hopanoids: (1) an increased abundance in Proterozoic rocks compared to Phanerozoic rocks ( Summons et al. , 1999 ); (2) increased abundance in rocks recording oceanic anoxic events in the Phanerozoic ( Knoll et al. , 2007 ; Talbot et al. , 2008 ; Cao et al. , 2009 ; Kasprak et al. , 2015 ); and (3) higher abundance of hpnP genes in environments where anoxic conditions prevail ( Ricci et al. , 2013 ). The recovery of 2-methyl hopanoids from anoxic mats in LSS is consistent with these lines of evidence, and the isolation of a cyanobacterium with the genetic machinery to synthesize 2-methyl hopanoids will facilitate future studies aimed at determining their functions in adaptation and/or metabolism. Strain hensonii performs anoxygenic photosynthesis Because the Leptolyngbya sp. strain hensonii lives under anoxic and sulfidic conditions in situ ( de Beer et al. , 2017 ) and the genome encodes at least one SQR protein, we hypothesized that the isolate could perform anoxygenic photosynthesis. Indeed, in the presence of sulfide and the PSII inhibitor DCMU, the isolate assimilated 5.7 (±0.71) μmol C mg dry weight −1 suggesting anoxgenic photosynthetic activity ( Figure 3 ). The capability to perform anoxygenic photosynthesis was confirmed by microsensor-based measurements in the absence and presence of DCMU ( Figure 4 ). In fact, microsensor-based measurements indicate that no induction time is necessary for anoxygenic photosynthesis ( Figure 4 ), even if strain hensonii had been grown aerobically before exposure to sulfide. This is in contrast to other characterized Cyanobacteria capable of performing anoxygenic photosynthesis that require ~2 h in the presence of sulfide before performing this activity ( Oren and Paden, 1978 ; Cohen et al. , 1986 ; Klatt et al. , 2015a ). Oxygenic and anoxygenic photosynthesis were never observed to occur simultaneously during our experiments because oxygenic photosynthesis was inhibited by H 2 S concentrations of 1 μ M or lower based on the detection limit of the specific H 2 S sensor used ( Figure 4 ). Regulation of anoxygenic photosynthesis In addition to oxygenic photosynthesis, several Cyanobacteria are capable of using sulfide as an electron donor, that is, performing anoxygenic photosynthesis using only PSI ( Cohen et al. , 1975a , 1975b ; de Wit and van Gemerden, 1987 ); however, the mechanism for regulating anoxygenic photosynthetic activity is different between isolates ( Cohen et al. , 1986 ; Garcia-Pichel and Castenholz, 1990 ; Klatt et al. , 2015a ; 2016a ). Below we report specific activity patterns dependent on H 2 S concentration and irradiance of strain hensonii in laboratory experiments and discuss plausible regulation mechanisms (see the SOM for additional details and discussion). Light and H 2 S concentration Anoxygenic photosynthesis was regulated by both irradiance and H 2 S concentration—GAP increased with increasing H 2 S concentration until a light-dependent maximum (GAP max ; dashed horizontal lines in Figure 5 ) was reached. The initial increase of GAP over low H 2 S concentrations and the saturation effect at higher concentrations resembled H 2 S-dependent Michaelis–Menten kinetics. The initial slope of the increase and the maximum GAP were also light-dependent. All patterns were strictly dependent on H 2 S concentration and were not affected by the temporal dynamics of exposure to sulfide (see H 2 S dynamics in Figures 4b and c ). As expected, the inhibition of GOP by DCMU did not have an effect on GAP (compare open and closed symbols in Figure 5 ) because GOP was also inhibited by H 2 S. H 2 S dependency suggests kinetic regulation of GAP The increase of GAP with H 2 S until a light-dependent maximum (GAP max ; Figure 5 ) is consistent with previous observations ( Cohen et al. , 1986 ; Garcia-Pichel and Castenholz, 1990 ; Klatt et al. , 2015a , 2016a ) and can be explained using a previously described model of the anoxygenic photosynthetic electron transport reactions ( Klatt et al. , 2015a ). According to the model, the H 2 S oxidation rate by SQR is concentration-dependent and SQR donates electrons to the PQ. Reoxidation of PQ is governed by light harvested in PSI. Thus, H 2 S oxidation proceeds at a rate that depends on the affinity of SQR for H 2 S and the oxidized part of the PQ pool ( k AP ; Figure 1 ; Table 2 ; SOM) and is consequently governed by H 2 S concentration and the availability of oxidized PQ. Light dependency suggests multiple sulfide-oxidizing enzymes Irradiance had two effects on GAP: (a) it determined GAP max and (b) it affected the initial slope of GAP increase with H 2 S concentration ( Figure 5 ). The first effect can be explained by considering that rates of H 2 S oxidation can only increase with H 2 S concentration until PSI becomes a bottleneck for electron transport reactions ( Klatt et al. , 2015a ). Specifically, the light energy harvested in PSI dictates the maximum electron transport rate in the irradiance range below light saturation ( k tot ; Figure 1 ; Table 2 ; SOM), which also represents GAP max ( Figure 5 ). Intriguingly, the light-dependent slope of the increase in GAP with H 2 S concentration could not be explained using the previously described model for anoxygenic photosynthesis in Cyanobacteria. Different light-dependent slopes of GAP have been observed in a cyanobacterium, but these could be explained by GAP and GOP occurring simultaneously, with the two photosynthetic modes competing for the PQ pool ( Klatt et al. , 2016a ). However, GAP and GOP are not performed concurrently in strain hensonii, suggesting light must have more complex, previously unconsidered, effects on GAP. To fit our data with the model of the anoxygenic photosynthetic electron transport reactions, we had to suspend a basic assumption: a steady pool of a single sulfide-oxidizing enzyme. Specifically, we had to make one of two plausible assumptions instead: (a) The isolate is equipped with one or multiple types of SQR and the abundance of these enzymes is dependent on irradiance. For instance, if, at higher light intensities the synthesis of SQR is upregulated and thus has a higher v max (see k AP_B in Table 2 ; SOM), the result would be more active SQRs and an increased maximum rate of H 2 S oxidation, manifested in a steeper initial slope in GAP (gray lines in Figure 5 for model output). (b) There are two types of sulfide-oxidizing enzymes (SQR and unidentified sulfide oxidase ‘USO’ in Figure 1 and Supplementary Figure S4 ) with different affinities for H 2 S, with SQR donating electrons to the PQ pool and ‘USO’ donating electrons into the electron transport chain at some other level, most likely to cytochrome b 6 f or plastocyanin (or cytochrome c 553 ), which are encoded in the genome of hensonii (see ‘USO’ and k AP2 in Figure 1 and Table 2 ; black lines in Figure 5 and Supplementary Figure S4 for model output; and SOM for more details). Although the genome is estimated to be 96% complete, we cannot rule out the possibility that the isolate encodes a second SQR. We still consider hypothesis (a) unlikely because, to the best of our knowledge, the regulation of SQR activity and/or abundance by light has not been observed in other phototrophs. Assumption (b) on the other hand eliminates the need for a direct effect of irradiance on enzyme activity or regulation of transcription to explain the light and H 2 S dependency of GAP. Other sulfur-oxidizing enzymes including flavotocytochrome c , dissimilatory sulfur reductase or sulfite:cytochrome c oxidoreductase were not encoded by the draft genome. Still, multiple studies point towards enzymes other than SQR involved in cyanobacterial AP. Namely, pure cultures of Phormidiaceae cyanobacterium SAG 31.92 (formerly Microcoleus chthonoplastes strain 11) oxidize sulfide to thiosulfate ( de Wit and van Gemerden, 1987 ), while other anoxygenic Cyanobacteria (i.e., Oscillatoria spp.) oxidize sulfide to elemental sulfur that accumulates extracellularly ( Cohen et al. , 1975a , 1975b ; Castenholz and Utkilen, 1984 ) or to or sulfite ( Rabenstein et al. , 1995 ). For now, the second sulfide-oxidizing enzyme, however, remains hypothetical. Clearly, both hypotheses (a) and (b) invite future testing. Independent of whether using assumption (a) or (b) outlined above, we consistently obtained the best fit for our experimental data (lines in Figure 5 and Supplementary Fig S4 ) by assuming that the Leptolyngbya sp. strain hensonii’s SQR for H 2 S with apparent K M values between 0.05 and 0.2 m M (see SOM for details). SQRF from the GSB Chlorobaculum tepidum was found to have K M values in the millimolar range ( Chan et al. , 2009 ). Conversely, the well-characterized SQR of Geitlerinema sp. PCC 9228 (type A SQR), which has been implicated in anoxygenic photosynthesis, has a high affinity for sulfide ( Oren and Paden, 1978 ). Regardless, the K M of SQRs for sulfide can vary substantially. Effects of H 2 S on oxygenic photosynthesis Both metabolically versatile and obligate oxygenic Cyanobacteria inhabit microbial mats characterized by fluctuating redox conditions and intermittent exposure to H 2 S. Diverse metabolic processes in Cyanobacteria can be affected by H 2 S—most prominently, sulfide can inhibit oxygenic photosynthesis ( Oren et al. , 1979 ) by poisoning PSII. In strain hensonii, GOP was instantaneously inhibited by H 2 S ( Figure 4c ). Oxygenic photosynthesis did not recover until H 2 S concentrations remained at ~0 μ M for ~30 min ( Figure 6 ) and the rate of GOP recovery decreased with increasing light intensity. The mechanism of inhibition is thought to be the interaction of H 2 S with the OEC in PSII ( Cohen et al. , 1986 ; Garcia-Pichel and Castenholz, 1990 ). To test if the kinetics of the OEC inhibition mechanism can explain the delayed recovery of oxygenic photosynthesis, we used a previously described model of OEC inhibition ( Klatt et al. , 2015b ). We found that the light-independent ~30 min delay of GOP recovery in strain hensonii can be understood by assuming that H 2 S only slowly dissociated from the OEC even after external H 2 S was depleted—that is, the back reaction to an active non-inhibited OEC ( k S2 in Figure 1 and Table 2 ) is slow. As soon as non-inhibited OEC is available, oxygenic photosynthesis can resume. To explore the light dependency of the recovery rate of GOP, we introduced degradation and repair rates of PSII (D 1 subunit) into the model ( k D and k R , respectively, in Figure 1 ; Table 2 ; see SOM for more details). We assumed that the rate of degradation is dependent on light intensity and the level of OEC inhibition. This is because excitation energy harvested in PSII cannot be used efficiently for photochemical reactions if a part of the OEC pool is inhibited. The ‘unused’ fraction of energy is expected to enhance degradation. In other words, H 2 S inhibition of the OEC enhances photoinhibition. Upon reinstatement of the complete pool of uninhibited OEC, light intensity becomes the only factor controlling recovery of GOP. If the light intensity is high, the rate of PSII degradation will still substantially exceed the rate of PSII repair, causing slow recovery. In contrast, low light intensities will allow for a rapid decrease in photoinhibition rates and consequently a fast recovery of oxygenic photosynthesis. The assumptions that (i) the 30-min delay in recovery is caused by OEC inhibition kinetics and (ii) the recovery rate of GOP depends on OEC inhibition are not independent—both are caused by an interplay between the kinetics of OEC inhibition and photoinhibition reactions based on the model depicted in Figure 1 (and described in the SOM). The results of the implementation of this concept into the numerical model are in remarkable agreement with the experimental data (lines in Figure 5 ). Thus, we propose that GOP inhibition is solely controlled by inhibition kinetics and does not invoke additional regulatory mechanisms, such as H 2 S-driven degradation of PSII and a delayed resynthesis of PSII. Still, future studies of this physiology including transcriptomic studies are necessary to fully elucidate the mechanism of GOP inhibition. Effect of H 2 S on reactions downstream of PSI Besides the direct regulatory effects on the initial oxidation reactions of oxygenic and anoxygenic photosynthesis, our data suggest that sulfide also affects reactions downstream of PSI, likely reactions of the Calvin cycle. H 2 S appears to both enhance and inhibit these reactions, with the balance between these contrasting effects depending on light and H 2 S conditions. Inhibition of anoxygenic photosynthesis at non-optimal H 2 S concentrations was previously observed by Cohen et al. (1986) . Intriguingly, in Leptolyngbya sp. strain hensonii, the inhibition was light-dependent. During exposure to the optimal light intensity for GOP (137 μmol photons m −2 s −1 ), GAP max was reached at ~44 μ M H 2 S, followed by a pronounced decrease in GAP with increasing H 2 S concentration ( Figure 5 ). The pronounced decrease of GAP was, however, not observed during exposure to lower light intensities ( Figure 5 ). Because light has an effect on this inhibition, a simple substrate inhibition of SQR cannot account for the decrease in GAP. Using our model, we found that light-dependent inhibition by H 2 S can best be explained by assuming that H 2 S inhibits a reaction downstream of PSI (e.g., in Figure 1 and Table 2 , see lines in Figure 5 for model output and SOM for more details), which only has a role when the maximum GAP is not exclusively controlled by irradiance, that is, at light intensities where the rate of CO 2 fixation limits the overall electron transport rate. The enhancement of reaction rates downstream of PSI became apparent in the observation that rates of anoxygenic photosynthesis can exceed the rates of oxygenic photosynthesis ( Figure 7 , note that GAP in Figure 5 is up to 200% of GOP max ). During exposure to 36 μmol photons m −2 s −1 , GAP did not exceed GOP at any H 2 S concentration ( Figure 5 and Figure 7 ). However, the maximum GAP (in electrons) was roughly two times higher than GOP max during exposure to 137 and 180 μmol photons m −2 s −1 ( Figure 7 ). An enhancement of photosynthetic rates by sulfide was confirmed by 13 C-bicarbonate incubations in the absence and presence of DCMU ( Figure 3 ). In the presence of sulfide (without DCMU), Leptolyngbya sp. strain hensonii incorporated higher amounts of 13 C-bicarbonate (12.9 (±1.61) μmol C assimilated mg dry weight −1 ) compared to cells that received no sulfide (9.07 (±0.92) μmol C assimilated mg dry weight −1 ) ( Figure 3 ). The lower assimilation in cells that received both DCMU and sulfide is presumably because DCMU prevented the switch to oxygenic photosynthesis upon depletion of sulfide. We again used our model to identify the most likely mechanism for the enhancement of GAP. We found the best agreement with our experimental data by assuming that (i) H 2 S upregulates rates downstream of PSI ( γ in in Table 2 and Figure 1 ; SOM for more details) and (ii) the increase in electron transport rate is further supported by excitation energy transfer from PSII to PSI, which is regulated by the redox state of the PQ pool ( β in k tot in Table 2 ; Figure 1 ; SOM for more details). Thus, H 2 S has no enhancing effect at low light intensities because light harvested in PSI limits electron transport rates ( Figure 7 ). Around the optimal light intensity GOP becomes rate limited by CO 2 fixation reactions in the Calvin cycle ( Sukenik et al. , 1987 ; Cardol et al. , 2011 ) and enhancement and inhibition can take effect (see lines in Figures 5 and 7 for model output). Based on these data, we propose that H 2 S has two regulatory effects downstream of PSI: It enhances photosynthetic rates at saturating light intensities if concentrations of H 2 S are below 44 μ M . Above this threshold, inhibitory effects outweigh the enhancing effect of H 2 S on reactions downstream of PSI. The mechanisms behind enhancement and inhibition warrant further research. Summary: complex response based on simple mechanisms The physiological responses of hensonii to H 2 S are complex: (i) No induction time is necessary for anoxygenic photosynthesis, which suggests that the sulfide-oxidizing machinery is constitutively expressed. (ii) The rates of anoxygenic photosynthesis are regulated by both H 2 S and irradiance. Specifically, rates of anoxygenic photosynthesis increase with H 2 S at a light-dependent slope until light limitation occurs or until inhibitory effects of H 2 S occur, which are more pronounced at higher irradiance. (iii) Under the optimal light conditions, rates of anoxygenic photosynthesis are nearly double that of oxygenic photosynthesis. We suggest that (ii) and (iii) can be explained based on concerted responses of multiple elements involved in oxygenic and anoxygenic photosynthesis: the kinetics of sulfide oxidation by SQR and an ‘USO’, enhanced excitation energy transfer from PSII to PSI upon exposure to sulfide, and enhancing and inhibitory effects of sulfide on reactions downstream of PSI, most likely in the Calvin cycle. (iv) O 2 production is inhibited by H 2 S concentrations <1 μ M and remains inhibited for ~30 min even after depletion of sulfide, wherein the recovery rate of oxygenic photosynthesis after this lag phase is dependent on irradiance. Intriguingly, these observations can be explained by considering the kinetics of OEC inhibition and relaxation, and the kinetics of photoinhibition, that is, PSII/D1 degradation and repair. Therefore, the activity patterns of strain hensonii in response to sulfide and irradiance are thus likely based on relatively simple, instantaneous mechanisms that do not necessarily involve adjustments of the enzyme equipment. Ecophysiology of strain hensonii In pure culture, strain hensonii requires no induction time to perform anoxygenic photosynthesis and consumes sulfide until oxygenic photosynthesis is no longer inhibited. The switch between oxygenic and anoxygenic photosynthesis in strain hensonii is subject to complex regulatory pathways, but essentially depends on light and sulfide. In this respect, our results are consistent with previous characterizations of Cyanobacteria inhabiting sulfidic environments. However, the sluggish recovery (~30-min delay) of oxygenic photosynthesis following depletion of H 2 S is not consistent with the observed success of the isolate in situ . Based on the abundance of 16S rRNA gene sequences affiliated with hensonii recovered from LSS mat ( Hamilton et al. , 2017 ) and the physiology of the strain described here, it is likely that this organism has a key role in shaping this highly dynamic mat microenvironment. In situ , the cyanobacterial layer of the LSS mat transitions elegantly between photosynthetic modes over the diel light cycle: In the early morning anoxygenic photosynthesis dominates. In the evening, the cyanobacterial layer transitions back to anoxygenic photosynthesis. To illustrate that the physiology of strain hensonii is not consistent with in situ observations, we used our model to simulate the activity of the isolate in LSS over a diel cycle ( Figure 8 ). In our hypothetical biofilm, we assumed that irradiance in the late morning is high enough for complete depletion of H 2 S in the uppermost layers because sulfide supply from underneath is capped by cyanobacterial anoxygenic photosynthesis in deeper layers as has been observed in natural systems including LSS ( Klatt et al. , 2016b ; de Beer et al. , 2017 ). Our model predicts that when sulfide becomes locally depleted in the uppermost layer due to anoxygenic photosynthesis in deeper layers, there is a delay between anoxygenic and oxygenic photosynthetic activity because strain hensonii cannot switch instantaneously from anoxygenic to oxygenic photosynthesis. It seems highly unlikely that a cyanobacterium exhibiting 30 min of photosynthetic inactivity at high light is competitive in the environment and the lag was not observed in situ in LSS ( de Beer et al. , 2017 ). To understand how strain hensonii can still be successful in the environment, we need to consider that anoxygenic photosynthesis in the cyanobacterial layer of the LSS mat is fueled by three sources of H 2 S: diffusion from underlying sediment, diffusion from the water column and locally produced H 2 S within the cyanobacterial layer of the mat ( de Beer et al. , 2017 ). Locally produced sulfide could cryptically fuel cyanobacterial anoxygenic photosynthesis ( de Beer et al. , 2017 ). This means that anoxygenic photosynthesis at high light could be operational even though H 2 S concentrations approach <1 μ M —concentrations low enough to allow for oxygenic photosynthesis to start. Thus, strain hensonii could remain photosynthetically active throughout the photoperiod. This implies a very close beneficial interaction with sulfate reducing bacteria. The result is a cyanobacterial dominated mat in a delicately poised environment, the productivity of which is largely controlled by local sulfate reduction. Microbial mat systems likely represent hotspots of evolution including sulfur cycling processes and photosynthesis ( Nisbet and Sleep, 2001 ). Even the earliest oxygenic phototrophs were likely exposed to intermittently sulfidic conditions in the immediate microenvironment despite largely ferruginous conditions in the oceans during the Archean and much of the Proterozoic ( Lyons et al. , 2014 ). Early Cyanobacteria might have had to develop strategies to cope with H 2 S toxicity that have been refined over the following billions of years ( Castenholz, 1977 ; Garlick et al. , 1977 ; Oren et al. , 1979 ; Cohen et al. , 1986 ; Miller and Bebout, 2004 ). In LSS and in the hensonii isolate, these strategies are not necessary—the cyanobacterial part of the mat performs anoxygenic photosynthesis until enough sulfide is consumed to enable oxygenic photosynthesis, whereas the Chlorobi-dominated deeper mat continuously performs anoxygenic photosynthesis due to sulfide production from locally closely associated sulfate reducing organisms. The physiology of the hensonii strain is consistent with ecological success in this environment: (i) no induction time is necessary for anoxygenic photosynthesis; (ii) rates of anoxygenic photosynthesis are regulated by both H 2 S and irradiance; (iii) O 2 production is inhibited by H 2 S concentrations as low as 1 μ M and the recovery rate of oxygenic photosynthesis is dependent on irradiance; (iv) rates of anoxygenic photosynthesis can be nearly double those of oxygenic photosynthesis. While the evolutionary history of metabolically versatile Cyanobacteria remains unknown, our data highlight the possibility of coevolution of sulfate reduction and cyanobacterial anoxygenic photosynthesis in microbial mat systems where local sulfur cycling is fueled by a dense biofilm population."
} | 10,238 |
30995278 | PMC6469795 | pmc | 7,777 | {
"abstract": "Recent work on microbiomes is revealing the wealth and importance of plant-microbe interactions. Microbial symbionts are proposed to have profound effects on fitness of their host plants and vice versa, especially when their fitness is tightly linked. Here we studied local adaptation of host plants and possible fitness contribution of such symbiosis in the context of abiotic environmental factors. We conducted a four-way multi-year reciprocal transplant experiment with natural populations of the perennial grass Festuca rubra s . l . from northern and southern Finland, Faroe Islands and Spain. We included F . rubra with and without transmitted symbiotic fungus Epichloë that is vertically transmitted via host seed. We found local adaptation across the European range, as evidenced by higher host fitness of the local geographic origin compared with nonlocals at three of the four studied sites, suggesting that selection pressures are driving evolution in different directions. Abiotic factors did not result in strong fitness effects related to Epichloë symbiosis, indicating that other factors such as herbivory are more likely to contribute to fitness differences between plants naturally occurring with or without Epichloë . Nevertheless, in the case of asymmetric symbiosis that is obligatory for the symbiont, abiotic conditions that affect performance of the host, may also cause selective pressure for the symbiont.",
"conclusion": "Conclusions Our study shows that adaptive evolution in contrasting climatic environments has resulted in local adaptation across the European range in the perennial host grass F . rubra . We found that large-scale abiotic environments did not result in strong differences in fitness between genotypes naturally occurring with or without Epichloë in the absence of high herbivory pressure. In the case of tight fitness linkage, however, it should be noted that selection against nonlocal host genotypes indirectly also decreases fitness of nonlocal symbiont genotypes and thus possibly contributing to the evolution of the symbiont. Future studies should strive for combining reciprocal transplantation experiments with reciprocal inoculations to unravel more complex interactions between host and symbiont genotypes and natural environments.",
"introduction": "Introduction Variability in direction and magnitude of natural selection is a major force shaping biodiversity [ 1 ]. As a result, natural populations encountering differing selection pressures become genetically differentiated and locally adapted [ 2 – 4 ]. Local adaptation is traditionally defined as higher fitness of local than nonlocal individuals in a given environment [ 5 ]. Selective agents driving local adaptation consist of both abiotic and biotic factors, and the latter become especially apparent when local populations of closely interacting species coevolve [ 6 , 7 ]. Plant-associated symbionts have the potential to be highly beneficial for the fitness of their hosts, as has been shown for nitrogen-fixing rhizobia and mycorrhizal fungi. This makes symbiotic associations between plants and microbes excellent study systems for examining how patterns of local adaptation are shaped by symbiosis, especially because plants as sessile organisms need to adapt to surrounding environmental conditions. Estimating the role of symbiotic associations in local adaptation should involve natural environments, where fitness benefits are determined by resource acquisition and allocation. For example, using local and nonlocal soils and reciprocal inoculation, it has been shown that local soil and local genotypes of arbuscular mycorrhizal fungi promote resource acquisition on Andropogon gerardii [ 8 ] At its most extreme, coevolution of hosts and symbionts can result in obligatory associations where survival or reproduction are not possible without the symbiotic partner, and fitness of the symbiont and the host become tightly linked. In these cases, selection against nonlocal host plants results in potential fitness reduction for the symbiont. However, the role of vertically transmitted symbionts in local adaptation of their hosts to abiotic environment is still largely unknown. Systemic fungal symbionts of grasses of the genus Epichloë (Ascomycota; Clavicipitaceae) are an example of asymmetric interactions, where the fungus grows asymptomatically between host cells inside aboveground tissues of the plant. Epichloë reproduces asexually by growing hyphae in newly produced tillers and seeds of the host grass, resulting in vertical transmission, and making them entirely dependent on their host [ 9 ]. Epichloë species are specialized symbionts of grasses with a shared coevolutionary history with their hosts and are transmitted in host maternal lines [ 9 – 11 ]. In agricultural grasses, Epichloë have been viewed as mutualists mostly due to the herbivore-deterring alkaloids that they produce [ 12 ]. Studies on natural populations have shown that asymmetric symbiosis that is facultative to the host plant can range from mutualistic to parasitic [ 13 ]. Harmful effects on the host plant are most evident in sexual strains of Epichloë species that produce spore-forming structures called stromata–a condition known as choke disease–that prevents or hampers development of seeds on the host plant [ 14 ]. However, even asexual vertically transmitted Epichloë species (formerly Neotyphodium , [ 15 ]) can be harmful to the host if costs of harboring the symbiont exceed the benefits [ 16 – 18 ]. This balance could be altered in novel environments, where allocation of host resources can change and potentially result in costs of harboring Epichloë or benefits of increased resistance to abiotic stress. As symbiosis is obligatory for reproduction and persistence of Epichloë , adaptive evolution of both parties is potentially heavily affected by host plant performance. Because of the tight and asymmetric fitness linkage, adaptation of the host plant to local conditions (temperature, precipitation and annual variation in day length) can play an important role in evolution of grass- Epichloë symbiosis. Local adaptation in plants is often associated with differentiation in flowering responses to temperature and photoperiod, and responses to these factors can influence potential for vertical transmission via successful seed production, making environmental factors influencing host plant performance indirectly governing also fitness of the symbiont. Local adaptation of the host can therefore be beneficial for the symbiont, but unless Epichloë provides fitness benefits for the host or especially if it is costly, plants with Epichloë could be selected against. Natural selection can promote occurrence of Epichloë even when the fungus is not transmitted to all offspring if patterns of selection vary in heterogeneous environments [ 13 , 19 ]. Natural grass populations have been found to consist of plants with and without Epichloë at variable frequencies and they might be completely absent in some areas [ 20 – 22 ]. This is in part due to often incomplete vertical transmission, resulting in tillers and seedlings without Epichloë even when associations are mutualistic [ 23 , 24 ]. Loss of the symbiont can be associated with absence of selective advantage and potentially also from genetic mismatches between host and symbiont that can arise from evolutionary conflicts between reproductive modes and genetic variation. These conflicts could be prevalent when cross-pollination of flowers introduces new host genotype combinations in seeds that can prevent growth of the vertically transmitted Epichloë species that cannot actively choose their hosts [ 25 ]. We used natural populations of an outcrossing perennial grass, Festuca rubra L. sensu lato (Poaceae, red fescue) and its symbiont Epichloë festucae (Leuchtm., Schardl, & Siegel), as a model to study local adaptation in host plants and whether naturally occurring plants with or without symbiont show different fitness responses. Classical reciprocal transplant experiments where individuals from different environments are reciprocally transplanted in native environments of each origin allows to test for local adaptation, evidenced by higher fitness of the local population compared with each of the nonlocal populations [ 5 ]. To our knowledge, few reciprocal transplant studies with multiple sites spanning a large geographic area have been conducted–especially in the context of how fitness of the host can be modulated by the symbiont at native sites of natural host populations in the field. Our prediction was that local host populations have become locally adapted and host genotypes naturally harboring E . festucae (referred to as Epichloë from here on) could have reduced fitness due to costs of symbiosis, increased fitness due to resistance to abiotic stress or show no differences related to the tested abiotic environments when compared with naturally Epichloë -free genotypes. Although positive and negative effects of fungal symbionts including Epichloë on growth and reproduction, photosynthetic rate, abiotic stress tolerance, and competitive ability have been documented [ 26 , 27 ], most of these studies have been conducted with cultivars or in agricultural, nutrient-rich environments or greenhouse conditions [ 28 – 30 ]. Use of natural populations and environments can demonstrate ecologically relevant fitness differences, and whether hosts harboring the symbiont are favored by selection in nature. We conducted a four-way reciprocal transplant experiment across a broad geographic scale in Europe (northern Finland, Faroe Islands, southern Finland and Spain) and estimated fitness by quantifying several fitness components over three years at each site. We aimed at answering the following questions: first, do we find evidence for local adaptation of the host on abiotic environments on a large geographic scale? Our hypothesis was that in a reciprocal transplant experiment in home environments of each geographic origin in the field, local plants would have higher fitness than nonlocals, and tested this hypothesis both at the level of estimated cumulative fitness and individual fitness components (survival, biomass, flowering propensity and number of flowering culms) in each year. Second, how does Epichloë symbiosis contribute to host plant fitness in natural environments? More specifically, we tested whether naturally occurring host genotypes with or without Epichloë show different fitness responses in local or novel abiotic environments.",
"discussion": "Discussion Local adaptation across Europe Our large-scale, multi-year reciprocal transplant experiment revealed local adaptation in F . rubra across Europe, as evidenced by higher fitness in local plants compared with nonlocals in northern Finland, southern Finland and Spain. These findings demonstrate the role of natural selection in shaping genetic and phenotypic differentiation in this widespread host grass species. Evidence for local adaptation was supported by multiple fitness components and cumulative fitness estimates. Other studies on grassland plants have documented local adaptation across Europe in some but not all studied species [ 35 , 36 ]. In our study case local adaptation of the host can have consequences for evolution of both partners as fungal symbiont Epichloë is entirely dependent on the grass. Therefore, local adaptation of the host grass will benefit symbiotic partners, and causes natural selection acting against nonlocal host plant genotypes to also decrease performance of nonlocal Epichloë strains. Comparisons across large geographic distances often show local adaptation and differences in selection pressures (e.g. [ 37 – 40 ]). At higher latitudes, plants need to be adapted to strong seasonal changes in temperature, including long winters with temperatures below freezing and variation in day length and light quality. We found that in northern Finland, fitness advantage of the local origin was due to higher survival and flowering propensity than the nonlocal origins. This could be due to differences in photoperiod responses that are required for flowering induction and preparation for overwintering. In F . rubra as well as in other perennial grasses, flowering induction occurs in two steps where consecutive periods of short days, cold temperatures (vernalization) and long days are required [ 41 ]. It is also possible that floral development is in general slower in plants adapted to a longer growing season, and flowering culms in plants from other geographic origins did not have enough time to develop. Plants originating from northern Finland had surprisingly high performance at all sites, indicating that for example responses to photoperiod did not lower their fitness in nonlocal environments. Long-term survival of these northern genotypes was low for example in Spain, possibly due to drought stress during hot and dry summers. Plants from Spanish semiarid grasslands cope with this situation by means of summer dormancy, but seashore populations of F . rubra have been found to remain green throughout the growing season [ 42 ]. Common garden experiments with pasture grasses have shown that plants from different geographic origins differ in their responses to climatic extremes, such as drought, that are associated with climate change [ 43 , 44 ]. At the Faroe Islands, local plants were outperformed by nonlocal plants in all fitness components except biomass, but the Faroese plants had relatively low survival and reproductive success also at nonlocal sites. At the Faroe Islands, the surviving local individuals seemed to be able to utilize the long growing season in the cool and humid oceanic climate, resulting in larger biomass compared to nonlocals. Larger vegetative size and low flowering propensity and number of flowering culms could indicate that the Faroese plants differ in their allocation to sexual vs vegetative reproduction, as has been found in sea shore populations of F . rubra in Spain [ 42 ]. It is also possible that strong selective pressures related to for example temperature extremes such as cold winters might not have a large role in shaping Faroese populations. This can also have resulted in presence of maladaptive alleles via gene flow either from other regions or cultivars of the same species. There might also be more fine-scale environmental differences across the Faroe Islands that would be revealed by reciprocal transplantations between the specific islands. Furthermore, as our study focused on large-scale climatic differences in abiotic factors between the regions, inclusion of effect of competition with surrounding vegetation could reveal local adaptation also in Faroese plants, if their higher biomass production would be correlated with better competitive ability. However, as F . rubra occurs in habitats with relatively low competition, inclusion of a competition treatment would have significantly changed our results. Role of Epichloë symbiosis in fitness variation of the host Vertical transmission mode of the symbiont is predicted to be associated with mutualistic interactions [ 45 ], predicting that Epichloë should be generally promoting fitness of their hosts. This is supported by data from agronomical systems where abundance of nutrients can contribute to beneficial effects of symbiosis to the host, as has been documented for example in perennial ryegrass ( Lolium perenne ) and tall fescue ( Festuca arundinacea ) [ 46 ]. In a study with both wild grasses and cultivars of tall fescue, an overall beneficial effect of Epichloë was reported in a transplantation experiment, but similarly as indicated in our present study on F . rubra , fitness effects depended on the environment and host plant genotype and varied between years and fitness components [ 47 ]. Herbivory is the most studied factor contributing to evolution of the mutualistic association due to alkaloid compounds produced by Epichloë [ 9 ]. Fitness benefits for the host grass are determined by resource acquisition and allocation, especially when the symbiont is using resources for production of protective alkaloids requiring nitrogen [ 48 ]. In natural populations and environments the defensive role may be more variable and context dependent, as levels of alkaloid production profiles and their success for preventing herbivory can vary [ 49 ]. In addition, presence of Epichloë can also result in reduced vegetative biomass, as was found for example in a natural population of Festuca arizonica in a field experiment [ 48 ]. We focused here on the role of large scale abiotic factors driving evolution of local adaptation and found no strong fitness differences between plants with or without Epichloë , indicating that herbivory rather than abiotic factors is driving local evolution involved in Epichloë symbiosis. However, in some cases novel environmental conditions can induce gain or loss of fitness in plants with Epichloë in the studied environments. Loss of fitness in Faroese plants harboring Epichloë in our study at the site in northern Finland could be due to breakdown of mutualism in nonnative grass- Epichloë genotype combinations in stressful conditions, associated with expression changes in a set of fungal genes involved in enhanced nutrient uptake and degradation [ 50 ]. On the contrary, in Spain survival of the same Faroese genotypes with Epichloë was improved to some degree. This could potentially result from improved resistance to drought in Faroese plants with Epichloë as in Spain plants have to cope with seasonal droughts and with very intensive sunlight year-round. Further studies on drought and salt stress resistance could reveal whether thick and waxy leaves of the Faroese plants with Epichloë would confer drought tolerance and enable persistence of green leaves throughout the growing season also in dry habitats such as Spanish grasslands. Plants with Epichloë from northern Finland showed a clear increase in reproductive fitness and biomass production compared with plants without Epichloë when transplanted in southern Finland where the growing season is longer than in their native environment, although no individuals with Epichloë have been found in natural F . rubra populations in this region. Fitness comparisons of local host plants genotypes with or without Epichloë indicated that abiotic factors did not seem to impose selective pressure on Epichloë symbiosis, especially at native sites. In another study with Spanish F . rubra at the same experimental site in Spain, plants with Epichloë had greater phosphorus content than plants without Epichloë [ 51 ], potentially yielding fitness differences in a longer term. Local grazing pressures by large vertebrate grazers not tested here are likely to contribute more to selective advantage of symbiosis, as Epichloë occur at high frequencies at collection sites of the studied geographic origins with heavy grazing in northern Finland (reindeer), Faroe Islands (sheep) and Spain (cattle). Even in the absence of fitness benefits, mathematical models based on metapopulation theory have predicted that vertically transmitted Epichloë species can be maintained in populations even in the absence of fitness benefits to the host and when Epichloë is not transmitted to all developing seeds [ 19 ]. However, as fitness effects of Epichloë on F . rubra have been found to change depending on plant age and from year to year [ 16 , 18 , 49 , 51 ], studies examining survival and germination success of seeds with or without Epichloë could provide more evidence for selective advantage depending on the environment. Differences in germination success could also contribute to resulting frequencies of Epichloë occurrence in grass populations, if seedlings with or without Epichloë are more successfully recruited [ 24 ]. Possible fitness consequences for the fungal symbiont Selective forces driving evolution of the fungal partner Epichloë are tightly correlated with host fitness, and persistence and vegetative reproduction of the host enables survival and growth of Epichloë . Local adaptation of the host plant in our study has strong implications for fitness of the fungal symbiont Epichloë , as nonlocal fungal genotypes are selected against when survival of the nonlocal hosts is low and reduced probability to flower results in prevention of vertical transmission via seed. This scenario is possible because the fungal symbiont Epichloë is entirely dependent on the grass and unable to switch between hosts due to predominant vertical transmission. Therefore, local adaptation of the host grass will benefit both symbiotic partners and causes natural selection acting against nonlocal host plant genotypes to also decrease performance of nonlocal Epichloë strains. Microbial local adaptation to their host’s internal environment can be tested by reciprocal inoculation between host and microbe origins. Most studies on microbial local adaptation to date have been conducted on host-pathogen systems [ 52 , 53 ]. In mutually beneficial symbiotic interactions, local mycorrhizal fungi have been shown to contribute to host fitness in local and nonlocal environments [ 8 , 54 ]. Our present study included only plants naturally occurring with or without Epichloë , as this allowed determining how selection acts on natural genotype combinations in the wild. However, in this system it is also possible to grow the same host plant genotypes with and without Epichloë where the symbiont has been experimentally removed but requires careful control of how the removal treatment (heating seeds or fungicide application) could affect host plant fitness. Studies involving experimental inoculation of selected Epichloë strain in seedlings without Epichloë , would enable testing for different grass- Epichloë genotype combinations, and even three-way interactions (host genotype x Epichloë genotype x environment) in the wild. Also, in order to better estimate the role of production of anti-herbivore compounds in natural grass- Epichloë populations, studies are currently on the way to characterize alkaloid production profiles of Epichloë originating from different regions."
} | 5,602 |
36281261 | PMC9587336 | pmc | 7,779 | {
"abstract": "Metabolic engineering involves the manipulation of microbes to produce desirable compounds through genetic engineering or synthetic biology approaches. Metabolomics involves the quantitation of intracellular and extracellular metabolites, where mass spectrometry and nuclear magnetic resonance based analytical instrumentation are often used. Here, the experimental designs, sample preparations, metabolite quenching and extraction are essential to the quantitative metabolomics workflow. The resultant metabolomics data can then be used with computational modelling approaches, such as kinetic and constraint-based modelling, to better understand underlying mechanisms and bottlenecks in the synthesis of desired compounds, thereby accelerating research through systems metabolic engineering. Constraint-based models, such as genome scale models, have been used successfully to enhance the yield of desired compounds from engineered microbes, however, unlike kinetic or dynamic models, constraint-based models do not incorporate regulatory effects. Nevertheless, the lack of time-series metabolomic data generation has hindered the usefulness of dynamic models till today. In this review, we show that improvements in automation, dynamic real-time analysis and high throughput workflows can drive the generation of more quality data for dynamic models through time-series metabolomics data generation. Spatial metabolomics also has the potential to be used as a complementary approach to conventional metabolomics, as it provides information on the localization of metabolites. However, more effort must be undertaken to identify metabolites from spatial metabolomics data derived through imaging mass spectrometry, where machine learning approaches could prove useful. On the other hand, single-cell metabolomics has also seen rapid growth, where understanding cell-cell heterogeneity can provide more insights into efficient metabolic engineering of microbes. Moving forward, with potential improvements in automation, dynamic real-time analysis, high throughput workflows, and spatial metabolomics, more data can be produced and studied using machine learning algorithms, in conjunction with dynamic models, to generate qualitative and quantitative predictions to advance metabolic engineering efforts.",
"conclusion": "4 Conclusion: future challenges and directions There are various challenges faced in the development of metabolomics approaches for metabolic systems engineering. These include metabolite coverage, automation, and throughput. Increasing the coverage of metabolites can compensate the restricted throughput of LC-MS methods whereas the use of robotic liquid handler platforms and microfluidics allow for high throughput analysis of metabolites as discussed in the previous section. However, microfluidic approaches still lack the sensitivity to quantitate metabolites of low amounts. Further developments in this area would drive the production of large datasets of high-quality data that could be coupled to ML algorithms which can enhance the understanding of biological networks to improve the production of a phenotype during strain development. Transfer learning, an area of ML, could be used to make inferences and predictions on a biological system based on observations made in other biological systems ( Camacho et al., 2018 ). This would reduce the need in generating data from many different systems as various biological systems share similar characteristics. However, the challenge does arise on how best to apply the knowledge learnt in a particular system to a novel system for which data is limited. ML has also been utilized to execute automated metabolite identification in conventional metabolomics ( da Silva et al., 2015 ; Monge et al., 2019 ). Such ML approaches can possibly be applied to aid the current bottleneck of spatial metabolite identification in imaging MS. The use of ion mobility spectrometry (IMS) as a separation technique in imaging MS is another possibility of improving metabolite identification. ML has been used successfully to predict the collisional cross section of metabolites separated in IMS ( Zhou et al., 2016 ) and therefore ML can potentially be applied to IMS coupled to imaging MS to drive metabolite identification. Furthermore, ML approaches have also been used in pathway engineering of microbes. For instance, an ensemble ML approach has been developed to optimize inputs, such as biomolecules, promoter constructs, and nutrient levels, for synthetic biology and metabolic engineering applications ( Radivojević et al., 2020 ). AI/ML are also used for designing both de novo and retrosynthetic pathways ( Lee et al., 2019 ; Lin et al., 2019 ). The prospect of combining real-time metabolomics data together with the spatial distribution of metabolites to better understand the complexity of biology networks is intriguing. Further advancements in single-cell metabolomics could also enhance our understanding of cell-cell heterogeneity and the complexity of their interactions, which would be beneficial in future systems metabolic engineering applications.",
"introduction": "1 Introduction Systems metabolic engineering drives the transformation of microorganisms into effective bio-factories that produce large amounts of target molecules for various industries such as food, biotechnology, and pharmaceuticals ( Choi et al., 2019 ). The integration of -omics data with computational tools used in systems biology defines systems metabolic engineering. This systems-level metabolic engineering integration allows the creation of new metabolic pathways and products and the rewiring of regulatory circuits ( Lee et al., 2011 ), which in turn aids in improving strain development. Metabolomics plays a crucial role amongst the -omics technologies. As the metabolome is downstream of the proteome, it is the endpoint of biological processes, hence, reflecting cell responses and phenotypical interplays from genetic and environmental perturbations ( Griffin, 2006 ). Furthermore, metabolites also maintain intercellular signalling, energy balance and other cellular functions ( Rubakhin et al., 2011 ; Amantonico et al., 2010 ). With improvements in the quality and coverage of metabolomics technologies, changes in metabolite levels due to complex biological interactions can be determined and can be integrated into models to determine the physiological state of a metabolic network ( Töpfer et al., 2015 ; Volkova et al., 2020 ). Through systems-level metabolic engineering, with the focus on integrating metabolomics data with modelling approaches, the efficiency of microbial cell factories can be enhanced. Moreover, integrating metabolomics data together with fluxomics, transcriptomics, and proteomics into mathematical models seeks to promote rational approaches to improve strain and cell-line development ( Roume et al., 2013 ; Vorreiter et al., 2016 ). In this review, we focus on metabolomic approaches and the integration of modelling strategies. Insights on the progress and limitations of metabolomic data generation, dynamic and constraint-based modelling approaches will also be shared. Finally, the challenges faced and the potential of machine learning in systems metabolic engineering as a strategy to produce efficient bio-factories will be discussed."
} | 1,829 |
26063391 | PMC4523385 | pmc | 7,780 | {
"abstract": "The flavodiiron proteins (FDPs) are involved in the detoxification of oxidative compounds, such as nitric oxide (NO) or O 2 in Archaea and Bacteria. In cyanobacteria, the FDPs Flv1 and Flv3 are essential in the light-dependent reduction of O 2 downstream of PSI. Phylogenetic analysis revealed that two genes ( flvA and flvB ) in the genome of Chlamydomonas reinhardtii show high homology to flv1 and flv3 genes of the cyanobacterium Synechocystis sp. PCC 6803. The physiological role of these FDPs in eukaryotic green algae is not known, but it is of a special interest since these phototrophic organisms perform oxygenic photosynthesis similar to higher plants, which do not possess FDP homologs. We have analyzed the levels of flvA and flvB transcripts in C. reinhardtii cells under various environmental conditions and showed that these genes are highly expressed under ambient CO 2 levels and during the early phase of acclimation to sulfur deprivation, just before the onset of anaerobiosis and the induction of efficient H 2 photoproduction. Importantly, the increase in transcript levels of the flvA and flvB genes was also corroborated by protein levels. These results strongly suggest the involvement of FLVA and FLVB proteins in alternative electron transport.",
"introduction": "Introduction Chlamydomonas reinhardtii is a soil-dwelling green alga with great flexibility in its photosynthetic machinery and metabolism, which are employed to cope with changing light, carbon and nutrient supplies and oxic/anoxic conditions. During photosynthesis, specialized antenna complexes harvest and transfer light energy to the PSII and PSI reaction centers, where primary charge separation initiates photosynthetic linear electron flow by oxidizing water at PSII and reducing NADP + to NADPH downstream of PSI. These electron transfer reactions are coupled with proton pumping across the thylakoid membrane, and the resulting proton gradient, ΔpH, drives the ATP synthesis. Photosynthetic organisms have developed different photoprotective mechanisms and alternative electron transport pathways to prevent the over-reduction of the photosynthetic electron transport chain and to maintain an optimal NAD(P)H/ATP ratio under different environmental conditions (reviewed in Peltier et al. 2010 , Cardol et al. 2011 , Shikanai 2014 ). In cyanobacteria, flavodiiron proteins (FDPs, also called A-type flavoproteins, Flvs) function as a strong electron sink, redirecting excess electrons to O 2 in a non-harmful way (reviewed in Allahverdiyeva et al. 2015a , Allahverdiyeva et al. 2015b ). Since C. reinhardtii possesses two genes with high homology to Synechocystis sp. strain PCC 6803 (hereafter, Synechocystis ) flv genes, it is highly conceivable that the proteins encoded by these genes are also involved in photosynthetic electron transport in C. reinhardtii . FDPs are a family of enzymes with nitric oxide (NO)/O 2 -reductase activity and have a modular structure with a N-terminal metallo-β-lactamase-like domain and a C-terminal flavodoxin-like domain as core units ( Vicente et al. 2002 ). The metallo-β-lactamase module harbors a non-heme di-iron center with histidine and carboxylate residues as ligands; this is the active site of NO/O 2 reduction ( Silaghi-Dumitrescu et al. 2003 ). At the C-terminus, the FMN prosthetic group is embedded and acts as the electron donor for the di-iron domain. In FDP monomers, these two redox centers are too distant from each other to perform electron transfer ( Vicente et al. 2008 ). However, the monomers can build a ‘head-to-tail’ dimer structure for efficient electron transfer. This arrangement brings the di-iron center of each monomer in close contact with the FMN moiety from the other monomer (Vicente et al. 2008). In organisms that conduct oxygenic photosynthesis, including cyanobacteria, green algae, mosses and lycophytes, an additional NAD(P)H:flavinoxidoreductase module is fused at the C-terminus of the FDPs. These oxygenic photosynthetic organisms always possess at least two different FDPs, which are grouped into the two clusters A and B ( Zhang et al. 2009 ). It is noteworthy that genes encoding FDP homologs have not been detected in the sequenced genomes of diatoms, haptophytes or higher plants, Picea sitchensis being an exception. An ancient plant, P. sitchensis possesses a single gene with homology to flv ; however, the enzyme encoded by this gene lacks the additional C-terminal domain that is typical of all other oxygenic photosynthetic organisms ( Allahverdiyeva et al. 2015a ). Most studies conducted so far on the function of FDPs in photosynthetic organisms have been focused on cyanobacteria. The genome of Synechocystis, a non-N 2 -fixing, unicellular cyanobacterium, contains four genes ( sll1521 , sll0219 , sll0550 and sll0217 ) encoding a family of FDPs: Flv1, Flv2, Flv3 and Flv4, respectively. A reverse genetics approach applied to Synechocystis has demonstrated the essential function of Flv1 and Flv3 proteins in the light-dependent reduction of O 2 , also known as the Mehler-like reaction ( Helman et al. 2003 ). Recently, it has been found that Flv1 and Flv3 proteins are crucial for safeguarding the photosynthetic apparatus, particularly the PSI complex, under fluctuating light intensities, mimicking natural light conditions ( Allahverdiyeva et al. 2013 , Allahverdiyeva et al. 2015b ). The other two FDPs, Flv2 and Flv4, are not involved in O 2 photoreduction ( Helman et al. 2003 , Allahverdiyeva et al. 2015a ). Instead, these proteins function as a heterodimer in the photoprotection of PSII under CO 2 -limiting and high light conditions by releasing excess excitation pressure at the acceptor side of PSII to a currently unknown electron acceptor ( Zhang et al. 2009 , Zhang et al. 2012 ), in co-operation with phycobilisomes ( Bersanini et al. 2014 , Chukhutsina et al. 2015 ). The filamentous heterocystous N 2 -fixing cyanobacterium, Anabaena sp. strain PCC 7120 (hereafter Anabaena ), possesses six FDPs. Flv1A and Flv3A proteins are specific to vegetative cells and probably function in the Mehler-like reaction, whereas Flv2 and Flv4 proteins presumably mediate photoprotection of PSII, similar to their role in Synechocystis ( Ermakova et al. 2013 , Ermakova et al. 2014 ). The additional set of two FDPs in Anabaena , Flv1B and Flv3B, are heterocyst specific ( Ermakova et al. 2013 ). It has been shown that Flv3B protects nitrogenase by performing light-induced O 2 uptake and maintaining micro-oxic conditions inside of the heterocysts, while the role of Flv1B remains unknown ( Ermakova et al. 2014 ). In the eukaryotic green alga C. reinhardtii , two flv genes have been identified as paralogs in each cluster: flvA (Cre12.g531900) and flvB (Cre16.g691800). Despite a lack of sufficient experimental data, the high homology between the cyanobacterial and algal FDP proteins makes the involvement of FDPs in O 2 photoreduction highly likely ( Zhang et al. 2009 , Peltier et al. 2010 , Cardol et al. 2011 , Dang et al. 2014 ). In this work, we analyzed the expression patterns of C. reinhardtii flvA and flvB at the transcript and protein levels under different environmental conditions, including acclimation to different light intensities, CO 2 concentrations and sulfur deprivation. Our results strongly support the involvement of the FLVA and FLVB proteins in alternative electron transfer.",
"discussion": "Discussion Analysis of putative reference genes As a first approach to obtain information about the function of FDPs in C. reinhardtii , we applied RT–qPCR to determine the response of flv transcript levels to varying environmental conditions. The determination of appropriate reference genes for each organism under particular environmental conditions is crucial to employing the correct normalization strategy to transcript analysis ( Huggett et al. 2005 , Guenin et al. 2009 ). In this study, we analyzed several candidate reference genes, and the cblp and ubc8 genes were determined to be the most suitable for the interpretation of the transcript data obtained after the shift of algal cultures from high CO 2 and standard growth light conditions to low CO 2 and/or high light conditions ( Fig. 1 A–C). The situation was different when the sulfur deprivation protocol was applied ( Melis et al. 2000 ) for the initiation of long-term H 2 photoproduction. Under nutrient deprivation conditions, the sealed algal cultures pass through several physiological stages, resulting in massive changes in cellular metabolism from oxygenic photosynthesis to anaerobic photo-fermentation ( Atteia et al. 2013 , Catalanotti et al. 2013 , Allahverdiyeva et al. 2014 ). In this case, the cblp and act genes were the most stable reference genes ( Fig. 1 D). This study confirmed that there are no universal reference genes, and the choice of appropriate reference genes varies depending on the environmental conditions and the nature of the analyzed target genes. FDPs possibly work as an alternative electron sink in C. reinhardtii FDPs are known to function in alternative electron transport routes in cyanobacteria (reviewed in Allahverdiyeva et al. 2015a , Allahverdiyeva et al. 2015b ). The presence of homologs of the genes coding for FDPs in C. reinhardtii suggests a possible involvement of their products in photosynthetic electron transport. However, the function of FDPs in C . reinhardtii has not been addressed thoroughly and needs to be elucidated. RT–qPCR and Western blot analysis demonstrated that flvA and flvB were significantly up-regulated on both the transcript and protein levels after the change in CO 2 (shifting the cultures from HC to LC) ( Figs. 3 , 5 ) and/or light (shifting the cultures from GL to HL) regimes ( Figs. 4 , 5 ). The strongest up-regulation of the FLVA protein was observed after the shift to high light, whereas the FLVB protein was up-regulated under all three different environmental conditions tested in the present study. The treatment of cells with both high light and/or limited CO 2 concentrations led to a decrease in photosynthetic activity ( Fig. 2 ). The exposure of cells to high light causes an increase in NAD(P)H levels, while the lower CO 2 availability led to a higher ATP demand ( Kramer and Evans 2011 ). During evolution, photosynthetic organisms have developed sophisticated mechanisms to dissipate excess reducing power in harmless ways and to balance possible mismatches in production and demand of ATP and NAD(P)H, the ratio of which changes upon environmental cues through the regulation of linear and alternative electron transport pathways ( Peltier et al. 2010 , Cardol et al. 2011 , Kramer and Evans 2011 ). Recent studies with cyanobacteria have demonstrated the function of FDPs as powerful electron sinks under stress conditions: Flv2 and Flv4 are strongly up-regulated and involved in the photoprotection of PSII under ambient CO 2 and high light conditions ( Zhang et al. 2009 , Zhang et al. 2012 , Bersanini et al. 2014 ), whereas Flv1 and Flv3 proteins can release electron pressure after PSI, thus safeguarding PSI under fluctuating light intensities ( Allahverdiyeva et al. 2013 ). Flv1 and Flv3 proteins act as a strong electron sink, redirecting about 20–60% of electrons originating from PSII to O 2 during illumination under air-level CO 2 and under strong Ci deprivation, respectively ( Allahverdiyeva et al. 2011 ). Importantly, O 2 photoreduction by the FDP pathway generates water without the formation of reactive oxygen species (ROS) ( Vicente et al. 2002 ), thus also contributing to ATP synthesis. The data obtained with C. reinhardtii flvA and flvB genes strongly resembles the gene expression pattern of Anabaena flv1A and flv3A , where both genes were strongly up-regulated at low CO 2 and moderately up-regulated at high light conditions ( Ermakova et al. 2013 ). Moreover, accumulation of the flv3 transcript and a strong up-regulation of the Flv3 protein have been observed in Synechocystis cells under low CO 2 conditions, whereas the flv1 and flv3 transcripts in Synechocystis did not show a remarkable induction under high light conditions ( Zhang et al. 2009 ). The limited information about the response of flv1 transcripts to different environmental cues is probably due to low transcript abundance of this gene in Synechocystis ( Zhang et al. 2009 , Allahverdiyeva et al. 2015a ). A recent study showing an up-regulation of FLVA and FLVB proteins in the pgrl1 mutant of C. reinhardtii under low CO 2 as well as high light conditions indicates that FDPs could function as an electron valve to compensate for the lack of, or impaired, cyclic electron flow ( Dang et al. 2014 ). Interestingly, under the combined HLLC stress condition used in that study, the up-regulation of FDPs in the pgrl1 mutant was transient and disappeared after 48 h. Instead, the elevated H 2 O 2 level indicated a replacement of the FDP pathway by the true Mehler reaction and the formation of ROS ( Dang et al. 2014 ). Similarly, our results with C. reinhardtii wild type during HLLC stress showed an up-regulation of both FDPs during the first 48 h ( Fig. 5 ). This implies an important function for these proteins upon changes in environmental conditions. Our expression analysis of FLVA and FLVB, together with previous results, suggests that FDPs in C. reinhardtii also play an important role as alternative electron sinks in order to prevent redox poise at the photosynthetic electron transport chain. The possible electron donor of FDPs in C. reinhardtii is not known yet. Based on in vitro studies on recombinant Synechocystis Flv3 proteins, it was concluded that FDPs function as an NAD(P)H-O 2 -oxidoreductase ( Vicente et al. 2002 ). However, this is not the case for Synechocystis Flv2 and Flv4 proteins functioning at the PSII acceptor side. Recently, ferredoxin 1 (FDX1) was found to interact with FLVB, thus opening up a new discussion about the possibility of FDX1 as an electron donor to the FDPs proteins in C. reinhardtii ( Peden et al. 2013 ). FDPs participate in photosynthetic acclimation of C. reinhardtii to sulfur deprivation During acclimation to sulfur deprivation, algae experience a strong metabolic shift from oxygenic photosynthesis, where CO 2 is assimilated and starch accumulated, towards anaerobic photo-fermentation, where starch reserves are metabolized to produce ATP and NAD(P)H. The anaerobic re-oxidation of NAD(P)H involves several fermentative pathways that produce organic acids (acetate, formate, lactate, malate and succinate), ethanol, H 2 and CO 2 (reviewed in Atteia et al. 2013 , Catalanotti et al. 2013 ). Some enzymes of fermentative metabolism, such as the [Fe–Fe]-hydrogenases and pyruvate formate-lyase, are sensitive to O 2 remaining in the chloroplast ( Atteia et al. 2013 ). The acclimation to sulfur deprivation that triggers H 2 photoproduction in algae can be divided into several phases ( Kosourov et al. 2002 ) ( Fig. 6 A). During the photosynthetic stage (phase I) of acclimation to sulfur deprivation (0 to ∼10 h) the O 2 concentration in the bioreactor rises until respiratory processes take over (phase II). Anaerobiosis is established at approximately 40 h after the shift to sulfur deprivation (phase III). The up-regulation of both FLVA and FLVB proteins demonstrates a correlation with the presence of O 2 in the culture, with the maximum FDP amount observed approximately 24 h after the shift ( Fig 6 C, D). The FDP up-regulation during the photosynthetic and respiratory phase of H 2 photoproduction indicates that O 2 photoreduction via FDPs is important in the acclimation to these conditions. It has been postulated that the decrease of O 2 after a shift to sulfur deprived medium is mainly due to an increase in mitochondrial respiration ( Melis et al. 2000 , Melis 2007 , Ghirardi et al. 2010 ). Our results indicate that FDPs contribute to the establishment of anaerobiosis by functioning in light-induced O 2 uptake. The increased levels of FDPs in the chloroplast during the first phases of sulfur deprivation may accelerate the establishment of anaerobiosis and therefore help to ensure the function of the fermentative pathways within a shorter time period. In the later phase IV, while H 2 is produced in anaerobiosis, the FDPs are down-regulated ( Fig. 6 ). Taken together, we propose that FDPs in C. reinhardtii function as an alternative electron sink during oxygenic photosynthesis by actively assisting to decrease the O 2 level inside the chloroplast at the onset of anaerobiosis and are replaced by [Fe–Fe]-hydrogenases later on, when anaerobiosis is fully established. In both cases, FDPs and [Fe–Fe]-hydrogenases support electron flow in thylakoids for the production of ATP at the expense of reducing power accumulated downstream of PSI and, thus, also protect the photosynthetic electron transport chain from over-reduction. This rapid acclimation to anaerobiosis is likely to be advantageous for the soil-dwelling C. reinhardtii , which regularly faces anoxic or micro-oxic conditions in nature."
} | 4,311 |
33501265 | PMC7805676 | pmc | 7,781 | {
"abstract": "We analyze the efficacy of modern neuro-evolutionary strategies for continuous control optimization. Overall, the results collected on a wide variety of qualitatively different benchmark problems indicate that these methods are generally effective and scale well with respect to the number of parameters and the complexity of the problem. Moreover, they are relatively robust with respect to the setting of hyper-parameters. The comparison of the most promising methods indicates that the OpenAI-ES algorithm outperforms or equals the other algorithms on all considered problems. Moreover, we demonstrate how the reward functions optimized for reinforcement learning methods are not necessarily effective for evolutionary strategies and vice versa. This finding can lead to reconsideration of the relative efficacy of the two classes of algorithm since it implies that the comparisons performed to date are biased toward one or the other class.",
"conclusion": "Conclusions We analyzed the efficacy of modern neuro-evolutionary strategies for continuous control optimization on the MuJoCo locomotion problems, that constitute a widely used benchmark in the area of evolutionary computation and reinforcement learning, and on additional qualitatively different problems. The term modern evolutionary strategies indicates algorithms that compute the interrelated dependencies among variations of better individuals, or that use a form of finite difference method to estimate the local gradient of the fitness function. The results obtained on the MuJoCo, Long double-pole and Swarm foraging problems indicate that these methods are generally effective. The comparison of the results obtained with different algorithms indicate that the OpenAI-ES algorithm outperforms or equals the CMA-ES, sNES, and xNES methods on all considered problems. Overall, the data collected, the ablation studies and the experiments conducted by varying the population size indicate that the efficacy of the OpenAI-ES method is due to the incorporation of optimization and normalization techniques commonly used in neural network research. More specifically, the efficacy of the method can be ascribed to the utilization of the Adam stochastic optimizer, which operates effectively also in the presence sparse gradients and noisy problems and avoids an uncontrolled growth of the size of the connection weights. Moreover, the efficacy of the method can be ascribed to the usage of normalization techniques that preserve the adaptability of the network and reduce overfitting. Finally, we demonstrate that the reward functions optimized for the PPO are not effective for the OpenAI-ES algorithm and, vice versa, the reward functions optimized for the latter algorithm are not effective for the former algorithm. This implies that the reward function optimized for a reinforcement learning algorithm are not necessarily suitable for an evolutionary strategy algorithm and vice versa. Consequently, this implies that a proper comparison of algorithms of different classes should involve the usage of reward functions optimized for each algorithm. Indeed, comparisons carried out by using reward functions optimized for one method only could be biased. The usage of deterministic policies (commonly used in evolutionary methods) vs. stochastic policies (commonly used in reinforcement learning methods) seems to be an important cause of the differences observed between the OpenAI-ES and the PPO algorithms with respect to the sensitivity to the reward function. Whether the usage of different reward functions is necessary for all evolutionary and reinforcement learning algorithms or only for some algorithms deserve further investigations. Similarly, the identification of the characteristics that make a reward function suitable for a specific algorithm deserves further investigations.",
"introduction": "Introduction Model-free machine learning methods made significant progress in the area of sequential decision making which involves deciding from experience the sequence of actions that can be performed in a certain environment to achieve a goal. In the area of reinforcement learning (Sutton and Barto, 2018 ), progress has been achieved primarily by combining classic algorithms with deep learning techniques for feature learning. Notable examples are agents trained to play Atari games on the basis of raw pixels input (Mnih et al., 2015 ) and simulated robots capable of performing locomotion and manipulation tasks (Schulman et al., 2015a , b ; Andrychowicz et al., 2019 ). Recently, similar progress has been made in the area of evolutionary computation through neuro-evolutionary methods (Stanley et al., 2019 ), also indicated as direct policy search methods (Schmidhuber and Zhao, 1998 ). In particular, in a recent paper Salimans et al. ( 2017 ) demonstrated how neural network controllers evolved through a specific natural evolutionary strategy achieve performance competitive with the reinforcement learning methods mentioned above on the MuJoCo locomotion problems (Todorov et al., 2012 ) and the Atari games from pixel inputs (Mnih et al., 2015 ). In this work Salimans et al. ( 2017 ) also demonstrated for the first time that evolutionary strategies can be successfully applied to search spaces involving several hundred thousand parameters and can complete the evolutionary process in few minutes thanks to the their highly parallelizable nature. However, the relation between the OpenAI-ES algorithm introduced by Salimans et al. ( 2017 ) and other related algorithms such as CMA-ES (Hansen and Ostermeier, 2001 ) and Natural Evolutionary Strategies (Wierstra et al., 2014 ) is still to be clarified. In particular, whether or not the former method is more effective than the other related methods, and/or whether the advantage of the method introduced by Salimans et al. ( 2017 ) comes from the usage of the virtual batch normalization technique (Salimans et al., 2016 , 2017 ) that can be applied also to the other methods. Such et al. ( 2017 ) compared the OpenAI-ES method with related algorithms. They used a classic evolutionary strategy (see the section Methods) and obtained performance similar to those reported by Salimans et al. ( 2017 ) on 13 selected Atari games, but lower performance on the MuJoCo humanoid locomotion problem. The classic method resulted less sample efficient than the natural evolutionary strategy used by Salimans et al. ( 2017 ) on this problem. Mania et al. ( 2018 ) demonstrated how a simplified evolutionary strategy is sufficient to solve the MuJoCo locomotion problems and outperform state-of-the-art policy gradient methods. Henderson et al. ( 2018 ) stressed the importance of considering the variability among replications and the impact of hyper-parameters to evaluate the efficacy of alternative methods. Other works pointed out that the considered problems admit compact solutions. In particular, Rajeswaran et al. ( 2017 ) demonstrated how the MuJoCo locomotion problems can be solved with shallow networks. Such et al. ( 2017 ) demonstrated that some Atari games admit simple solutions, an issue highlighted also in other works (e.g., Wilson et al., 2018 ). In this paper we compare systematically the performance of the evolutionary strategy proposed by Salimans et al. ( 2017 ) with other related methods in order to verify the relative efficacy of available algorithms on continuous optimization problems. To avoid biases caused by the usage of a specific class of problems, we extend the test with additional and qualitatively different problems (see below). Finally, we analyze the role of the reward function and critical hyper-parameters. As we will see, the evolutionary strategy proposed by Salimans et al. ( 2017 ) outperforms or equals related approaches in all problems and is relatively robust with respect to the setting of hyper-parameters. The advantage of this method is not only due to the usage or virtual batch normalization that have been applied to all methods in our analysis. It can rather be ascribed to the efficacy of the Adam stochastic optimizer (Kingma and Ba, 2014 ) which avoids an uncontrolled growth of the size of the connection weights. Finally, we show how the contribution of virtual batch and weight decay normalization is minor in simple problems, but crucial in more complex ones. The analysis of the role of the reward function indicates that functions optimized for reinforcement learning are not necessarily effective for evolutionary strategies and vice versa. Indeed, the performance of evolutionary strategies can improve dramatically with the usage of suitable reward functions. This finding should lead to a reconsideration of the relative efficacy of the two classes of algorithm since it implies that the comparisons performed to date are biased toward one or the other class."
} | 2,213 |
21304466 | PMC3182658 | pmc | 7,783 | {
"abstract": "Electrospun nanofiber scaffolds have been shown to accelerate the maturation, improve the growth, and direct the migration of cells in vitro . Electrospinning is a process in which a charged polymer jet is collected on a grounded collector; a rapidly rotating collector results in aligned nanofibers while stationary collectors result in randomly oriented fiber mats. The polymer jet is formed when an applied electrostatic charge overcomes the surface tension of the solution. There is a minimum concentration for a given polymer, termed the critical entanglement concentration, below which a stable jet cannot be achieved and no nanofibers will form - although nanoparticles may be achieved (electrospray). A stable jet has two domains, a streaming segment and a whipping segment. While the whipping jet is usually invisible to the naked eye, the streaming segment is often visible under appropriate lighting conditions. Observing the length, thickness, consistency and movement of the stream is useful to predict the alignment and morphology of the nanofibers being formed. A short, non-uniform, inconsistent, and/or oscillating stream is indicative of a variety of problems, including poor fiber alignment, beading, splattering, and curlicue or wavy patterns. The stream can be optimized by adjusting the composition of the solution and the configuration of the electrospinning apparatus, thus optimizing the alignment and morphology of the fibers being produced. In this protocol, we present a procedure for setting up a basic electrospinning apparatus, empirically approximating the critical entanglement concentration of a polymer solution and optimizing the electrospinning process. In addition, we discuss some common problems and troubleshooting techniques.",
"discussion": "Discussion Note: The majority of the examples presented here deal with electrospinning poly-L-lactic acid (PLLA) nanofibers. This is simply because PLLA is the most commonly spun polymer in our laboratory. However, we have also successfully used these methods to electrospin other polymers (e.g., PLGA, PCL, PS) and believe that the techniques presented here are easily applicable to the majority of mid- to high-molecular weight polymer solutions."
} | 554 |
38688974 | PMC11061196 | pmc | 7,784 | {
"abstract": "This study integrated bacterial community and soil chemicals to characterize the soil ecosystem in an open upland field managed by six controlled fertilizer programs using the minimum amount of pesticides. Amplicon sequencing the 16S rRNA gene revealed that inorganic nitrogen fertilizer and compost altered the diversity and structure of the soil bacterial community throughout buckwheat ( Fagopyrum esculentum Moench ‘Hitachiakisoba’) cultivation. The bacterial community comprised three clusters that contained bacteria that are prevalent in soils fertilized with nitrogen (cluster 1, 340 taxa), without nitrogen and compost (cluster 2, 234 taxa), and with compost-fertilized (cluster 3, 296 taxa). Cluster 2 contained more taxa in Actinobacteriota and less in Acidobacteriota , and cluster 3 contained more taxa in Gemmatimonadota compared with the other clusters. The most frequent taxa in cluster 1 were within the Chloroflexi phylum. The bacterial community structure correlated with soil chemical properties including pH, total organic carbon, SO 4 2− , soluble Ca 2+ . A co-occurrence network of bacterial taxa and chemicals identified key bacterial groups comprising the center of a community network that determined topology and dynamics of the network. Temporal dynamics of the bacterial community structure indicated that Burkholderiales were associated with buckwheat ripening, indicating plant-bacteria interaction in the ecosystem.",
"introduction": "Introduction Biogeochemical activities of soil microbes decompose organic matter and contribute to carbon and nitrogen cycles in ecosystems 1 , 2 , and in anthropogenically-established agricultural fields. Farmer-friendly inorganic fertilizers have been supplying missing nutrients and increased crop productivity at low cost without being labor-intensive for over a century 3 . However, the intensive or long-term application of the inorganic fertilizers disturbs microbial communication in soil ecosystems, and adversely affects crop yield and quality 4 , 5 . Although no technology is yet available to isolate and characterize all the soil microorganisms involved in these phenomena, analysis of bacterial communities based on sequencing 16S ribosomal RNA (rRNA) genes in soil ecosystems has identified numerous uncultivated microbes 6 and revealed a global diversity of soil microbial communities 7 . Therefore, a better understanding the structure and function of the soil microbial community and its relationships with crops and soil nutrients is required to develop sustainable agricultural practices. Agricultural fields that have been systematically managed over the long term are stable ecosystems that can serve as models, because interactions among a vigorous microbial community, soils, and crops should be reproducible 8 , 9 . The purpose of this study was to determine the temporal and fertilizer-dependent dynamics of soil bacterial communities in buckwheat fields under long-term management. In our target field, systematic management over three decades with minimum pesticides has resulted in stable crop production, indicating stable control of the bacterial community structure and soil properties. The open upland ecosystem is another unique feature of the test field and a promising model that fills the gap between actual farmland and systematic bacterial community studies in laboratories and/or greenhouses. Therefore, analysis of these test fields should reveal the dynamics of soil microbial communities in upland agricultural fields that depend on fertilization protocols and crop growing periods. This study investigated six test plots under different fertilization conditions in a test field during a single season of buckwheat ( Fagopyrum esculentum ) cultivation. Buckwheat is a popular crop from which edible seed flour is derived. We analyzed extensive datasets of bacterial community structures, soil chemical properties, and crop phenotypic traits indicative of changes caused by fertilizer programs and cultivation stages in a test field. Network analyses indicated that pH, total organic carbon (TOC) and SO 4 2− are core factors of the bacterial community structure. The network comprises a center and three groups, in which distinct bacteria maintain specific topologies and dynamics.",
"discussion": "Discussion Accumulating evidence emphasizes the importance of microbiomes in controlling crop productivity that is a consequence of complex associations among microbiomes, plants, and abiotic factors. We investigated these associations in a unique upland fertilizer test field that has been stably managed for over 30 years. This allowed the generation of datasets of bacterial abundance and chemical components in soil. Soil properties clearly differed among the six test plots according to fertilizer protocols. Phosphate increased buckwheat growth (Fig. 1 c,e), which confirmed previous results derived from the same field between 1992 and 1999 10 . Bacterial communities that were altered by crop cultivation periods became apparent. These findings revealed bacterial communities that participate in upland soil fertility and crop nutrition. Hierarchical clustering sorted soil bacteria into three clusters in the manner dependent on nitrogen fertilizer and compost (Fig. 4 ). Three groups determined by co-occurrence network analyses largely overlapped with the clusters (Fig. 5 ). Composting was associated with more diverse microbial taxa than other soils (Table 1 , Fig. 3 d) and included 101 taxa that were specific to test plot C (Fig. 3 f). The frequency of the phylum Gemmatimonadota was high in plot C soil (Fig. 3 e, top) and that of taxa identified in this phylum belonged to cluster 3 (Supplementary Table S4 ). The paucity of taxa in plots NK, NP, and NPK suggested that nitrogen fertilizer affects bacterial diversity (Fig. 3 d,f). The bacterial community structure correlates with inorganic nitrogen and organic fertilizers in other fields 8 , 18 . Taxa in the phylum Actinobacteriota were more diverse in cluster 2 than in the other clusters (Fig. 3 f). Five of eight predominant taxa in cluster 2 belonged to Actinobacteriota (Supplementary Table S4 ), indicating that these bacteria dominated soil without either nitrogen fertilizer or compost. Taxa in the phylum Chloroflexi were more prevalent in cluster 1 (Fig. 3 f and Supplementary Table S4 ), indicating that Chloroflexi dominated when soil was fertilized with nitrogen. Bacterial network hubs connecting the intra-and inter-groups were shared in all groups with Acidobacteriota and Proteobacteria , and specifically in groups 0, 1, and 2 with Actinobacteriota and Chloroflexi (Fig. 5 d and Supplementary Table S6 ). These phyla were diverse and ubiquitous not only in soils from the test plots (Fig. 3 f) but also in other types of soils 7 . Consequently, nitrogen fertilizer and compost were major determinants of the bacterial community structure in soils from the six test plots. This study revealed linkage between bacterial community structure and pH in PCA and co-occurrence networks. The significance of pH to the soil bacterial community structure in the test plot soils indicates that it is an important predictor of bacterial community structure 19 , 20 . Our finding of lower soil pH in the nitrogen-supplemented NK, NP, and NPK plots is consistent with that fact that nitrogen fertilizer decreases the pH of various types of soils 11 . The results of the clustering analysis indicated that bacterial communities differed between soils fertilized with and without nitrogen fertilizer (Fig. 4 h and Supplementary Table S3 ). This indicates that the link between pH and soil bacterial community structures is a consequence of nitrogen fertilization. Buckwheat tolerates acidic soils (pH 5.5‒6.0), in which plant biomass and seed yield increase 21 , 22 . The increased plant biomass might be linked to bacterial community structures in acidic soil. However, further investigation is required to identify specific bacteria that are associated with soil pH. Topological analysis located group 0 at the center of the co-occurrence network (Fig. 5 ). The betweenness centrality parameter identified 25 nodes as intergroup or intragroup hubs in the network. The sole hub of chemical components was SO 4 2− (Fig. 5 c,d; “M”), which is a counter ion of the nitrogen fertilizer (NH 4 ) 2 SO 4 . This is consistent with the indirect function of SO 4 2− as a consequence of nitrogen/sulfate fertilizers. Sulfates derived from inorganic nitrogen fertilizers might have lowered the soil pH in the test plots. By contrast, SO 4 2− can specifically maintain soil ecosystems through microbial activities that immobilize it to organic matter 23 . Such immobilization prevented an immediate increase in soil SO 4 2− levels after nitrogen application and this induced them to increase later at the F stage (Figs. 2 a, 6 d and Supplementary Fig. S3 ). This agrees with the findings that the NP and NPK plots supplemented with nitrogen accumulated high levels of TOCs and changed the SO 4 2− contents more dynamically. The stage-dependent difference during cultivation revealed nine bacteria that were prevalent at the ripening stage (Fig. 6 d). The frequencies of only three bacteria increased at the flowering stage, suggesting a unique plant-bacteria interaction in soils where buckwheat ripens. Two bacteria associated with ripening belonged to the large Comamonadaceae and Burkholderiaceae families. The order Burkholderiales is a dominant component of many soil ecosystems and includes species that promote plant growth, are endophytic 24 , and solubilize phosphate to a plant-available form 25 . These mechanisms potentially explain the association between ripening buckwheat and bacteria. This notion has attracted much attention in terms of understanding soil ecosystems in the field, especially in test plots consisting of andisols with an extremely high capacity to adsorb phosphoric acid. Another ripening-associated bacterium belongs to the genus Nitrospira that includes nitrogen dissimilatory and nitrite-oxidizing bacteria, as well as complete ammonia oxidizers (comammox) 26 , suggesting a relationship with plants via nitrogen dynamics in soil. The soils in the six test plots should be active in terms of nitrification because they accumulated little ammonium even on the second day of (NH 4 ) 2 SO 4 application (Supplementary Fig. S3 ). These results suggested a correlation between nitrifying bacteria and ripening buckwheat. However, their direct role in ripening awaits further investigation because nitrification is the result of complex interactions among available ammonium, plant root exudates, soil properties, comammox, and archaea 2 . This study determined the impact of fertilizers on plant growth, soil bacterial community structure and chemical properties. Long-term fertilizer programs differentiated soil properties in the six test plots, generated bacteria that are key for a community structure, and promoted their temporal interactions with plants. A pioneer study in an open field system in Rothamsted, UK started in 1843 to investigate wheat production with rotations of potato, oats, beans, and other crops 9 , 27 . Buckwheat, rye, sweet potato, ground nuts, wheat and potato are uniquely rotated in the test field assessed herein. Among these crops, we focused on buckwheat, of which 1.6 million tons were produced globally during 2019 (FAOSTAT) 28 . Our findings will guide the development of future agricultural technology to control bacterial community structures and improve buckwheat productivity. We plan to investigate ecosystems in the test plots rotating the six crops. The data should reveal unique ecosystems associated with stably managed crop rotations and facilitate the development of practical strategies to design and control bacterial community structures and plant cultivation."
} | 2,986 |
36519191 | PMC9948233 | pmc | 7,785 | {
"abstract": "Abstract Biology leverages a range of electrical phenomena to extract and store energy, control molecular reactions and enable multicellular communication. Microbes, in particular, have evolved genetically encoded machinery enabling them to utilize the abundant redox‐active molecules and minerals available on Earth, which in turn drive global‐scale biogeochemical cycles. Recently, the microbial machinery enabling these redox reactions have been leveraged for interfacing cells and biomolecules with electrical circuits for biotechnological applications. Synthetic biology is allowing for the use of these machinery as components of engineered living materials with tuneable electrical properties. Herein, we review the state of such living electronic components including wires, capacitors, transistors, diodes, optoelectronic components, spin filters, sensors, logic processors, bioactuators, information storage media and methods for assembling these components into living electronic circuits.",
"conclusion": "CONCLUSION Recent works in the fields of electromicrobiology, synthetic biology and bioelectronics have yielded a wealth of information about the electrical properties of cells and biomolecular components. It is an exciting time where rapid advances in knowledge and engineering of such components are enabling the integration of these biological entities with electrical devices to generate living electronics. It is encouraging to see the recent trend for reporting on the intrinsic parameters (e.g. conductivity and charge carrier mobility) of conductive biomolecules and multicellular networks. As knowledge in these areas continues to expand, it is important for the community to continue, whenever possible, to communicate these intrinsic properties so that they can be directly compared and evaluated across studies. Taking advantage of the natural and engineered sensing capabilities of these electroactive biomolecules and microbes has led to the development of new biosensors directly capable of providing an electronic readout. The development of bioactuator components for electrically controlling the behaviour of biomolecules and microbes is allowing for closed‐loop biohybrid devices that electrically communicate enabling them to sense, compute and actuate in response. Additionally, the recent trends for controlling the assembly of biomolecules and cells hold great promise for integrating multiple disparate living electronic components into more complex electrical circuits with predictable and tuneable behaviours. By combining these assemblies with biofabricated metallic and semiconductor nanomaterials, living electronic systems with both solid‐state and biological functionalities can be developed. The bodies of work discussed herein show the different avenues in which living electronic components can be constructed and characterized, highlighting the diverse specialities and points of view required to push the overall field forward. However, this then creates new space for these specialities and their resulting disparate developments to converge in overcoming challenges and creating new technologies thought impossible when attempted separately. By combining the different facets of electromicrobiology, synthetic biology and bioelectronics, perhaps novel living electronics with improved biocompatibility, lower environmental footprints and hybrid biological and solid‐state capabilities can be constructed.",
"introduction": "INTRODUCTION The fields of bioelectronics, synthetic biology and electromicrobiology are converging through the development of new living electronics where biological entities (e.g. biomolecules, cells or cellular communities) are directly integrated as electrical components into electronic circuits (Dunn, 2020 ). Biohybrid devices have been built that integrate a diversity of living electronic components spanning many length scales. Research in the field of bioelectronics is yielding materials innovations that improve the abiotic: biotic interface between biological entities and electrodes (Tseng et al., 2020 ). Synthetic biologists are leveraging concepts from electrical engineering as analogies when designing DNA‐, RNA‐ and protein‐based logic circuits to program predictable and dynamic phenotypes in cells and biomolecular networks (Brophy & Voigt, 2014 ; Gao et al., 2018 ; Green et al., 2017 ). Meanwhile, advances in electromicrobiology are providing a mechanistic understanding of molecular components underlying the ability of microorganisms to electrically interact with insoluble materials and other cells in their environment (Kracke et al., 2015 ). The cross‐disciplinary integration of the recently gained knowledge in each of these fields is uniquely poised to produce novel living electronics that leverage the unique capabilities of biological and solid‐state systems. Herein, we summarize recent advances in the characterization of biological entities as electrical components, the integration of biological entities into electrical devices, and provide a prospective for new types of electrical devices that could be constructed using living components. Living electronics offer unique capabilities including the ability to self‐assemble, self‐power, repair, biodegrade, process molecular information and store information in low‐energy structures. Living electronics can be produced with renewable feedstocks at ambient temperature and pressures, thereby offering a path towards sustainable production of electronics. To leverage these unique capabilities the field needs a framework to integrate these components into conventional electrical circuits."
} | 1,407 |
31618939 | PMC6843568 | pmc | 7,786 | {
"abstract": "Energy harvesting from human-body-induced motion is mostly challenging due to the low-frequency, high-amplitude nature of the motion, which makes the use of conventional cantilevered spring-mass oscillators unrealizable. Frequency up-conversion by mechanical impact is an effective way to overcome the challenge. However, direct impact on the transducer element (especially, piezoelectric) increases the risk of damaging it and raises questions on the reliability of the energy harvester. In order to overcome this shortcoming, we proposed a transverse mechanical impact driven frequency up-converted hybrid energy harvester for human-limb motion. It utilizes the integration of both piezoelectric and electromagnetic transducers in a given size that allows more energy to be harvested from a single mechanical motion, which, in turn, further improves the power density. While excited by human-limb motion, a freely-movable non-magnetic sphere exerts transverse impact by periodically sliding over a seismic mass attached to a double-clamped piezoelectric bimorph beam. This allows the beam to vibrate at its resonant frequency and generates power by means of the piezoelectric effect. A magnet attached to the beam also takes part in generating power by inducing voltage in a coil adjacent to it. A mathematical model has been developed and experimentally corroborated. At a periodic limb-motion of 5.2 Hz, maximum 93 µW and 61 µW average powers (overall 8 µW·cm −3 average power density) were generated by the piezoelectric and the electromagnetic transducers, respectively. Moreover, the prototype successfully demonstrated the application of low-power electronics via suitable AC-DC converters.",
"conclusion": "5. Conclusions and Future Works This paper presents a human-limb motion driven, piezoelectric and electromagnetic hybrid energy harvester that utilized the frequency up-conversion technique by the transverse impact mechanism. Instead of using any resonant structure (e.g., compliant cantilever beam), a freely movable non-magnetic metallic sphere was used as the low-frequency oscillator, which overcomes the limitations of designing energy harvesters for human-body-induced motion. Use of two transducers allows simultaneous power generation from a single mechanical excitation, which increases the power density of the harvester. The theoretical model was derived based on its working principle, and then a macroscale prototype was fabricated and tested. A series of tests were carried out to partially optimize its parameters and to observe its output performances. The piezoelectric and electromagnetic transducers of the prototype energy harvester simultaneously generated maximum 93 µW and 61 µW average powers, respectively, while excited by human-limb motion at ~2 g peak acceleration. Analysis of the measured voltage and acceleration data shows that the frequency was up-converted to 818 Hz from 5.2 Hz human-limb motion. In order to utilize the harvested energy for practical low-power electronics applications, suitable AC–DC converters (rectifier for PE and voltage multiplier for EM transducers) were constructed and demonstrated. For a functional volume of 19.2 cm 3 , the average power density of the hybrid energy harvester prototype was 8 µW cm −3 , which is ~1.5× higher than its electromagnetic only counterpart (5.4 µW cm −3 ). However, the generated power and the power density of the harvester was still low as the size, mass, and diameter of the ball, height and curvature of the attached mass-top, stiffness of the piezoelectric beam, etc. were chosen arbitrarily, which are all significantly related to power generation. Further optimization of these parameters would be able to deliver higher power within a reduced volume. A more portable design and lighter packaging material should be adopted for its intended use. Our future work will include further optimization of the design parameters (e.g., spring stiffness, mechanical and electrical damping, transverse impact, magnet-coil assembly, etc.) through finite element analysis (FEA) tools, and to fabricate a compact and smaller device with improved output performances to be efficiently used in powering portable and wearable smart devices from human-body-induced motion.",
"introduction": "1. Introduction With the ongoing development of microelectronic technologies, multiple low-power consuming wireless sensor devices are being embedded within hand-held and wearable consumer electronics. These devices are mainly powered by an external power source (e.g., electrochemical batteries) of the consumer electronics and their continuous use allows it to run out of power quickly. Compared to device technologies, the development of the power sources (i.e., batteries) are still slower, even though the devices require less power to operate. Electrochemical batteries have a limited lifespan, and require periodic charging that is inconvenient or sometimes impossible. Moreover, since most batteries contain toxic chemicals, disposal of the expired batteries produces hazardous waste that enhances environmental pollution and poses threats to human and animal health. Therefore, there is great interest in developing self-powered electronics for uninterruptible and long-lasting operation by eliminating the need for recharging or replacing the power source. In recent years, energy harvesting from surrounding energy sources (e.g., light, heat, sound, vibration, etc.) has drawn much attraction to address these circumstances [ 1 , 2 , 3 ]. Among these sources, vibration is the most attractive physical energy source due to its versatility, incorruptibility, and abundance in nature [ 4 ]. However, different vibration sources (e.g., human and machine motion, water and wind flow, rotary motion etc.) generate vibrations of different frequencies and amplitudes, and mostly exhibit low-frequency, large-amplitude characteristics with various cyclic movements in different directions [ 5 , 6 , 7 ]. These vibrations, in the form of kinetic energy, can effectively be converted into electrical energy by employing compatible electromechanical transduction mechanisms that include piezoelectric [ 8 ], electromagnetic [ 9 ], electrostatic [ 10 ], magnetostrictive/magnetoelectric [ 11 ], and triboelectric [ 12 ] mechanisms. The performance of a vibration energy harvester greatly depends on the characteristics of vibration, the type of transducer, and how the transducer is coupled to the mechanical system. Generally, vibration energy harvesters utilize an inertial mechanism employed by a cantilevered spring-mass system, having a specific resonant frequency. Harvested energy (power) is at its maximum when the harvester’s resonant frequency matches the applied vibration frequency. Unfortunately, the power output decreases dramatically as the frequency of excitation (i.e., the resonant frequency of the harvester) decreases [ 13 ]. Moreover, employing a cantilevered spring-mass system for low-frequency (<10 Hz) energy harvesting is quite challenging due to the size constraints for specific application. Human-body-induced motion (e.g., walking, running, shaking limbs, etc.) also generates low-frequency (<6 Hz) vibrations, which do not allow the cantilever structure to be employed conveniently [ 14 ]. Hence, efficient energy harvesting from human-body-induced motion for hand-held and wearable smart devices requires clever design choices. Micro/nano-structured triboelectric nanogenerators [ 15 , 16 ], flexible piezo-composite based piezoelectric nanogenerators [ 17 , 18 ] etc. have shown great application potential in wearable biomechanical energy harvesting and motion sensing. However, they require huge efforts in material development, which was not of our interest. Our primary interest was to design and develop inertial based, low-frequency (e.g., human-body-induced motion) energy harvesters. The mechanical frequency up-conversion mechanism [ 19 ], among numerous design approaches over the past few years, has become the mainstream approach for human-motion based energy harvesting. It allows the transducer element (in the form of a spring-mass system) to actuate at its own resonant frequency (considerably high) by a low-frequency oscillatory or rotary system that responds to the external low-frequency vibration generated by human-motion. Commonly used methods of mechanical frequency up-conversion include mechanical impact and plucking [ 20 , 21 , 22 , 23 ]. Impact excitation transfers an instantaneous momentum into the transducer element whereas plucking excitation implies a slow deflection of the transducer element followed by its sudden release. In general, these methods exert direct force straight to the transducer element that could potentially lead to damage, especially in the case of piezoelectric devices. In order to overcome these issues with piezoelectric energy harvesters, we introduced the transverse impact-based frequency up-conversion mechanism in a human handy-motion driven electromagnetic energy harvester by employing a double-clamped FR4 cantilever beam as a high-frequency oscillator and a freely movable sphere as a low-frequency oscillator [ 24 ]. The transverse impact mechanism meets the reliability challenge and the freely-movable sphere allows the device to operate efficiently at extremely low frequencies (with sufficiently large amplitudes) of handy-motion vibration, meaning its non-resonant behavior [ 25 ]. However, the device generates low power and its average power density is poor. In order to improve its performance, we attempted to hybridize our previous work by incorporating a piezoelectric transducer without cost to the harvester volume. A hybrid energy harvesting technology combines two or more types of transducers that simultaneously capture energy from the same excitation [ 26 , 27 , 28 ]. In this paper, we present the theoretical modeling and experimental characterization of a piezoelectric (PE) and electromagnetic (EM) hybrid energy harvester for human-limb motion by utilizing the transverse mechanical impact-based frequency up-conversion strategy. Transverse impact, created by a sliding sphere over the parabolic tip of a mass attached to a clamped–clamped piezoelectric beam, eliminates the reliability issue from rapid damage of the piezoelectric cantilever due to direct impact. Moreover, simultaneous power generation from both PE and EM transducers offers a higher power density. A theoretical model for the hybrid generator under transverse impact was developed and experimentally validated with a prototype device. The proposed approach has the potential of reliable operation under low-frequency and high-amplitude excitation of human-body-induced motion toward the development of self-powered portable and wearable smart devices.",
"discussion": "4. Experimental Results and Discussion 4.1. Optimal Overlap and Damping Measurements In order to generate maximum possible voltage and power from the prototype, it was important to determine the optimum overlaps between the magnet and coil as well as between the freely-movable sphere and parabolic-top of the proof-mass. The optimum magnet-coil overlap was determined by a benchtop test setup [ 24 ] using an electrodynamic shaker whereas the overlap between the sphere and the mass-top was determined by the human-limb vibration test setup (due to the limitation of the shaker to generate low-frequency, large-amplitude excitation). As seen from Figure 5 a, the optimum magnet-coil overlap was −1 mm. The lateral gap between the magnet and coil was also 1 mm. Since the absolute values were not primarily of interest in determining the optimum magnet-coil overlap, normalized values were used. Figure 5 b shows the change in the open circuit voltages generated by both the PE and EM transducers with the change in the overlap between the sphere and the mass-top. As seen from the figure, the sphere could not make significant contact with the parabolic top when the overlap was 0.2 mm as the clearance between the ball and inner surface of the channel was 0.2 mm. On the other hand, the sphere could not slide over the mass-top and was captured in the middle when the overlap was 0.5 mm because the speed/force of the sphere was not sufficient to pass through. The 0.4 mm overlap between the sphere and mass-top was considered as the optimum value since the open circuit voltages were the maximum for both the PE and EM transducers. The error bars in Figure 5 b indicate the range of voltages generated for multiple attempts as the characteristics of the excitation (frequency and amplitude) applied by human-limb were not always the same. The damping behavior of both PE and EM transducers were determined by the impulse response test using an electrodynamic shaker [ 24 ]. A high amplitude impulse (30.3 ms −2 with 50 ms pulse period and 500 µs pulse width) was applied to the harvester. Then, the mechanical damping ratio ( ζ m ) and total damping ratio ( ζ T ) of both transducers were estimated from the open circuit and loaded impulse response signals, respectively. The logarithmic decrement method was used to calculate the damping ratio as\n (11) ζ = 1 2 π ln ( a 1 a 2 ) \nwhere a 1 and a 2 are the amplitudes of two consecutive peaks in the impulse response plot of the transducer. Subtraction of the mechanical damping ratio ( ζ m ) from the total damping ratio ( ζ T ) gives the electrical damping ratio ( ζ e ). By conducting this experiment, the mechanical damping ratio was found to be 0.011. On the other hand, the electrical damping ratio for the piezoelectric transducer and electromagnetic transducer were 0.017 and 0.016, respectively. It should be noted that the electrical damping values were determined by the impulse response across the corresponding optimum load resistances of the transducers, which were determined by measuring the voltage across various load resistors and calculating the power delivered to them. The power is experimentally equal to V p − p 2 / 4 R l , where V p − p is the peak–peak value of the measured voltage across each load resistance R l . 4.2. Transducer Outputs Figure 6 illustrates the measured peak–peak voltages and peak powers delivered to the load resistances connected to the PE and EM transducers, while the prototype was excited by human-limb motion. Error bars indicate the range of voltage and power values measured for multiple attempts. On average, the maximum of 0.98 mW and 0.64 mW peak power were delivered to 40 kΩ and 85 Ω load resistances connected to the PE and EM transducers, respectively. Note that the optimum load resistances for the PE and EM transducers were in the range of kΩ and Ω, respectively. The resistance of the coil was measured as 84 Ω, which closely matches that of the measured optimum load of 85 Ω. On the other hand, the source resistance ( R source ) of a piezoelectric material depends on its vibration frequency ( f ) and capacitance ( C ), according to R source \n = 1/(2π fC ). The capacitance of the piezoelectric beam (doubly clamped parallel bimorph) was measured as 16 nF and the frequency of its vibration was calculated as 815 Hz (measured as 818 Hz). This gives the calculated R source as 38.3 kΩ, which closely matches the measured optimal load resistance of 40 kΩ. Figure 7 shows the instantaneous voltage and power waveforms across 40 kΩ optimum load resistances of the PE transducer. The voltage and power waveforms generated by the EM transducer also followed the same trend. The maximum peak–peak voltages across the corresponding optimum load resistances generated by the PE and EM transducers were 12.53 V and 0.47 V, respectively. However, the peaks of both voltage and power waveforms decayed exponentially with time due to the damping, which, in turn, reduced their rms (1.92 V for PE and 72 mV for the EM transducers) and average values (93 µW for the PE and 61 µW EM transducers), respectively. Peak amplitudes of the instantaneous power were reduced to almost zero as the time passed, and before the next impact occurred. As a result, the values of average power reduced dramatically. It is to be noted that the waveforms were collected simultaneously, therefore, the overall damping was composed of mechanical damping and the electrical damping of both transducers, as discussed earlier. As seen from the figure, the amplitude decays were not perfectly exponential due to process variation in assembling the harvester components. Two consecutive maximum peaks were generated in one cycle of the applied excitation since the sphere exerted transverse impact on the mass-top twice during its back and forth movement in one cycle. It should be noted that there was no significant change in the peak values of the voltage and power with the change in the frequency of human-limb motion as the variation in the acceleration amplitude was small, however, the values of the rms voltage and average power output changed with the change in the frequency of excitation [ 34 ]. This occurred because of the change in the time interval between two consecutive impacts with the change in the frequency and was also due to the exponentially decaying behavior of the voltage waveform generated by the transducer. The input excitation characteristics (frequency and amplitude of human-limb motion) were measured along each axis of the accelerometer mounted on the prototype during the test. As the harvester prototype was driven along the accelerometer’s Y-axis, the peak acceleration amplitude was maximum in this direction (~2 g), whilst those in other directions were relatively low (~0.95 g along X-axis and ~0.75 g along Z-axis). Data were collected at the 50 Hz sampling rate. The frequency components of both applied acceleration and the generated voltage waveforms were determined by Fast Fourier Transform (FFT) analysis. Figure 8 shows that the frequency of the applied excitation was 5.2 Hz whereas the frequency of the voltage waveform generated by the PE transducer (same for the EM transducer) was 818 Hz, indicating the frequency up-conversion behavior of the harvester. 4.3. AC–DC Conversion The voltage generated by the proposed harvester has alternating (AC) characteristics due to the time-varying characteristic of the input excitation. Most electronic devices are driven by DC voltage source. Therefore, AC–DC conversion is necessary before using the harvested energy. Generally, a full bridge rectifier using four diodes is used to rectify the ac voltage generated by the harvester unit. In our prototype harvester, the voltage generated by the EM transducer was very low when compared to that of the PE transducer. Therefore, a conventional bridge rectifier cannot satisfy the need for rectification and significant voltage generation to drive an electronic load. This is why, a 4-stage Villard’s voltage multiplier circuit was used with the EM transducer whereas a bridge rectifier, on the other hand, was used with the PE transducer for AC–DC conversion, as shown in Figure 9 a. The voltage multiplier rectifies the voltage output with voltage multiplication based on the number of stages used [ 35 ]. The bridge rectifier used four Schottky barrier diodes whereas the voltage multiplier circuit used four pairs of Schottky barrier diodes (HSMS-2852-BLKG, Broadcom Inc., San Jose, CA, USA) and 10 µF, 50 V capacitors, soldered on a printed circuit board (PCB) designed by a professional PCB design tool (Proteus 8.0). The outputs of both bridge rectifier and multiplier circuit were connected to a 33 µF, 50 V storage capacitor (C s ) to accumulate the rectified and multiplied DC electrical energy that was used to power a number of parallelly connected LEDs that demonstrated its application potential, as shown in Figure 9 b. Figure 10 a shows the output AC voltage waveforms of the PE and EM transducers of the prototype harvester (with the rectifier and multiplier connected) while excited by human-limb motion, to be converted to DC and stored in the storage capacitor (C s ). The charging characteristics of the storage capacitor (C s ) was also observed at the same time, as presented in Figure 10 b. The charging behavior is influenced by the inherent output characteristics (voltage and current) of the piezoelectric and electromagnetic transducers where the voltage determines the maximum limit of charging and the current determines the charging speed. As a result, the high output current and low output voltage of the electromagnetic transducer charges the capacitor relatively faster than the low output current and high output voltage of the piezoelectric transducer. When the DC outputs from both transducers were coupled together, the storage capacitor was charged even faster and reached over 2 V DC voltage and was able to turn on the LEDs used as the electronic load."
} | 5,210 |
33324382 | PMC7726332 | pmc | 7,787 | {
"abstract": "Bacteria in the genus Geobacter thrive in iron- and manganese-rich environments where the divalent cobalt cation (Co II ) accumulates to potentially toxic concentrations. Consistent with selective pressure from environmental exposure, the model laboratory representative Geobacter sulfurreducens grew with CoCl 2 concentrations (1 mM) typically used to enrich for metal-resistant bacteria from contaminated sites. We reconstructed from genomic data canonical pathways for Co II import and assimilation into cofactors (cobamides) that support the growth of numerous syntrophic partners. We also identified several metal efflux pumps, including one that was specifically upregulated by Co II . Cells acclimated to metal stress by downregulating non-essential proteins with metals and thiol groups that Co II preferentially targets. They also activated sensory and regulatory proteins involved in detoxification as well as pathways for protein and DNA repair. In addition, G. sulfurreducens upregulated respiratory chains that could have contributed to the reductive mineralization of the metal on the cell surface. Transcriptomic evidence also revealed pathways for cell envelope modification that increased metal resistance and promoted cell-cell aggregation and biofilm formation in stationary phase. These complex adaptive responses confer on Geobacter a competitive advantage for growth in metal-rich environments that are essential to the sustainability of cobamide-dependent microbiomes and the sequestration of the metal in hitherto unknown biomineralization reactions.",
"introduction": "Introduction Metal micronutrients such as nickel (Ni II ), cobalt (Co II ), manganese (Mn II ), and iron (Fe II ) are essential for life yet toxic above relatively low concentrations ( Buccella et al., 2019 ). Not surprisingly, microorganisms have evolved numerous adaptive responses to import the essential metals from the environment while preventing their excessive intracellular accumulation and intoxication ( Chandrangsu et al., 2017 ). Metal homeostasis is primarily achieved by the antagonistic activities of metal importers and exporters ( Chandrangsu et al., 2017 ). Cells often use high affinity transporters to import the metals with specificity and rely on specialized proteins and chaperones to integrate them into pathways dedicated to the synthesis of metalloproteins and enzyme cofactors ( Buccella et al., 2019 ). Collectively, biometals contribute to the synthesis of up to one third of the cell’s proteome and to metabolic functions essential to the growth and survival of the cell ( Buccella et al., 2019 ). Each of these metals must be available in just the right intracellular concentration (i.e., the cellular metal quota) to prevent intoxication ( Outten and O’Halloran, 2001 ). Thus, dedicated metalloregulatory systems monitor the intracellular metal levels and modulate the expression of transporters and other proteins essential for metal homeostasis ( Chandrangsu et al., 2017 ). Metal exporters provide the primary mechanism to eliminate excess metal ( Chandrangsu et al., 2017 ) but the cellular response to metal intoxication is often more extensive, as cells have to cope with the direct and indirect impacts of the reactive metals on proteins and DNA. For example, Co II can bind and inactivate numerous proteins non-specifically, displace other metals (particularly, Fe II ) from prosthetic groups and metal-binding sites, and generate free radicals ( Valko et al., 2005 ). Its high affinity for thiol groups disrupts disulfide bonds in proteins, reduces the free thiol pool and can interfere with sulfur assimilation ( Barras and Fontecave, 2011 ). Hence, Co II intoxication causes generalized damage in the cells, requiring extensive reprogramming to cope with multiple stressors. The essentiality yet toxicity of metal micronutrients such as Co II exerts selective pressure on microorganisms to tune their metabolism to the fluctuating availability of the metal species from geochemical sources. Yet, many aspects of the biological cycling of metal micronutrients remain relatively obscure. This is particularly true for Co II , a metal micronutrient that some microorganisms assimilate to produce enzyme cofactors (cobamides) in the cobalamin (vitamin B 12 ) family involved in metabolic reactions essential to all living cells ( Shelton et al., 2019 ). Genes encoding cobamide-dependent enzymes are widespread in prokaryotes but only a fraction of surveyed genomes have complete pathways for de novo cobamide synthesis ( Zhang et al., 2009 ; Shelton et al., 2019 ). As a result, most microorganisms need to salvage cobamides from the environment, a nutritional dependency that drives syntrophic interactions with cobamide producers ( Seth and Taga, 2014 ). Cobamide-dependent microbiomes depend on the ability of cobamide producers to import and assimilate the soluble Co II cation. The divalent species, however, readily oxidizes to Co III on the surface of Mn IV oxide particles ( Crowther et al., 1983 ; Kay et al., 2001 ). Co II mobility in soil and sediment systems is also limited by the tendency of the metal to coprecipitate with Fe III and Mn IV oxide minerals ( Krupka and Serne, 2002 ). Additionally, Fe III and Mn IV oxides sorb large amounts of the metal cation, sequestering it in solid phases that reduce its bioavailability ( Backes et al., 1995 ). The absorption and co-precipitation of most of the available Co II into Fe III and Mn IV minerals gives Fe III and Mn IV -reducing bacteria, such as those in the genus Geobacter , a competitive advantage for growth in cobamide-dependent microbiomes ( Figure 1 ). These bacteria gain energy for growth from the reductive dissolution of the metal oxides, which are reactions that solubilize Fe II and Mn II and remobilize Co II and Co III ( Reguera and Kashefi, 2019 ). Geobacter species are also important drivers of organic matter degradation, a process that generates organic chelators with affinity for Co III . This keeps the trivalent species in solution and available for use as an electron acceptor ( Reguera and Kashefi, 2019 ). The dissimilatory reduction of chelated forms of Co III by Geobacter reduces Co III to Co II ( Reguera and Kashefi, 2019 ). The low reduction potential of the Co II species (−0.28 V versus standard hydrogen electrode, SHE) and its toxicity to bacteria at relatively low concentrations have been assumed to prevent its biological reduction to Co 0 ( Hau et al., 2008 ; Cosert and Reguera, 2019 ). Despite its toxicity, Geobacter species, including the model laboratory strain Geobacter sulfurreducens , assimilate Co II to synthesize cobamides, which they secrete to sustain several syntrophic partners ( Yan et al., 2012 ) ( Figure 1 ). These syntrophic interactions are favored in local epigenetic zones enriched in Fe III and Mn IV oxides, where Co II preferentially accumulates ( Burkhardt et al., 2009 ). This raises yet unexplored questions about the cellular tolerance of Geobacter species for Co II and the mechanisms that allow these microorganisms to survive and even thrive in Co II -rich environments. FIGURE 1 Known contribution of Geobacter species to the cycling of cobalt (Co). Geobacter bacteria reduce chelated and mineral forms of Co III to Co II , whose bioavailability is limited by the tendency of the Co species to adsorb and/or co-precipitate with Fe III and Mn IV oxides. The reduction of Fe III and Mn IV oxides by Geobacter bacteria solubilizes Co II for the synthesis of cobamides that support the growth of syntrophic partners. We gained insights into the environmental controls of Geobacter activities in cobamide-driven microbiomes by investigating the adaptive responses of G. sulfurreducens to growth and reproduction in the presence of Co II . Consistent with environmental exposure, we demonstrate high Co II resistance in this laboratory strain and describe pathways for protein and DNA repair, cell envelope modifications, and biofilm formation that allow the cells to effectively cope with Co II stress. Importantly, we show that metal acclimation activates respiratory chains that could participate in the reductive precipitation of the metal on the cell’s surface to alleviate toxicity. These adaptive responses allow Geobacter species to grow in Co II -rich environments, sustaining the productivity of the native microbiomes and contributing to hitherto unknown reactions of the Co cycle.",
"discussion": "Discussion The high Co II tolerance and complex acclimation response of G. sulfurreducens is consistent with selection mechanisms during long-term environmental exposure to the metal. Fe III and Mn IV oxides form heterogenous mixes with natural organic matter and metal micronutrients ( Huang and Zhang, 2020 ) that provide optimal conditions for the growth of Geobacter species ( Reguera and Kashefi, 2019 ). The high reactivity of the Fe III and Mn IV (hydr)oxides sequesters Co II and other metal cations in the mineral phases ( Backes et al., 1995 ; Krupka and Serne, 2002 ), concentrating them in the metal oxide-rich epigenetic zones ( Burkhardt et al., 2009 ). The reductive dissolution of the metal-bearing minerals mobilizes the metal cations ( Huang and Zhang, 2020 ) and increases their concentration in the pore-water to toxic levels ( Weber et al., 2009 ). Cu II , for example, can be mobilized to levels (∼20 μM) above the minimum concentration (10 μM) that inhibits the growth of G. sulfurreducens in the laboratory ( Kimber et al., 2020 ). Yet, this bacterium grew from low cell densities, albeit with trade-offs in growth efficiency, in the presence of up to 1 mM CoCl 2 ( Figure 3 ). Furthermore, it was relatively unaffected when exposed to the same metal concentrations during the exponential phase of growth ( Figure 7 ). We attributed this to the expression in exponentially growing cells of the biofilm EPS ( Rollefson et al., 2011 ), which can shield the cells from metal infiltration. Cell density can also affect cellular metabolism and the secretion of metabolites that change the chemical speciation, bioavailability, and toxicity of metals ( Franklin et al., 2002 ). Furthermore, increases in cell numbers activate stress responses that acclimate the population and increase tolerance to a number of stressors ( Li et al., 2001 ). By contrast, cells inoculated at low densities must first reprogram their physiology to acclimate to and initiate growth in the presence of the metal stressor. Acclimation is evident in the extended periods of growth arrest ( lag phase) that G. sulfurreducens cultures initially experienced when growing with sublethal concentrations of CoCl 2 ( Figure 3 ) and in the multiple cellular pathways that were activated to couple growth to Co II detoxification ( Figure 4 ). The transcriptomic studies provided insights into the extensive transcriptional reprogramming that allowed G. sulfurreducens to cope with Co II stress ( Figure 5 ). Transcript levels for Co II importers remained constant, consistent with the absence in G. sulfurreducens of transcriptional regulators (e.g., CzrA and CoaR) that other bacteria use to directly control Co II uptake for metal homeostasis ( Waldron and Robinson, 2009 ). Instead, G. sulfurreducens acclimation involved metal (PII-NG) and heme (GSU0356 histidine kinase) sensors and a transcriptional regulator of central metabolism (HgtR) ( Figure 5 ). Cells also upregulated a CzcABC pump for proton-driven export of metal traversing the outer membrane, a canonical mechanism used by other Gram-negative bacteria to increase metal resistance ( Ma et al., 2020 ). In addition, Co II upregulated a periplasmic glutaredoxin, which repairs and rejoins cysteines oxidized by Co II to refold proteins to their native and functional conformation ( Ezraty et al., 2017 ). The activation of a periplasmic MauG-like di-heme cytochrome c peroxidase (GSU1538) suggested that Co II accumulated in the periplasm at levels sufficiently high to generate H 2 O 2 ( Barras and Fontecave, 2011 ). Di-heme cytochrome c peroxidases detoxify H 2 O 2 in the periplasm by reducing it to two H 2 O molecules ( Pettigrew et al., 2006 ). This reaction receives electrons from a dedicated electron donor such as the monoheme cytochrome GSU2513, which was also upregulated by Co II ( Table 1 ). Without the peroxidase-cytochrome pair, H 2 O 2 would oxidize solvent-exposed [4Fe-4S] 2+ clusters in proteins, producing inactive [3Fe-4S] 3+ species that abolish the redox activity of the metalloprotein ( Imlay, 2008 ). The detoxification of H 2 O 2 is also important to prevent Fenton-like reactions that generate highly toxic ∙OH radicals and exacerbate oxidative stress ( Leonard et al., 1998 ). Despite mechanisms for periplasmic detoxification, Co II may have infiltrated the cytoplasm and damaged essential macromolecules. The presence of cytoplasmic chelators such as glutathione facilitates reactions between Co II and H 2 O that generate ROS and oxidatively damage DNA ( Leonard et al., 1998 ). Co II can also bind components of the CRISPR Cascade complex that mediates antiviral defense, changing its specificity for target DNA and stimulating its RNA-independent DNA cleavage activity ( Sundaresan et al., 2017 ). To cope with DNA damage, G. sulfurreducens activated the expression of UvrD, a helicase of the nucleotide excision repair pathway ( Kamarthapu and Nudler, 2015 ) and transcription-coupled repair ( Epshtein et al., 2014 ). The latter is particularly important to maintain the transcriptional activity of the cell during metal intoxication. This is because UvrD associates with NusA to backtrack RNAP when stalled at a DNA lesion. The helicase then recruits the UvrAB repair complex to the damaged site ( Epshtein et al., 2014 ). This intervention allows the RNAP to resume transcription as soon the lesion is repaired ( Kamarthapu and Nudler, 2015 ). The Irving-Williams series (Mn II < Fe II < Co II < Ni II < Cu II > Zn II ) predicts greater stability for Co II than Fe II or Mn II complexes independently of the ligand ( Hill and Sadler, 2016 ). As a result, Co II intoxication preferentially impacts Fe II and Mn II metalloproteins. To prevent the retention of the toxic metal in the metalloproteome, G. sulfurreducens downregulated non-essential proteins with Fe II prosthetic groups ( Figure 5 ). Nearly all of the downregulated proteins contained Fe-S clusters or metallocenters coordinating Fe II atoms ( Table 2 ). The chemical similarities with Fe II facilitate the infiltration of Co II into Fe-S clusters but the greater electron density of Co II alters the coordination of the metal with the enzyme and its activity ( Thorgersen and Downs, 2007 ; Waldron and Robinson, 2009 ). Co II is also able to compete with Fe II for binding to the porphyrin ring of heme groups such as those in cytochromes ( Thorgersen and Downs, 2007 ). This could be catastrophic in the periplasm, where heme-containing respiratory chains are particularly abundant. Co II -hemes are weaker transporters of charges than the native Fe II -hemes ( Majtan et al., 2011 ), impairing, or even abolishing, respiratory growth. To compensate for this, G. sulfurreducens downregulated non-essential heme-containing proteins such as the cytochrome bd oxidase subunits CydAB required for aerobic respiration ( Figure 5 ). Similarly, cells downregulated genes encoding the formate dehydrogenase complex (the Fe-S cluster protein FdnH and the cytochrome b FdnI) and the secretory accessory protein FdnT, as these proteins are only needed for formate-dependent growth. Cells also downregulated an Rrf2 protein (GSU1639), which uses cysteine residues to bind Fe-S clusters and co-regulate Fe-S cluster biosynthesis and Fe II homeostasis ( Schwartz et al., 2001 ). The high affinity of Co II for cysteines may prevent Rrf2 protein from sensing Fe-S cluster availability in the cytoplasm. To prevent further deregulation of Fe II homeostasis, cells downregulated the rrf2 gene ( Table 2 ). The principles of the Irving-Williams series ( Hill and Sadler, 2016 ) also explain the high affinity of Co II for Fe II -heme. Downregulating non-essential proteins with Fe II -hemes can provide some partial relief ( Table 2 ). However, Co II can also infiltrate the Fe II -hemes during their biosynthesis and prevent their incorporation into proteins. This leads to the accumulation of free Co II -hemes in the cytoplasm and cytotoxicity ( Lin and Everse, 1987 ). The upregulation of a heme-containing histidine kinase (GSU0356) ( Table 1 ) could provide a mechanism to sense the impact of Co II on the heme pool and coordinate the heme detoxification response, as reported in other bacteria ( Anzaldi and Skaar, 2010 ). The advantage of this heme-sensing mechanism is that cells can simultaneously co-regulate heme biosynthesis to Co II and Fe II homeostasis ( Dailey et al., 2017 ). We initially reasoned that Co II infiltration in the free hemes could have increased the intracellular levels of Fe II and exacerbate metal toxicity ( Lin and Everse, 1987 ; Anzaldi and Skaar, 2010 ). For example, free Fe II , like Co II , can generate ROS via Fenton chemistry and cause intracellular damage ( Everse and Hsia, 1997 ). However, although Co II and Fe II intoxication had overlapping transcriptional responses, most of the shared gene targets were reversely regulated ( Figure 4C ). Thus, cells faced conditions of Fe II limitation during Co II intoxication. The accumulation of Co II in the periplasm could competitively exclude Fe II from import across the inner membrane, reducing its intracellular availability. Furthermore, once removed from metalloproteins and prosthetic groups, Fe II can be sequestered non-specifically by cytoplasmic chelators, effectively reducing its intracellular availability. In addition to mechanisms for metal detoxification in the periplasm and cytoplasm, G. sulfurreducens induced pathways that could have promoted the extracellular immobilization of the metal. For example, cells upregulated outer membrane lipoproteins that could have modulated the permeability of the outer membrane ( Nikaido, 2003 ) and/or function as adhesins to promote cell-cell aggregation ( Konovalova and Silhavy, 2015 ). Additionally, Co II triggered the expression of EPS-associated proteins (PEP-CTERM proteins) typically expressed by biofilm-forming bacteria ( Haft et al., 2006 ). The synthesis by planktonic cells of G. sulfurreducens of the biofilm EPS (Xap) precedes biofilm formation and allows the cell to anchor to the Xap matrix cytochromes needed for metal reduction ( Rollefson et al., 2011 ). This redox activity could allow the planktonic cells to reductively precipitate Co II on the cell surface, generating the metal nanoclusters visualized by TEM ( Figure 6A ). The mineral particles resolved by TEM formed on discreet foci on the cell surface, similarly to the distribution of outer membrane cytochromes of the Pcc complexes ( Qian et al., 2007 ). Furthermore, the Pcc outer membrane cytochromes can bind and reductively precipitate divalent metal cations to their elemental form (e.g., Hg II to Hg 0 ) ( Hu et al., 2013 ). A similar reaction could allow the cytochromes to reductively precipitate Co II to Co 0 on the cell surface. The Pcc outer membrane cytochrome complexes contain periplasmic and extracellular c -type cytochromes within an outer membrane porin to electronically connect periplasmic carriers to extracellular electron acceptors ( Shi et al., 2016 ). The upregulation by Co II of a respiratory cytochrome bc complex (Cbc5) could provide a mechanism for energy conservation from the reduction of Co II at the Pcc foci ( Figure 5 ). The Cbc5 complex is anchored to the inner and outer membranes and could interact with the periplasmic cytochrome of the Pcc complex to complete the electron transfer pathway to Co II ( Figure 5 ). Although none of the Pcc genes were differentially expressed by Co II , we confirmed the upregulation of the Pcc outer membrane c- cytochrome OmcC (GSU2731) when the false discovery rate (FDR) threshold was increased from 0.05 to 0.08. This could indicate that some cells may be upregulating the PccC cytochrome. Alternatively, cells may constitutively produce the Pcc complexes under the culture conditions used in our study. Experimental testing of this hypothesis is warranted. The expression of lipoprotein adhesins and a redox-active EPS could also have allowed cells to aggregate and form biofilms ( Figure 7A ), an adaptive response that confers on G. sulfurreducens increased resistance to soluble, toxic metals ( Cologgi et al., 2014 ). The downregulation of a cytoplasmic diguanylate cyclase (DGC) with a canonical GGDEF domain (GSU1643) ( Table 2 ) in Co II -stressed cells may have reduced the intracellular levels of c-di-GMP in order to regulate the planktonic-to-biofilm transition. Most DGC enzymes contain sensory domains that modulate the synthesis of the bacterial second messenger bis-(3′,5′)-cyclic dimeric guanosine monophosphate (c-di-GMP) to specific input signals, including metals. For example, Zn II reversibly binds the subunits of the E. coli DgcZ dimer (formerly YdeH) to allosterically regulate the synthesis of c-di-GMP ( Zähringer et al., 2013 ). The Geobacter DGC enzyme does not have metal-binding domains but has instead the N -terminal phosphoreceiver (REC) domain of DGCs in the PelD superfamily ( Table 2 ). The best studied PelD-like DGC is WspR, the response regulator of the Wsp chemosensory pathway that regulates cell-cell aggregation and biofilm formation in Pseudomonas aeruginosa ( D’Argenio et al., 2002 ). Phosphorylation of the receiver domain in the WspR dimer activates the synthesis of c-di-GMP and autoaggregative/biofilm phenotypes ( Hickman et al., 2005 ). Mg II cations bind near the receiver’s active site of the WspR dimer and contribute to its activity ( De et al., 2008 ). The downregulation in Co II -stressed cells of the DGC enzyme could reflect a feedback mechanism to the infiltration of Co II in the protein ( Waldron and Robinson, 2009 ). Alternatively, Co II -stressed cells may have downregulated the WspR-like DGC to reduce GTP demand for c-di-GMP and increase the availability of the nucleotide triphosphate for EPS synthesis ( Rehm, 2010 ). The EPS matrix can then promote cell-cell aggregation and biofilm formation as a protective mechanism against metal toxicity ( Cologgi et al., 2014 ). Biofilm formation in G. sulfurreducens embeds the cells in an electroactive matrix of cytochromes and conductive pili that effectively immobilizes soluble metals ( Cologgi et al., 2014 ). The conductive pili are particularly important to overcome metal toxicity in biofilms because they provide a large redox surface area for the extracellular immobilization and reductive precipitation of toxic metals ( Cologgi et al., 2011 , 2014 ). The pilus surface is decorated with specialized motifs optimal for the coordination of divalent metal cations ( Feliciano et al., 2015 ). These metal traps have high affinity for Co II and, at high enough potentials, can reductively precipitate it as Co 0 nanoparticles ( Cosert and Reguera, 2019 ). Furthermore, the conductive pili are retractable appendages ( Speers et al., 2016 ), a dynamic feature that allows cells to detach the minerals and recycle the structural peptides in the membrane for a new cycle of pilus polymerization and metal reduction ( Reguera, 2018 ). We did not identify in the Co II transcriptome any of the genes encoding proteins of the pilus biosynthetic apparatus ( Table 1 ) nor did we observe pilus filaments by TEM ( Figure 6 ). This was not unexpected because we used growth temperatures (30°C) that prevent pilus assembly in planktonic cells ( Reguera et al., 2005 ; Cologgi et al., 2011 ). Under these conditions, cytochrome respiratory chains involving outer membrane Pcc complexes provided the primary pathway for extracellular electron transfer in Co II -stressed cells. Thus, Pcc cytochromes could have promoted the mineralization of Co II on discreet surface foci as a detoxification mechanism ( Figure 6 ). The presence of metal nanoclusters on the surface of Co II -treated cells suggests that hitherto unknown biological reactions could contribute to the geochemical cycling of this important metal. We estimated that, on average, cells removed from the solvent 25 μM concentrations of Co II ( Figure 6C ). As a comparison, the intracellular Co II quota is in the low to sub-μM range and typically below the limits of detection of mass spectrometry assays ( Outten and O’Halloran, 2001 ). Co II biomineralization may be more significant in biofilms, thanks to the concentration in the biofilm matrix of conductive pili ( Cologgi et al., 2014 ; Steidl et al., 2016 ) with high affinity motifs for Co II binding and reduction to Co 0 ( Cosert and Reguera, 2019 ). These adaptive responses confer on Geobacter a competitive advantage for growth in metal-rich environments despite the mobilization of Co II during the reductive dissolution of metal oxide mineral phases. The ability of Geobacter bacteria to reductively precipitate Co II could also alleviate metal stress on syntrophic partners that depend on interspecies cobamide transfer to sustain their metabolism. Furthermore, the formation of Co 0 nanoparticles effectively metallizes the cell surface and could allow Geobacter cells to gain energy from the reduction of low potential electron acceptors and to transfer respiratory electrons to syntrophic partners. Hence, Co II mineralization may help define the niche space of Geobacter -driven microbiomes and provide molecular markers to predict the impact of their activities in the fate of this and other essential elements."
} | 6,492 |
30719335 | PMC6358281 | pmc | 7,788 | {
"abstract": "We have developed an integrated, multienzyme functionalized membrane reactor for bioconversion of a lignin model compound involving enzymatic catalysis. The membrane bioreactors were fabricated through the layer-by-layer assembly approach to immobilize three different enzymes (glucose oxidase, peroxidase and laccase) into pH-responsive membranes. This novel membrane reactor couples the in situ generation of hydrogen peroxide (by glucose oxidase) to oxidative conversion of a lignin model compound, guaiacylglycerol-β-guaiacyl ether (GGE). Preliminary investigation of the efficacy of these functional membranes towards GGE degradation is demonstrated under convective flow mode. Over 90% of the initial feed could be degraded with the multienzyme immobilized membranes at a residence time of approximately 22 s. GGE conversion product analysis revealed the formation of oligomeric oxidation products upon reaction with peroxidase, which may be a potential hazard to membrane bioreactors. These oxidation products could further be degraded by laccase enzymes in the multienzymatic membranes, explaining the potential of multi enzyme membrane reactors. The multienzyme incorporated membrane reactors were active for more than 30 days of storage time at 4 °C. During this time span, repetitive use of the membrane reactor was demonstrated involving 5–6 h of operation time for each cycle. The membrane reactor displayed encouraging performance, losing only 12% of its initial activity after multiple cycles of operation.",
"conclusion": "4. Conclusions Transformation of lignin macromolecules to value-added small molecules through enzymatic degradation on a membrane platform was the primary objective of this study. Successful fabrication of multienzyme (laccase, peroxidase and glucose oxidase) immobilized PVDF microfiltration membrane bioreactors was demonstrated. The functionalized membranes were used for oxidative degradation of the lignin model compound GGE. Our multienzyme immobilized membranes, engineered through the layer-by-layer assembly method, were capable of breaking the main chain linkage in lignin-type molecules. Preliminary investigation revealed over 90% of initial GGE degradation with the multienzyme immobilized membranes under the optimum flow rate. A combination of HPLC, LC-MS analysis on the GGE conversion product confirmed the formation of oligomeric oxidation products upon reaction with peroxidase. The laccase enzymes present in the bioreactors were able to further degrade these oligomeric units, validating the potential use of multienzyme membrane systems. Retention of enzymatic activity (towards GGE degradation) of the membrane reactors was established up to multiple cycles of repetitive use. The membrane reactors were active for about a month time of storage at 4 °C. This study opens up perspectives for further enhancement of multienzyme membrane reactor systems and indicates their potential for applications in the field of the biodegradation of renewables.",
"introduction": "1. Introduction Integration of enzymatic catalysis with membrane technology has attracted growing attention to facilitate functionalized membranes as bioreactors [ 1 , 2 , 3 , 4 ]. Synthetic membranes provide a versatile platform for immobilization of bio-catalysts (enzymes), thereby overcoming the inadequacies of soluble enzymes such as instability, difficult recovery, trouble in handling and non-reusability [ 5 , 6 ]. Enzymatic reactions, on the other hand, encourage emerging technologies incorporating multienzyme systems, making catalytic strategies operative and sophisticated [ 7 , 8 , 9 ]. Immobilized multienzymatic systems that exploit the selectivity of biocatalysts have been developed from time to time [ 10 , 11 ]. In this respect, functionalized membranes with porous support and functional polymer matrices may be an ideal platform for multiple enzyme immobilization and thus aid in developing bioreactors for enzymatic reactions [ 12 , 13 ]. Lignocellulosic feedstocks have received continuous attention as renewable biomass for the generation of biofuels and fine chemicals [ 14 ]. In particular, the highly abundant polymer, lignin, deserves more attention than only getting used for low-value applications such as low-grade fuel [ 15 ]. Plenty of research has been conducted that reports ways to exploit the prospective of lignin as a resource for value-added chemicals [ 14 , 16 , 17 ]. However, practical utility is far away owing to the challenges involved during lignin depolymerization. Most of the methods reported for lignin valorization, such as pyrolysis, catalytic oxidation and/or hydrolysis under supercritical conditions, etc., are either energy-consuming or environmentally unfavorable [ 18 ]. In nature, lignin is degraded by a pool of extracellular ligninolytic enzymes such as peroxidases and laccases over a period of many years [ 19 ]. One of the novel approaches to mimic natural ways of lignin depolymerization is involving multienzymatic reactions. Membrane-based multienzyme systems can be constructed by carrying out sequential deposition onto the membrane pores. The layer-by-layer (LbL) adsorption technique is a general and versatile tool for the controlled fabrication of surfaces and pores by the consecutive deposition of alternatively-charged polyelectrolytes [ 20 , 21 , 22 ]. Efforts have been made to fabricate multienzyme surfaces through LbL techniques for bioprocessing applications [ 23 , 24 , 25 , 26 , 27 ]. Recent interest in enzymatic methods for lignin biodegradation has focused on using enzymes such as peroxidases, laccase, phenol oxidases, etc. [ 28 , 29 , 30 , 31 ]. Peroxidases and laccases exhibit low substrate specificity and relatively wide pH of action and considered as a versatile tool towards oxidative processes [ 4 , 32 , 33 , 34 , 35 , 36 , 37 , 38 ]. However, the exploitation of their potentiality is limited, especially in the case of peroxidases, by their lower stability under harsh operative conditions such as high temperature, presence of surfactant, organic media and elevated level of hydrogen peroxide. Although there are quite a few reports on the use of peroxidase for the lignin oxidative process [ 39 ], horseradish peroxidase (HRP) has been shown to catalyze spontaneous polymerization of a variety of aromatic compounds [ 40 , 41 ]. Such undesired polymeric by-products have to be filtered out of the reactor solution in order to avoid negative feedback to the biocatalysts. On the other hand, in the case of membrane bioreactors, such polymeric products need to be avoided to circumvent clogging and fouling of the membrane. One such technique is to use a multienzyme system to convert such poly-/oligo-meric products formed by peroxidases into simpler ones. We report here a composite membrane with horseradish peroxidase (HRP) and laccase immobilized on it via the LbL assembly technique and its performance towards the degradation of a lignin model compound. The hypothesis is that by the use of such a multienzyme immobilized membrane system, any unnecessary by-products can simultaneously be converted to smaller molecules, thereby prohibiting membrane fouling, as well as enzyme inhibition. In essence, the HRP enzyme partially degrades/modifies the substrate (guaiacylglycerol-β-guaiacyl ether) to an oligomeric unit, which is then degraded by laccase to monomeric units ( Scheme 1 ). As peroxidase enzyme needs hydrogen peroxide as one of the substrates, glucose oxidase (GO) was incorporated as a third enzyme for the in situ generation of hydrogen peroxide. Alternatively for membranes with only laccase and HRP on them, hydrogen peroxide was added to the feed. The main aim of the current study is to understand the activity of such multienzyme-functionalized membranes towards degradation of an aromatic phenolic lignin model compound, specifically guaiacylglycerol-β-guaiacyl ether (GGE). Within that context, our specific goals are: (i) fabrication and characterization of the functionalized membrane by alternating the attachment of cationic and anionic polyelectrolytes via the LbL assembly technique; (ii) investigation of the activity of the membrane reactors towards the degradation of a lignin model under convective flow conditions and analysis of the degradation products; and (iii) evaluation of the long-term performance of the enzyme-functionalized membranes.",
"discussion": "3. Results and Discussions 3.1. Characterizations of the Functionalized PVDF Membranes The functionalized membranes were characterized by ATR FT-IR, FIB-SEM, XPS and water permeability studies to assess the success of polymerization of PAA in the pores of PVDF membranes. ATR FT-IR spectroscopy was used for initial verification of successful fabrication of the PVDF and functionalized PVDF membranes. Figure 1 compares the spectra of a PVDF membrane (PVDF) as supplied, a PAA-PAH-functionalized PVDF membrane (PVDF-PAA-PAH) and an enzyme-functionalized PVDF membrane (PVDF-PAA-PAH-ENZ). The characteristic absorption peaks of the CF 2 groups of the PVDF chains lie in the region of 1050–1280 cm −1 for all the membranes [ 42 ]. The appearance of new peaks at 1720 cm −1 and 1544 cm −1 ( Figure 1 , red line) corresponding to the carbonyl stretch and anti-symmetric stretching of carboxyl groups (COOH), respectively, verify successful polymerization of acrylic acid in the membrane matrix [ 43 ]. Enzymes usually have two absorption bands, near 1645 cm −1 and 1540 cm −1 , corresponding to the peptide backbone amide I and amide II modes, respectively [ 44 ]. Our enzyme-functionalized PVDF membranes showed similar peaks ( Figure 1 , black line). The peak at 1642 cm −1 was due to the stretching vibration of the C=O amide bond, and the 1545 cm −1 peak was assigned to the combination of the bending vibration of the N-H bond and the stretching vibration of the C–N bond of the peptide backbone, signifying immobilization of enzymes on the membrane support [ 35 ]. The peak around 2920 cm −1 and 2850 cm −1 , in all the samples, was attributed to the asymmetric and symmetric C–H stretch. The band peaks in the range of 3500–3300 cm −1 were associated with a combination of the N–H stretch and O–H stretch belonging to the protein and the membrane matrix. The surface morphologies of the PVDF and functionalized PVDF membranes were characterized by scanning electron microscopy (SEM). FIB (focused-ion-beam) SEM was used to access the inside of the fabricated membrane pores as shown in Figure 2 . The bare PVDF membranes ( Figure 2 a) are fairly porous with an average pore size of 225 ± 50 nm. After the PAA polymerization, the morphology of the membrane pores changed substantially, shrinking the pores to an average size of 100 ± 5 nm ( Figure 2 b). Successive modification of the membrane after immobilizing PAH and the enzymes shrunk the pores further, reducing the average diameter to 70 ± 12 nm, as can be seen in Figure 2 c. Figure 2 d shows the inside morphology of the pores of the functionalized membrane using FIB-SEM. The distinction between the PVDF background and the polymeric layer immobilized on it is very clear from the FIB-SEM; however, it is hard to distinguish between an enzyme layer and a PAA-polymeric layer. The surface compositions of bare PVDF membranes and enzyme-functionalized PVDF membranes (PVDF-PAA-PAH-ENZ) were explored by X-ray photoelectron spectroscopy (XPS). Survey spectra of these membranes exhibit significant N and considerably higher C content in the functionalized PVDF membrane ( Figure 3 b) compared to the bare PVDF membrane ( Figure 3 a). The N content in the functionalized membranes is due to the presence of the polyallylamine, as well as from the peptide backbone of the enzymes, consistent with our earlier findings [ 45 ]. Deconvolution of the C1s peak near 290 eV for the functionalized membrane yielded three peaks corresponding to carbon in C–C (285 eV), C–O (287 eV) and the C=O (289 eV) functionalities ( Figure 3 c), in addition to a signature of C–F attributed to the PVDF support [ 45 , 46 ]. Deconvolution of the N1s core-level spectra of the functionalized membrane resulted in two peaks at 399 eV and 401 eV, and these were assigned to C–N (amine/amide) from the enzyme and ammonium salt (C–NH 3 + ) from the polyallylamine layer, respectively. Finally we profiled the effect of pH on the water permeability of PVDF-PAA-functionalized membranes ( Figure 4 ), which serves as a definitive test of PAA functionalization inside the membrane pores [ 47 , 48 ]. The flux of the membrane decreased with increasing pH due to the expected swelling of the PAA hydrogel inside the pores [ 49 ]. The water permeability data were fit with Equation (S1) [ 50 ] yielding a pKa of 5.6. 3.2. Reactivity of the Membrane Bioreactors towards GGE Degradation/Modification The multienzyme immobilized membranes were fabricated as described in the Materials and Methods Section. The initial activity of the laccase and HRP enzymes were measured in the solution state prior to immobilization by the conventional colorimetric assay in the presence of 2,2´-azinobis(3-ethylbenzothiazoline-6-sulfonic acid)-diammonium salt (ABTS) (please see Supporting Information Figure S7 ). It should be noted here that the presence of multiple enzymes in the membrane makes it difficult to measure the individual activity of the immobilized enzymes. However, the loading of the enzyme into the membrane matrix was confirmed by the Bradford protein assay of the enzyme feed and the permeate during enzyme immobilization. In general, 35–60% of each enzyme could be loaded on the membranes. The applicability of the enzyme-functionalized membrane towards degradation of the lignin model compound was demonstrated by passing an aqueous solution of GGE through the membrane in a dead-end cell. The multienzyme immobilized membranes had three enzymes on them (Lac, HRP and GO) to be able to work as a bioreactor. It should be noted here that laccase uses oxygen, whereas the enzyme HRP uses hydrogen peroxide as the electron accepting secondary substrate during the oxidation of the primary substrate. A continuous oxygen atmosphere for the laccase enzyme was maintained by air flow from the air gas tank, which also maintained necessary positive pressure for controlling the flow rate. During the experiments the necessary concentration of hydrogen peroxide was maintained by adding glucose in to the feed, which upon reaction with GO produces hydrogen peroxide in situ. HPLC and LC-MS analyses were used to monitor the diminution of the GGE content in the permeate, as well as to detect the presence of various oxidation products of GGE (please see Supporting Information, Figure S2 ). Figure 5 a portrays the degradation of GGE by enzymatic membranes as a function of different flow rates. The data in Figure 5 a were from the Lac-HRP-GO membrane (Feed 3.1 mM GGE) in the presence (blue diamonds) or absence (red circles) of glucose (3 mM) in the feed. The approximate amount of enzymes on the membrane used in this experiment were laccase 5.6 mg, peroxidase 5.3 mg and glucose oxidase 3.9 mg. The data indicate that with glucose in the feed, close to 95% of the GGE could be degraded at a flow rate of 15 L·m −2 ·h −1 (LMH) under an applied pressure of around 4 bars. At such a slow flow rate (high residence time, 22 s; Figure 5 b), comparable degradation could also be achieved without glucose in solution (~85%, 17-s residence time). Comparison at a higher flow rate (lower residence time) reveals that the membrane worked much better when all three enzymes work simultaneously. A degradation as high as 90% was achieved at 64 LMH (5s residence time, Figure 5 b) with glucose in the feed. In contrast, in the absence of glucose in the feed, the membrane could degrade only 65% of the initial GGE under a similar flow rate. The data in Figure 5 c are the GGE degradation profile from an independent laccase immobilized membrane. These data closely resemble the data from the multienzyme membrane reactor without glucose in the feed ( Figure 5 a, red circle) to prove that in the absence of the necessary substrate, hydrogen peroxide, the multienzyme membrane reactor could act as a single enzyme membrane reactor. This demonstrates the efficacy of the multienzyme membrane towards the degradation of lignin model compounds. Figure 5 d shows the reusability (three repeated cycles) of the multienzyme-functionalized membrane as assessed by retention of its capacity to catalyze the conversion of GGE under convective flow mode. In this case, a membrane with laccase (3.3 mg) peroxidase (2.4 mg) was used, and a stoichiometric amount of hydrogen peroxide was added into the feed (3.1 mM of GGE in water). Each cycle was comprised of passing 100 milliliters of feed over five hours of operation. As evident from the Figure 5 d, our enzyme-functionalized membrane displayed encouraging retention of activity, losing only 12% of its initial activity after multiple cycles of operation. This is consistent with our earlier observation [ 45 ], where we have discussed the similar stability of the enzyme immobilized membrane matrix. Effort was made to characterize the degraded products from the enzymatic reactions. LC-MS analysis of the reaction products of GGE conversion revealed multiple oxidized products with m / z ranging from 251 [M + Na] + up to 979 [M + Na] + . Enzymatic conversion of GGE (MW 320) to multiple polymeric oxidized products with higher molecular weight has been reported by other researchers, as well [ 51 ]. It is noteworthy to point out that the generation of such oligomeric products is enzyme dependent. Our findings on this are that horseradish peroxidase generated the high molecular weight dimeric ( m / z 661, B ) and trimeric ( m / z 979, C ) products to a greater extent than the low molecular weight ( m / z 251, A ). However, laccase produced a higher amount of A ( m / z 251) than B ( m / z 661), and none of the higher ones were seen. To be specific, laccase produced 99% of A and only 0.8% of B of the total GGE conversion product. In contrast, with HRP only 43% of the total GGE conversion product was A . The formation of B , in this case, was ten-times more than that produced from the laccase reaction. Moreover, about 2% of the trimeric adduct C was formed with HRP, which was not seen in the laccase reaction. All the data are tabulated in Table 1 . The effect was also seen in the case of the multienzyme membrane with the formation of 62% of the degradation product A compared to only 0.7% and 0.4% of the oligomeric products B and C respectively. Based on GGE degradation patterns, it can be concluded that reactions performed by HRP resulted in oxidative oligomerization, probably formed through C–C coupling of the phenolic units [ 52 ]. Laccase, on the other hand, degraded the GGE and the oligomeric products from the HRP reactions through an alkyl-phenyl ether bond cleavage reaction [ 38 ]. The fact that laccase was able to break down such oligomers to monomeric units justifies the unification of multiple enzymes with the membrane reactor to protect its activity. Various such multienzymatic systems have been studied from time to time [ 7 , 8 , 9 , 39 , 53 ]. While immobilized enzymes generally have better stability over the solution phase, the shortening of the diffusion time of the substrate or transformed substrate from one enzyme to another enzyme in multienzymatic systems makes them more potent with higher observed catalytic activity. Jia et al. recently discussed a comparison of the efficiency of substrate conversion by such a multienzymatic system to free enzyme, and in a few cases, a decrease in performance was also observed [ 8 ]."
} | 4,940 |
38459056 | PMC10924106 | pmc | 7,790 | {
"abstract": "Bacteria have developed various defense mechanisms to avoid infection and killing in response to the fast evolution and turnover of viruses and other genetic parasites. Such pan-immune system ( defensome ) encompasses a growing number of defense lines that include well-studied innate and adaptive systems such as restriction-modification, CRISPR-Cas and abortive infection, but also newly found ones whose mechanisms are still poorly understood. While the abundance and distribution of defense systems is well-known in complete and culturable genomes, there is a void in our understanding of their diversity and richness in complex microbial communities. Here we performed a large-scale in-depth analysis of the defensomes of 7759 high-quality bacterial population genomes reconstructed from soil, marine, and human gut environments. We observed a wide variation in the frequency and nature of the defensome among large phyla, which correlated with lifestyle, genome size, habitat, and geographic background. The defensome’s genetic mobility, its clustering in defense islands, and genetic variability was found to be system-specific and shaped by the bacterial environment. Hence, our results provide a detailed picture of the multiple immune barriers present in environmentally distinct bacterial communities and set the stage for subsequent identification of novel and ingenious strategies of diversification among uncultivated microbes.",
"introduction": "Introduction Bacteria are under constant threat of infection by a variety of genetic parasites such as bacteriophages (henceforth called phages) 1 . As a result of this strong selective pressure, they have evolved multiple sophisticated defense mechanisms capable of regulating the flux of genetic information spread by mobile genetic elements (MGEs) via horizontal gene transfer (HGT) 2 – 4 . The complete set of bacterial defense systems’ repertoire can be designated as their defensome. Several bacterial defense systems have been discovered and extensively discussed in the literature, revealing two major groupings based on their components and modes of action: innate (non-specific) and adaptive immune systems 5 , 6 . Typical examples of innate immunity include prevention of phage adsorption 7 , restriction-modification (R–M) systems that use methylation to recognize self from non-self-DNA 8 , and abortive infection (Abi), in which the infected cell commits suicide before the invading phage can complete its replication cycle 9 . Recent efforts to de-novo identify microbial defense systems resulted in the discovery of several additional innate immune mechanisms with a wide range of genetic architectures 3 , 4 , highlighting the strong selective pressure imposed by genetic parasites on microbial communities. Adaptive immune systems, on the other hand, are so far exclusively represented by clustered, regularly interspaced short palindromic repeats (CRISPR)-Cas, a family of defense systems that provides acquired immunity through the acquisition of short DNA sequences from MGEs that are incorporated into the host genome as spacers 10 . Large-scale efforts for defense system mapping have been recently propelled by the development of bioinformatic tools such as DefenseFinder 11 and PADLOC 12 that rely on a profuse collection of HMM profiles and specific decision rules for each known defense system. Such mapping has been mainly conducted in bacterial species from reference genome databases (e.g., NCBI RefSeq) that are known to overrepresent acute/common human pathogens and organisms that can largely be cultivated in laboratory 11 – 13 . While extremely insightful, such studies provide a limited snapshot of the bacterial defensome, as they miss the uncharted fraction of environmental microbial diversity that remains uncultured. The current global Earth microbiome has been estimated at ~5 × 10 30 prokaryotic cells 14 scattered throughout a wide range of environments, including deep oceanic and continental subsurfaces, upper oceanic sediment, soil, and oceans as the most densely populated cases. In many environments, 99% of microbes are yet uncultured 15 , while cultured representatives belong overwhelmingly to the phyla Actinobacteria, Bacteroidetes, Firmicutes, and Proteobacteria. For nearly 4 billion years, bacteriophages have co-evolved with bacteria, with recent estimates pointing to the presence of ~10 31 viral particles in the biosphere 16 , and up to 10 23 infection events per second taking place just in the global ocean 17 . During the last decade, extensive progress in high-throughput sequencing technologies and computational methods enabled culture-independent genome-resolved metagenomics to recover draft or complete metagenome-assembled genomes (MAGs) 18 – 20 . The latter have advanced our understanding of the diversity, abundance, and functional potential of microbiota and phageome composition and corresponding ratios across different environments. A healthy adult human gut, for example, is a reservoir for ~4 × 10 13 bacterial cells (mostly Firmicutes and Bacteroidetes) 21 , and low (10 −3 −1) virus-to-prokaryote ratios (VPRs) 22 . In contrast, marine ecosystems typically show larger VPRs (between 8 × 10 − 3 −2.15 × 10 3 , mean of 21.9), followed by soil environments which show the largest ratios (between 2 × 10 − 3 − 8.2 × 10 3 , mean of 704) (reviewed in ref. 23 ). We hypothesize that the strong VPR dynamics across temporal and spatial scales is likely to profoundly shape the defensome arsenal across biomes. In this study, we conducted a large-scale in-depth investigation on the abundance, distribution, and diversity of the defensome in complex bacterial communities from three key environments: soil, marine, and the human gut. We tested the association between defensome and different mechanisms of genetic mobility, the former’s colocalization in defense islands, and assessed the mutational landscape of high-frequency single nucleotide polymorphisms (SNPs) and insertion-deletions (indels) across defense gene families. These results provide a unique view of the interplay between microbial communities and their phage invaders, and will pave the way to the identification of hitherto unknown defense systems and/or other phage-resistance mechanisms across the enormous diversity of yet-uncultivated microbial populations.",
"discussion": "Discussion In this study, we present a large-scale analysis of the abundance and diversity of defensomes of genomes/species from complex microbial communities and three representative biomes: soil, marine, and the human gut. Our results on the quantification of the defensome in marine MAGs lend support to a scenario of a limited defense arsenal in this environment (Fig. 2b ). The latter can be accounted by a variety of potential explanations namely: (i) the fact that oligotrophic open oceans typically show an overrepresentation of clades characterized by heavily streamlined genomes 57 (e.g., Dadabacteria, Chloroflexota) (Supplementary Fig 1 and Supplementary Data 2 ), and thus, more likely to opt for more transient defense systems and little metabolic plasticity to better cope with the limiting environment of the surface ocean; (ii) the dominantly planktonic lifestyle and low cell-density in the marine environment (at least for the free-living fractions accounted for in our MAG dataset) which may in itself, or through a reduced frequency of HGT, contribute to a more limited anti-MGE arsenal; (iii) the fact that the large majority of HMMs currently available to detect defense systems were initially developed on the basis of genetic data that overrepresents not only cultivable bacteria, but also lineages (e.g., Escherichia , Bacillus , Pseudomonas ) that are more distantly related to those that make up the global ocean microbiome (Supplementary Data 3 ). On a broader view, our results qualitatively match those recently obtained for RefSeq genomes in terms of the most abundant systems (R–Ms, CRISPR-Cas) and overall diversity of families identified 11 . The enhanced granularity offered by our cross-environment comparison revealed a few curious differences at the level of preferential ‘second line’ defense families. One of such differences concerns SoFIC and CBASS which are present in roughly 20% of soil and marine MAGs (mainly in Chitinophagales and Caulobacterales), but considerably less predominant (~8%) in human gut MAGs (mainly in Verrucomicrobiales, Enterobacterales, and Bacteroidales) (Fig. 2a ). Inversely, the abortive infection system Rst_PARIS is present in 20% of human gut MAGs (mainly in Bacteroidales) but is virtually absent in soil or marine environments (Fig. 2a ). While R–Ms (and to a lesser extent CRISPR-Cas) are largely ubiquitous, our results are supportive of a “second line” of defense systems (SoFIC, CBASS, Rst_Paris, etc.) that is also mostly non-species-specific, differentially favored across distinct environments, and privileged by distinctive strategies of genetic mobility (Fig. 4c , see below). As we move down the ladder of defense system abundance, we face an increasing variety of cryptic, presumably highly specialized, and more species / population-specific systems. By further splitting our dataset into sub-environments or by geographic location, we observed significant differences in defense system abundance (Fig. 3 ). And while the increased densities observed at serpentine systems and across the Arctic Ocean can be explained by the extreme conditions experienced at such environments and a subsequent phage-bacteria imbalance, the more subtle variations in defense system density in human gut MAGs across multiple countries and the panoply of confounding variables associated, preclude the identification of more explanatory scenarios. Higher densities of defense genes were consistently observed in (or in the close vicinity) of MGEs compared to those found in the chromosome (excluding MGEs) (Fig. 4a ). Such colocalization facilitates the rapid acquisition and/or diversification of the defensome to provide resistance against multiple other MGEs. It was recently suggested that the carrying of certain defense systems by MGEs by a given bacterial host, may not always relate with the latter’s need for protection, but instead, in the best interest of the MGE itself in order to overcome or displace antagonistic MGEs 51 . Our observation of a complex and heterogeneous distribution of defense gene families across different classes of MGEs supports such a hypothesis and suggests an exploitation of MGEs by defense genes/systems for purposes other than host defense. It ultimately highlights the need to better understand the molecular, and evolutionary interactions between the threesome host-phage-mobilome. Genes acquired by HGT, and MGEs in particular, tend to integrate in a small number of chromosome hotspots to decrease the fitness cost of their integration. Successive rounds of integration/excision/partial deletion of MGEs, when accompanied by the co-option of defense genes/systems, may result in the formation of defense islands. While initially thought that the latter were merely “genomic junkyards” in which the defense genes that are frequently acquired via HGT accumulate because insertion in these regions is unlikely to be deleterious, it has now become clear that there is a specific selective advantage in such clustering of genes, such as functional cooperation between different defensive modules and generation of novel functions. When compared across environments, defense islands did not show significant differences in terms of size, relative abundances of major defense families, or at the topmost abundant COG functional categories for genes classified as ‘non-defensive’ (Fig. 5a, c and Supplementary Fig. 8c ). While many of these genes seem to encode factors involved in genetic mobility, others have hitherto unknown functions. In this line, an interesting next step would be to build upon our precise delimitation of defense islands in such a large and phylogenetically diverse MAG dataset and use a previously developed colocalization framework 3 to leverage the identification of novel defense systems. A significant overrepresentation of several defense families (e.g., Hachiman, R–M, Thoeris) was observed in defense islands (compared to non-island regions). Yet, for certain of these families, such overabundance did not translate into a higher likelihood of colocalization with the remainder of the defensome (and vice-versa). These observations point to the possibility of previously unappreciated epistatic interactions or increased probability of functional diversification for a selected subset of families of defense genes/systems in defense islands. In this regard, the extent to which non-canonical HGT mechanisms (e.g., gene transfer agents, nanotubes, membrane vesicles) and MGE-independent mechanisms of diversification (e.g., homologous recombination) respectively shape the movement of defense genes and the evolution of defense islands remains unclear. Under the Red Queen evolutionary dynamics, the coevolution between opposing hosts and parasites portrays evolution as a never-ending evolutionary arms-race between defense and counter-defense strategies. Such antagonistic coevolution pervades evolutionary change through multiple ingenious strategies, including: (i) point mutations in phage DNA recognition sites to reduce the likelihood of restriction by R–M systems 58 ; (ii) phase-variation/inversions/point mutations in MTases, REases or S modules leading to altered R–M systems’ specificity 55 , 59 ; (iii) ON/OFF switch in CRISPR immunity through mutations in cas genes 60 ; among others. Thus, not only turnover and recombination, but also rapid sequence evolution of certain defense genes/systems through mutation are key factors shaping the host-parasite evolutionary trajectory. Such diversification strategies are a function of the size and the diversity of the defensome gene pool in a bacterial population, and will shape how the latter remains evolutionarily responsive to temporally or spatially variable selection imposed by phages. Different defense genes are expected to evolve at different rates. For example, significant differences in purifying selection have been described across different Types of R–M REases and MTases, highlighting distinct signatures of adaptive evolution 13 . To gain a birds-eye-view of potentially coexisting subpopulations bearing substantial defense gene-level diversity, we built upon a metagenome read recruitment approach. This allowed us to identify a subset of defense genes having a higher-than-expected frequency of SNPs + indels, globally evolving under strong purifying selection, and a heterogeneous landscape of mutation types profoundly affected by the environment (and thus by population structure). Whereas for some of these genes we can point out determinants capable of explaining such observations—namely the presence of domains known for their predisposition to genetic variation (e.g., the motility-associated motA domain 61 in zorA , or the ftsk translocase domain 62 in sspH )—the lack of substantial functional and mechanistic insights on the remaining ones (and on their systems) precludes further meaningful ascertainments. It is important to appreciate that our computational analysis is challenged by a few difficult-to-control confounding variables and limitations that are worth discussing. The first, concerns the imbalance in our dataset between the number of samples recovered from each biome, as well as their geographic distribution. While the number of soil and marine MAGs analyzed was essentially the same, human gut MAGs were roughly one order of magnitude greater. From the geographic standpoint, marine samples have a global representation, but soil and human gut microbiome data are greatly biased towards the US and China. These observations highlight a critical need for thorough geographic sampling, more global representation of participants in microbiome studies, and fairer access to genomics resources, especially in resource-poor countries. A second confounding variable, likely more relevant, concerns the fact that MAG binning methods using short reads tend to miss certain low-abundance or difficult-to-resolve MGE families. The fact that defense genes are often carried or colocalize with MGEs, necessarily indicates that our results (i) may have a bias in the ratio of defense genes inside versus outside the mobilome, and (ii) are most likely a partial underestimated picture of the real defensome abundance. Future inclusion of long-read data will enable reference-quality genome reconstruction from metagenomes, and further improve our observations. Third, our observations are not representative of all bacterial communities and are likely influenced by the characteristics of the sampled environments. Still, the stringent dataset filtering used in our study in terms of MAG completeness and N 50 (with associated controls shown in Supplementary Fig. 1 ), together with a previous demonstration on the accuracy of MAG size estimates (that are part of our dataset) compared with reference genomes 26 , makes us have good reasons to think that our analyses constitute a reasonable proxy of the defense landscape diversity carried by such populations, and of the complex interplay underlying their interactions at the intra- and inter-environment level. Lastly, while this study provides novel and intriguing insights into the defensome colocalization, it does not address the specific mechanisms and interactions between different systems, nor the interplay with phage counter-defense strategies 63 , 64 . The efforts recently undertaken to identify novel defense mechanisms in typically easily cultivable bacteria must now be followed by initiatives to expand the search to uncultivated microbes in complex microbial communities, to understand how such mechanisms collaborate or antagonize with one another, how they co-opt or are co-opted by MGEs, and how they are shaped by the surrounding environment. Our work provides a first stepping stone in such a direction."
} | 4,549 |
36514486 | PMC9731846 | pmc | 7,792 | {
"abstract": "Current agricultural practices heavily rely on the excessive application of synthetic pesticides and fertilizers to meet the food demands of the increasing global population. This practice has several drawbacks including its negative impact on the environment and human health. Recently, the use of natural products has gained interest as alternatives to these synthetic agrochemicals due to their selective working mechanisms and biodegradability. In order to efficiently produce these natural agrochemicals, engineering microorganisms is emerging as an increasingly viable approach, and it is anticipated that it will have a significant market share in the near future. This approach manipulates the metabolism of microbes to manufacture the desired natural compounds from low-cost starting materials. This review discusses recent examples of this approach. The produced natural products can serve as biopesticides or plant growth regulators for the sustainable improvement of plant growth and disease control. The challenges in further developing these strategies are also discussed.",
"conclusion": "5 Conclusions and perspectives Natural agrochemicals that are produced from engineered microbes are an attractive alternative to the synthetic agrochemicals. However, there are only a small number of published reports of these agrochemicals being applied in the fields despite their successful and effective laboratory trials. Farmers have developed a strong trust in the effectiveness and quick results of chemical fertilizers and pesticides, making it difficult to adopt and/or replace them with alternative techniques. Furthermore, the titers of the majority of natural agrochemicals remain low, affecting their large-scale production. New fermentation strategies, gene editing tools, and novel computational tools would be useful to improve the product titer. Advanced sequencing techniques can be used to better understand the microbial community of soil and discover novel plant-beneficial microbes which can be further engineered. Protein engineering approaches could be useful to enhance the activity of the enzymes or to eliminate feedback regulation on enzymes. Protein-ligand docking for virtual screening of new plant hormones may be another potential direction for future research. Based on the crop of interest, the properties of the plant hormones can be improved by designing and introducing modifications using computational simulations. It would be interesting to engineer microbes that can harvest sunlight and improve the net photosynthetic rate when deployed in the field. Another interesting idea is to engineer microbes that might improve water absorption and retention in arid lands. The plant-associated microbes from desert plants might be isolated and subjected to adaptive evolution in dry conditions followed by genome sequencing. The identified mutations might be introduced into model microbes, which might then be applied in water-scarcity lands. The development of biosensor strains that can sense plant's needs and then release agrochemicals in a controlled manner would be another attractive strategy. For successful deployment of the above-mentioned engineered microbes in the field, the selection strategy must include evaluation of the microbes’ ability to survive in the environment of interest, including tolerance to the associated stressors such as aridity, heat, and low light. Another major risk is the possibility of gene transfer into native microbes [ 45 ], which must be carefully evaluated during the selection process. Other considerations are their biocompatibility, non-pathogenicity, ability to colonize the plant rhizosphere, effectiveness in competing with the existing microbes, enhanced shelf-life, and eco-friendliness. We envision that a large number of natural agrochemicals can be produced by using engineered microbes as discussed in this review. These strategies could play an important role in the sustainable agriculture over the coming years.",
"introduction": "1 Introduction The rapid growth of the human population and the resulting rise in food demands have imposed a large burden on agriculture [ 1 ]. In the past, plant productivity and crop yields have been considerably enhanced through the extensive use of synthetic fertilizers and pesticides. However, significant concerns were raised regarding their undesired impacts on the environment and ecosystems [ 2 ]. Many of these non-natural chemicals remain in the soil for an extended amount of time [ 3 ], affecting soil fertility and long-term agricultural productivity. In addition, chemical pesticides frequently result in the death of non-targeted beneficial organisms and the development of pesticide resistance [ 4 ]. Many pesticides are also lost to the atmosphere or water resources causing environmental pollution [ 5 , 6 ]. Some pesticide residues could also contaminate the produce and pose health threats to the farmers who apply them due to their prolonged exposure [ [7] , [8] , [9] ]. Because of the above concerns, the use of natural products to control plant diseases has recently received increasing attention. These products are non-toxic, selective towards target pests, and biodegradable [ 10 , 11 ]. An example is azadirachtin, a natural insecticide derived from neem oil, which is widely used in organic farming [ 12 , 13 ]. Many of these molecules are extracted from plants. Unfortunately, they often incur high extraction and purification cost due to their low content in plant materials [ 14 ]. With the recent advances in metabolic engineering, some of these plant natural products can be sustainably produced from affordable building blocks by using engineered microorganisms [ 15 ]. In this review, we summarize recent examples of this approach ( Table 1 ), primarily concerning production of biopesticides and plant hormones ( Fig. 1 ). We also share our personal views on the opportunities and challenges that may be faced in the future when pursuing this research direction. Table 1 Recent examples of natural agrochemicals produced by engineered microorganisms. Table 1 Agrochemicals Type/Action Microbial hosts Titer/Yield Ref. Cinnamaldehyde Nematicide Escherichia coli 75 mg/L from 20 g/L glucose (Shake flask) (Bang, Hyun Bae et al., 2016) Thaxtomin A Herbicide Streptomyces albus J1074 222 mg/L from 10 g/L cellobiose (Shake flask) (Jiang, Guangde, et al., 2018) Cepacin A Anti-oomycete Burkholderia ambifaria NA (Mullins, Alex J. et al., 2019) Indole-3-acetic acid (IAA) Plant hormone E. coli 744 mg/L from 20 g/L of glucose (Shake flask) (Guo, Daoyi, et al., 2019) Indole-3-acetic acid (IAA) Plant hormone Cupriavidus pinatubonensis JMP134 NA (Zuniga, Ana, et al., 2018) Gibberellins (GAs) Plant hormone Yarrowia lipolytica 13 mg/L GA 3 and 18 mg/L GA 4 (24-well plate) (Kildegaard, Kanchana R. et al., 2021) Abscisic acid (ABA) Plant hormone Y. lipolytica 264 mg/L (24-well plate) (Arnesen, Jonathan Asmund et al., 2022) Strigolactones (SLs) Plant hormone E. coli – Saccharomyces cerevisiae consortium 48 μg/L 5-deoxystrigol (5DS) from 40 g/L Xylose (Shake flask) (Wu, Sheng, et al., 2021) NA: Not available. Fig. 1 Engineering microbes to produce natural products for improving plant growth. These microbes are cultivated on renewable and cost-effective substrates to produce natural compounds with the potential to function as biocontrol agents. Biopesticides protect crop health by killing phytopathogens. Plant hormones play versatile roles in regulating plant metabolism and improving resilience to abiotic and biotic stresses. Application of prebiotics to stimulate the growth of plant-beneficial microorganisms could be a future research direction. Fig. 1"
} | 1,925 |
31831639 | null | s2 | 7,794 | {
"abstract": "The gut microbiota produce hundreds of molecules that are present at high concentrations in the host circulation. Unraveling the contribution of each molecule to host biology remains difficult. We developed a system for constructing clean deletions in "
} | 63 |
34467342 | PMC8395622 | pmc | 7,796 | {
"abstract": "Surfaces with microscale\nroughness can entail dual-scale hierarchical\nstructures such as the recently reported nano/microstructured surfaces\nproduced in the laboratory ( Wang\net al. Nature 2020 , 582 , 55 − 57 32494077 ). However,\nhow the dual-scale hierarchical structured surface affects the apparent\nwetting/dewetting states of a water droplet, and the transitions between\nthe states are still largely unexplored. Here, we report a systematic\nlarge-scale molecular dynamics (MD) simulation study on the wetting/dewetting\nstates of water droplets on various dual-scale nano/near-submicrometer\nstructured surfaces. To this end, we devise slab-water/slab-substrate\nmodel systems with a variety of dual-scale surface structures and\nwith different degrees of intrinsic wettability (as measured based\non the counterpart smooth surface). The dual-scale hierarchical structure\ncan be described as “nanotexture-on-near-submicrometer-hill”.\nDepending on three prototypical nanotextures, our MD simulations reveal\nfive possible wetting/dewetting states for a water droplet: (i) Cassie\nstate; (ii) infiltrated upper-valley state; (iii) immersed nanotexture-on-hill\nstate; (iv) infiltrated valley state; and (v) Wenzel state. The transitions\nbetween these wetting/dewetting states are strongly dependent on the\nintrinsic wettability ( E in ), the initial\nlocation of the water droplet, the height of the nanotextures ( H 1 ), and the spacing between nanotextures ( W 1 ). Notably, E in – H 1 and E in – W 1 diagrams show that\nregions of rich wetting/dewetting states can be identified, including\nregions where between one to five states can coexist.",
"conclusion": "Conclusion A large-scale MD study on the wetting/dewetting\nbehavior of a water\ndroplet on dual nano/near-submicrometer scale hierarchical surfaces\nis presented. Our model systems are based on a thin slab of nanotexture-on-near-submicrometer-hill\nwith different strengths of surface–water interaction. Three\ntypical topologies of nanotextures are considered. Overall, the final\nwetting/dewetting state of the water droplet is dependent on the droplet\ninitial location on the surface. If starting with the Cassie state,\nfour wetting/dewetting states are observed in our MD simulations,\nin the (forward) sequence of the Cassie, the infiltrated upper-valley,\nthe immersed nanotexture-on-hill, and the Wenzel states, as the water–surface\ninteraction increases. On the other hand, if starting with the Wenzel\nstate, the four states in the (backward) sequence of the Wenzel, the\ninfiltrated valley, the infiltrated upper-valley, and the Cassie states\nare observed as the water–surface interaction decreases. The\nforward/backward sequences represent sequential transitions from the\nCassie to the Wenzel state and vice versa, and are not symmetric.\nIn both sequences, the immersed nanotexture-on-hill state and the\ninfiltrated valley state are unique intermediate states. In both forward/backward\nsequential wetting/dewetting transitions, the corresponding three\ncritical intrinsic adsorption energies to characterize transitions\nare determined. These critical intrinsic adsorption energies appear\nto be correlated with the height of the nanotextures ( H 1 ) and the intrananotexture distance ( W 1 ). On the basis of the final wetting/dewetting states\noriginated from the two limiting initial conditions, two wetting/dewetting\ndiagrams are plotted on the E in – H 1 plane and on the E in – W 1 plane, respectively. These\nnew insights into the multiple wetting/dewetting states on dual-scale\nhierarchical structured surfaces will not only deepen our understanding\nof generic wetting/dewetting phenomena on rough surfaces but also\nhave important implications for the fabrication of self-cleaning and\nanti-icing surfaces with high water/ice repellence and durability,\nas well as to the manipulation of the wetting properties of rough\nsurfaces through controlling the synergetic interplay of the dual\nnano- and submicrometer structures by design.",
"introduction": "Introduction Knowledge of the wettability\nof surfaces has important implications\nin a wide range of scientific and technological fields, such as microfluidics, 1 − 3 heat transfer, 4 , 5 surface self-cleaning, 6 surface coating, 7 inkjet\nprinting, 8 drag reduction, 9 and biomimetics. 10 In addition\nto the dependence of chemical composition, the wettability of a surface\nis also dependent on the physical topography of the surface. 11 For instance, very recently, we reported that\nthe closed-loop topology of the nanowalls/nanochannels can result\nin unusual wetting behavior, i.e., multiple Wenzel states with apparent\ncontact angles > 0° on an orderly structured solid surface\nwhose\nflat counterpart is superhydrophilic with a contact angle of 0°. 12 Examples of dual- or multiscale hierarchical\nsurface structures\nthat have a significant effect on surface wettability are prevalent\nin nature, including lotus leaves, Viola tricolor, water strider legs,\nfishing spider Dolomedes triton, and shark skin. 13 − 17 Previous experimental studies have shown that these\nnatural surfaces exhibit superhydrophobic (or superoleophobic) properties,\nowing to the synergetic effects of the dual or multiscale hierarchical\nstructures. Jiang and co-workers revealed a novel nanoscale structure\non micrometer-scale papillae for the lotus surface. Branchlike nanostructures\nwith a diameter of ∼120 nm were found on randomly distributed\npapillae with diameters ranging from 5 to 9 μm. 18 Additionally, hierarchical structured surfaces are also\ncommonly present in many biomimetic products and other artificial\nmaterials obtained via chemical or physical methods 10 , 19 − 22 such as chemical vapor deposition (CVD), 23 nano-self-masking fabrication, 22 , 24 femtosecond\nlaser surface processing (FLSP), 25 − 27 and biotemplating nanofabrication. 19 Notably, hybrid nanostructured/microstructured\nlayers can be produced on the surfaces of metals, semiconductors,\npolymers, and glasses. 25 , 28 , 29 Some of these specially made surfaces enable self-cleaning of dust\ncontaminants. Understanding the physics of wetting phenomena\nassociated with\nwater droplets is of fundamental importance in the design of nonsmooth\nsurfaces with a desired wettability. Intense experimental effort has\nbeen devoted to studying the wetting process and the role of the hierarchical\ngeometry. 19 , 22 , 29 , 30 Chen et al. 22 categorized\ndroplet impingement on biomimetic lotus-like hierarchical superhydrophobic\nsurfaces into three states, dewetting, microwetting, and nanowetting\nstates, based on measuring contact-angles using high speed photography.\nMcCarthy et al. 19 pointed out that the\nmicroscale and nanoscale components of the lotus-like hierarchical\nsuperhydrophobic surfaces play distinct, but complementary, roles\nin repelling wetting from droplet impact. The nanoscale component\nprovides a high antiwetting capillary pressure, while the microscale\ncomponent impedes the development and propagation of pressures associated\nwith liquid compression. However, to date, no direct experimental\nobservation of wetting states on the dual-scale hierarchical surfaces\nat nanometer/submicrometer scales have been reported. Few computer\nsimulations of wetting states on a dual nano/microscale\nhierarchical surface have been reported, due to the very high computing\ncost required to model large-scale systems with billions of atoms\nand molecules. To reduce the system size, previous molecular dynamics\n(MD) simulations typically involved dual subnano/nanoscale hierarchical\nsurfaces. 31 , 32 Because the two length scales are relatively\nclose, the wetting behaviors on the subnano/nano two parts can interfere\nwith one another and thus it can be harder to identify distinct wetting\nstates on the dual-scale structured surface. Additionally, most previous\nMD simulations adopted a single-scale structured surface, e.g., a\nrough surface with a single length scale of nanopillared or nanotrapezoidal\ntextures. 33 − 36 These studies observed only two common wetting/dewetting states:\nthe Wenzel state (in which water droplets are in full contact with\nthe rough surface) and the Cassie–Baxter (or Cassie) state\n(in which water droplets are in contact only with the tops of the\nstructured surface). For more realistic and common dual-scale (or\nmultiscale) hierarchical surfaces, the presence of at least of two\nlength scales (nano, micro, and larger etc.) renders the wetting/dewetting\nstates of water droplet much more complex. 19 , 22 , 30 Moreover, it is not fully understood how\nthe dual-scale hierarchical surface interplays with the intrinsic\nwettability, and the associated physical mechanics underlying the\ntransition between wetting/dewetting states remains to be explored. Here, we report a comprehensive large-scale MD simulation study\nof the wetting/dewetting behavior of a water droplet on a dual nano/near-submicrometer\nscale hierarchical structured surface. To accommodate these much finer\nlength scales compared to previous MD simulations, 31 , 32 a slab droplet/substrate system was devised and used. Five physical\nparameters associated with the slab droplet/substrate system were\nfound to strongly affect the transition between various wetting states:\n(i) the topology of the nanotexture; (ii) the intrinsic adsorption\nenergy ( E in ), 12 i.e., the adsorption energy of a water molecule on the corresponding\nsmooth surface, which is used to characterize the water–smooth\nsurface interaction; (iii) the initial location of the droplet; (iv)\nthe height of nanotextures ( H 1 ); and (v)\nthe spacing between each unit of nanotextures ( W 1 ). From our MD simulations, we identified five distinct wetting\nstates for a water droplet. One state occurred uniquely when the initial\nstate was Cassie, and another occurred uniquely when the initial state\nwas Wenzel. The dependence of the critical intrinsic adsorption energies\non H 1 and W 1 was quantitatively analyzed. Importantly, the wetting diagrams plotted\non the E in – H 1 and E in – W 1 planes were obtained, where between one to five states\nwere found to coexist. Lastly, we discuss the implication of the obtained\nresults to the wetting behavior on larger dual-scale structured surfaces.\nIn view of the wide presence of rough surfaces with dual-scale or\nmultiscale hierarchical structures in nature and industrial productions,\nit is of both fundamental and practical importance to explore the\nnew wetting/dewetting states of a water droplet on a dual-scale hierarchical\nsurface.",
"discussion": "Results and Discussion In the MD simulations with the mW water model, initially, a water\ncuboid consisting of 47 616 water molecules was placed on the\ntop of the nanotextures, i.e., the water droplet was initially in\nthe Cassie state. Then the system was relaxed for a simulation time\nof 6.5–30 ns until the wetting state of the water droplet reached\nequilibrium. This series of MD simulations showed that the final state\nof the water droplet was dependent on the intrinsic water-surface\ninteraction E in = 4.748ε SW , where the strength of the LJ interaction parameter ε SW varied from 0.19 to 4.82 kJ/mol. Note again that here intrinsic\nadsorption energy E in was used to characterize\nthe water–smooth surface interaction. Parallel Nanowalls For the water droplet on the surface\nwith parallel nanowalls, snapshots of the equilibrated water droplets\non surfaces with different E in are shown\nin Figure 2 . Four different\npartial wetting/dewetting states were observed as E in increased from 0.92 to 22.91 kJ mol –1 : Cassie state ( Figure 2 A), where the water droplet stays on the top of the nanotextures;\ninfiltrated upper-valley (IFUV) state ( Figure 2 B), where only the upper section of the valley\nbetween the near-submicrometer hills is infiltrated with water; immersed\nnanotexture-on-hill (IMNTH) state ( Figure 2 C), where nanotextures on the near-submicrometer\nhills are fully immersed in the water droplet but the nanotextures\nin the valley are not; and Wenzel states ( Figure 2 D), where all the nanotextures and valleys\nare completely immersed in the water droplet. The transition from\none state to another, among the four partial dewetting/wetting states,\ncould be controlled by changing the parameter E in . Accordingly, three critical intrinsic adsorption energies\n( E in CW ) for the state-to-state transition were\nestimated: (1) below E in CW1 = 9.16 ± 0.46 kJ mol –1 , the water droplet adopted the Cassie state; (2) above this up to E in CW2 = 10.08 ± 0.46 kJ mol –1 , the water droplet\nadopted the IFUV state; (3) above this up to E in CW3 = 20.16 ±\n0.46 kJ mol –1 , the water droplet adopted the IMNTH\nstate; and (4) above E in CW3 the water droplet adopted the Wenzel\nstate. Figure 2 Snapshots of the equilibrated water droplet on the model surface\nwith parallel nanowalls, after MD time ≥ 6.5 ns. (A–D)\nStarting from the initial configuration shown in Figure 1 B with the water slab (red\ncuboid) in the Cassie state; (E–H) starting from the initial\nconfiguration shown in panel (D) with the water slab in the Wenzel\nstate. The intrinsic adsorption energy E in and the intrinsic contact angle θ in for the smooth\nsurface are shown beneath each snapshot. The final equilibrium state\nof the water droplet is described in the bottom of each panel. In addition, another initial configuration for\nthe water droplet\nbeing in the Wenzel state was also considered, as shown in Figure 2 D. In this series\nof MD simulations, four wetting/dewetting states were recognized as E in reduced from 22.91 to 0.92 kJ mol –1 : the Wenzel state ( Figure 2 E); the infiltrated valley (IFV) state ( Figure 2 F), where only the valley between the near-submicrometer\nhills was infiltrated by water and no nanotextures were filled by\nwater; the IFUV state ( Figure 2 G); and the Cassie state ( Figure 2 H). This new sequence of wetting states is\nalmost like a backward sequence of the previous one, in which the\ngradual decrease of E in resulted in sequential\ntransition from the Wenzel to the Cassie state. For these state transitions,\nthe corresponding three critical intrinsic adsorption energies ( E in WC ) were as follows: (1) above E in WC1 = 8.25 ± 0.92 kJ mol –1 , the water droplet adopted the Wenzel state; (2)\nbelow this down to E in WC2 = 5.96 ± 0.46 kJ mol –1 , the water droplet adopted the IFV state; (3) below this down to E in WC3 = 3.67 ± 0.46 kJ mol –1 , the water droplet\nadopted the IFUV state; and (4) below E in WC3 the water droplet\nadopted the Cassie state. Here, the IFV state was a new wetting state\nnot seen in the sequence where the Cassie state was the initial configuration\nfor the water droplet and E in increased.\nThis apparent hysteresis behavior for the wetting/dewetting state\ntransition is a manifestation of the kinetic factor and coexistence\nof multiple states. Furthermore, in order to confirm that these\nnewly observed sequences\nof wetting/dewetting states are generic, independent of the water\nmodel, benchmark MD simulations were also performed using the fully\natomistic SPC/E water model and the same dual-scale hierarchical surface.\nDue to higher computational cost of SPC/E, H 1 was reduced to 12 Å and W 2 to 76 Å, with a smaller (slab) water droplet containing 12 251\nwater molecules. This series of independent MD simulations (with LJ\nparameters in the same range as used for the MD/mW simulations) showed\nthat all the five distinct wetting/dewetting states can be also observed\nin the forward and backward sequences. For example, Movies S1 – S4 , respectively,\ndemonstrate the trajectories of the formation of the four states,\nCassie, IFUV, IMNTH, and Wenzel states, with a Cassie initial configuration\nand with increasing the surface water interaction; while Movies S5 – S8 , respectively, display the trajectories of the formation of the\nfour states, Wenzel, IFV, IFUV and Cassie states, with a Wenzel initial\nconfiguration and with decreasing surface water interaction. Dispersed\nNanopillars For the water droplet on the\nsurface with dispersed nanopillars, snapshots of the equilibrated\nwater droplets on surfaces with different E in are shown in Figure 3 . The initial configuration for the water cuboid was on the nanopillars\n( Figure 3 Ai). Once\nthe water droplet reached equilibrium, three distinct wetting/dewetting\nstates were observed in a forward sequence: Cassie ( Figure 3 Aii), IMNTH ( Figure 3 Aiii), and Wenzel ( Figure 3 Aiv) states. On the\nother hand, if the initial configuration for the water droplet was\na Wenzel state ( Figure 3 Bi), after equilibration, three distinct wetting/dewetting states\nwere observed in a backward sequence: Wenzel ( Figure 3 Bii), IFUV ( Figure 3 Biii), and Cassie ( Figure 3 Biv) states. Contrary to the parallel nanowalls\ntopography, here the IFUV and the IFV wetting states in the forward\nand the backward sequence, respectively, were not observed. However,\nwhen the nanopillars had a shorter height of H 1 = 12 Å, all four wetting/dewetting states were observed\nin the forward sequence: Cassie ( Figure 3 Cii), IFUV ( Figure 3 Ciii), IMNTH ( Figure 3 Civ), and Wenzel ( Figure 3 Cv) states. Figure 3 Snapshots of a water droplet on the model\nsurface with dispersed\nnanopillars topography, (A, B) with H 1 = 16 Å and (C, D) with H 1 = 12\nÅ. (Ai–Di) Initial configurations of the water droplet\nand (ii–vi of A–D) remaining snapshots of the corresponding\nequilibrium wetting/dewetting states at different values of E in . In panels (A) and (C), the Cassie state\nwas the initial configuration; in panels (B) and (D), the Wenzel state\nwas the initial configuration. The intrinsic adsorption energy E in , intrinsic contact angle θ in , and name of the final equilibrium wetting/dewetting state are shown\nbeneath each snapshot. Note that (Diii) displays a new wetting state\ndifferent from the already identified five states but only observed\nin this case, and thus it is not specially named. The absence of the IFUV state at H 1 =\n16 Å but re-emergence at H 1 =\n12 Å reflects an interplay between two factors: the spatial volume\nof the connected nanochannels in the nanotexture and the surface tension\nof the infiltrated water, when the water starts to infiltrate into\nthe upper region of the valley between near-submicrometer hills. When\nthe height of the nanopillars was lower ( H 1 = 12 Å), the surface tension of the infiltrated water in the\nupper region of the valley became lower (compared to that at H 1 = 16 Å), thereby weakening the wetting\ntendency of water on the nanopillars and rendering the IFUV state\na stable state. A similar explanation applied to the absence of the\nIFV state at H 1 = 16 Å ( Figure 3 B) with re-emergence\nat H 1 = 12 Å ( Figure 3 D) in the reverse sequence. Interestingly,\nfor a reverse sequence at H 1 = 12 Å,\nbesides the already identified Wenzel ( Figure 3 Dii), IFV ( Figure 3 Div), IFUV ( Figure 3 Dv), and Cassie ( Figure 3 Dvi) states, a new wetting state ( Figure 3 Diii) appeared, where\nthe valleys between the near-submicrometer hills and nanopillars on\nthe near-submicrometer hills were infiltrated by water whereas the\nnanopillars within the valleys were not immersed in water. This new\nwetting state also resulted from the interplay between the two competing\nfactors discussed above. Checkerboard-like Nanopillars For\na water droplet on\nthe surface with checkerboard-like nanopillars, snapshots of the equilibrated\nwater droplets on surfaces with different E in are shown in Figure 4 . Again, two nanotexture heights at H 1 = 16 and 12 Å, respectively, were considered. At H 1 = 16 Å, in the forward sequence, four equilibrium\nstates were identified: Cassie state ( Figure 4 Aii), IFUV state ( Figure 4 Aiii), IMNTH state ( Figure 4 Aiv), and Wenzel state ( Figure 4 Av). The reverse sequence identified\nWenzel state ( Figure 4 Bii), IFV state ( Figure 4 Biii), IFUV state ( Figure 4 Biv), and Cassie state ( Figure 4 Bv). All the wetting/dewetting states observed\nin both forward and backward sequences were the same as those observed\non the surface with parallel nanowalls. At H 1 = 12 Å, the same set of states were also observed in\nthe forward sequence ( Figure 4 C) and backward sequence ( Figure 4 D) as those observed at H 1 = 16 Å, although the critical intrinsic adsorption\nenergies were different. Figure 4 Simulations of a water droplet on the model\nsurface with checkerboard-like\nnanopillars, (A, B) with H 1 = 16 Å\nand (C, D) with H 1 = 12 Å. (Ai–Di)\nInitial configuration of the water droplet and (ii–v of A–D)\nremaining snapshots of the corresponding equilibrium wetting/dewetting\nstates at different values of E in . In\npanels (A) and (C), the Cassie state was the initial configuration;\nin panels (B) and (D), the Wenzel state was the initial configuration.\nThe intrinsic adsorption energy E in , intrinsic\ncontact angle θ in , and name of the final equilibrium\nwetting/dewetting state are shown beneath each snapshot. Comparison of the Effects of the Three Topographies The\ncritical intrinsic adsorption energies obtained for the three\nprototype surface topographies are compared in Figure 5 A and B. The results demonstrate that the\ncritical intrinsic adsorption energy E in CW1 shows little\ndependence on the topology of the nanotexture. E in CW2 for the parallel\nnanowalls nanotexture is almost the same as that for the checkerboard-like\nnanotexture but is slightly larger (∼1.97 kJ mol –1 ) than that for the dispersed nanopillars nanotexture ( Figure 5 A). E in CW3 is 5.27 and\n7.78–7.92 kJ mol –1 smaller for the dispersed\nnanopillars and checkerboard-like at H 1 = 12–16 Å, respectively, indicating that the parallel\nnanowalls can prevent the infiltration of the water droplet into the\nvalley. Figure 5 Variation of the critical intrinsic adsorption energies for the\nthree surfaces. Black open, red solid, and violet open symbols (circles,\nparallel nanowalls; stars, dispersed nanopillars; squares, checkerboard-like\nnanopillars) represent the critical adsorption energies of the first,\nthe second, and the third transition from the Cassie to Wenzel state\n(CW1, CW2, and CW3) or from the Wenzel to Cassie states (WC1, WC2,\nand WC3), respectively. (A, C) from the Cassie to Wenzel state, E in CW ; (B, D) from the Wenzel to Cassie state, E in WC ; vs either H 1 at fixed W 1 =\n20 Å or W 1 at fixed H 1 = 12 Å. The solid curves represent exponential-function\nfittings to the critical intrinsic adsorption energy vs the height\nof the nanotextures H 1 (A, B) or vs the\nintrananotexture distance W 1 (C, D). Black\nline, E in CW1 or E in WC1 ; red line, E in CW2 or E in WC2 ; violet line, E in CW3 or E in WC3 . Left panel pictures display\nthe zoom-in views of the selected areas marked by the green dashed\nrectangles. On the other hand, E in WC1 for the\nnanotexture of parallel nanowalls\nwas about 0.69 (0.46) kJ mol –1 larger than that\nfor the dispersed nanopillars nanotexture and 2.06 (2.06) kJ mol –1 smaller than that for the checkerboard-like nanotexture\nat H 1 = 12 (16) Å ( Figure 5 B). E in WC2 for the parallel\nnanowalls nanotexture was nearly the same as that for the checkerboard-like\nnanotexture at both H 1 = 12 and 16 Å,\nwhich was 0.80 (1.37) kJ mol –1 smaller than that\nfor the dispersed nanopillars at H 1 =\n12 (16) Å. The difference in E in WC3 for the three topographies\nvaried within 1.37 kJ mol –1 , in the order of parallel\nnanowalls < checkerboard-like nanopillars < dispersed nanopillars.\nHowever, each of the six intrinsic adsorption energies obtained from\nthe three types of topographies exhibited a similar trend as the height\nof nanotextures varied. In view of the similar trend and generally\nsmall differences (except for E in CW3 ) among the intrinsic adsorption\nenergies from the three topographies, as well as the same five wetting/dewetting\nstates, the next section focuses on the parallel nanowalls nanotexture\nas the representative model to further illustrate the effects of structural\nparameters, H 1 , W 1 , W 2 , and W 3 , on the\nwetting/dewetting transition. Critical Intrinsic Adsorption\nEnergies and Diagrams of Wetting\nStates For H 1 changing from 4\nto 20 Å at fixed W 1 = 20 Å, Figure 5 (A) shows the corresponding\ncritical adsorption energies ( E in CW1 , E in CW2 , and E in CW3 ) for the sequential transitions from the Cassie to the Wenzel states.\nInterestingly, all the three critical adsorption energies increase\nas H 1 increases. By fitting exponential-function\ncurves to the critical adsorption energies vs H 1 (see solid line in Figure 5 A), the following relations are obtained: 1 2 3 Here, the units for all the critical\nadsorption\nenergies E in CWi (or E in WCi ) are kJ mol –1 and for H 1 are Å. The relations\n( Figure 5 B) between\nthe critical adsorption energies and H 1 for the reverse sequence of transitions from the Wenzel\nto the Cassie states ( E in WC1 , E in WC2 , and E in WC3 ) are as follows: 4 5 6 Note that E in WC1 , E in WC2 , and E in WC3 generally increase as H 1 increases,\nand these trends are the same as those for E in CW1 , E in CW2 , and E in CW3 . The dependence of the critical adsorption energy on the intrananotexture\ndistance W 1 was also investigated at fixed H 1 = 12 Å. Figure 5 C shows the exponential-function fitting\nto the critical adsorption energies ( E in CW1 , E in CW2 , and E in CW3 ) vs W 1 for the sequence of transitions\nfrom the Cassie to the Wenzel state. Clearly, E in CW2 (above which\nthe water droplet prefers the IMNTH states) decreased as W 1 increased. This indicates that increasing W 1 leads to a lower critical intrinsic adsorption energy\nfor transitions from IFUV to IMNTH states. Similarly, E in CW1 (above\nwhich the water slab prefers the IFUV state) exhibited a decreasing\ntrend as W 1 increased, since the width\nof the upper region of the valley, determined by the distance between\nthe two nanowalls adjacent to the valley and on the near-submicrometer\nhills, generally increased as W 1 increased.\nIn contrast, E in CW3 exhibited little dependence on W 1 and the width of the upper region of the valley between\nnear-submicrometer hills. The obtained relations are given as follows: 7 8 9 The coefficients of the\nexponential term in eqs 7 and 8 are 21.59 and 10.62, respectively, reflecting\nthe dependence of E in CW1 and E in CW2 on W 1 . In contrast,\nthe much smaller coefficient of 0.001 in eq 9 reflects the near-independence of E in CW3 on W 1 . As a comparison, Figure 5 D shows the exponential-function\nfitting to the critical adsorption energies ( E in WC1 , E in WC2 , and E in WC3 ) vs W 1 for the sequence of transitions\nfrom the Wenzel to the Cassie state. E in WC1 and E in WC2 decreased with the increase of W 1 , while E in WC1 also exhibited little dependence on W 1 . The obtained relations are given below: 10 11 12 Note that E in CW1 and E in WC3 correspond\nto two reverse sequences of transitions (transition from the Cassie\nto the IFUV state vs transition from the IFUV to the Cassie state).\nThe curve of E in WC3 exhibits clear hysteresis compared with\nthat of E in CW1 ( Figure 5 A vs B, C vs D], similar to the transition between\ntwo common wetting/dewetting states, the Wenzel and Cassie states,\non single-scale structured surfaces. 35 The simulation results demonstrate that all the critical adsorption\nenergies for both the sequential transition from the Cassie state\nto the Wenzel state and the sequential transition from the Wenzel\nstate to Cassie state increase as H 1 increases\nwhile they decrease as W 1 increases. This\nis because the critical adsorption energy reflects or is related to\nthe free energy barrier from one wetting state to another and the\nenergy barrier is mainly contributed by the change of the surface\ntension energy of the droplet which increases with increasing H 1 while it decreases with increasing W 1 . Additionally, it is worth noting that the\nwater droplet can stay in a non-Wenzel wetting state, e.g., the IMNTH\nstate, even when the intrinsic contact angle is 0° (at E in = 18.67 kJ mol –1 ; see Figure 5 A and 5 C). This phenomenon can be attributed to the topography of\nthe parallel nanowalls nanotexture which has the same effect in the\ndirection normal to the nanowalls as the closed-loop nanowalls. As\nreported previously by us, 12 the topology\nof closed-loop nanowalls/nanochannels on a surface can introduce an\nunbalanced force applied to all water molecules along the triple-phase\ncontact line (TCL) which hinders the outward diffusion of the TCL,\nthereby leading to the phenomenon of the topological wetting state\n(i.e., the water droplet is in a Wenzel state with a contact angle\n> 0° even when the intrinsic contact angle of the water droplet\nis 0°). Likewise, the parallel nanowalls here also introduce\nan unbalanced force on all water molecules along the TCL in parallel\nto the nanowalls, thereby blocking the outward diffusion and infiltration\nof water into the deeper valley and resulting in the IMNTH state with\na nonzero contact angle (rather than the Wenzel state) even when θ in = 0°. This interesting wetting behavior indicates that\nthe parallel nanowalls topography can lead to apparently higher hydrophobicity\nfor rough surfaces. Overall, the wetting of a water droplet\non the dual nano/near-submicrometer\nscale hierarchical surface can be classified into multiple states,\ndepending on how the water droplet infiltrates into the cavities between\nthe nanotextures and the valley between the near-submicrometer hills.\nStarting from the Cassie state, the top of the nanotextures and the\nnear-top section of the valley are in direct contact with the water\ndroplet. With gradually increasing E in , the next rising state is the IFUV state due to the wider distance\nbetween the hills (e.g., Figure 2 B), followed by the IMNTH state (e.g., Figure 2 C). The nanotextures within\nthe valley can be immersed in water once contacted by the water droplet.\nHowever, due to the dewetting capillary pressure from the valley,\nthey tend to be spared from contacting the water droplet if the initial\nstate is a Cassie state. Once E in is further\nincreased to overcome the free-energy barrier due to the dewetting\ncapillary pressure, water can infiltrate into the valley and immerse the nanotextures\nin the valley as well, resulting in the final Wenzel state (e.g., Figure 2 D). Conversely,\nstarting with the Wenzel state, as E in is gradually reduced, the water droplet is withdrawn\nfirst from the smaller scale nanotexture to reduce the tension while\nthe valley is still filled by water, which is the IFV state (e.g., Figure 2 F). IFV seems to\nbe an intermediate state where the water–surface interaction\nis not strong enough to overcome the dewetting capillary pressure\ndue to the nanocavity within the nanotextures. As E in is further reduced, the water droplet is withdrawn\nfrom both the valley and nanotexture cavities ( Figure 2 G) to reduce tension. The dewetting capillary\npressure appears to increase as the height of nanotextures increases\nbut decrease as the intrananotexture distance increases (see CW2 curves\nin Figure 5 A and C).\nThus, it is expected that, on the dual-scale nano/near-submicrometer\nsurface, if the width of the valley is wide enough to render the dewetting\ncapillary pressure lower than that due to the cavity between nanotextures,\nthe IFV state ( Figure 2 F) tends to form if the Cassie state is the initial configuration.\nThis deduction is further confirmed by a test simulation with a larger\nvalley width ( W 3 = 204 Å) which renders\nthe dewetting capillary pressure sufficiently lower, leading to the\nIFV state ( Figure S1 ), consistent with\nprevious experimental deduction. 22 Note\nthat the valley appears to result in an extra dewetting capillary\npressure to reduce the tendency of wetting the nanotextures within\nthe valley. Meanwhile, the smaller nanotexture enables less contact\nif the water droplet is in the IFV state or the IFUV state, leading\nto a much smaller interaction between the water droplet and the dual-scale\nstructured rough surface compared to that between the water droplet\nand the corresponding flat surface. The two factors together cause\nthe dual nano/near-submicrometer scale hierarchical surface to favor\na higher dewetting tendency. Besides the effects of H 1 and W 1 , the effects of the\nwidth of the hill ( W 2 ) and the valley\n( W 3 ) on the critical intrinsic adsorption\nenergies have also been analyzed (see the Supporting Information for details). To summarize, with the Cassie\nand the Wenzel states being viewed\nas the two extreme initial conditions for a water droplet, the final\nwetting/dewetting states at any E in can\nexhibit two limiting cases: either a strongly dewetting state or a\nstrongly wetting state. Between these two limiting states, other intermediate\nstates as shown above can be attained. For example, at a certain E in , if a Cassie initial configuration results\nin an IFUV state whereas a Wenzel initial configuration results in\na Wenzel state, then IFUV, IMNTH, IFV, and Wenzel states can coexist\nat this given E in due to different starting\nlocations of the water droplet. On the other hand, if both the Cassie\ninitial configuration and the Wenzel initial configuration can result\nin an IFUV state, then only the IFUV state would exist, regardless\nof any starting location of the water droplet. One special case is\nthat if a Cassie initial configuration results in an IFUV state whereas\na Wenzel initial configuration results in an IFV state, the IMNTH\nstate would be excluded because the water–surface interaction\ncannot overcome the dewetting capillary pressure due to the nanotexture.\nThus, in this case, only IFUV and IFV states can coexist. From these\nanalyses, the order of wetting strength among the five above-identified\nwetting states is determined as shown in Figure 6 A, according to which intermediate states\ncan be easily obtained after demarcating the two extreme wetting/dewetting\nconditions. Furthermore, the wetting/dewetting states of the water\ndroplet can be changed by changing the E in and the length of H 1 and W 1 of the nanotexture. A wetting/dewetting diagram on the E in – H 1 plane,\naccording to eqs 1 – 6 for W 1 = 20 Å,\ncan be obtained ( Figure 6 B). For H 1 in the range of 4–20\nÅ, three monostate regions are identified: the Cassie state at\nlow E in , the IFUV state in a small region\nat relatively high E in , and the Wenzel\nstate at much higher E in ; and three bistate-coexistence\nregions, one tristate coexistence region, one tetrastate coexistence\nregion, and one pentastate coexistence region. Likewise, the wetting/dewetting\ndiagram can be also plotted on the E in – W 1 plane ( Figure 6 C) according to eqs 7 – 12 for H 1 = 12 Å. In this diagram, there are three\nmonostate regions, three bistate coexistence regions, one tristate\ncoexistence region, and one tetrastate coexistence region. Within\neach multistate coexistence region, the final wetting state of the\nwater droplet is dependent on its initial configuration. To verify\nthis conclusion, a new series of MD simulations was performed using\nthe five different prototype wetting states identified in Figure 2 as initial configurations\nand with the same values of E in and H 1 ( E in = 8.70 kJ\nmol –1 and H 1 = 20 Å),\nwhich, together, reflect a point (P) in the pentastate coexistence\nregion (II) in Figure 6 B. The obtained equilibrium state of the water droplet was identified\nas one of the five corresponding wetting/dewetting states, and, in\nall, a total of five proposed states were identified ( Movies S9 – S13 ), confirming the above conclusion. Additionally, for a surface with\ndifferent structural parameters, although the values of the critical\nadsorption energies and the boundaries of wetting-state regions in\nthe wetting/dewetting diagram could be different, the coexistence\nof multiple wetting states can still arise. Figure 6 (A) Schematic rising\nsequence of the five wetting/dewetting states\nvs increasing E in . According to this sequence,\nintermediate wetting states can be obtained after departing from the\ntwo limiting wetting/dewetting conditions. Wetting/dewetting diagrams\n(B) on the E in – H 1 plane for W 1 = 20 Å\nand (C) on the E in – W 1 plane for H 1 = 12 Å.\n(I) Wenzel & IMNTH & IFUV & IFV; (II) Wenzel & IMNTH\n& IFUV & IFV & Cassie; (III) Cassie & IFUV & IFV;\n(IV) Cassie & IFV & IFUV; (V) IFUV & IFV; (VI) IFUV. Finally, we note that although dimensions of the\nsystems used in\nMD simulation are much smaller than those in realistic applications,\nmany wetting/dewetting phenomena exhibit length-scale independence.\nFor example, the computed contact angle of nanodroplets and measured\ncontact angle of macroscopic droplets are nearly the same on the same\ntype of flat surface, regardless of the huge size difference. Here,\nour atomic simulation study focuses on the dual-scale nano/near-micrometer\nroughness. The latter is already notably larger, compared to the dual-scale\nsubnano/nanoscale roughness reported in previous simulation study, 31 , 32 where the minor (nanotexture) and major (hill) structures were considered.\nWith the much large model system considered here, we are able to observe\nthe generic behavior of nanodroplets on the dual-scale structured\nsurface (for example, whether water enters the gap between nanopillars)\nwhile avoiding the apparent interference between the wetting behaviors\non dual-scale surfaces with minor and major structures not largely\nseparated in the length scale. This generic behavior shows little\ndependence on the actual size of the droplet. Therefore, we expect\nthat the obtained results can mimic those wetting/dewetting behaviors\nof larger water droplets on the realistic surfaces with dual-scale\nroughness. Furthermore, to investigate the effect of size on the wetting\nstates on dual-structured surfaces, we built a larger model system\nwith larger structural parameters, which results in consistent conclusions.\nThe system size has a total length of 1056 Å, as shown in Figure S2A . A droplet with 90 954 water\nmolecules is placed on the surface, and the whole system contains\n98 424 atoms. When the droplet is initially in the Cassie state,\nthe equilibrium states are Cassie, IFUV, IMNTH, and Wenzel states\nwith increasing intrinsic water–surface interaction ( Figure S2Bi–Bv ), as characterized by intrinsic\nadsorption energy E in . If the droplet\nis initially in the Wenzel state, the equilibrium states are Wenzel,\nIFV, IFUV, and Cassie states with decreasing E in ( Figure S2Ci–Cv ). These\nare all consistent with the results obtained from the 528 Å length\nmodel used in the simulations illustrated above, although the critical\nadsorption energy which is determined by the structure dimensions\nsuch as W 1 and H 1 (as shown Figure 5 ) could be different due to different structural parameters.\nMoreover, the wetting details do not depend on the whole system size\nwhen the scale is larger than a certain scale (∼several nanometers).\nTherefore, the various wetting/dewetting states observed in our simulations\nare expected to be also seen in larger-sized realistic systems."
} | 9,845 |
24957038 | PMC4101518 | pmc | 7,797 | {
"abstract": "Recently, cyanobacteria have become one of the most attractive hosts for biochemical production due to its high proliferative ability and ease of genetic manipulation. Several researches aimed at biological production using modified cyanobacteria have been reported previously. However, to improve the yield of bioproducts, a thorough understanding of the intercellular metabolism of cyanobacteria is necessary. Metabolic profiling techniques have proven to be powerful tools for monitoring cellular metabolism of various organisms and can be applied to elucidate the details of cyanobacterial metabolism. In this study, we constructed a metabolic profiling method for cyanobacteria using 13 C-labeled cell extracts as internal standards. Using this method, absolute concentrations of 84 metabolites were successfully determined in three cyanobacterial strains which are commonly used as background strains for metabolic engineering. By comparing the differences in basic metabolic potentials of the three cyanobacterial strains, we found a well-correlated relationship between intracellular energy state and growth in cyanobacteria. By integrating our results with the previously reported biological production pathways in cyanobacteria, we found putative limiting step of carbon flux. The information obtained from this study will not only help gain insights in cyanobacterial physiology but also serve as a foundation for future metabolic engineering studies using cyanobacteria.",
"conclusion": "4. Conclusions Triple quadrupole mass spectrometry-based target metabolic profiling method for cyanobacteria was developed. By using 13 C-labeled cell extracts as internal standards, we successfully determined the absolute concentrations of 84 metabolites in three different cyanobacterial type strains used in this study. The overview of respective metabolic profiling results showed that Synechocystis strains had a different tendency from Synechococcus strains in terms of the distribution of some metabolites such as amino acids and sugar phosphates. Moreover, the growth rate of the three investigated strains seemed to be correlated with adenylate energy charge value rather than TCA cycle activity. This might indicate that the cyanobacterial TCA cycle does not have a role in supplying energies for the microorganism’s cellular metabolism. Finally, we suggested a candidate for the limiting step in cyanobacterial biological production under photoautotrophic conditions. The data obtained and insights revealed in this study are expected to contribute to cyanobacterial physiological and industrial researches.",
"introduction": "1. Introduction Cyanobacteria are ubiquitous, globally important photosynthetic microorganisms that are capable of providing oxygen to the atmosphere by using solar energy. Recently, photosynthetic microorganisms have been considered as attractive candidates for biomass resources due to their high photosynthetic ability. In particular, cyanobacterial biofuel production derived from environmental CO 2 has been considered as one of the effective ways to reduce CO 2 emissions and construct a sustainable low-carbon society system. Cyanobacteria naturally produce valuable materials including lipids [ 1 ], sugars [ 2 ] and some kinds of polymers [ 3 ] directly from atmospheric CO 2 although they do not have the inherent ability to produce industrially important materials such as higher alcohols. However, currently available tools for genetic modification allowed metabolic engineers to use cyanobacterium as a host for inserting exogenous genes for production of industrially important compounds. In fact, recent researches in cyanobacterial biological production have successfully achieved the production of targeted materials such as fatty acids [ 1 ], sugars [ 2 ], hydrogen [ 4 ], acetone [ 5 ], mannitol [ 6 ], ethylene [ 7 , 8 ], ethanol [ 9 ], isoprene [ 10 ], 3-hydroxybutyrate [ 11 ], 2,3-butanediol [ 12 ], isobutyraldehyde, isobutanol [ 13 ], isopropanol [ 14 ], 1-butanol [ 15 ], and 2-methy-1-butanol [ 16 ]. However, most of the bioprocess trials were not satisfactory for industrial application. An inadequate systematic metabolome catalogue of commonly used cyanobacterial strains might cause misunderstanding of the metabolome state, which in turn may lead to a misguided metabolic engineering strategy. Therefore, a robust metabolic profiling system of cyanobacteria must be initially established. Metabolic profiling has been established as a powerful tool for gaining insights in cellular metabolism [ 17 , 18 ]. Although relative metabolite concentration changes can be informative, levels of absolute metabolite concentrations can be critical. Especially for understanding metabolic dynamics, it is a prerequisite for several approaches in metabolic analysis [ 19 ]. In particular, this information is useful for strain improvement via metabolic engineering as it is important to monitor actual changes in the altered cellular metabolism after gene manipulation [ 20 , 21 ]. Recently, advances in chromatography and mass spectrometry technologies have allowed metabolomics to play a more important role in biological studies since they enabled simultaneous measurements of numerous cellular metabolites. As a result, these technologies are increasingly being widely applied in mapping the changes in altered or natural cellular metabolisms. Currently, the cyanobacterial strains Synechococcus elongatus PCC7942, Synechococcus sp. PCC7002, and Synechocystis sp. PCC6803 (hereafter PCC7942, PCC7002, PCC6803, respectively) are most often used as background strains for metabolic engineering since the whole genome sequence and established tools for gene manipulation are available. However, without major phenotypic differences and available metabolome information of candidate strains, the choice of background strain will probably rely subjectively on the researcher's experience and intuition. In this study, therefore, we introduced an analytical procedure for molar-based widely targeted metabolic profiling analysis to characterize the difference in the metabolomes of three cyanobacterial strains with potential for biological production namely PCC7942, PCC7002, and PCC6803. To perform widely targeted metabolic profiling in the multiple reaction monitoring (MRM) mode, gas chromatography–triple quadrupole mass spectrometry (GC/QqQ-MS) was used for analyzing amino acids and organic acids, while analysis of intermediates of central metabolism such as sugar phosphates and cofactors was performed using reserved phase ion-pair liquid chromatography (RP-IP-LC)/QqQ-MS. Triple quadrupole mass spectrometer is suitable for metabolic profiling because of its high sensitivity and selectivity as well as its wide dynamic range that makes it possible to give accurate quantitative information [ 22 , 23 ]. Moreover, to obtain the absolute concentration of each metabolite, we used 13 C-labeled cell extracts as internal standards, a technique known to effectively reduce errors due to variations occurring in the analysis and sample processing [ 24 ]. This method is restricted to studies using culturable organisms because all metabolites in the internal standard need to be fully or partially labeled. However, when dealing with a large number of samples, our method is expected to be less tedious than other high accuracy quantitative methods such as standard addition method. Finally, we compared and discussed the molar-based differences in the cellular metabolism of the three cyanobacterial strains in their mid-exponential phase. These results provide informative data and an objective viewpoint for further metabolic engineering researches in cyanobacteria [ 25 ].",
"discussion": "2. Results and Discussion 2.1. Optimization of 13 C-internal Standard Amount for Quantitation Regarding the internal standard, equal or close peak intensities of targeted analytes are desirable. Here, we use the uniformly 13 C-labeled cell extracts as the internal standard and their preparation method is described in the Experimental Section. To determine the appropriate 13 C-IS concentration, four different amounts of 13 C-IS were added to the extraction solvent with PCC7002 cells equivalent to 3 mg dry weight. The addition of 5 μL, 10 μL, 20 μL, 50 μL IS were tested for this purpose. Extraction and MS/MS analysis were performed with the method described in the Experimental Section. Then, the absolute values of log 10 ( 12 C/ 13 C) were calculated by following the Equation:\n (1) \n The Abs value represents 10-fold differences between the monoisotopic peak area and fully-labeled peak area. According to the Abs value, 50 μL of 13 C-IS was the most suitable amount to be added in the samples ( Figure 1 ). Thus, we added 50 μL of 13 C-IS to the extraction solvents in all experiments including the calibration curve construction. Figure 1 Comparison of area ratios of monoisotopic peaks to their corresponding fully-labeled peaks with different amounts of 13 C-internal standards. Absolute values of log 10 ( 12 C/ 13 C) are represented with a heat map. Abbreviations: 3PGA, 3-phosphoglycerate; Acetyl-CoA, acetyl coenzyme A; ADP, adenosine diphosphate; ADP-Glc, ADP- glucose; AMP, adenosine monophosphate; ATP, adenosine triphosphate; CMP, cytosine monophosphate; CTP, cytosine triphosphate; DHAP, dihydroxyacetone phosphate; Disaccharide-P, disaccharide phosphate; F1P, fructose 1-phosphate; F6P, fructose 6-phosphate; FAD, flavin adenine dinucleotide; FBP, fructose bisphosphate; FMN, flavin mononucleotide; G1P, glucose 1-phosphate; G6P, glucose 6-phosphate; GAP, glyceraldehyde 3-phosphate; GDP, guanosine diphosphate; GMP, guanosine monophosphate; GTP, guanosine triphosphate;HMBPP, 4-Hydroxy-3-methyl-but-2-enyl pyrophosphate; IMP, inosine monophosphate; IPP, isopentenyl pyrophosphate; DMAPP, dimethylallyl pyrophosphate; MEP, 2-C-methylerythritol 4-phosphate; Mn6P, mannose 6-phosphate; NAD, nicotinamide adenine dinucleotide; NADP, nicotinamide adenine dinucleotide phosphate; PEP, phosphoenolpyruvate; R1P, ribose 1-phosphate; R5P, ribose 5-phosphate; Ru5P, ribulose 5-phosphate; RuBP, ribulose bisphosphate; S7P, sedoheptulose 7-phosphate; TMP, thymidine monophosphate; UDP-Glc, UDP-glucose; UMP, uridine monophosphate; UTP, uridine triphosphate; XMP, xanthosine monophosphate. 2.2. Overview of the Metabolic Differences between Type Strains Using our targeted metabolic profiling method, the absolute cellular concentrations of 84 metabolites were successfully determined in the three cyanobacterial strains. All metabolite concentrations were standardized by g-dry cell weight ( Table 1 ). Fifteen metabolites (organic acids and amino acids) were analyzed by both GC/QqQ-MS and LC/QqQ-MS systems. Although calculated from different calibration curves, the absolute concentration of the same metabolites derived from two different analytical systems showed similar values and standard deviation ( Figure 2 a). This result indicates that each analytical system is sufficiently accurate on its own to represent the same data sets. In this study, we selected one data set that has smaller standard deviation. metabolites-04-00499-t001_Table 1 Table 1 Absolute concentrations of each metabolite in μmol/g-drycell weight. S.D. represents standard deviation. Metabolite PCC6803 PCC7002 PCC7942 Average S.D. Average S.D. Average S.D. 2-Isopropylmalate 4.11 × 10 −1 7.75 × 10 −2 1.34 × 10 −1 2.01 × 10 −2 3.22 × 10 –2 1.18 × 10 −2 2-Oxoglutarate 2.94 × 10 −1 7.06 × 10 −2 1.77 × 10 0 7.23 × 10 −1 1.77 × 10 −1 5.14 × 10 −2 3PGA 1.48 × 10 1 1.86 × 10 0 2.02 × 10 1 4.87 × 10 0 7.13 × 10 0 1.18 × 10 0 4-Aminobutanoate 9.67 × 10 −3 2.09 × 10 −3 9.23 × 10 −3 2.40 × 10 −3 1.42 × 10 −2 1.01 × 10 −2 6-Phosphogluconate 4.66 × 10 0 8.35 × 10 −1 1.99 × 10 0 1.11 × 10 0 1.87 × 10 −1 4.76 × 10 −2 Acetyl CoA 4.43 × 10 −1 4.29 × 10 −2 6.22 × 10 −2 8.02 × 10 −3 6.54 × 10 −2 9.12 × 10 −3 ADP 6.51 × 10 0 7.58 × 10 −1 2.68 × 10 0 1.46 × 10 −2 2.38 × 10 0 6.44 × 10 −1 ADP-Glc 7.00 × 10 −2 4.13 × 10 −2 6.13 × 10 −1 2.66 × 10 −1 3.88 × 10 −2 1.76 × 10 −2 Alanine 4.53 × 10 0 1.16 × 10 0 6.50 × 10 −1 1.48 × 10 −1 8.12 × 10 −1 6.74 × 10 −1 α-Glycerophosphate 4.77 × 10 −1 7.50 × 10 −2 9.83 × 10 −1 1.94 × 10 −1 2.22 × 10 −1 3.74 × 10 −2 AMP 1.07 × 10 1 1.48 × 10 0 4.56 × 10 0 4.09 × 10 −1 3.92 × 10 0 1.09 × 10 0 Arginine + Ornithine 2.12 × 10 1 5.02 × 10 0 4.31 × 10 0 6.72 × 10 −1 3.52 × 10 0 9.79 × 10 −1 Asparagine 4.34 × 10 −1 8.01 × 10 −2 2.66 × 10 −2 3.97 × 10 −3 8.68 × 10 −2 3.49 × 10 −2 Aspartate 7.14 × 10 0 1.20 × 10 0 2.69 × 10 0 6.09 × 10 -1 2.53 × 10 0 1.07 × 10 0 ATP 2.14 × 10 0 5.58 × 10 −1 3.09 × 10 0 3.48 × 10 −1 1.90 × 10 0 8.72 × 10 −1 Citrate 2.16 × 10 0 7.83 × 10 −2 4.22 × 10 0 1.43 × 10 0 3.57 × 10 −1 8.69 × 10 −2 CMP 3.43 × 10 −1 7.71 × 10 −2 2.17 × 10 −1 4.07 × 10 −2 2.71 × 10 −1 3.67 × 10 −2 CTP 5.97 × 10 −1 1.08 × 10 -1 6.52 × 10 −2 9.81 × 10 −3 2.97 × 10 −2 1.13 × 10 −2 Cystein 4.79 × 10 −1 2.30 × 10 −1 5.28 × 10 −1 1.73 × 10 −1 2.32 × 10 −1 1.89 × 10 −1 Cytidine 1.28 × 10 −2 4.16 × 10 −3 2.11 × 10 −2 5.56 × 10 −3 3.40 × 10 −2 7.35 × 10 −3 DHAP 4.30 × 10 −1 7.71 × 10 −2 5.48 × 10 −1 2.91 × 10 −1 1.15 × 10 −1 1.67 × 10 −2 Disaccharide-P 6.03 × 10 −4 3.27 × 10 −4 1.21 × 10 −2 9.63 × 10 −3 1.56 × 10 −3 1.05 × 10 −3 F1P 1.12 × 10 −2 3.71 × 10 −3 1.90 × 10 −2 4.37 × 10 −3 6.99 × 10 −3 1.04 × 10 −3 F6P 4.61 × 10 −1 9.81 × 10 −2 5.19 × 10 −1 2.13 × 10 −2 2.75 × 10 −1 7.97 × 10 −2 FAD 9.98 × 10 −2 3.19 × 10 −2 1.60 × 10 −1 3.38 × 10 −2 1.98 × 10 −1 2.80 × 10 −2 FBP 2.40 × 10 −1 1.07 × 10 −1 3.39 × 10 −1 1.44 × 10 −2 2.08 × 10 −2 1.37 × 10 −3 FMN 4.77 × 10 −2 1.95 × 10 −2 4.81 × 10 −2 7.94 × 10 −3 5.26 × 10 −2 2.18 × 10 −2 Fumarate 1.63 × 10 −1 2.27 × 10 −2 9.08 × 10 −2 9.45 × 10 −3 5.11 × 10 −2 9.58 × 10 −3 G1P 6.91 × 10 −2 1.04 × 10 −2 1.41 × 10 −1 1.45 × 10 −2 9.23 × 10 −2 4.99 × 10 −2 G6P 1.84 × 10 0 3.10 × 10 −1 4.19 × 10 0 7.29 × 10 −1 1.57 × 10 0 3.43 × 10 −1 GAP 2.63 × 10 −1 7.81 × 10 −2 5.59 × 10 −1 2.24 × 10 −1 1.88 × 10 −1 4.91 × 10 −2 GDP 1.38 × 10 0 1.73 × 10 −1 9.38 × 10 −1 1.77 × 10 − 1 3.46 × 10 0 4.42 × 10 0 Glutamate 2.13 × 10 2 4.79 × 10 1 4.25 × 10 1 6.52 × 10 0 3.39 × 10 1 8.11 × 10 0 Glutamine 3.47 × 10 0 4.61 × 10 −1 5.14 × 10 −1 1.37 × 10 −2 6.34 × 10 −1 2.12 × 10 −1 Glycerate 2.53 × 10 −2 9.28 × 10 −3 2.06 × 10 −2 5.33 × 10 −4 2.09 × 10 −2 1.93 × 10 −2 Glycine 1.67 × 10 1 4.51 × 10 0 2.17 × 10 0 8.78 × 10 −1 1.78 × 10 1 2.66 × 10 1 Glycolate 7.51 × 10 −2 2.03 × 10 −2 3.18 × 10 −2 1.47 × 10 −2 5.44 × 10 −2 1.49 × 10 −2 GMP 7.30 × 10 −1 2.05 × 10 −1 3.75 × 10 −1 3.47 × 10 −2 3.77 × 10 −1 6.40 × 10 −2 GTP 8.31 × 10 −1 2.35 × 10 −1 9.61 × 10 −1 1.93 × 10 −1 4.11 × 10 −1 1.31 × 10 −1 Guanosine 7.26 × 10 −2 3.51 × 10 −2 2.91 × 10 −1 1.92 × 10 −1 9.43 × 10 −2 3.71 × 10 −2 Histidine 1.10 × 10 −1 3.46 × 10 −2 7.03 × 10 −2 2.20 × 10 −2 6.45 × 10 −2 2.58 × 10 −2 HMBPP 9.27 × 10 −2 2.66 × 10 −2 5.40 × 10 −2 1.90 × 10 −2 3.32 × 10 −2 4.33 × 10 −3 Homoserine 3.05 × 10 −2 1.16 × 10 −2 1.32 × 10 −2 1.33 × 10 −3 1.51 × 10 −2 3.85 × 10 −3 Hydroxyproline 9.27 × 10 −1 3.05 × 10 −1 4.44 × 10 −1 1.72 × 10 −1 7.65 × 10 −1 5.11 × 10 −1 IMP 4.64 × 10 −1 3.84 × 10 −2 2.48 × 10 −1 9.97 × 10 −2 1.72 × 10 −1 6.00 × 10 −2 Inosine 1.46 × 10 −2 5.54 × 10 −3 9.41 × 10 −2 8.20 × 10 −2 6.80 × 10 −2 2.51 × 10 −2 IPP + DMAPP 1.40 × 10 −2 1.60 × 10 −3 9.64 × 10 − 3 1.79 × 10 −3 1.63 × 10 −2 1.28 × 10 −3 Isocitrate 1.45 × 10 0 1.25 × 10 −1 3.04 × 10 0 1.43 × 10 0 1.09 × 10 −1 4.96 × 10 −3 Isoleucine 3.70 × 10 −1 9.37 × 10 −2 7.14 × 10 −2 1.37 × 10 −2 1.50 × 10 −1 1.16 × 10 −1 Leucine 7.72 × 10 −1 1.55 × 10 −1 1.26 × 10 −1 2.32 × 10 −2 1.76 × 10 −1 1.33 × 10 −1 Lysine 1.57 × 10 −1 4.40 × 10 −2 1.02 × 10 −1 1.11 × 10 −2 9.32 × 10 −2 3.62 × 10 −2 Malate 1.82 × 10 −1 2.24 × 10 −2 6.36 × 10 −2 4.72 × 10 −3 2.08 × 10 −1 7.92 × 10 −2 MEP 8.38 × 10 −2 1.64 × 10 −2 3.94 × 10 −2 9.17 × 10 −3 3.83 × 10 −2 8.08 × 10 −3 Methionine 6.13 × 10 −1 1.10 × 10 −1 2.67 × 10 −1 4.40 × 10 −2 2.30 × 10 −1 4.51 × 10 −2 Mn6P 2.92 × 10 −1 4.19 × 10 −2 2.53 × 10 −1 7.48 × 10 −2 2.02 × 10 −1 3.06 × 10 −2 NAD 5.14 × 10 −1 1.26 × 10 −1 3.84 × 10 −1 2.97 × 10 −2 5.64 × 10 −1 1.04 × 10 −1 NADP 6.14 × 10 −1 1.26 × 10 −1 6.82 × 10 −1 1.38 × 10 −1 3.79 × 10 −1 8.28 × 10 −2 Nicotinate 3.56 × 10 −4 7.66 × 10 −5 7.72 × 10 −4 3.27 × 10 −4 4.62 × 10 −4 1.86 × 10 −4 Orotate 1.03 × 10 −2 1.58 × 10 −3 5.97 × 10 −3 2.34 × 10 −3 1.75 × 10 −3 7.23 × 10 −4 Pantothenate 4.91 × 10 −3 1.02 × 10 −3 2.13 × 10 −4 3.22 × 10 −5 4.56 × 10 −4 2.59 × 10 −4 PEP 2.51 × 10 0 7.59 × 10 −1 2.27 × 10 0 7.18 × 10 −1 1.47 × 10 0 2.45 × 10 −1 Phenylalanine 1.99 × 10 −1 6.09 × 10 −2 7.73 × 10 −2 1.86 × 10 −2 7.63 × 10 −2 4.77 × 10 −2 Proline 9.14 × 10 −1 1.76 × 10 −1 5.06 × 10 −2 9.05 × 10 −3 1.38 × 10 −1 1.23 × 10 −1 Pyridoxamine-5P 4.58 × 10 −2 5.05 × 10 −3 3.32 × 10 −2 2.92 × 10 −3 3.17 × 10 −2 3.82 × 10 −3 Pyroglutamate 2.81 × 10 0 7.22 × 10 −1 5.95 × 10 −1 8.06 × 10 −2 1.51 × 10 0 1.41 × 10 0 Pyruvate 1.05 × 10 1 1.38 × 10 0 1.81 × 10 1 4.20 × 10 0 5.76 × 10 0 1.12 × 10 0 R1P 3.62 × 10 −3 7.87 × 10 −4 2.90 × 10 −3 1.36 × 10 −3 9.68 × 10 −4 2.08 × 10 −4 R5P 1.95 × 10 −1 6.88 × 10 −2 1.52 × 10 −1 6.33 × 10 −2 5.82 × 10 −2 8.74 × 10 −3 Ru5P 4.04 × 10 −1 6.01 × 10 −2 2.19 × 10 −1 4.25 × 10 −2 4.48 × 10 −2 5.98 × 10 −3 RuBP 2.92 × 10 −1 5.35 × 10 −2 2.37 × 10 −1 5.29 × 10 −2 8.39 × 10 −2 2.74 × 10 −2 S7P 3.57 × 10 0 1.10 × 10 0 5.60 × 10 0 2.50 × 10 0 1.28 × 10 0 4.41 × 10 −1 Serine 7.12 × 10 −1 2.63 × 10 −1 3.11 × 10 −1 9.71 × 10 −2 9.12 × 10 −1 1.01 × 10 0 Shikimate 1.32 × 10 −3 5.41 × 10 −4 1.66 × 10 −3 1.13 × 10 −3 1.19 × 10 −3 7.67 × 10 −4 Succinate 3.23 × 10 −1 7.42 × 10 −2 2.15 × 10 −1 6.81 × 10 −2 1.97 × 10 −1 4.21 × 10 −2 Threonine 8.39 × 10 −1 1.99 × 10 −1 2.95 × 10 −1 8.09 × 10 −2 3.52 × 10 –1 2.68 × 10 −1 TMP 2.71 × 10 −2 6.61 × 10 −3 2.23 × 10 −2 6.80 × 10 −3 3.57 × 10 −2 2.71 × 10 −3 Tryptophan 2.19 × 10 −1 4.26 × 10 −2 6.95 × 10 −2 7.88 × 10 −3 8.85 × 10 −2 1.18 × 10 −2 Tyrosine 3.04 × 10 0 6.83 × 10 −1 6.09 × 10 −1 6.31 × 10 −2 5.62 × 10 −1 2.60 × 10 −1 UDP-Glc 2.24 × 10 0 2.22 × 10 −1 6.98 × 10 −1 7.48 × 10 −2 1.05 × 10 0 1.97 × 10 −1 UMP 7.44 × 10 −1 1.52 × 10 −1 6.05 × 10 −1 2.43 × 10 −1 7.79 × 10 −1 1.58 × 10 −1 Uridine 2.16 × 10 −2 9.04 × 10 −3 5.21 × 10 −2 3.01 × 10 −2 8.93 × 10 −2 2.92 × 10 −2 UTP 5.54 × 10 −1 1.49 × 10 −1 3.19 × 10 −1 4.17 × 10 −2 1.75 × 10 −1 4.96 × 10 −2 Valine 9.81 × 10 −1 2.00 × 10 −1 2.09 × 10 −1 3.94 × 10 −2 3.60 × 10 −1 2.13 × 10 −1 XMP 4.73 × 10 −2 1.37 × 10 −2 8.66 × 10 −2 3.06 × 10 −2 1.01 × 10 −1 2.73 × 10 −2 Figure 2 ( a ) Comparison of the absolute concentration values from GC/QqQ-MS and LC/QqQ-MS systems. The bar graphs represent the mean values of triplicate samples and error bars indicate the standard deviations. These metabolites were successfully detected by both systems and resulted in similar values of absolute concentration; ( b ) Comparison of the metabolic profiles of the three cyanobacterial strains, namely, PCC7002, PCC7942 and PCC6803 under photoautotrophic conditions. Z-scored data were hierarchically clustered and the results are represented with a heat map. For biological production, high proliferative ability and capability for genetic modification are important ( Table 2 ). The cyanobacterial strains used in this study are genetically well-studied and the whole genome information is available. Therefore, they have potential to be attractive host strain candidates for metabolic engineering. From our results, the metabolic profile of the PCC6803 strain clearly indicated that most amino acids were present in higher amounts and this tendency was also observed in the other Synechocystis strain: Synechocystis sp. PCC6714 ( Supplementary Figure 1 ). Although it is known that the two types of Synechocystis strains, PCC6803 and PCC6714, have several physiological differences such as salinity response, they are closely related from the viewpoint of their genome (16S rDNA identity 99.4%) [ 26 ]. Nevertheless, the metabolic profiles of these two strains are clearly different from each other in terms of several metabolites ( Supplementary Figure 1 ). These results confirm that metabolomics has the ability for high-resolution phenotypic analysis that can differentiate genetically similar strains, which can be informative when choosing the appropriate host for biological production. On the other hand, some intermediates of the primary carbon metabolic pathway such as Calvin and TCA cycles were notably accumulated only in PCC7002 ( Figure 2 b). It was previously reported that the TCA cycle activity exhibited a strong correlation with the growth rate of Saccharomyces cerevisiae [ 27 ] and this phenomenon is expected to occur in other aerobic microorganisms. However, we could not find any relationship between the level of TCA cycle intermediates and the growth rate in three cyanobacterial strains. This might be due to the “unusual” TCA cycle of cyanobacteria [ 28 ] and their distinct energy acquisition strategy derived from the photosystem. Although a novel insight for cyanobacterial cyclic TCA cycle was reported wherein succinic semialdehyde was suggested to be a new intermediate of TCA cycle [ 29 ], the metabolite could not be detected in either strain. It is probable that succinic semialdehyde is not accumulated under constant light condition, although this assumption needs confirmation. metabolites-04-00499-t002_Table 2 Table 2 Phenotypic characterization of the strains used in this study. Strain Growth Rate (h −1 ) 1 Genetic Modification Utilizable Organic Substrate Synechocystis sp. PCC6803 0.107 ± 0.003 [ 30 ] Glucose 2 Synechococcus sp. PCC7002 0.165 ± 0.002 [ 6 ] Glucose, Glycerol Synechococcus elongatus PCC7942 0.123 ± 0.002 [ 31 ] None 1 The growth rate is measured and expressed as the mean ± S.D. ( n = 3); 2 In case of the glucose-tolerant strain. 2.3. Characterization of Each Strain by Molar-Based Metabolite Distribution and AEC Values The absolute cellular concentrations of 84 metabolites are described in the pie graph of each strain ( Figure 3 a). Glutamate was the most abundant metabolite in all strains, especially in Synechocystis strains; occupying approximately 60% of the total detected metabolites. In these graphs, five metabolite groups were classified namely: amino acids, sugar phosphates, nucleotides, organic acids and others based on their chemical properties or roles in metabolism. Among the highly observed metabolites in the cyanobacterial strains, glutamate, arginine and ornithine are related to nitrogen metabolism [ 32 ], 3PGA is a product of carbon fixation reaction, three adenylate nucleotides work as energy currencies, and pyruvate is a final product of glycolysis. These compounds play important roles in energy acquisition or nutrient storage. In spite of the similar lineup of metabolites, the molar-based group distribution of PCC7002 was significantly different from the others. In order to minimize the effects of different culture conditions on metabolome, we cultivated marine cyanobacterium PCC7002 at the same temperature and light conditions with the other two strains except for modified BG-11 medium. Although the temperature in our condition is relatively low compared to the optimal condition, there is no significant difference in the doubling time of PCC7002 between our condition (4.26 ± 0.06 h) and the optimal condition (4 ± 0.3 h) [ 33 ]. However, it is necessary to point out that low temperature has been shown to cause nitrogen limitation in PCC7002 [ 33 ], which probably explained the notable accumulation of 2-oxoglutarate and low levels of glutamate and other related amino acids in PCC7002 under our condition. The adenine nucleotides ATP, ADP and AMP stoichiometrically couple with global metabolic pathways of a living cell in which metabolically available energy is momentarily stored in the form of energy-rich phosphate bonds. Therefore, the intracellular adenylate energy condition is expected to reflect phenotypic or metabolic differences and give an informative guide for elucidation of metabolism in various strains of cyanobacteria. To evaluate the intracellular energy condition, we calculated the adenylate energy charge (AEC) in terms of absolute concentrations using the Equation\n (2) \n The AEC value has been proven to be a metabolic regulatory parameter for various enzymatic reactions and their related pathways in vivo [ 34 , 35 ]. The lowest AEC value was observed in Synechocystis . In contrast, PCC7002 had the highest value among the three strains ( Figure 3 b) although the AEC values (with a mean of 0.359) were relatively low compared with those of non-photosynthetic microorganisms, for example, Escherichia coli (0.90–094), Saccharomyces cerevisiae (0.84–0.93) [ 36 ]. Based on the AEC values and metabolite distribution data, there seems to be a relationship between th AEC value and accumulation of phospho-related metabolites. More interestingly, we found very similar tendencies between AEC values and the growth rate of mid-exponential phase in the three investigated strains and as well as in PCC6714 ( Supplementary Figure 2 ). The growth rate of each strain was estimated by its growth curves during exponential growth phase ( Supplementary Figure 3 ). Since the Pearson correlation coefficient using mean values was 0.906, AEC values were strongly correlated with growth rate, but not with the absolute concentration of ATP. This result indicates that cyanobacterial growth during mid-log phase is directly or indirectly dominated by cellular energy states. Although a well-correlated relationship between AEC and growth in Escherichia coli has been reported in previous research [ 36 ], we did not expect this result because cyanobacteria have a completely different energy strategy from aerobionts such as E.coli . Cyanobacteria mainly obtain ATP from photosynthesis but most aerobionts get ATP from cellular respiration. Since ATP has been known as an activator of ribulosebisphosphate carboxylase/oxygenase which is a key enzyme of calvin-benson cycle [ 37 ], cellular energy state should have affected growth by a different mechanism from aerobionts. Although an approach to enhance the proliferative ability—which is important for cyanobacterial biological production—by optimizing the upstream cultivation parameters has been described [ 38 ], there is hitherto no approach based on cellular energy state. Our result demonstrated a new possibility of improving cyanobacterial growth from a completely different aspect. On the other hand, the information on organic carbon distribution is likely to play an important role in selecting appropriate strains for specific biochemical production because several metabolites can be precursors of important bioproducts. Figure 3 ( a ) Composition of the measured metabolome in three cyanobacteria. The pie graph shows the distributions of different metabolites in mid-exponential phase; ( b ) Comparison of adenylate energy charge and growth rate in each strain. The bar graphs indicate the mean values of triplicates and error bars indicate the standard deviation. 2.4. Overview and Prospect for Cyanobacterial Biological Production As described previously, there are many challenges in cyanobacterial biological production. Most of these researches in cyanobacteria are using genetically modified strains with introduced exogenous genes because the wild type strain has low or no ability to produce valuable biofuels such as alcohols or some kinds of carbohydrates. However, the end products are derived from naturally occurring precursors. Assuming a simple Michaelis-Menten enzymatic reaction without the allosteric effect, the reaction rate unambiguously depends on the molar concentration of the substrate. Therefore, monitoring such kind of precursors and determining the metabolic rate-limiting points are important and essential considerations in metabolic engineering even when using wild type strains. In order to visualize the previous reports of cyanobacterial production, we summarized the modified metabolic pathways, which described the relationship between primary metabolites and various targeted products ( Figure 4 ). This network clearly showed that most targeted products were closely linked to pyruvate or acetyl-CoA, and the level of these two metabolites vary significantly among the three cyanobacterial type strains. In addition, the ratios of pyruvate/acetyl-CoA are commonly high but different among the strains. In the case of PCC7002, the amount of pyruvate was approximately 300-fold relative to that of acetyl-CoA. The reaction from pyruvate to acetyl-CoA, which is highly regulated by complex cell mechanisms [ 39 ], might be the limiting step in various metabolic pathways. For further improvement of cyanobacterial biological production through acetyl-CoA, a better alternative pathway which is independent from pyruvate dehydrogenase would be preferable. For example, non-oxidative glycolysis pathway via phosphoketolases seems to be one of the possibilities for avoiding this limiting step [ 40 ]. Figure 4 Overview of the modified metabolic pathways related to previously reported cyanobacterial biological production. The targeted products are underlined while the metabolites in red can be quantitated by our method. The vertical axes of the bar graphs show the intercellular concentrations in μmol/g-dry cell weight of each metabolite (mean ± S.D., n = 3)."
} | 7,482 |
34828122 | PMC8622738 | pmc | 7,798 | {
"abstract": "Machine learning methods, such as Long Short-Term Memory (LSTM) neural networks can predict real-life time series data. Here, we present a new approach to predict time series data combining interpolation techniques, randomly parameterized LSTM neural networks and measures of signal complexity, which we will refer to as complexity measures throughout this research. First, we interpolate the time series data under study. Next, we predict the time series data using an ensemble of randomly parameterized LSTM neural networks. Finally, we filter the ensemble prediction based on the original data complexity to improve the predictability, i.e., we keep only predictions with a complexity close to that of the training data. We test the proposed approach on five different univariate time series data. We use linear and fractal interpolation to increase the amount of data. We tested five different complexity measures for the ensemble filters for time series data, i.e., the Hurst exponent, Shannon’s entropy, Fisher’s information, SVD entropy, and the spectrum of Lyapunov exponents. Our results show that the interpolated predictions consistently outperformed the non-interpolated ones. The best ensemble predictions always beat a baseline prediction based on a neural network with only a single hidden LSTM, gated recurrent unit (GRU) or simple recurrent neural network (RNN) layer. The complexity filters can reduce the error of a random ensemble prediction by a factor of 10. Further, because we use randomly parameterized neural networks, no hyperparameter tuning is required. We prove this method useful for real-time time series prediction because the optimization of hyperparameters, which is usually very costly and time-intensive, can be circumvented with the presented approach.",
"conclusion": "12. Conclusions This research aimed to improve neural network time series predictions using complexity measures and interpolation techniques. We presented a new approach not existing in the recent literature and tested it on five different univariate time series. First, we interpolated the time series data under study using fractal and linear interpolation. Second, we generated randomly parameterized LSTM (one may understand the randomly parameterized LSTM neural network as a sort of a brute-force neural network approach). Neural networks and a step-by-step approach predicted the data under study, i.e., each dataset, and consequently, a non-interpolated dataset, a linear-interpolated dataset, and a fractal-interpolated dataset. Lastly, we filtered these random ensemble predictions based on their complexity, i.e., we kept only the forecasts with complexity close to the original complexity of the data. By applying the filters to the randomly parameterized LSTM ensemble, we reduced the error of the randomly parameterized ensemble by a factor of 10. The best filtered ensemble predictions consistently outperformed a single LSTM prediction, which we use as a reasonable baseline. Filtering ensembles based on their complexities has not been done before and should be considered for future ensemble predictions to reduce the costs of optimizing neural networks. In terms of interpolation techniques, we found that fractal interpolation works best under the given circumstances. For the complexity filters, we identified a combination of a Single-Value-Decomposition-based, e.g., SVD entropy or Fisher’s information, and another complexity measure, e.g., the Hurst exponent, to perform best. We conclude that interpolation techniques generating new data with a complexity close to that of the original data are best suited for improving the quality of a forecast. We expect the presented methods to be further exploited when predicting complex real-life time series data, such as environmental, agricultural, or financial data. This is because researchers are often confronted with a meager amount of data, and the given time series properties/complexity may change over time. Thus, new sets of hyper-parameters may have to be found. Using a random neural network ensemble and then filtering the predictions with respect to the complexity of older data circumvents this problem.",
"introduction": "1. Introduction Machine learning and neural networks are today’s state of the art when it comes to predicting time series data. Applications feature various research areas and tasks, such as future population estimates, predicting epileptic seizures [ 1 ], or estimating future stock market prices. All machine learning approaches depend on the quality and quantity of the available data, i.e., their complexity or randomness and the actual amount of data, and the algorithm’s right parameterization. The three main reasons for machine learning approaches to perform poorly are: An insufficient amount of data, i.e., the data are not fine-grained or long enough; Random data, with its blueprint, the Brownian motion (cf. [ 2 ]); Bad parameterization of the algorithm. A way to enrich a dataset and increase its fine-grainededness is to use interpolation. In the simplest case, this is done using a linear interpolation [ 3 ]. Complex real-life data are generally considered non-linear and originate from systems where the interactions and corresponding agents are not known in detail. Therefore, linear interpolation approaches do not depict the complexity and non-linearity of real-life systems. One approach to consider the complexity of the data under study is fractal interpolation [ 4 ]. The challenge is to find the right interpolation technique for the task at hand, i.e., what interpolation technique is best suited for improving machine learning time series predictions? There are many ways to measure the complexity or inherent information of time series data. This article will refer to these measures of signal complexity as complexity measures since they measure how complex the data under study are. Given that, the question is how can the complexity of the data under study be taken into account to improve predictions? Finding the right set of hyperparameters for any given machine learning algorithm, especially for neural networks, can be a tedious and time-consuming task. In addition, there is usually more than one set of hyperparameters with high performance for a given dataset. Here, the hypothesis is that differently parameterized algorithms, again especially neural networks, capture different aspects of the data. However, is there a way to take into account different aspects of the data to reduce the effort of finding the right set of parameters? We present an approach that addresses those three issues. The overall hypothesis here is that neural network time series predictions can be improved by considering the complexity of the data under study. First, we interpolate the data under study using a fractal interpolation adapted to the data complexity. Second, we use randomly parameterized long short-term memory (LSTM [ 5 ]) neural networks to make ensemble predictions. Third, we use complexity properties of the data for filtering the ensemble predictions to improve their accuracy. Using the presented ideas, we can reduce the error of a random long short term memory (LSTM [ 5 ]) neural network ensemble prediction on average by a factor of 10 by filtering the results depending on their complexity. Further, we set a baseline by using an LSTM, a GRU, [ 6 ], and a recurrent neural network, [ 7 ], with only one hidden layer to make predictions of the time series data under study. The best ensemble predictions always outperformed these baseline predictions. The discussed filter methods can be applied to any ensemble predictions and therefore be used to improve existing ensemble approaches. We show that the fractal interpolation approach, which considers the complexity of the data under study, is the preferred method to improve neural network time series predictions. Therefore, one should consider the discussed interpolation techniques when facing an insufficient amount of data. Given that we used a randomly parameterized neural network implementation, this approach is a more generalized one and can be applied to any univariate time series dataset without further optimization, i.e., we can circumvent the optimization procedure. In addition, since the ensemble consists of several independent neural networks, this approach can, in principle, be parallelized infinitely to speed up the calculations. We further show how different filters perform for different datasets and give recommendations on which filters to employ to improve ensemble predictions. The remainder of the paper is organized as follows. Section 2 gives an overview on related research. In Section 3 , we describe the used methodology and discuss how the different techniques fit together. Section 4 presents all used datasets and Section 5 the corresponding interpolation techniques. In Section 6 , we discuss the used complexity measures. Section 7 shows the neural network implementation and the ensemble scheme. The error metrics are discussed in Section 8 . Section 9 describes the used ensemble (-complexity) filters. In Section 10 , we present LSTM, GRU and recurrent neural network predictions as a baseline to compare our ensemble predictions to. Section 11 discusses the results. We further conclude our findings in Section 12 . In addition, we added an Appendix to collect peripheral plots and data to keep the main text focused.",
"discussion": "11. Results and Discussion We linear- and fractal-interpolated five different time series data. Afterward, we did a random ensemble prediction for each, consisting of 500 different predictions for each interpolation technique (and non-interpolated time series data). The results of these random ensembles can be found in Appendix E in Table A5 and Table A6 . We further filtered these predictions using complexity filters (see Section 9 ) to finally reduce the number of ensemble predictions from 500 to 5, i.e., to 1%. The best five results for all time series data and each interpolation technique, regarding the RMSE and the corresponding error (see Section 8 ) are shown in Table 5 for the monthly international airline passengers dataset. Table A1 , Table A2 , Table A3 and Table A4 , which feature the results for all other datasets, can be found in Appendix B . The corresponding plots for the three best predictions of each time series data can be found in Appendix C . We highlighted the overall best three results as bold entries. The results show that the interpolated approaches always outperformed the non-interpolated ones when it comes to the lowest RMSEs. Further, the ensemble predictions could significantly be improved using a combination of interpolation techniques and complexity filters. 11.1. Interpolation Techniques Regarding the different interpolation techniques of the overall three best results for all time series data, i.e., a total of 15 predictions, we find 9 fractal-interpolated predictions and 6 linear-interpolated predictions. Though the linear-interpolated results outperformed the fractal-interpolated ones in some cases, we conclude that fractal interpolation is a better way to improve LSTM neural network time series predictions. The reason for this is: Taking into account the results shown in Figure 7 and Table 5 , though the RMSE of the linear-interpolated result is lower (best result, lowest RMSE) than that of the second and third best ones (the fractal-interpolated results), the corresponding error of the RMSE is higher. Taking a closer look at the different ensemble predictions of Figure 7 , we can see that the quality of the single predictions for the linear interpolated case is lower in terms of how close the actual curve data are to the different ensemble predictions. Therefore, the authors guess that this advantage of the linear-interpolated results vanishes when the statistic, i.e., the number of different ensemble predictions, increases. This behavior can be found for the monthly international airline passenger dataset, the monthly car sales in Quebec dataset, and the CFE specialty monthly writing paper sales dataset. 11.2. Complexity Filters Of these 75 best results for all interpolation techniques and different data, only 13 are single filtered predictions. A significant 62 are double-filtered predictions (i.e., two different complexity filters were applied). Not a single unfiltered prediction made it into the top 75 results. We, therefore, suggest always using two different complexity filters for filtering ensembles. When it comes to the specific filters used, we cannot find a pattern within the 15 best results, as only the combinations SVD entropy & Hurst exponent and Lyapunov exponents & Hurst exponent occur more than once, i.e., each occurred only two times. Examining the 75 best results, though, we get a different picture. Here, we find 7 occurrences of the combination Shannon’s entropy & Fisher’s information followed by 6 occurrences of Shannon’s entropy & SVD entropy. Further, taking into account that SVD entropy and Fisher’s information behave similarly (as both are based on SVD, see Section 6 ), we find that 57 of the best 75 results contain at least one SVD-based, i.e., Single-Value-Decomposition-based, complexity measure. Therefore, we suggest using an SVD-based complexity measure in combination with the Hurst exponent Shannon’s entropy. The authors’ recommendation is that the best combination is SVD entropy & Hurst exponent. 11.3. Remarks and Summary Summing up the results of this research, we draw the following conclusions: Random ensemble predictions can significantly be improved using fractal and linear interpolation techniques. The authors recommend using a fractal interpolation approach as the shown results feature a more stable behavior than those for the linear interpolation; Random ensemble predictions can significantly be improved using complexity filters to reduce the number of predictions in an ensemble. Taking into account the unfiltered and non-interpolated results shown in Table A5 and Table A6 and comparing them to the best results, shown in Table 5 and Table A1 , Table A2 , Table A3 and Table A4 , we see that the RMSE was reduced by a factor of ≈ 10 on average; The best results of the random ensemble, i.e., the single step-by-step predictions always outperformed the baseline predictions, Table 2 , Table 3 and Table 4 , and Appendix D . Here, we note that the given baseline predictions are probably not the best results that can be achieved with an optimized LSTM neural network but are still reasonable results and serve as baseline to show the quality of the ensemble predictions; Though the unfiltered results ( Table A5 and Table A6 ) suggest a trend and a minimum for the errors depending on the number of interpolation points, this trend vanishes when applying complexity filters. Therefore, we could not find a trend for the number of interpolation points for any interpolation technique and any complexity filters. Though this research used univariate time series data for analysis, our approach can be extended to arbitrary dimensional multivariate time series data. We expect multivariate prediction approaches to benefit from this research greatly. For multivariate time series, different features may have different complexity properties. Thus, one may employ tools such as effective transfer entropy, [ 49 ], which is a complexity measure specifically dealing with multivariate time series data, or other complexity measures fit to deal with multivariate problems. Further criteria for the best fit based on correlations present in multivariate data may be found regarding the fractal interpolation for multivariate time series data. The limitations of the presented approach hide in the parameter range of the neural network implementation. Though we can set arbitrary ranges to parameterize the neural network, computation costs can be reduced significantly if a good range for a specific dataset is known or can be guessed. Further research of the presented framework may include switching the LSTM layers to feed-forward neural network layers or simple recurrent neural network (RNN, i.e., non-LSTM layers) layers. Here, one can adopt the ideas of time-delayed recurrent neural networks, [ 50 ], or time-delayed feed-forward neural networks, [ 51 ]. For both approaches, one can choose the input of the neural network to match the embedding of the time series, i.e., use the estimated time-delay and embedding dimension as done for a phase space reconstruction of a univariate time series data (see Appendix F ), as done in [ 52 ]."
} | 4,173 |
39215347 | PMC11363401 | pmc | 7,799 | {
"abstract": "Background Cabbage Fusarium wilt (CFW) is a devastating disease caused by the soil-borne fungus Fusarium oxysporum f. sp. conglutinans (Foc). One of the optimal measures for managing CFW is the employment of tolerant/resistant cabbage varieties. However, the interplay between plant genotypes and the pathogen Foc in shaping the rhizosphere microbial community, and the consequent influence of these microbial assemblages on biological resistance, remains inadequately understood. Results Based on amplicon metabarcoding data, we observed distinct differences in the fungal alpha diversity index (Shannon index) and beta diversity index (unweighted Bray–Curtis dissimilarity) within the rhizosphere of the YR (resistant to Foc) and ZG (susceptible to Foc) cabbage varieties, irrespective of Foc inoculation. Notably, the Shannon diversity shifts in the resistant YR variety were more pronounced following Foc inoculation. Disease-resistant plant variety demonstrate a higher propensity for harboring beneficial microorganisms, such as Pseudomonas , and exhibit superior capabilities in evading harmful microorganisms, in contrast to their disease-susceptible counterparts. Furthermore, the network analysis was performed on rhizosphere-associated microorganisms, including both bacteria and fungi. The networks of association recovered from YR exhibited greater complexity, robustness, and density, regardless of Foc inoculation. Following Foc infection in the YR rhizosphere, there was a notable increase in the dominant bacterium NA13, which is also a hub taxon in the microbial network. Reintroducing NA13 into the soil significantly improved disease resistance in the susceptible ZG variety, by directly inhibiting Foc and triggering defense mechanisms in the roots. Conclusions The rhizosphere microbial communities of these two cabbage varieties are markedly distinct, with the introduction of the pathogen eliciting significant alterations in their microbial networks which is correlated with susceptibility or resistance to soil-borne pathogens. Furthermore, we identified a rhizobacteria species that significantly boosts disease resistance in susceptible cabbages. Our results indicated that the induction of resistance genes leading to varied responses in microbial communities to pathogens may partly explain the differing susceptibilities of the cabbage varieties tested to CFW. \n Video Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s40168-024-01883-0.",
"conclusion": "Conclusions This study revealed that differences in specific resistance genes between CFW resistant and susceptible cabbage genotypes, along with inoculation with the Foc pathogen, significantly influenced the structure and composition of the rhizosphere microbial community. We successfully established a correlation between microbial communities and differential resistance levels to CFW. The FOC1 gene induced specific taxa in resistant variety YR and susceptible variety ZG, particularly an increase in beneficial bacteria in YR. Furthermore, the rhizosphere microbes in YR exhibited a more proactive response to Foc inoculation, effectively enhancing the recruitment of beneficial bacteria while simultaneously inhibiting the proliferation of harmful fungi. This work contributes to a better understanding of the phylogenetic mechanisms underlying the recruitment of beneficial rhizosphere microbiota and highlights the importance of considering microbial communities in the selection of CFW-resistant varieties. This opens up new avenues for breeding crops that more effectively utilize microbiota for protection and improved disease management.",
"discussion": "Discussion The rhizosphere microbiota is important for enhancing the adaptation and productivity of plant hosts and plays a key role in plant resistance to biotic and abiotic stresses [ 8 , 56 , 57 ]. It has attracted significant attention in recent decades. It is known that different plant genotypes can produce different rhizosphere microbiomes [ 58 , 59 ]. However, there is a lack of detailed study on the variability of rhizosphere microbial communities within the same plant species attributable to distinct disease resistance genes. Additionally, the differential responses of rhizosphere microbiota in cabbage varieties, resistant versus susceptible, to Foc infection, and the correlation between the levels of CFW resistance in cabbage and its rhizosphere microbial composition, remain underexplored. This study established that genotypic differences, as determined by a single disease resistance gene ( FOC1 ), along with inoculation with Foc, significantly influenced the rhizosphere microbiota of both the resistant variety YR and susceptible variety ZG. Notably, the microbial community diversity in YR underwent a more pronounced alteration following Foc inoculation compared to ZG, as evidenced by both alpha and beta diversity analyses. As plants grow, they deliver large amounts of nutrients to their roots, which are released into the soil through the root system. These root exudates selectively recruit beneficial microbes from the soil to colonize their root system; this phenomenon is more pronounced when plants are under stress [ 8 , 60 ]. Our findings indicated a significant association between the FOC1 gene and the enrichment of specific taxa within the rhizosphere microbiota. Notably, a greater number of bacterial zOTUs were enriched in YR compared to ZG, and these enriched bacterial zOTUs are classified as potentially beneficial bacteria. For instance, when exposed to the pathogen Rhizoctonia solani , sugar beet roots attract Chitinophaga into the endosphere, suppressing the fungal pathogen [ 61 ]. Sphingomonas can promote plant growth [ 62 ]. Conversely, a higher number of fungal zOTUs were enriched in ZG, encompassing both potentially pathogenic as well as beneficial. Intriguingly, beneficial bacteria predominated among the bacterial zOTUs enriched in both YR and ZG, with a notably higher enrichment observed in YR. Similarly, pathogenic fungi were the dominant component in the enriched fungal zOTUs of both YR and ZG, with ZG exhibiting a greater enrichment of pathogenic fungal species compared to YR. Although potentially pathogenic fungi were found in the YR, fungal zOTU belongs to the genus Fusarium , is well known for including phytopathogenic species, and was only enriched in ZG [ 63 ]. After inoculation with Foc, the YR variety exhibited an enrichment of 63 bacterial zOTUs, whereas 53 bacterial zOTUs were notably absent, compared to the condition before inoculation. In contrast, the ZG variety displayed an equal number of both enriched and absent bacterial zOTUs post-inoculation, compared to the condition before inoculation. Interestingly, most of these zOTUs enriched by YR and ZG belonged to genera beneficial to plants, which also indicated that YR has a stronger ability to recruit beneficial bacteria compared to ZG. Additionally, following inoculation with Foc, the YR variety demonstrated an enrichment of 21 fungal zOTUs and a concurrent absence of 39 fungal zOTUs, compared to the condition before inoculation. In comparison, the ZG variety exhibited an enrichment of 12 fungal zOTUs, alongside the absence of 25 fungal zOTUs, compared to the condition before inoculation. Furthermore, the zOTUs absent in each variety, which predominantly belong to pathogenic fungal genera, revealed a notable tendency in the YR variety to reduce the recruitment of pathogenic fungi following infestation with Foc. This observation implies that both the resistant variety YR and the susceptible variety ZG exhibit a tendency to recruit beneficial microbes while concurrently reducing the recruitment of pathogenic fungi in response to infestation by the pathogenic fungus Foc. This phenomenon has also been confirmed in similar studies [ 64 , 65 ]. Additionally, Boer et al. investigated the antagonistic effect of individual and mixed strains of four bacteria on plant pathogens and found that more bacteria in the soil could lead to stronger competition with pathogens for resources [ 66 ]. In contrast, the presence of pathogenic fungi other than the genus Fusarium may facilitate the development of CFW. For example, Rhizophlyctis , which is enriched in ZG and can degrade cellulose, may promote the colonization of cabbage roots by pathogenic fungi. [ 67 ]. Correspondingly, Khoury and Alcorn observed that infection by Rhizoctonia solani in two cotton varieties resulted in root lesions, a condition that may potentially compromise the effectiveness of physical barriers in these plants, thereby facilitating colonization by Verticillium albo-atrum [ 68 ]. Competitive and cooperative interactions between microbial species can influence plant health [ 9 , 69 , 70 ]. In this study, we found that FOC1 also has a significant influence on the rhizosphere microbial co-occurrence network of cabbage. Both before and after Foc inoculation, the microbial interaction network of the YR variety exhibited increased complexity, as evidenced by a greater number of nodes and edges. It has been demonstrated that soil microbial communities with more intricate networks (with more nodes and edges) confer greater benefits to plants compared to simpler networks [ 71 ]. Complex networks help to better respond to environmental changes or suppress soil-borne pathogens. For example, the microbial network of healthy soil is more complex than that of soil susceptible to bacterial wilt [ 72 ]. Additionally, inoculation of grassland soil into sterilized greenhouse soil increased the complexity of the chrysanthemum rhizosphere microbial network and suppressed the development of pathogenic fungi Olpidium [ 73 ]. The average degree and network density are important indicators of the network’s effectiveness. After inoculation with Foc, the average degree and network density were increased in both varieties, but the YR network had a higher average degree and network density than the ZG network both before and after inoculation. YR networks with higher average degrees and network density indicated high connectivity and community efficiency. The combination of higher average degrees and network density indicated that microbial communities were more responsive to environmental perturbations, and highly connected networks provided greater functional redundancy [ 74 – 76 ]. This was in alignment with the results of our alpha and beta diversity analyses. In this sense, the success of pathogen invasion can be reduced if the rhizosphere microbial community is highly connective and efficient [ 77 ]. Agler et al. introduced the concept of “microbial hubs” to describe the presence of highly interconnected species within plant microbial networks [ 78 ]. Their findings proposed that these extensively interconnected species played a crucial role in plant health, serving as intermediaries between plants and their microbiome. Microbial hubs may play a pivotal role in sustaining disease-suppressive soil conditions, enhancing nutrient absorption, augmenting the efficacy of biocontrol agents, and facilitating the mediation of defense signals among plants [ 78 – 80 ]. There were two hub taxa identified as Mycobacterium and Chitinophaga in the YR bacterial network (Fig. 4 ), with Mycobacterium and Chitinophaga being considered potentially beneficial bacteria [ 81 ]. In contrast, there was only one hub taxa identified as Chitinophaga in the ZG bacterial network (Fig. 4 ). After inoculation with Foc, the hub taxa of YR and ZG were replaced by Caulobacter (bzOTU_581), Pseudomonas (bzOTU_1647), Chitinophaga (bzOTU_72) and Novosphingobium (bzOTU_187), Ilumatobacter (bzOTU_1496), Mucilaginibacter (bzOTU_122), respectively (Fig. 4 ). Meanwhile, Pseudomonas (bzOTU_1647) and Chitinophaga (bzOTU_72) were enriched in YR after inoculation with Foc, while hub taxa in the ZG network were not enriched in ZG after inoculation with Foc. Consequently, we hypothesized that Pseudomonas (bzOTU_1647) and Chitinophaga (bzOTU_72) played a critical role in the functional network after YR inoculation with Foc. We have successfully isolated Pseudomonas brassicacearum NA13 from Foc-inoculated YR and sequenced the whole genome. Our results indicated that the NA13 strain plays a beneficial role both in inhibiting the growth of the pathogen Foc and in promoting plant defense responses. The genomic characterization of Pseudomonas brassicacearum NA13 was associated with plant immune responses and the transport of bacterial secondary metabolites, which may contribute to the protection of cabbage against Foc infestation. The rhizosphere microbiota are abundant in microbe-associated molecular patterns (MAMPs) that initiate the primary layer of plant immune defense, limiting pathogen proliferation. MAMPs such as bacterial flagellin, chitin deacetylase, and chitin activate strong, tissue-specific immune responses in the roots of Arabidopsis thaliana . These responses operate through pathways independent of salicylic acid (SA) and jasmonic acid (JA) signaling, demonstrating a highly specialized defense mechanism [ 82 ]. NA13 undoubtedly possesses a range of components acting as MAMPs (e.g., flagellin). Meanwhile, there were clusters of related genes that code for the secondary metabolites syringomycin, 2,4-diacetylphloroglucinol and hydrogen cyanide, which are well-known antifungal compounds [ 83 , 84 ]. Concurrently, the presence of NA13 was observed to induce a robust defense response in plants. In the JA/ET and SA signaling pathways, the marker genes PR4 (JA/ET) and PR1 (SA) were significantly upregulated, indicating a robust activation of the plant's defense mechanisms. Intriguingly, WRKY70, a negative regulator within the SA biosynthesis, along with negative regulators CTR1 of the ET signaling pathway, demonstrated a marked downregulation. In addition to these findings, it has been well established that root-associated microbes trigger Induced Systemic Resistance (ISR), a defense mechanism that is distinct from the SA pathway and often associated with JA and ET signaling pathways [ 85 ]. ISR activation by beneficial microbes like NA13 enhances the plant's ability to fend off a wide range of pathogens through a priming effect, where the plant’s immune system is put on alert and responds more robustly to subsequent pathogen attacks. This modulation of gene expression represents a noteworthy shift in the plant's immune defense strategy, underlining the pivotal role of NA13 in orchestrating these changes. The significant alterations in the expression of these defense signaling pathway marker genes signify the activation of plant immune defenses, highlighting the intricate interplay between microbial presence and plant physiological responses. This indicated that (1) NA13 is capable of activating the plant’s immune response to pathogens, and (2) this bacterium could use secondary metabolites to have beneficial effects on the plant, such as directly inhibiting the growth of pathogens. The resistance gene FOC1 was classified as a TIR-NBS-LRR type R gene [ 6 ]. In plant immune responses, the activation of R genes triggers downstream immune signaling, which regulates plant resistance to pathogens. For instance, the TIR domain of the tobacco N gene is essential and sufficient for its association with the pathogen-derived elicitor p50, conferring resistance to tobacco mosaic virus [ 86 ]. The plant defense system relies on the mitogen-activated protein kinase (MAPK) signaling cascade, various transcription factors including NAC, MYB, and WRKY, pathogenesis-related (PR) genes, and the signaling networks for hormones such as salicylic acid (SA), jasmonic acid (JA), ethylene (ET), gibberellin (GA), and abscisic acid (ABA), all of which function both independently and in concert [ 87 – 90 ]. Zhou et al. demonstrated that the TIR-NBS-LRR type R gene GmTNL16 participates in soybean defense against Phytophthora via the JA and SA pathways [ 91 ]. Nevertheless, the intricate interactions and regulatory mechanisms among beneficial microbes such as NA13, the resistance gene FOC1 , the pathogen Foc, and the plant immune system remain to be fully elucidated. Further research is essential to uncover the complexities of these interactions and their implications for plant immunity. These findings not only provided deeper insights into the molecular mechanisms governing plant immunity but also underscored the potential of manipulating such pathways for enhancing disease resistance in crops. In summary, our results suggested that NA13 may play a key role in the three-way interaction with host plant immunity and fungal pathogen via a mechanism that enhances plant defence and inhibit Foc growth."
} | 4,211 |
34256567 | PMC8397335 | pmc | 7,800 | {
"abstract": "The control of liquid\nmotion on the micrometer scale is important\nfor many liquid transport and biomedical applications. An efficient\nway to trigger liquid motion is by introducing surface tension gradients\non free liquid interfaces leading to the Marangoni effect. However,\na pronounced Marangoni-driven flow generally only occurs at a liquid–air\nor liquid–liquid interface but not at solid–liquid interfaces.\nUsing superhydrophobic surfaces, the liquid phase stays in the Cassie\nstate (where liquid is only in contact with the tips of the rough\nsurface structure and air is enclosed in the indentations of the roughness)\nand hence provides the necessary liquid–air interface to trigger\nevident Marangoni flows. We use light to asymmetrically heat this\ninterface and thereby control liquid motion near superhydrophobic\nsurfaces. By laser scanning confocal microscopy, we determine the\nvelocity distribution evolving through optical excitation. We show\nthat Marangoni flow can be induced optically at structured, air-entrapping\nsuperhydrophobic surfaces. Furthermore, by comparison with numerical\nmodeling, we demonstrate that in addition to the Marangoni flow, buoyancy-driven\nflow occurs. This effect has so far been neglected in similar approaches\nand models of thermocapillary driven flow at superhydrophobic surfaces.\nOur work yields insight into the physics of Marangoni flow and can\nhelp in designing new contactless, light-driven liquid transport systems,\ne.g., for liquid pumping or in microfluidic devices.",
"conclusion": "Conclusions To\nmanipulate liquid flow by light, we brought the liquid, in our\ncase water, in contact with a superhydrophobic surface, in whose roughness\nfeatures air is entrapped. The superhydrophobic surface is composed\nof soot to absorb the light efficiently and convert it to a temperature\nincrease. A focused light is illuminating a spot on a superhydrophobic\nsurface. This generates a nonuniform temperature distribution, which\ninduces liquid flows. Even for a small temperature gradient, e.g.,\n0.2 K/mm generated on the solid surface, flows with a velocity of\na few micrometers per second are created. Our experimental and numerical\nstudy has shown that it is indeed possible to create Marangoni flows\non the superhydrophobic surface which drive the liquid. We have shown\nthat the heating also induces buoyancy-driven flows, which have so\nfar been neglected for superhydrophobic surfaces. Buoyancy-driven\nflows might act in the same direction as the Marangoni flows and are\ntherefore not necessarily easy to distinguish. For the flow cell chosen,\nwe have identified these two contributions and related their magnitudes\nwith respect to each other. This fundamental study helps to\nunderstand the mechanisms of fluid\ndynamics near superhydrophobic surfaces. From a practical view, optical\nactuation provides several advantages: on the one hand, by adjusting\nthe light spot size and intensity, the temperature gradient on superhydrophobic\nsurfaces can be easily controlled; on the other hand, the spatial\nadjustability of the light provides a flexibility in creating fluid\nmotions at different positions. This could in future work allow one\nto achieve precise control of the thermocapillary flow on superhydrophobic\nsurfaces via a simple illumination approach. Furthermore, a wide-field\nillumination, such as solar irradiation, could also be expected to\ntrigger flows with this optothermal setup. The design of the\nexperimental setup plays a critical role, both\non the macroscopic scale, i.e., with respect to the fluid reservoir\nof channel dimensions, as well as on the microscopic scale, i.e.,\nwith respect to the geometry and slip length of the superhydrophobic\nsurface. We discussed the influencing factors in terms of scaling\nrelations. This provides guidelines for designing thermocapillary-driven\nsystems that best exploit Marangoni forces at superhydrophobic surfaces.\nFurthermore, buoyancy effects might also be exploited by a suitable\ndesign of the setup. With the presented optical triggering of buoyancy-driven\nflows, new propulsion methods for fluids could be developed.",
"introduction": "Introduction Controlled liquid transport,\nespecially guiding liquid flows on\nthe microscale, is crucial for numerous applications, such as microfluidics\nfor chemical or biochemical analysis or inkjet printing techniques.\nHowever, traditional pressure-driven liquid transport becomes difficult\nat small scales. For instance, driving liquid flow through capillaries\nbecomes less and less energy efficient the smaller the capillaries\nare. This is because, according to the law of Hagen-Poisseuille, the\nflow at a given pressure difference scales with the fourth power of\nthe radius. With decreasing system size, the surface to volume\nratio of the\nfluid increases significantly and correspondingly controls the fluid\nflow. Hence, interfaces provide an excellent platform for manipulating\nflows at small scales. When a surface stress is generated at a fluid–fluid\ninterface, it spontaneously induces a deformation or dynamic motion.\nFor instance, surface-driven flows are induced by a surface tension\ngradient at the fluid–fluid interface, which is also known\nas the Marangoni effect. 1 It is generated\nwhen the fluid interface is subjected to a thermal or concentration\ngradient. The Marangoni effect has been reported to induce directional/convective\nflows in the liquid phase, 2 − 5 to guide droplets or floating objects at fluid interfaces, 6 − 9 and to determine various spreading dynamics phenomena, e.g., fingering\ninstabilities, 10 , 11 “Marangoni bursting”. 12 In addition to these observations, the Marangoni\neffect has been reported as a promising approach for the actuation\nof droplets or liquid films in microfluidic systems, 10 , 13 − 20 e.g., for directional transport of droplets on a nonuniformly heated\nsolid surface 16 , 17 , 21 or when a droplet is immersed in another immiscible liquid which\nis imposed with a temperature gradient. 22 − 27 However, apart from systems such as droplets or films, where\nfree\nfluid–fluid interfaces are immanently present, the Marangoni\neffect has been rarely reported to trigger continuous flows in confined\nliquid systems, e.g., pure liquid flows in enclosed microfluidic devices\nsuch as channels, due to the lack of free liquid interfaces. The liquid\nbeing in contact with the solid wall, or the liquid–solid interface,\nis normally considered as a no-slip boundary (i.e., the velocity is\nzero). To overcome this problem, the use of a structured, air-entrapping\nsuperhydrophobic surface has been proposed to introduce a liquid–fluid\ninterface and thereby a thermocapillary-driven Marangoni flow close\nto a solid interface. 38 Superhydrophobic surfaces have attracted\nintensive attention due to their liquid repellency as well as drag\nreduction properties. 28 − 35 The liquid phase, which is in contact with these surfaces, cannot\npenetrate the features of the surface roughness, which leads to an\nair phase remaining in the valleys of the surface roughness (or the\nso-called Cassie state). 36 , 37 This provides a discontinuous\nliquid–air interface close to the solid surface. Baier et al.\nhave theoretically shown that a temperature gradient imposed along\na superhydrophobic surface can induce directional thermocapillary-driven\nflows in the liquid phase along the direction the temperature decreases. 38 Later, Amador et al. demonstrated this principle\nexperimentally, obtaining a thermocapillary flow with a speed of 10–50\nμm/s in a submillimeter-wide channel with superhydrophobic sidewalls\nand an imposed temperature gradient of 2 K/cm. 39 These two works have shown the possibility of triggering\nliquid motion by the thermocapillary effect. Beyond the basic\nprinciple, a smart and precise control of thermocapillary\nflows over superhydrophobic surfaces has not yet been realized. An\noptothermal approach shows the possibility to achieve this goal without\nany complicated heating and cooling setups. While heating and cooling\nelements are typically fixed in place, an activation by light could\nbe much more flexible, e.g., by moving a light beam along a superhydrophobic\nsurface, as well as by varying the light intensity and illuminated\narea and, hence, the temperature gradient. In the experimental work\nof Amador et al., 39 it was shown that a\nflow could be created by coating half of the channel with black ink.\nHowever, no precise control of the temperature gradients have been\nrealized. Furthermore, the physical details of this kind of flow have\nnot yet been investigated. At small dimensions, the thermocapillary\nflow is typically considered\nthe main source of fluid propulsion. Other thermally induced physical\neffects, especially temperature-induced change in the liquid density\n(buoyancy force), are usually neglected. However, this may not correct\nfor thermocapillary flows over superhydrophobic surfaces. Because\nof the small slip length of the superhydrophobic surface (usually\na few micrometer), 30 , 40 the corresponding thermocapillary\nflow is much weaker than on a free liquid interface. Similar to the\ncase of buoyant-thermocapillary instabilities in a liquid phase with\na free upper interface 41 − 45 or in binary droplets, 46 − 49 we may need to consider both thermal effects. However,\nto the best of our knowledge, no theoretical models and experiments\non thermocapillary flows at superhydrophobic surfaces including buoyancy\nhave so far been reported. In this study, we want to clarify this\nissue to better understand the flow mechanism over superhydrophobic\nsurfaces. Therefore, we study an optical approach to manipulate\nliquid motion\nby exploiting the thermocapillary effect along the air–water\ninterfaces of a rough superhydrophobic surface. We constructed a superhydrophobic\nsurface with a photothermal material. Nonuniform heating of the surface\nunder illumination with light allows the generation of a nonuniform\ntemperature distribution at the surface and in the liquid phase. This\ndistribution leads to the generation of surface tension gradients\nat the liquid–air interface near the superhydrophobic surface\ndriving a thermocapillary flow ( Figure 1 ). Furthermore, density gradients in the liquid phase\nmay occur, which can additionally trigger buoyancy-driven flows. Our\nexperimental approach of having the heated surface at the top allows\nus to distinguish between these two effects. Combining the experimental\nobservations with theoretical simulations enables us to study the\nunderlying physics. This work provides a basic understanding of thermally\ntriggered flows on superhydrophobic surfaces. Figure 1 Concept of optothermally\ninduced thermocapillary flow in a liquid\nin contact with a superhydrophobic surface in the Cassie state. (a)\nSchematic of the thermocapillary driven flow induced by nonuniform\nlight irradiation on the surface covered with soot (at the top). Due\nto the photothermal effect, the substrate is heated and conducts the\nheat to the liquid. The nonuniform temperature distribution at the\nliquid interface causes a gradient in the surface tension σ\nat the liquid–air interface ( or ⟨ ∂T ⟩ r in the r direction), which\nleads to a surface flow ( U ). (b) Microstructures\nof the soot surface imaged by SEM and schematic of the discontinuous\nliquid–air interface at the superhydrophobic surface.",
"discussion": "Results and Discussion Temperature\nDistribution and Flow Pattern The laser\nbeam focused on the soot surface (spot diameter ∼350 μm)\ncreates an axisymmetrically heated spot due to its Gaussian shape\nin its energy distribution. When the surface is exposed to air, this\npeak is sharp, corresponding to the focused beam ( Figure S4 ). The presence of water leads to broadening and\nflattening of the heated spot ( Figure 2 a). The energy distribution in the laser beam seems\nthen not to play a role. Figure 2 (a) Illustration of the temperature profile\nnear the superhydrophobic\nsurface from the water side, measured by the IR camera. (b) Measured\ntemperature differences on the soot surface: in the r -direction, relative to the temperature far away from the heated\nspot (equals room temperature). In the z -direction,\ndifferences between top and bottom of the cell at a given distance\nfrom the center of the heated spot. (c) Schematic and visualization\nof the fluid pattern in the liquid layer. Images of fluorescently\nlabeled particles (white dots) superimposed with velocity vectors\nobtained from PIV analysis (green) in the three regions indicated\nby colored rectangles. (Scale bars: 50 μm). The heat produced at the soot surface under the laser irradiation\nis conducted through the liquid, causing only a small temperature\ndifference in the z -direction, which points perpendicular\nto the surface. The temperature distribution in the liquid layer was\naxisymmetric to the center of the laser beam in the r -direction and is shown in Figure 2 b. Near the soot surface, a temperature gradient in\nthe r -direction of Δ T /Δ r ∼ 0.25 K/mm was obtained. In the z -direction, the difference between the top and bottom surfaces was\nΔ T z ∼ 0.1\nK. Heat conduction through the water phase is hence an important effect.\nIt both distributes the heat laterally along the substrate and vertically\nto it. The first one influences the temperature gradient driving the\nMarangoni flow. The latter locally heats up the water up to a significant\ndepth. By adding fluorescent particles as tracers, the thermally\ninduced\nfluid motion in the liquid was recorded by confocal microscopy. Three\nregions marked by color-coded rectangles in Figure 2 c were observed: near the top interface with\na horizontal distance from the hot spot center (red, r = 500 μm, z ∼ 0), near the top interface\nin the heated spot region (yellow, r ∼ 0, z ∼ 0) and near the bottom substrate (blue, r ∼ 0, z ∼ 1000 μm).\nNear the superhydrophobic surface, directional flows are generated\nalong the superhydrophobic surface due to the directional temperature\ngradient ( Figure 2 c).\nNear the top interface (yellow, r ∼ 0, z ∼ 0), the tracers move axisymetrically outward\nin the direction of decreasing temperature. Near the bottom glass\nsurface ( r ∼ 0, z ∼\n1000 μm), the tracers move inward to feed the outward flow at\nthe upper interface. Without the laser switched on, we observed only\nrandom (Brownian) motion in our cell. Simulation Results With the use of the experimental\ntemperature distribution as input ( Figure 2 b for the soot-based superhydrophobic surface, Figure S6 for the other surfaces), finite element\nsimulations were performed for the three cases listed above ( Figure 3 ). The axisymmetric\ntemperature distribution is shown in Figure 3 a. In all three cases, axisymmetric toroidal\neddies are generated in the liquid phase, which however differ in\ntheir velocity distribution. For pure Marangoni flow without buoyancy\n[case (1), Figure 3 b], the flow is concentrated close to the superhydrophobic surface.\nThe highest velocity is at the boundary, which drags the rest of the\nliquid along. Because the system is closed, there is a backflow through\nthe lower part of the water reservoir. For pure buoyancy [case (2), Figure 3 c], the convective\nrolls are more equally distributed between the upper and lower part\nof the reservoir. The highest velocity in the upper part is at about\n1/4 of the height of the reservoir. The effect of the slip boundary\ncondition at the superhydrophobic surface is only small because the\nslip length of the soot surface is small. Case (3) ( Figure 3 d) is a combination of the\ntwo previous ones with both the Marangoni stress and the buoyancy\ncausing an outward flow in the upper part of the reservoir. The velocity\ndistribution is nearly a superposition of the two other cases. Since\nthe velocities in the two individual cases are of a similar order\nof magnitude, the maximum velocity is obtained at a position in between\nthe superhydrophobic boundary and the position for pure buoyancy.\nIf the strength of either Marangoni stress or buoyancy is modified,\nthe position of the maximum velocity also shifts accordingly. In the\nfollowing, we will use these 3 types of flow to identify the processes\noccurring in the experiment. Figure 3 Finite element simulation of the fluid motion\nunder local heating\nof the superhydrophobic soot surface at the top. (a) Temperature distribution\nin the simulated system, (b–d) Flow vectors and velocity magnitude\nfor (b) case (1): Marangoni stress at the superhydrophobic surface\nwithout buoyancy forces in the liquid; (c) case (2): no Marangoni\nstress, but buoyancy forces in the liquid; and (d) case (3): Marangoni\nstress at superhydrophobic surface and buoyancy forces in the liquid.\n(Related parameters: b = 0.22 μm; ∂σ / ∂T = −0.155 mN/m K; β = 2.570\n× 10 –4 K; and T 0 = 293.15 K.) Velocity Distribution and\nMechanisms To verify the\nmechanism of triggering Marangoni flows by photothermal stimuli, flow\nvelocity measurements on surfaces with different wettability and morphology\nwere performed: superhydrophobic soot ( b l –soot ∼ 0.22 μm),\nsuperhydrophilic soot ( b hydrophilic =\n0), and superhydrophobic pillar ( b l –pillar ∼ 3.2 μm) surfaces.\nHere, superhydrophilic soot surfaces having the same surface properties\nand morphology as the superhydrophobic soot surfaces except for the\nwettability were tested first to compare the velocity distribution\nclose to the surface and second to examine the thermally induced buoyancy\neffect when no Marangoni stresses are present. Superhydrophobic pillar\nsurfaces were examined to study the role of surface morphology. Their\nslip length is supposedly 1 order of magnitude larger than the superhydrophobic\nsoot surface. The corresponding experimentally determined velocity\nprofiles in the liquid layer, are shown in Figure 4 . They are measured perpendicular to the\ninterface at r = 500, 1500, and 2500 μm from\nthe center of the heated spot. Figure 4 Photothermally induced velocity profiles\nin the liquid being in\ncontact with solid substrates with different wettability and slip\nlength. Symbols: measurements, Lines: calculations. (a) Cassie state\non superhydrophobic surfaces. Experimental: Superhydrophobic soot\nsurface, calculations: case 3 [Marangoni-stresses (“σ”)\nand buoyancy (“ g ”)]. Inset: simulated\nvelocity magnitude. (b) Cassie state: close-up near the superhydrophobic\nsoot surface for r = 500 μm. Experimental:\nsuperhydrophobic soot surface. Calculations: cases 2 (slip, g ) and 3 (σ, g ); (c) Wenzel state\non superhydrophilic surface. Experimental: superhydrophilic soot surface.\nCalculation: no-slip boundary at the upper interface and buoyancy\nconsidered in the fluid (no-slip, g ). Inset: simulated\nvelocity magnitude; (d) comparison of the experimentally determined\nvelocity profiles near the upper interface of the liquid layer ( z < 50 μm; r = 500 μm). The experimentally determined velocity distribution\non the superhydrophobic\nsoot surface ( Figure 4 a) shows a mean velocity value u ∼ 4 μm/s\nclose to the surface and increases with z until a\nmaximum value u max ∼ 8 μm/s\nappears around z ∼ 150 μm. Hence, the\nvelocity profile does not correspond to the classic theoretical scenario\nwhere buoyancy is neglected and the maximum velocity appears directly\nat the surface (calculation case 1, Figure 4 b). Rather, there is a basic agreement with\ncase 3, considering both Marangoni stresses and buoyancy. The position\nof the maximum velocity depends on the choice of the effective slip\nlength of the surface with the maximum being closer to the surface,\nthe larger the slip length is. With the slip length estimated from\nthe soot geometry of 0.22 μm, the calculated velocity profile\nshows a similar trend as the experimentally determined one. However,\nwe observed an approximately factor of 2 larger velocity at the interface\nand correspondingly a velocity maximum closer to the interface. It\nhas to be considered that first the boundary condition employed for\nthe calculations is likely overestimating the boundary velocity due\nto Marangoni stresses, as discussed in the previous chapter. Second,\nthere might be surface-active contaminations present at the air–water\ninterface of the superhydrophobic surface. Contamination of the interface\nhas been shown to effectively reduce slippage at the interface and\nwould therefore also reduce the Marangoni velocity. 53 , 62 − 64 Third, an estimation of the effective slip length\nof the randomly structured soot surface is per se not exact. All these\neffects may reduce the real value of the velocity at the interface.\nHence with these considerations, the relation between the calculated\nvelocity profile for a slip length of 0.22 μm and the experimental\nobservation is reasonable. By comparing the shapes of the velocity\nprofiles from the simulations with those from the experiments, we\ncan determine an effective slip length of 0.13 μm in the simulations.\nThis leads to a good agreement between calculations and experiments.\nOverall, this shows that indeed buoyancy plays an important role in\nour system and cannot a priori be neglected in such small-scale systems.\nA similar observation has been recently made by Li et al. 48 for droplets. Furthermore, we demonstrate\nthat there is indeed a Marangoni flow\npresent at the superhydrophobic surface and that there is not only\nbuoyancy-driven motion. Figure 4 b shows a close-up of the velocity profile near the superhydrophobic\nsurface ( r = 500 μm). There is a finite velocity\nat the water–air interface. Calculations considering buoyancy\nand a slip boundary condition with b = 0.22 μm\nbut no Marangoni stresses (case 2) predict a much lower velocity at\nthe interface. Hence, there must be Marangoni flow in the experiments.\nThe same argument holds for the pillar surface ( Figure S8 ). In contrast, in control experiments with\nhydrophilic soot surfaces,\npure buoyancy was observed ( Figure 4 c). Here, Marangoni stresses are absent due to the\nmissing air–water interface. The velocity maximum is shifted\naway from the wall as compared to the superhydrophobic case, where\nthe superposition of buoyancy and Marangoni effects moves the maximum\ncloser to the surface. In the experiments, we still detected\na low velocity of ≈1\nμm/s close to the surface. This velocity stems from the diffusion\nof the tracer particles, which have a random orientation. On the basis\nof the Stokes–Einstein relation, 65 , 66 a diffusion\nvelocity of the order of micrometer per second is to be expected (and\nin agreement with the observations). Overall, the agreement between\nthe experimentally measured velocities and the calculations for pure\nbuoyancy is very good. The velocity maximum is located at about z = 250 μm, has a similar magnitude, and the velocity\ndirectly at the hydrophilic surface tends to zero. The dependence\nof the Marangoni velocity is shown in Figure 4 d. From the no-slip scenario\nin the hydrophilic case, the Marangoni velocity increases with increasing\nslip length. The increase observed in experiments between the superhydrophobic\nsoot and the superhydrophobic pillar surfaces is however not as strong\nas predicted by the calculations. A slip length of b l –pillar ∼ 3.2\nμm leads to a predicted Marangoni velocity of 110 μm/s.\nThis is a significantly larger overprediction as for the superhydrophobic\nsoot. The same reasons for overprediction also apply here: the boundary\ncondition itself as well as possible contamination of the air–water\ninterface. Further reasons may lie in the different geometric designs\nof the interfaces, which might be prone to effects of surface contamination\nto different degrees. After switching on the illumination, the\nflow first starts slowly\nand then stabilizes to a constant velocity after a time of about 60\ns. This is exemplarily shown in Figure 4 b for times t = 60s and t = 120 s after the start of the illumination. The flow development\ncorresponds to the temperature response of the surface, which needs\nabout 60 s to heat up to its final temperature ( Figure S5 ). Therefore, all the experimental data were recorded\nafter at least 60 s illumination when the flow became stable. Scaling\nRelations To analyze the relation between thermocapillary\nand buoyancy flows, a natural approach is to compare the corresponding\ncharacteristic velocities. The Marangoni velocity for a superhydrophobic\nsurface u M scales as 12 . The magnitude\nof the buoyancy-driven velocity u b is\ndetermined by a balance of buoyancy forces\nand viscous friction. The buoyancy force per unit volume is F b = βρ 0 Δ Tg , where β is the coefficient of thermal\nexpansion, ρ 0 the initial density, and g the acceleration due to gravity. Δ T is the\nhorizontal temperature difference at the upper liquid surface. Viscous\nforces scale as F v ∼ μU / L with the characteristic velocity U and characteristic length scale L . In\nthe present case, the latter corresponds to the height of the fluidic\nchamber H . Overall, balancing both contributions\nyields a characteristic buoyancy-driven velocity u b ∼ βρ 0 g Δ TH 2 /μ.\nHence, 4 With the typical values\nof our experiments\nand Δ T = 0.1 K, values of u M / u b in the range of 0.05–0.7\nare obtained, depending on the considered case (slip length and temperature\ngradient), indicating buoyancy being of somewhat larger influence\nthan thermocapillary flow. Yet, the individual dependencies in eq 4 are delicate and, as will\nbe discussed, need further investigation. Equation 4 is similar to the inverse of the dynamic\nBond number Bo = Ra / Ma , which has been used to characterize the ratio of Marangoni to buoyancy\nflows at free water–air interfaces. 43 , 67 , 68 Here, Ra is the Rayleigh\nnumber and Ma the Marangoni number. For free interfaces,\nthe dynamic Bond number does not include a slip length, but the characteristic\nlength L x in the direction\nparallel to the interface at the same place. In the classical setups,\nthis is both the fluid container size and the length over which the\ntemperature gradient is applied. Theoretically, increasing the\nslip length should also help in increasing u M / u b . Performing a series of numerical\nsimulations with varying slip lengths and evaluating the typical velocities\nfor u M and u b yields a velocity ratio u M / u b that consistently with eq 4 , increases linearly with the slip length b ( Figure S9 , at fixed thicknesses). Nevertheless,\nwe did not find this linear dependency on the slip length in our experiments,\neven when considering that the temperature gradient was slightly smaller\nfor the pillared surface. Further investigations are required to elucidate\nthe role of surface geometry or contaminations as described above. Furthermore, it has been observed for Marangoni flow at free interfaces\nthat the ratio between thermocapillary flow and buoyancy experimentally\nrather follows an L x / H 3 -relation than a 1/ H 2 -relation as in the reverse Bond number. 68 Varying the height H in the numerical simulation\nsimilarly indicates a b / H 2 -relation. These observations are in line with recent findings for\ndroplets indicating that the classical Bond number is not enough to\ncharacterize the interplay of thermocapillary and buoyancy. 48 , 69 With respect to the question of how systems exploiting thermocapillary\nflow at superhydrophobic surfaces should be designed in order to take\ngood advantage of thermocapillary flow, a computational variation\nof the system height illustrates the strong increase of buoyancy with\nthe height H ( Figure S10a ). For the present superhydrophobic soot surface ( b ∼ 0.1 μm), a desired dominance of the Marangoni flow\nover buoyancy would require system heights in the low μm regime.\nIf larger slip lengths could be achieved, the critical thickness of\nthe system could be raised to a few millimeters ( Figure S10b ). Further conclusions can be drawn from\nthe Marangoni number Ma , which characterizes the\nstrength of Marangoni flow.\nThe Marangoni number is a measure of the heat transport by convection\ndue to surface tension gradients to the bulk heat transport by conduction\nand as such defined as Ma = UL /α,\nwith α being the thermal diffusivity and U and L the characteristic velocity and length scale. In our case, for a superhydrophobic surface\nand L = H for the convection. The\nMarangoni\nnumber for a superhydrophobic surface hence reads 5 This is, in principle, the same expression\nas derived by Yariv, 55 , 56 only Yariv specifically considered\nlongitudinal grooves with a pitch p that occurs instead\nof the slip length b . Since b ∼ p , both expressions are equivalent. For the present system, Ma shp = 0.1–0.37. Hence, Marangoni flow\nis present, yet its strength is not extremely large. The Marangoni\nnumber also illustrates the influence of the heat conduction through\nthe water, which was already observed experimentally. It could be\nfurther enhanced by increasing the slip length of the surface, increasing\nthe temperature gradient or by employing omniphobic surfaces with\ndifferent fluids of lower α."
} | 7,265 |
35635769 | PMC9327519 | pmc | 7,801 | {
"abstract": "Abstract Plant–soil feedbacks (PSFs) are considered a key mechanism generating frequency‐dependent dynamics in plant communities. Negative feedbacks, in particular, are often invoked to explain coexistence and the maintenance of diversity in species‐rich communities. However, the primary modelling framework used to study PSFs considers only two plant species, and we lack clear theoretical expectations for how these complex interactions play out in communities with natural levels of diversity. Here, we extend this canonical model of PSFs to include an arbitrary number of plant species and analyse the dynamics. Surprisingly, we find that coexistence of more than two species is virtually impossible, suggesting that alternative theoretical frameworks are needed to describe feedbacks observed in diverse natural communities. Drawing on our analysis, we discuss future directions for PSF models and implications for experimental study of PSF‐mediated coexistence in the field.",
"introduction": "INTRODUCTION It has become well understood that reciprocal interactions between plants and the soil biota, collectively termed plant–soil feedbacks (PSFs), play an important role in structuring the composition and dynamics of plant communities. PSFs operate alongside other factors, including abiotic drivers (Bennett & Klironomos, 2019 ) and above‐ground trophic interactions (Van der Putten et al., 2009 ), but are thought to be a key mechanism generating negative frequency‐dependent feedbacks that promote coexistence and maintain plant diversity (Bever et al., 2015 ; Kulmatiski et al., 2008 ; Van der Putten et al., 2013 ). The existence of PSFs has long been known (Bever, 1994 ; Van der Putten et al., 1993 ), but our understanding of their importance—particularly in relation to patterns of coexistence—has developed rapidly in recent years (Crawford et al., 2019 ; Klironomos, 2002 ; Mangan et al., 2010 ; Petermann et al., 2008 ). Broad interest in PSFs was ignited by the development of simple mathematical models that illustrated the potential of PSFs to mediate plant coexistence (Bever, 2003 ; Bever et al., 1997 ; Ke & Miki, 2015 ). These models have played a crucial guiding role for a wide range of empirical studies, as well (Kulmatiski et al., 2008 , 2011 ; Pernilla Brinkman et al., 2010 ). The first, and still most widely known and used, model for PSFs was introduced by Bever and colleagues in the 1990s (Bever, 1992 , 1999 ; Bever et al., 1997 ). In this framework, often referred to simply as the Bever model, each plant species is assumed to promote the growth of a specific soil component (i.e. associated bacteria, fungi, invertebrates, considered collectively) in the vicinity of individual plants. In turn, the fitness of each plant species is determined by the relative frequency of different soil components. Starting from minimal assumptions, Bever et al. ( 1997 ) derived a set of differential equations to capture these dynamics. PSFs can be either positive (fitness of a plant species is higher in soil conditioned by conspecifics, compared to heterospecific soil) or negative (a plant species experiences lower relative fitness in its own soil). Bever et al. introduced a single quantity to summarise whether community‐wide PSFs are positive or negative, and showed that this value characterises the dynamical behaviour of the model. In the original Bever model of two plant species, positive PSFs lead to priority effects, and consequently the exclusion of one species, while negative PSFs result in neutral oscillations. It is thus widely suggested that negative PSFs help sustain coexistence in real‐world plant communities (Kulmatiski et al., 2008 ; Van der Putten et al., 2013 ), perhaps with spatial asynchrony playing a role in stabilising the cyclic dynamics (Bever, 2003 ; Revilla et al., 2013 ). Subsequent studies have generalised PSF models to include, for example more realistic functional forms (Eppinga et al., 2006 ; Umbanhowar & McCann, 2005 ), more explicit representations of the soil community (Bever et al., 2010 ), spatial structure (Eppinga et al., 2006 ; Molofsky et al., 2002 ; Suding et al., 2013 ) or additional processes such as direct competitive interactions between plants (Bever, 2003 ). However, the original Bever model remains an important touchstone for the theory of PSFs (Abbott et al., 2021 ; Ke & Miki, 2015 ; Ke & Wan, 2020 ) and informs empirical research through the interaction coefficient, I s , derived by Bever et al., which is commonly measured and used to draw conclusions about coexistence in experimental studies. Despite the ubiquity of this model, and the fruitful interplay of theory and experiment in the PSF literature, extensions to communities with more than two or three species have appeared only rarely and recently (but see Eppinga et al., 2018 ; Kulmatiski et al., 2011 ; Mack et al., 2019 ). While PSF models motivate hypotheses and conclusions about species‐rich natural communities, there is much still unknown about the behaviour of these models with natural levels of diversity (Van der Putten et al., 2013 ). Here, we extend the Bever model to include any number of plant species, and show that the model is equivalent to a special form of the replicator equation studied in evolutionary game theory (Hofbauer & Sigmund, 1998 ). In particular, this model corresponds to the class of bimatrix games, where there are two players (here, plants and soil components) which interact with asymmetric strategies and payoffs. The replicator dynamics of bimatrix games are well‐studied, allowing us to characterise many properties of the Bever model with n plant species. Surprisingly, using this equivalence, we show that coexistence of more than two species in this model is never robust.",
"discussion": "DISCUSSION The Bever model has played a central role in motivating PSF research, and continues to guide both theory and experiment in this fast‐growing field (Abbott et al., 2021 ; Bever et al., 2015 ; Kandlikar et al., 2019 ; Ke & Wan, 2020 ). Here, we extend the Bever model to include any number of plant species, and highlight its equivalence to bimatrix game dynamics. Taking advantage of the well‐developed theory for these dynamics, we are able to characterise the behaviour of this generalised Bever model in detail. Our central finding is that there can be no robust coexistence of plant species in this model. Regardless of the number of species, n , the model never exhibits equilibrium coexistence or other attractors. Coexistence can be attained through neutral oscillations, but these dynamics lack any restoring force, meaning diversity would quickly be eroded by stochasticity or exogenous forcing. In this respect, the generalised model behaves similarly to the classic two‐species system. However, unlike the two‐species model, oscillations with n > 2 species can only occur under very restricted parameter combinations. These parameterisations are vanishingly unlikely to arise by chance and highly sensitive to small deviations. Thus, coexistence of more than two species is neither dynamically nor structurally stable. This result may seem surprising, because a significant body of experimental evidence indicates that PSFs can and do play an important role in mediating the coexistence of more than two species in natural communities (Bever et al., 2015 ; Kulmatiski et al., 2008 ; Mangan et al., 2010 ; Petermann et al., 2008 ). Apparently, the picture suggested by the two‐species Bever model generalises in nature, but not in the model framework itself. We note that this framework was introduced as an intentional simplification to illustrate the potential role of PSFs in mediating coexistence, not to accurately model the biological details of PSFs. Indeed, the model has been wildly successful in spurring research into PSFs. Alongside extensive empirical study of these processes, other modelling approaches have emerged, accounting for more biological realism (e.g. Bever et al., 2010 ; Eppstein & Molofsky, 2007 ; Umbanhowar & McCann, 2005 ), or with the demonstrated capacity to produce multispecies coexistence (e.g. Bonanomi et al., 2005 ; Eppinga et al., 2018 ; Miller & Allesina, 2021 ). Some of these are minor modifications of the Bever model framework; others build on distinct foundations (Ke & Miki, 2015 ; Ke & Wan, 2020 ). Our results suggest that these various avenues are worth pursuing further. Our findings also help clarify important aspects of coexistence across modelling approaches. For example, Eppinga et al. ( 2018 ) and Mack et al. ( 2019 ) recently introduced a multispecies PSF model which can exhibit stable coexistence. Their model is inspired by the Bever model framework, but departs from it in two ways: by introducing more realistic soil dynamics, including a carrying capacity for soil communities, and by applying a separation of timescales, under the assumption that soil dynamics are very rapid compared to plant dynamics. Our analysis indicates that the second feature is unable to account for stabilisation. Regardless of the relative rates of plant and soil dynamics, the coexistence equilibrium of the generalised Bever model is never attractive (see also Figures S2 and S3). This is a fundamental feature of the model structure, not a result of particular parameter choices. When oscillations do exist in the Bever model, they are always neutral, meaning that their amplitude is fixed by the initial conditions of the system, and cannot diminish through the dynamics. These observations make clear that the crucial factor driving coexistence in the model of Eppinga, Mack and colleagues is self‐regulation within soil communities, not rapid soil dynamics. Indeed, our analysis suggests that the internal dynamics of plant or soil communities must interact with PSFs to maintain diversity in natural systems. We have shown that PSF models that are structurally similar to the Bever model—in which plant dynamics depend only on soil frequencies, and soil dynamics depend only on plant frequencies—are incapable of exhibiting stable coexistence of any number of species. Multispecies coexistence becomes possible when plants (Bever, 2003 ; Revilla et al., 2013 ), soils (Eppinga et al., 2018 ; Mack et al., 2019 ), or both experience an independent source of self‐regulation, which might arise from resource competition, physical limits to density or some other mechanism. PSFs are likely to matter most for the maintenance of diversity when they interact with these internal plant or soil dynamics in non‐trivial ways. In the case of combined plant competition‐feedback models (Bever, 2003 ), for example, we have already seen that no robust coexistence is possible when plants are competitively equivalent and experience ‘mean‐field’ interactions. On the other hand, when plant competition is dominated by strong intraspecific interactions, all plant species would coexist even in the absence of PSFs. Thus, PSFs can only contribute to the maintenance of diversity in such models by modifying competitive outcomes (Bever, 2003 ; Kandlikar et al., 2019 ; Revilla et al., 2013 )—that is, by interacting with the structure of the plant‐plant competitive network. These conclusions have practical implications for the study of PSFs in real‐world communities. The predictions of the Bever model are commonly used to guide the design and analysis of PSF experiments, especially in drawing conclusions about coexistence. Our analysis cautions that direct application of this model in multispecies communities might lead to incorrect inference. For example, attempts to parameterise the Bever model for three species using empirical data have produced predictions of non‐coexistence in plant communities that coexist experimentally (Kulmatiski et al., 2011 ). In many other studies, the interaction coefficient, I s , is calculated for species pairs and used to assess whole‐community coexistence (Bauer et al., 2017 ; Crawford et al., 2019 ; Fitzsimons & Miller, 2010 ; Kuebbing et al., 2015 ; Kulmatiski et al., 2008 ; Pendergast et al., 2013 ; Pizano et al., 2019 ; Smith & Reynolds, 2015 ; Suding et al., 2013 ) . However, we have seen that whole‐community coexistence is virtually impossible within the generalised model, and there is no guarantee that the pairwise coexistence conditions for this model will agree with n ‐species coexistence conditions in other frameworks (but see Mangan et al. ( 2010 ); Eppinga et al. ( 2018 )). For example, I s < 0 for all species pairs is neither necessary nor sufficient to produce coexistence in a metapopulation‐based model for PSFs (Miller & Allesina, 2021 ). Theory suggests that when PSFs do play a role in maintaining robust coexistence, interactions between plants and soil will necessarily be only part of the picture. On this point, we echo calls that have emerged in the empirical literature for more closely integrated study of PSFs and other processes, such as plant competition (Casper & Castelli, 2007 ; Lekberg et al., 2018 ) and more detailed soil biology (Bever et al., 2010 ; Hodge & Fitter, 2013 ). Our results strongly suggest that pairwise PSF measurements are insufficient to characterise plant coexistence and require contextualisation alongside these other ecosystem processes. Fundamentally, our analysis demonstrates that PSFs as envisioned in the classic Bever model cannot produce robust n ‐species coexistence in isolation. Our results also indicate basic structural features that are necessary for PSF models to support multispecies coexistence. Significantly, we find not only the absence of stabilisation in the Bever model, but generic instability . This suggests that, in diverse communities, other processes must exert a sufficiently strong influence on the community dynamics to overcome the baseline instability. We illustrate this idea in the Supplemental Methods, where we examine the effect of adding negative frequency dependence in the classic Bever model. Any amount of frequency‐ dependence stabilises neutral oscillations, but when these effects are weak, they cannot turn unstable equilibria into stable ones. The result is that the augmented model can only support multispecies coexistence when the rescaled zero‐sum game condition is met, and, as we have shown, this condition is never robust to small parameter variations. In this example, we consider negative frequency dependence, rather than density dependence (as in Bever, 2003 )), because it is difficult to compare the strengths of processes that mix units of frequency and density. This difficulty hints at a central limitation of classic PSF models, which are derived by projecting dynamics for plant and soil abundances onto the space of frequencies . (Eppinga et al., 2018 ; Ke & Wan, 2020 ; Kulmatiski et al., 2011 ; Revilla et al., 2013 ). The projected dynamics can mask unbiological outcomes in the original model (e.g. relative abundances oscillate around equilibrium while absolute abundances shrink to zero or explode to infinity). Indeed, the absolute abundance model (Equation 4 ) used to derive our n ‐species frequency dynamics (Equations 5 and 6 ) does not generally possess any fixed point, which is a basic requirement for species coexistence (Hutson, 1990 ; Hutson & Schmitt, 1992 ). The same is true for the model introduced by Eppinga et al. ( 2018 ) and Mack et al. ( 2019 ), even though this model exhibits stable dynamics for plant and soil frequencies. It is usually seen as desirable to study PSFs in the space of species frequencies, both because this facilitates connections to data, and because frequencies are considered appropriate units for analysing processes that stabilise coexistence (Adler et al. ( 2007 ); Eppinga et al. ( 2018 ), but see Kandlikar et al. ( 2019 ); Ke and Wan ( 2020 )). But models that introduce frequencies through a natural constraint, such as competition for finite space, will likely produce more realistic and straightforwardly interpretable dynamics. From a broader theoretical perspective, the qualitative change in model behaviour that we observe as the number of species increases from two to three or more is a striking phenomenon, but not an unprecedented one. Ecologists have repeatedly found that intuitions from two‐species models can generalise (or fail to generalise) to more diverse communities in surprising ways (Barabás et al., 2016 ; Smale, 1976 ; Strobeck, 1973 ) . Our analysis provides another illustration of the fact that ‘more is different’ (Anderson, 1972 ) in ecology, and highlights the importance of developing theory for species‐rich communities."
} | 4,181 |
31766417 | PMC6952951 | pmc | 7,804 | {
"abstract": "The microvalve for accurate flow control under low fluidic pressure is vital in cost-effective and miniaturized microfluidic devices. This paper proposes a novel microfluidic passive valve comprising of a liquid chamber, an elastic membrane, and an ellipsoidal control chamber, which actualizes a high flow rate control under an ultra-low threshold pressure. A prototype of the microvalve was fabricated by 3D printing and UV laser-cutting technologies and was tested under static and time-dependent pressure conditions. The prototype microvalve showed a nearly constant flow rate of 4.03 mL/min, with a variation of ~4.22% under the inlet liquid pressures varied from 6 kPa to 12 kPa. In addition, the microvalve could stabilize the flow rate of liquid under the time-varying sinusoidal pressures or the square wave pressures. To validate the functionality of the microvalve, the prototype microvalve was applied in a gas-driven flow system which employed an air blower or human mouth blowing as the low-cost gas source. The microvalve was demonstrated to successfully regulate the steady flow delivery in the system under the low driving pressures produced by the above gas sources. We believe that this new microfluidic passive valve will be suitable for controlling fluid flow in portable microfluidic devices or systems of wider applications.",
"conclusion": "4. Conclusions In summary, we proposed a novel microfluidic passive valve for a stable flow control in microfluidic environment. The valve was made up of a control chamber of ellipsoid surface and an elastic membrane containing two micro-holes, which would deflect to change the flow resistance of the control chamber under the pressurized liquid flowing through the micro-holes, thereby maintaining a constant flow rate totally independent of the varied inlet pressures. To investigate the flow performances of the valve, we fabricated a prototype microvalve using 3D printing and UV laser cutting technologies, and the flow rates of the prototype were measured accordingly under static and dynamic inlet pressure conditions. The experimental results showed a high throughput and nearly constant flow rate achieved in the prototype under an ultra-low threshold pressure. To further examine the flow stability of the valve, we employed it in a portable gas-driven flow system which was operated by an air blower and human mouth blowing, respectively. The system was found to output a stable liquid under the low driving pressures, which validated the applicability of the valve in portable microfluidics.",
"introduction": "1. Introduction The conceptual design of microfluidics or lab-on-a-chip associated with point-of-care test (POCT) applications, such as biological cell separation [ 1 , 2 ], nucleic acid diagnostic [ 3 , 4 ], bacteria detection, etc. [ 5 , 6 ] usually require high-throughput processing, autonomous actuation, and portability. As a key component of microfluidic POCT systems, a micro-scaled pumping unit for driving liquid should meet the technical functionality of the system. Nowadays, microvalves are being employed widely in pumping units for sophisticated flow control, and valve structure and actuation mechanism are key factors influencing their utility in limited resource applications. Generally, the geometry of the microvalve should be designed for a potential large-scale integration. The overall force for controlling liquid in the microvalve should be minimized for convenient actuation. In addition, for the sake of cost-effectiveness and portability, the microvalve should be wholly actuated on-chip with energy efficiency. Many microvalves with the functionality of on-chip actuation have been reported in previous literatures, especially the elastomeric membrane types which have been recognized as a cornerstone technology for enabling various applications in microfluidic systems [ 7 , 8 ]. These membrane valves are flexible in actuation, easy to manufacture by standard fabrication processes, and feasible for miniaturized integration with microfluidic devices for biological and biochemical applications. A typical membrane valve in structure would have three layers, including a control channel, a fluidic channel, and a thin elastic membrane, which can reversibly deflect to open or close the fluidic channel for flow control [ 9 ]. Generally, poly(dimethylsiloxane) (PDMS) is the most commonly used membrane material due to its good optical transparency and high elasticity for large deformations [ 10 ]. Other materials, such as thermal plastic polymer [ 11 , 12 ], shape memory alloy [ 13 , 14 ], glass [ 15 ], and more are also used in certain situations. As for the deflection of the membrane, various mechanical [ 16 , 17 ], electrostatic [ 18 , 19 ], pneumatic [ 20 , 21 ], magnetic [ 22 , 23 ], piezoelectric [ 24 ], or thermal [ 25 ] mechanisms have been proposed. Among the actuation mechanisms, pneumatic actuation is apparently the most commonly used technology. An active valve with pneumatic actuation usually employs off-chip apparatus such as an air-compressor, a pressure regulator, etc., and does the required flow regulation by the regulation of the air pressure. The valve can be combined to form complex microfluidic devices, such as peristaltic pumps [ 11 , 19 ] and mixers [ 26 ]. However, when a low-cost and portable microfluidic system such as in POCT applications requires totally on-chip activation, the active technology may be considered excessive. An alternative technology for liquid control is the passive actuation. In comparison to the active valve, a passive valve doesn’t require any external power, and it regulates the flow rate of liquid through autonomous adjustment of flow resistance. In addition, as it is capable of realizing self-adaptive resistance variations which completely compensate the fluidic pressure variations, a constant flow rate can, as a result, be achieved by the valve with a pre-determined threshold pressure [ 27 , 28 , 29 ]. Cousseau et al. developed a silicon membrane valve for drug delivery [ 30 ]. The valve comprised of a glass cover, a silicon membrane, and a bottom layer with a spiral channel. The valve was able to provide a constant liquid flow rate of 0.022 mL/min within an operating pressure range of 20 kPa to 50 kPa. Kartalov et al. proposed a PDMS push-up valve which was composed of a detour control channel, a membrane, and a fluidic channel [ 31 ]. The constant flow rate maintained by the valve was 0.033 mL/min, and the threshold pressure to achieve the flow rate was 103 kPa. Yang et al. designed a compliant flap and a rigid stopper which formed a restricted fluid path in a planar check valve, and the valve output a high flow rate of 1.2 mL/min with a threshold pressure of 100 kPa [ 32 ]. For fabricating passive valve with low threshold pressure, Doh et al. presented a parallel membrane valve which included two control channels, two vertical membranes, and a fluidic channel [ 33 ]. As the two membranes could restrict liquid in the fluidic channel with the autonomous deflection, the valve was capable of achieving flow regulation at a minimum pressure of 15 kPa. Our team had previously reported a parallel membrane valve with two horizontal membranes sandwiching a fluidic channel between them, and then all sandwiched between two control channels [ 34 , 35 ]. Due to the five-layer stacked architecture, the valve achieved a high flow rate of 2.79 mL/min with a low threshold pressure of 10 kPa. Thus, the valve could be used for high throughput sample processing in biological cell separation [ 36 , 37 ]. However, these existing passive valves can be difficult to integrate into useful microfluidic due to their high threshold pressures for flow regulation. For example, most mechanical and non-mechanical micropumps generate fluidic pressures lower than 10 kPa [ 38 , 39 ], which leaves the current valves with a major challenge when it comes to integrating them for such a low-pressure flow regulation. In this work, we developed a new microfluidic passive valve which can produce a constant flow rate of liquid under an ultra-low threshold pressure. To investigate the flow characteristics of the valve, we measured the flow rates of the valve under static and the time-dependent conditions of a varied set of inlet pressures. We also demonstrated some smart function abilities of the valve in combination with a gas-driven flow system for obtaining high throughput and stable flow delivery under low driving pressures.",
"discussion": "3. Results and Discussion 3.1. Flow Characterization Under Static Pressure As previously mentioned, the working principle of the valve is such that flow rate across the valve is auto-regulated by the deformations of the membrane. To investigate the flow characteristics of the valve, we measured the flow rates of the prototype valve at different inlet liquid pressures, and the corresponding effects of varied pressures on the flow rates were analyzed accordingly. Flow tests were at first performed under statically varied pressures as shown in Figure 6 a. In the experiment, the inlet liquid pressure of the valve was increased sequentially from 1 kPa to 17 kPa by 1 kPa step. The flow rate at each test pressure was measured and recorded in a minute. To quantitatively study the effect of the inlet pressure on the flow rate, we divided the flow rate curve into three phases based on the flow performances induced by the inlet pressures. In the first phase, the flow rate was in direct proportionality to the inlet pressure, and it increased steadily as the pressure increased from 1 kPa to 6 kPa. When the pressure was higher than 6 kPa, the flow rate began to show a significant nonlinear relationship with the inlet pressure. In the 6 kPa to 12 kPa pressure phase, we found that the flow rate was getting regulated, as it maintained a nearly constant value regardless of pressure change. To analyze the functionality of the valve, we calculated the mean flow rate and the flow variation across the above pressure ranges. The flow variation defined as the relative pulsation to the mean flow rate was calculated as a ratio of the bilateral tolerance of minimum to maximum flow to the overall mean flow rate. We obtained a mean flow rate of 4.03 ± 0.17 mL/min (flow variation ~4.22%) in the above pressure range. As the inlet pressure was afterwards increased from 12 kPa to 17 kPa, the flow rate started to slowly increase, and the flow rate could not be maintained constant anymore in the test phase. According to the experimental results, although the valve only regulated the constant flow rate for the inlet pressures between 6 kPa to 12 kPa, it still showed a significant flow autoregulation capability when the pressure was higher than the minimum threshold pressure of 6 kPa. In order to validate the flow regulation capability of the valve, we fabricated and tested a straight through device (membraneless), and it outputted a continuously increasing flow rate in the whole test process. In comparison to the membrane valve, the flow rate of the device at the inlet pressure of 17 kPa increased 123.5% (Δ Qs in Figure 6 a) against the flow rate at the pressure of 6 kPa, while the flow rate of the valve only increased by 23.7% (Δ Qv in Figure 6 a) in the above pressure range. Thus, we could conclude that the valve was totally self-adaptive, and its flow resistance varied in correspondence with the inlet pressure to regulate the flow rate by the autonomous deflection of the membrane, thereby realizing a constant flow rate in a passive manner. 3.2. Flow Characterization Under Dynamic Pressure It is necessary to investigate whether the valve is indeed capable of regulating flow rate through self-adaptive resistance variation when the inlet liquid pressure varies with the actuation time. To this objective, we examined the prototype valve under time-dependent varied pressures. The flow regulating performance of the prototype valve under different inlet pressures are shown in Figure 6 b. The inlet pressures were varied in sinusoidal wave mode and square wave mode, respectively. Both the sinusoidal wave pressures and the square wave pressures fluctuated from 6 kPa to 12 kPa within a time period of 10 s, and the flow rates were recorded at every 0.01 s. The pressure range was determined for the constant flow phase under the static experiments. The mean flow rates under both pressure variations were 4 mL/min and 3.98 mL/min, respectively, which were quite close to those of the static pressure test. The standard deviation of the flow rate under sinusoidal wave pressures was 0.31 mL/min, which was equivalent to a flow variation of 7.75%. In the square wave pressure experiment, we found out that flow rate abruptly increased and decreased to give peak flow and valley points when the pressures were on high and low pulsations, respectively. We thought the behavior of the flow pulsation was directly influential to the response of the self-adaptive resistance variation of the valve. As described in the figure, the response time for the flow rate stabilization under the pulsating pressures was less than 0.3 s, and the standard deviation of the stable flow rate was 0.36 mL/min with the flow variation of 9.04%. Although the flow variations under the time-dependent pressures were much higher than in the static test, the valve still showed a great capability of regulating the flow. As the inlet pressure was increased by 100% (6 kPa to 12 kPa), the flow rates only increased by 16.8% under the sinusoidal condition and 19.9% under square wave. In comparison with the previously reported passive valves, the valve proposed in this work achieved a higher flow rate at a much lower threshold pressure, as shown in Table 1 . This could be attributed to the special ellipsoid surface design incorporated in the control chamber. In most of the traditional membrane valves, the fluidic channel is designed with a rectangular cross-section which characteristically leaves a dead zone at the corners upon the deflection of the membrane. As the flow resistance of the valve should be increased to compensate the pressure increment, obtaining a required flow resistance quickly enough to compensate the pressure variations with such corners could be very difficult. To solve this problem, we designed an ellipsoid surface in the control chamber and two micro-sized holes in the membrane to allow liquid passage into the control chamber. The ellipsoidal surface design of the control chamber leaves quite an accommodating enclosure for the membrane to deflect into, thus facilitating a flow resistance development that could increase significantly, even under the low pressures. Furthermore, as the design in such of an inline configuration, liquid flows directly into the control chamber from the inlet, leaving no room for pressure loss in the valve, unlike in many traditional valves made up of exterior control channels. Therefore, the threshold pressure of our valve is less. In our study, we found the influence of the valve structure on the flow performance to be significant, precisely considering parameters such as diameter and depth of the ellipsoid surface of the control chamber, diameter of the liquid chamber, diameter and length of the outlet, thickness and Young’s modulus of the membrane, and diameter and location of the hole in the membrane. Further improvement on the flow performance of the valve could be achieved by optimizing the above parameters. 3.3. Low-Cost and Portable Gas-Driven Flow System For many small-scale biomedical applications, precise delivery of liquid at a low-cost manner is essential for the commercialization of these miniaturized devices and systems. As the commercial syringe pumps or peristaltic pumps are bulky in size while expensive in cost, it is necessary to develop a cost-effective small sized pump. Here, we came up with such a low-cost and portable gas-driven flow system which applied a microfluidic passive valve for accurate liquid control. The conceptualized system was composed of a pressure source, a sealed tank filled with deionized water and a microfluidic valve, shown in Figure 7 a. The pressure source employed was that of a human mouth whose blown air pressure was considered enough to drive the liquid. To examine the flow stabilization capability of the system, a flow sensor and an electronic balance were used to measure the time-dependent flow rates as being regulated by the valve. Figure 7 b shows the flow rates of the gas-driven flow system driven by the air blower. In the experiment, an operator held the air blower and slightly squeezed it to produce the pressurized air in the tank. The air pressure supplied by the air blower to the system was roughly in the range of 11.1 kPa to 14.2 kPa. The system was found to produce a stable flow rate after 0.5 s of actuation, and the mean flow rate of the system was 4.17 mL/min with a standard deviation of 0.23 mL/min, (or 5.52% of mean flow rate variation) across the time-varying pressures. For comparison, we also measured the flow rates of the system without the microfluidic valve, and the flow rate was 17.26 ± 3.1 mL/min with the flow variation of 17.96%. Figure 7 c shows the flow performance of the gas-driven flow system driven by a human mouth. The air pressure produced by a strong mouth blowing was about 6.1 kPa to 8.2 kPa, which was a little higher than the minimum threshold pressure of the valve for flow stabilization. It was found that the flow rate of the system was regulated to be stable after about 2 s of actuation, and the mean flow rate of the stable liquid was 3.81 ± 0.24 mL/min (flow variation of 6.3%). The system without the valve driven by mouth blowing showed some dramatic flow fluctuations, with a mean flow rate of 10.44 ± 2.05 mL/min (19.64% variation). For both the air blower system and the mouth blowing system without the valve, the flow variations were far higher than in the valve incorporated systems, just as the figures clearly reveal. Therefore, we could conclude that the gas-driven flow system integrated with a passive valve was capable of achieving a stable flow rate. Hence, as the valve was able to maintain a constant flow under such a low threshold pressure, we can envision its usefulness in many microfluidic applications. For example, it can be applied to provide accurate sample fluids for high throughput cell sorting and concentration [ 40 ]. As the processing rate of cell separation and concentration can be several mL/min, a valve with the function of high throughput flow control will be very effective for the miniaturization of the microfluidic cell sorting system [ 36 , 37 ]. In addition, the valve structure can be further optimized to reduce the flow rate for more available microfluidic applications, such as sample mixing [ 41 ], droplet manipulation [ 42 ], drug delivery [ 43 ], etc."
} | 4,729 |
30702898 | null | s2 | 7,805 | {
"abstract": "Ocean metaproteomics is an emerging field enabling discoveries about marine microbial communities and their impact on global biogeochemical processes. Recent ocean metaproteomic studies have provided insight into microbial nutrient transport, colimitation of carbon fixation, the metabolism of microbial biofilms, and dynamics of carbon flux in marine ecosystems. Future methodological developments could provide new capabilities such as characterizing long-term ecosystem changes, biogeochemical reaction rates, and in situ stoichiometries. Yet challenges remain for ocean metaproteomics due to the great biological diversity that produces highly complex mass spectra, as well as the difficulty in obtaining and working with environmental samples. This review summarizes the progress and challenges facing ocean metaproteomic scientists and proposes best practices for data sharing of ocean metaproteomic data sets, including the data types and metadata needed to enable intercomparisons of protein distributions and annotations that could foster global ocean metaproteomic capabilities."
} | 272 |
27118382 | PMC4847182 | pmc | 7,806 | {
"abstract": "Background It is known that during plant community assembly, the early colonizing species can affect the establishment, growth or reproductive success of later arriving species, often resulting in unpredictable assembly outcomes. These so called ‘priority effects’ have recently been hypothesized to work through niche-based processes, with early colonizing species either inhibiting the colonization of other species of the same niche through niche preemption, or affecting the colonization success of species of different niches through niche modification. With most work on priority effects performed in controlled, short-term mesocosm experiments, we have little insight in how niche preemption and niche modification processes interact to shape the community composition of natural vegetations. In this study, we used a functional trait approach to identify potential niche-based priority effects in restored semi-natural grasslands. More specifically, we imposed two treatments that strongly altered the community’s functional trait composition; removal of all graminoid species and removal of all legume species, and we compared progressing assembly with unaltered control plots. Results Our results showed that niche preemption effects can be, to a limited extent, relieved by species removal. This relief was observed for competitive grasses and herbs, but not for smaller grassland species. Although competition effects acting within functional groups (niche preemption) occurred for graminoids, there were no such effects for legumes. The removal of legumes mainly affected functionally unrelated competitive species, likely through niche modification effects of nitrogen fixation. On the other hand, and contrary to our expectations, species removal was after 4 years almost completely compensated by recolonization of the same species set, suggesting that priority effects persist after species removal, possibly through soil legacy effects. Conclusions Our results show that both niche modification and niche preemption priority effects can act together in shaping community composition in a natural grassland system. Although small changes in species composition occurred, the removal of specific functional groups was almost completely compensated by recolonization of the same species. This suggests that once certain species get established, it might prove difficult to neutralize their effect on assembly outcome, since their imposed priority effects might act long after their removal. Electronic supplementary material The online version of this article (doi:10.1186/s12898-016-0077-9) contains supplementary material, which is available to authorized users.",
"conclusion": "Conclusions In this study, we explored how priority effects within and among functional groups affect community assembly during natural plant community assembly. More specifically, our results show that, in a low nutrient, (semi-)natural grassland system, inhibitory priority effects acting through niche preemption can be slightly relieved by species removal. However, this relief depended on the competitive ability of the removed species, with relief only observed for more competitive grasses and herbs, but not for smaller grassland specialists. Although competition effects acting within functional groups (niche preemption) were observed for graminoids, they do not seem to apply to legumes. Indeed, the removal of legumes mainly affected functionally unrelated generalist species and megaphanerophytes, likely through the facilitative niche modification effects of nitrogen fixation after legume removal [ 14 ]. On the other hand, species removal was, contrary to our expectations, almost completely compensated by recolonization of the same species set, suggesting that the net community composition effects of species (group) removal is rather limited in this natural system. This additionally suggests that soil legacies are, at least up to a certain extent, important drivers of assembly patterns during natural grassland assembly. We can expect that, in the context of ecological restoration, if unwanted species get established, it might prove difficult to neutralize their effect on the community assembly outcome, since their imposed priority effects might act long after their removal through imposed soil legacies [ 16 ].",
"discussion": "Discussion General assembly patterns Changes in species richness and functional group composition through time observed in the C treatment can be interpreted as the natural assembly patterns in the studied grasslands. This natural assembly process is characterized by the replacement of generalist by specialist species and an increase in total vegetation cover through time. These results largely confirm the assembly patterns previously observed through a chronosequence approach in similar dry semi-natural grasslands [ 39 ]. At the functional trait level, the assembly patterns also partly confirmed the previous results of Helsen et al. [ 39 ]. However, opposite to the chronosequence study, the number of forest/shrub species showed a small increase, and annuals a strong decrease in species numbers through time in this study. Large herbs and grasses were found to remain relatively constant through time in this study, while a decrease in richness was found for these groups in the chronosequence study [ 39 ]. These differences might be caused by the smaller time scale in this study. Species removal effects on non-treatment species The initial treatments resulted in changes in the functional trait set of the grassland communities in the first years following manipulation, with, for the G treatment, a reduction of total cover, graminoid species richness and associated cover of the emergent groups that mainly consist of graminoid species. In the L treatment, manipulation initially resulted in a reduction of N fixating species, but had no significant effect on total cover. Interestingly, species richness was not affected by treatment, with similar levels of both specialist and generalist species across treatments. This is in accordance with other priority effect experiments, where species richness was found to converge, independent of initial differences in species richness and treatment [ 10 , 32 ]. Both the G and L treatments nevertheless affected species composition. More specifically, the removal of graminoids resulted in a small, but nonetheless significant increase in the number of species in EG 5 (large herbs and grasses). This is in accordance with our (inhibitory) niche preemption hypothesis, with species with similar niches as the removed graminoids benefitting from the treatment [ 26 , 28 ]. In other words, the removal of large competitive grasses resulted in a small increase of large competitive herbs, likely through colonization. The absence of a similar pattern for EGs 4 and 6, which also contain many graminoids, is likely caused by the fast recovery of graminoids species in these grasslands, well before other species can colonize due to reduced within-niche competition (see the ‘species removal effects on treatment species’ discussion section further). Alternatively, it could be argued that the inhibitory competitive priority effects are less pronounced within these EGs, which are characteristic of high stress-low competition communities. It has indeed been suggested that the strength of direct (inhibitory niche preemption) priority effects are dependent on soil nutrient levels, implying that priority effects in experimental studies (using optimal nutrient concentrations) are likely much stronger than those occurring in natural (nutrient poor) communities [ 10 , 36 ]. Since EG 5 mainly consists of relatively competitive species, this could explain the stronger effect of species removal for this specific group. Indeed, all else being equal, direct inhibitory niche preemption priority effects are expected to be more pronounced for competitive species that produce much biomass [ 24 , 36 ]. Since soil legacies are strongly species-specific, we also cannot exclude the possibility of differential facilitative or inhibitory soil legacy effects of graminoid species, indirectly promoting the establishment of species of EG 5 [ 21 , 22 ]. In accordance with our hypothesis, legume removal did not affect the species richness of EG 4 (or 7), which contain all legume species, but resulted in changes in unrelated functional groups (decrease of the number of large herbs and grasses, and an increase of megaphanerophytes). This suggests the occurrence of niche modification effects of legumes after their removal. As discussed earlier, the low overall competitive abilities of the legumes present in these grasslands (Additional file 2 ) likely explain why competitive exclusion (niche preemption) is limited within this functional group. Previous research has also shown that legumes do often not exert persistent inhibitory priority effects through size-asymmetric competition, and often facilitate higher biomass production of functionally different co-occurring plant species through nitrogen enrichment of the soil (facilitation) [ 19 , 49 ]. This facilitative niche modification effect can be especially effective for plants growing in nutrient poor grasslands, as is the case in this study. The removal of legumes in the L treatment likely resulted in open patches with increased nitrogen availability. Megaphanoreophyte seedlings seem to be better at establishing at these former legume sites, suggesting facilitative soil legacies through nitrogen enrichment [ 21 ]. In this scenario, the observed decrease in species of EG 5 might be partly caused by the decreased competitive success of large (herbs and grasses) against megaphanerophytes. Alternatively, the absence of legumes might have resulted in a lower overall availability of nitrogen in the community, an effect that will most strongly inhibit the growth of species that are not adapted to nutrient poor conditions, such as those of EG 5. Indeed, this EG mainly contains generalist species adapted to fast growth and relatively nutrient rich soils (Table 1 ; [ 39 ]). The positive effect on megaphanerophytes on the other hand might then suggest that legumes have a negative effect on tree and shrub seedlings through inhibitory niche modification effects, independent of their effect on nitrogen availability. Species removal effects on treatment species In this experiment, both graminoid and legume removal was after 4 years almost completely compensated by the recolonization of graminoids and legumes, respectively, strongly suggesting that niche processes shape community assembly and priority effects in certain semi-natural grassland systems, as previously argued by Helsen et al. [ 31 ]. Although this was largely expected for graminoids, we did not expect similar patterns to occur for legumes. More surprisingly, the species replacement rates among the treatments show that this recolonization is effectuated by largely the same set of species as those that were removed. This suggests that niche preemption through size-asymmetric competition is likely only partly driving these patterns, since we would have expected some levels of species replacement (within functional groups) in this case. Likely, the observed patterns are also partly driven by localized soil legacies that promote the colonization of the same species (facilitative), or prevent the colonization of other species (inhibitory legacy effects acting within a functional group). Although some studies demonstrated that within species plant-soil feedbacks can be inhibitory [ 21 , 35 ], other studies have indeed shown that many species exhibit weaker inhibitory, or even facilitative plant-soil feedbacks upon conspecifics compared to plant-soil feedbacks upon other species [ 23 , 34 ]. The observed patterns can, however, also be at least partly explained by other confounding factors. Since (dead) belowground biomass of the treatment species was not actively removed, possible priority effects of these species might have been much stronger than would have been the case after complete removal of the species. Indeed, inhibitory size-asymmetric competitive priority effects are not solely driven by aboveground biomass, but can also remain strong when aboveground biomass is periodically removed through mowing [ 19 ]. Furthermore, since all treatments were performed in relatively small plots within a larger grassland, the removed species are also present in the direct vicinity of the treatment plot, enhancing the chances of recolonization of the plot by the same species set, thus deflating replacement rates. This effect might have been especially strong for legumes, since only a relatively small number of species was present in these grasslands. Most of the graminoids were furthermore strongly clonal (Fig. 1 e, f), also allowing quick clonal recolonization of the plot by ramets present at the vicinity of the plot border. In conclusion, we believe that soil legacies likely resulted in reduced levels of species replacement, but that this affect was likely not as strong as suggest by the species replacement results. Treatment effects through time Although changes in the EG compositions across the different treatments persisted after 4 years (no significant interaction between time and treatment), we did also observe a fast recovery of the number and composition of both legumes and graminoids during the same time span. Contrary to our predictions, this suggests that the effect of specific functional group removal during grassland assembly does not result in alternative assembly pathways, through newly enforced priority effects of the secondary colonized species. These results more likely suggest that soil legacies result in, at least partial, maintenance of initial priority effects after species removal. This, in turn, allows the fast recolonization of the removed species, with only limited changes in overall species composition. These results are partly in agreement with the study of Plückers et al. [ 37 ], where initial differences in species richness and functional composition (forbs, grasses and legumes) through differential seeding, became very small after 4 years, with communities seemingly converging toward similar species richness and functional composition."
} | 3,585 |
36067284 | PMC9478672 | pmc | 7,807 | {
"abstract": "Significance To swim, bacteria must regulate a battery of motility genes in proper relation to other genes and the environments they encounter. To reveal how cells resolve this challenge, we studied the regulation of motility genes in the model organism Escherichia coli across growth conditions. By connecting gene expression with swimming behavior and growth, we illustrate how cells coordinate the regulation of swimming machinery with cell size such that the number of flagella per cell is maintained across conditions. The findings revise previous interpretations that saw swimming motility as a starvation response. Instead, cells are motile across growth conditions with size-dependent regulation, ensuring an efficient allocation of cellular resources to the synthesis of costly flagella machinery.",
"discussion": "Discussion In this study, we analyzed the regulation of motility genes by E. coli in different balanced growth conditions. We found that the fold change in gene expression per biomass compensates for the variation in cell size, resulting in the average number of flagella per cell remaining constant across growth conditions. This simple regulatory scheme ensures a fully motile population while keeping resource demands to synthesize and rotate flagella to a minimum. How do cells implement this regulation scheme? Future studies are needed to reveal further mechanistic insights, but our results point to the combined roles of transcriptional and posttranscriptional regulation in determining the abundance of the motility master regulator FlhDC. On the transcriptional level, cAMP-CRP–dependent activation on flhDC expression ( 46 , 47 ) may play an important role, as other cAMP-CRP–dependent genes are known to increase with decreasing growth rates under carbon limitation ( 27 , 48 ). In addition, the posttranscriptional regulation on flhDC expression might further be essential, as we found that modification of the 5′-UTR strongly affected flhDC expression. In this context, it is tempting to speculate about the physiological roles of small RNA species, which are being increasingly discovered and found to be involved in diverse regulatory tasks ( 49 , 50 ). Further, posttranslational regulation on flhD via the anti-FlhDC factor (YdiV) might be involved in adjusting the expression of fliA and other class II motility genes such that their expression scales with cell size ( 25 ). The findings reported here have implications for bacterial motility from an ecological perspective, particularly concerning its role in promoting fitness across different environments. Previous works have highlighted up-regulation as a fingerprint of anticipatory response, with motility triggered when nutrients run out ( 26 , 28 – 30 ). In contrast, we here propose that at least a part of the up-regulation of swimming in poorer growth conditions is not a starvation response per se, but an obligatory regulation to maintain sufficiently high flagella numbers and swimming as cell size changes. Future studies are needed to investigate how our findings merge with the ideas of anticipatory response, but the efficient regulation of motility genes to maintain swimming under growth-supporting conditions is in line with observations that bacterial cells quickly stop swimming ( 9 ), actively brake motor rotation ( 51 , 52 ), and even release their flagella upon entering starvation ( 53 , 54 ). Notably, the maintenance of cellular motility in growth-supporting conditions enables cell population to rapidly expand into unoccupied nutrient-rich territories, boosting overall population growth ( 9 ). The growth advantage of such a navigated range expansion relies on cells being motile across conditions, and a delayed onset of motility only in response to starvation would nullify the fitness advantage ( 9 ). Therefore, the efficient regulation of motility genes described here does not only minimize the resources required to build and fuel the motility machinery, but it also supports fast navigated range expansion, which further boosts fitness ( 9 , 21 , 33 ). The findings further provide a perspective on the relation between cell size and growth itself. Throughout the text, we have referred to the change in motility gene expression as an up-regulation in poor nutrient conditions. However, this change can also be viewed as a down-regulation in nutrient-replete conditions when cells grow fast. Given that the goal of the flagella regulatory system is to maintain the number of flagella per cell, we can view the decreased flagella expression at fast growth also as a consequence of increased cell size at fast growth. This view leads us to suggest a physiological rationale for E. coli ’s choice of cell size at different growth rates. It is generally preferrable for bacterial cells to keep a small biomass (i.e., cell size), as it promotes efficient diffusive transport, fast nutrient uptake, and strong dispersal ( 55 , 56 ). However, in favorable conditions allowing for rapid growth, the translational machinery per biomass is the most growth-limiting factor ( 57 , 58 ), and making cell size larger can be beneficial to alleviate this bottleneck: By increasing its size at fast growth, the cell effectively reduces the amount of flagella proteins that need to be synthesized, thus allowing more proteomic resources to be allocated toward ribosomes and other components of the translation machinery. Quantitatively, flagella proteins comprise ∼3.0% of the total protein mass in slow carbon-limited conditions and ∼0.7% in RDM ( 11 ). Thus, by increasing its cell size, E. coli manages to “save” 2.3% of the proteome that would have otherwise been tied up in flagella synthesis. To put this amount in perspective, the entire set of biosynthesis enzymes saved when cells are provided with all amino acids and nucleotides is only ∼11% of the proteome (comparing the proteome composition of cells grown in rich-defined medium supplemented with glucose to those grown in glucose-minimal medium). This saving accounts for a large share of the increase of growth rate from 1.0 1 / h in glucose-minimal medium to 1.8 1 / h in RDM ( 11 ), based on the well-established linear relation between ribosome content and growth rate, where every percent-of-proteome added to the protein synthesis machinery results in an ∼0.06 1 / h increase in growth rate ( 11 , 27 , 57 ). Thus, a 2.3% saving in proteome allocation to flagella synthesis would amount to a gain of ∼0.14 1 / h for growth in rich medium. In other words, had E. coli kept its size at the level observed in the poor-nutrient condition, then it would suffer a 0.14 1 / h reduction in growth rate (from the observed growth rate of 1.8 1 / h ) in rich medium, just due to motility expression alone. This proteome resource savings by a change of cell size should be similarly applicable to other cellular processes that demand protein expression on a per-cell basis, including cell division and cell pole maintenance. Therefore, increasing cell size at fast growth might be a simple and effective strategy to reduce competition for proteome resources at fast growth, for E. coli and possibly many other fast-growing bacterial species."
} | 1,810 |
26391740 | null | s2 | 7,808 | {
"abstract": "We have isolated a new extremely thermophilic fast-growing Geobacillus strain that can efficiently utilize xylose, glucose, mannose and galactose for cell growth. When grown aerobically at 72 °C, Geobacillus LC300 has a growth rate of 2.15 h(-1) on glucose and 1.52 h(-1) on xylose (doubling time less than 30 min). The corresponding specific glucose and xylose utilization rates are 5.55 g/g/h and 5.24 g/g/h, respectively. As such, Geobacillus LC300 grows 3-times faster than E. coli on glucose and xylose, and has a specific xylose utilization rate that is 3-times higher than the best metabolically engineered organism to date. To gain more insight into the metabolism of Geobacillus LC300 its genome was sequenced using PacBio's RS II single-molecule real-time (SMRT) sequencing platform and annotated using the RAST server. Based on the genome annotation and the measured biomass composition a core metabolic network model was constructed. To further demonstrate the biotechnological potential of this organism, Geobacillus LC300 was grown to high cell-densities in a fed-batch culture, where cells maintained a high xylose utilization rate under low dissolved oxygen concentrations. All of these characteristics make Geobacillus LC300 an attractive host for future metabolic engineering and biotechnology applications."
} | 331 |
34795326 | PMC8602335 | pmc | 7,810 | {
"abstract": "Drought severely restricts plant production and global warming is further increasing drought stress for crops. Much information reveals the ability of individual microbes affecting plant stress tolerance. However, the effects of emergent bacterial community properties on plant drought tolerance remain largely unexplored. Here, we inoculated Arabidopsis plants in vivo with a four-species bacterial consortium ( Stenotrophomonas rhizophila , Xanthomonas retroflexus , Microbacterium oxydans , and Paenibacillus amylolyticus , termed as SPMX), which is able to synergistically produce more biofilm biomass together than the sum of the four single-strain cultures, to investigate its effects on plant performance and rhizo-microbiota during drought. We found that SPMX remarkably improved Arabidopsis survival post 21-day drought whereas no drought-tolerant effect was observed when subjected to the individual strains, revealing emergent properties of the SPMX consortium as the underlying cause of the induced drought tolerance. The enhanced drought tolerance was associated with sustained chlorophyll content and endogenous abscisic acid (ABA) signaling. Furthermore, our data showed that the addition of SPMX helped to stabilize the diversity and structure of root-associated microbiomes, which potentially benefits plant health under drought. These SPMX-induced changes jointly confer an increased drought tolerance to plants. Our work may inform future efforts to engineer the emergent bacterial community properties to improve plant tolerance to drought.",
"introduction": "Introduction Due to the intricate natural environments, plants are faced with unfavorable conditions multiple times during their growth 1 . Drought is the most common environmental stress dramatically limiting plant growth and production in agriculture 2 . Climate change and global warming is accelerating the recurrence of serious drought events, causing severe ecological and food security issues 3 – 5 . Considering that such events are likely to further increase, there is an urgent need for the development of sustainable solutions to improve plant resistance against drought, such as the application of beneficial microbes 6 . Plants benefit from a wide variety of soil microbes 7 . Root-associated microbiomes play an important role in determining plant health and performance under various environmental conditions 8 , 9 , and the composition of root microbiome is affected by hosts and environmental factors 10 , 11 . Drought is one of the common environmental stresses, having significant effects on the soil microbiomes 12 , 13 . In addition to osmotic stress, drought often causes a strong impact on microbial composition due to increased soil heterogeneity, limited nutrient mobility and utility which aggravates plant stresses 14 . Some studies have found that certain specialized microbiomes might alleviate plant drought stress 15 – 17 . In turn, plant physiology and metabolism in response to drought stress can also alter the composition and structure of the microbiome with potential consequences for host adaptation and fitness 16 , 18 . Structural adaptations within bacterial communities in the root microbiome to abiotic and biotic stressors may provide plants with the potential to improve tolerance against drought, and further promote plant health 19 – 21 . Beneficial bacteria isolated from plant roots have been identified as plant growth-promoting rhizobacteria (PGPR) 22 , as they can directly and indirectly improve plant growth and performance under stress via promoting nitrogen fixation, increasing nutrient uptake, improving soil properties, inhibiting plant pathogens and enhance plant tolerance to drought 17 , 23 . For example, an early study found that Paenibacillus polymyxa increased drought tolerance in Arabidopsis thaliana by regulating the expression of gene ERD15 involved in the drought-stress response 24 . Although PGPRs have been widely studied in the past decade, most of these efforts so far have mainly focused on traits of single strain 25 – 27 , and are overlooking potential emergent properties of microbial communities on plant growth. Emergent properties of plant microbiota can be achieved when the bacterial community displays effects that are not observed from any of the community members when studied in isolation 28 . These emergent properties may result from the synergistic interactions among different species in the microbial community. As a result, it is possible that the functional capacity of the bacterial community or consortia is far beyond the sum of each individual due to beneficial interactions with each other 29 . Bacteria often live as biofilms and function as communities 30 , 31 . Multispecies biofilm is the naturally-occurring and dominant lifestyle of bacteria in nature and their interspecies interactions can lead to mutualistic relationships or competitive activities 32 – 34 . In our previous studies, we had observed strong synergistic effects of four soil-isolated bacterial strains on biofilm formation as these four strains significantly produced more biofilm when co-cultured together than the sum of four mono-species biofilms 35 . Later, a recent study further found that a three-species combination among four species, composed of Xanthomonas , Stenotrophomonas , and Microbacterium spp., which also showed increased biofilm production compared to their individual members, induced systemic resistance (ISR) in Arabidopsis thaliana against phytopathogens 20 . However, whether such synergistic effects of multispecies biofilms result in emergent properties on plant drought tolerance remains largely unexplored. It is known that the main component in biofilm is water (up to 97%), which has the potential to retain water for plants during drought 36 , 37 . Therefore, we selected this four-species consortium to investigate its potential impact on plant drought tolerance. We hypothesized that this four-species bacterial consortium would better protect plants from drought stress compared to single species, as these four strains together have the strongest synergy on biofilm formation and significantly produce more biofilm than the sum of their four single-species biofilms 35 . In this study, the model plant Arabidopsis thaliana , grown in vivo, was inoculated with four-species consortium (SPMX) composed of Stenotrophomonas rhizophila , Paenibacillus amylolyticus , Microbacterium oxydans , and Xanthomonas retroflexus to evaluate this consortium impact on plant drought tolerance. We tested more than 1200 Arabidopsis plants and showed that the SPMX inoculated together significantly improved plant survival under drought while no drought-tolerant effect was observed with single-strain inocula, indicating that the enhanced drought tolerance results from emergent properties of the four-species consortium rather than individual strains. Furthermore, we investigated SPMX-induced differences in plant physiology, drought-related gene expression and root-associated microbiomes, which might jointly help alleviate the negative effects of drought on plant performance. Understanding such emergent bacterial community properties may provide new opportunities to improve plant health and performance in the face of drought.",
"discussion": "Discussion Our data demonstrated that an addition of a four-species bacterial consortium (SPMX) could help improve plant performance and survival under drought. This induced drought tolerance was observed when plants were inoculated with four-strain SPMX together, while no drought tolerance was induced when each strain was inoculated individually, indicating that this enhanced drought tolerance resulted from emergent properties of the SPMX consortium. (Figs. 1 and 2 ). While the rhizobiome has previously been considered as an extended root phenotype 54 and several studies report the implication of root-associated microorganisms in drought tolerance 14 , 55 . This is, to our knowledge, the first demonstration of a minimal bacterial community emergent property leading to drought stress protection in a plant host. We therefore moved on to investigate the potential mechanisms underlying the SPMX-induced drought tolerance. Biofilm formed by SPMX may be partly responsible for this increased drought survival due to their known ability to produce high levels of hydrated polymers in the matrix to retain water 56 , 57 . Our confocal microscopy analysis revealed SPMX ability to form biofilm on root surface (Fig. 1g ). Recent studies also indicated the potential of biofilm in drought stress alleviation 15 , 58 . Besides, photosynthesis is the essential way for plants to obtain energy and its efficiency associates positively with chlorophyll contents 59 . In this study, the addition of SPMX increased chlorophyll contents of Arabidopsis under drought (Fig. 2 ), which would be helpful to maintain plant growth under drought due to potentially enhanced photosynthesis. ABA has been widely reported as a central regulator regulating the plant responses to drought stress via closing stomata to prevent water loss and inducing related genes to enhance drought tolerance 38 , 42 . Differences in expression levels of four ABA-related marker genes under drought suggested that the addition of SPMX affected the ABA signaling under drought. Upregulated ABA-biosynthetic gene NCED3 and ABA-responsive COR15 by SPMX addition (Fig. 3a, b ) reflected a possible enhanced ABA biosynthesis, which is known to increase plant drought tolerance 60 . ABA-responsive genes such as RAB18 , TSPO , and RD29B can be also induced by drought stress 61 – 63 (Fig. 3e ). Some studies have shown that RAB18 and TSPO were upregulated by environmental water-limited stress 43 – 45 . However, in our study, the expression of drought-responsive RAB18 and TSPO were significantly downregulated. The reduced expression of these two genes might reflect a decreased drought stress sensed by plants. This is probably due to the ability of the biofilm to retain water for plants, thereby reducing the water stress sensed by plants under drought. Combined, expression changes in these four maker genes suggested a possibility of increased drought tolerance and reduced drought stress when plants were inoculated with the SPMX consortium under drought. To study the root colonization of SPMX, we evaluated the relative abundance of each species in SPMX established in two rhizo-compartments via amplicon sequencing. Under watered conditions, Xr was the most abundant species in both rhizosphere and rhizoplane (Fig. 4 ). Intriguingly, we found that Mo became the most abundant rather than Xr during drought both in the rhizosphere and rhizoplane, while in our previous studies, Mo was always at the lowest ratio in SPMX-formed biofilm when no stress factor was active 64 . It suggested that Mo may play an important role under drought. It is worth-noting that the relative abundance of each SPMX strain was enhanced in the rhizoplane compared to that in the rhizosphere (Fig. 4 ), which may indicate an enhanced colonization ratio of SPMX in the rhizoplane compared to in the rhizosphere. It also surprised us that the low abundant SPMX could have such a significant impact on plant drought tolerance, which leads us to speculate whether the observed effects might also include an indirect effect derived from the addition of SPMX to the soil. Therefore, we further investigated possible rhizo-microbiome shifts caused by SPMX that may benefit plants against drought. Drought stress is considered as a negative abiotic factor that reduces microbial variety and abundance in the soil 12 , 14 , 65 . Our data further indicated that drought had a larger impact on microbiome structure in rhizosphere compared to that in rhizoplane due to more core phyla affected and changed in rhizosphere (Fig. 5c and Supplementary Fig. 6 ) (9 phyla in rhizosphere vs. 4 in rhizoplane). Similarly, under drought, 9 phyla in rhizosphere significantly responded to SPMX addition compared to those in rhizoplane (2 phyla) (Fig. 5c and Supplementary Fig. 7 ). In particular, our results showed that the SPMX stabilized the diversity of the root microbiome during drought (Fig. 5a ), which potentially benefited plant growth under drought 14 . Furthermore, SPMX significantly reshaped the root microbiomes during the drought (Fig. 5b, c and Supplementary Fig. 4 ). Interestingly, we found that these drought-impacted, SPMX-reshaped microbiomes were similar to the bacterial composition under watered conditions. This was reflected by the remarkable rise of the five drought-depleted phyla and drop of two dominant drought-enriched phyla when added SPMX under drought (Fig. 5c and Supplementary Fig. 7 ). Therefore, these RA-reversed phyla in response to SPMX addition under drought may further indicate a reduced water stress, as also reflected in the observed downregulation of the stress-responsive genes RAB18 and TSPO (Fig. 3c, e ), which likely benefits from water retained by the biofilm formed in the SPMX-inoculated condition. More work will be needed to confirm if this SPMX-formed biofilm observed here in vitro is also able to be formed in the soil. Alternatively, certain beneficial microbes are specifically recruited or enriched in presence of SPMX to help plants deal with drought stress. As we analyzed, Actinobacteria might be most influenced by the addition of SPMX under drought and may benefit plant tolerance to drought. Many of the known Actinobacteria strains were identified as PGPR to improve the plant’s multi-stressed tolerance and seedling vigor in water-restricted soil 66 , 67 . Furthermore, the enrichment for Actinobacteria during drought was not a random event but was most likely an intrinsic and natural microbial adaptation in response to drought 12 . The emergent properties of SPMX may strengthen such potentially mutually-beneficial relationships, although further evidence will be required to test this hypothesis."
} | 3,516 |
33645919 | PMC8601171 | pmc | 7,811 | {
"abstract": "Summary Directed evolution is a powerful method to optimize proteins and metabolic reactions towards user‐defined goals. It usually involves subjecting genes or pathways to iterative rounds of mutagenesis, selection and amplification. While powerful, systematic searches through large sequence‐spaces is a labour‐intensive task, and can be further limited by a priori knowledge about the optimal initial search space, and/or limits in terms of screening throughput. Here, we demonstrate an integrated directed evolution workflow for metabolic pathway enzymes that continuously generate enzyme variants using the recently developed orthogonal replication system, OrthoRep and screens for optimal performance in high‐throughput using a transcription factor‐based biosensor. We demonstrate the strengths of this workflow by evolving a rate‐limiting enzymatic reaction of the biosynthetic pathway for cis,cis ‐muconic acid (CCM), a precursor used for bioplastic and coatings, in Saccharomyces cerevisiae . After two weeks of simply iterating between passaging of cells to generate variant enzymes via OrthoRep and high‐throughput sorting of best‐performing variants using a transcription factor‐based biosensor for CCM, we ultimately identified variant enzymes improving CCM titers > 13‐fold compared with reference enzymes. Taken together, the combination of synthetic biology tools as adopted in this study is an efficient approach to debottleneck repetitive workflows associated with directed evolution of metabolic enzymes.",
"introduction": "Introduction Industrial biotechnology has offered commercialization of environmentally friendly transportation fuels, amino acids and value‐added chemicals by the use of fermentation feedstocks and microbial cell factories (Choi et al ., 2019 ). Yet, industrializing microbial cells for a broad range of applications within manufacturing, health and transportation industries often requires extensive engineering of both the microbial chassis and at the level of scaling up the fermentation processes (Van Dien, 2013 ; Nielsen and Keasling, 2016 ). Indeed, in the design of cell factories for fermentation‐based manufacturing of value‐added chemicals and therapeutics, biosynthetic pathways are often composed of enzymes from several different sources, and with enzyme activities and expression levels requiring careful balancing in order to achieve optimal pathway flux (Galanie et al ., 2015 ; Zhang et al ., 2020 ). While such multi‐dimensional optimization can be streamlined using design‐of‐experiment approaches and machine learning algorithms (Jeschek et al ., 2016 ; Xu et al ., 2017 ; Carbonell et al ., 2018 ), the regulatory and cellular complexity of living cells and the constraints in speed, scale, depth and costs of even rational trial‐and‐error engineering approaches challenge the development of microbial cell factories. As a complementary approach to bottom‐up rational engineering, evolution‐guided cell factory engineering has gained substantial traction over the last decade (Mundhada et al ., 2016 ; Sandberg et al ., 2019 ). Here, the key principles of evolutionary engineering includes targeted or genome‐wide genetic diversification coupled with screening of variant libraries (Packer and Liu, 2015 ). Numerous metabolic engineering studies have successfully applied directed evolution to improve product and feedstock tolerance and cell factory performance (Caspeta et al ., 2014 ; Park et al ., 2014 ; Mundhada et al ., 2016 ). While powerful, both the generation of large numbers of genetic variants and the development of proper selection regimes, as well as the cloning and transformation procedures associated with directed evolution cycles, are often time‐ and cost‐intensive. To overcome this, in vivo directed evolution uses endogenous or orthogonal cellular machineries to maintain high‐mutation rates without the need for iterative cycles of library cloning and transformation, apart from propagating the evolving population (Esvelt et al ., 2011 ; Ravikumar et al ., 2014 ; Crook et al ., 2016 ). One such system is OrthoRep enabling continuous generation of variant genes of interest expressed from a linear cytoplasmic chromosome that is propagated via an orthogonal error‐prone DNA polymerase (Ravikumar et al ., 2014 , 2018 ). With orthogonal in vivo evolution machineries at hand, any trait that can be coupled to growth (e.g. antibiotic resistance, tolerance to cultivation conditions and/or complementation of auxotrophies) enables facile identification of improved target genes without need for direct screening (Esvelt et al ., 2011 ; Ravikumar et al ., 2014 ; García‐García et al ., 2020 ; Rix et al ., 2020 ). However, for metabolic engineering, the expression of heterologous enzymes and proteins towards biobased production of value‐added chemicals seldom allows direct coupling of production to growth, or other high‐throughput screens, needed to capitalize on the massive diversity generated by in vivo evolution systems (Esvelt and Wang, 2013 ). Here, the recent development of biosensors based on allosterically regulated transcription factors (aTFs) can provide a complementary technology for coupling enzymatic activity or pathway flux with facile screening of large variant libraries in multiplex through fluorescence‐activated cell sorting (FACS) or growth (Raman et al ., 2014 ; Flachbart et al ., 2019 ). Briefly, such biosensors link binding of small‐molecule ligands to aTFs as input, with changes in expression of reporter genes or actuators as output (Mahr and Frunzke, 2016 ). Taken together, the coupling of continuous evolution systems with biosensing could allow metabolic engineers to cost‐effectively search for optimal pathway designs. Here, we combine the power of targeted in vivo mutagenesis using OrthoRep with high‐throughput biosensing for the rapid evolution of rate‐limiting metabolic reactions of the cis,cis ‐muconic acid (CCM) pathway (Weber et al ., 2012 ; Curran et al ., 2013 ; Suástegui and Shao, 2016 ). The 3‐step CCM pathway, consisting a dehydroshikimate dehydratase (AroZ), a multi‐subunit protocatechuic acid (PCA) decarboxylase and a catechol 1,2‐dioxygenase (Fig. 1A ) (Weber et al ., 2012 ; Curran et al ., 2013 ), has been extensively studied and optimized to support biobased production of plastics and coatings, following hydrogenation of CCM into adipic acid as a building block for nylon‐6,6 (Weber et al ., 2012 ; Curran et al ., 2013 ; Suastegui et al ., 2016 ; Leavitt et al ., 2017 ; Snoek et al ., 2018 ; Wang et al ., 2020 ). Importantly, for CCM pathway optimization, we have previously engineered the aTF BenM as a CCM biosensor in yeast (Skjoedt et al ., 2016 ), and this has further enabled optimization of yeast as a chassis for CCM production (Snoek et al ., 2018 ; Wang et al ., 2020 ), complemented by additional evolution‐ and machine learning‐guided optimization of the endogenous yeast aromatic amino acid pathway from which CCM is derived (Leavitt et al ., 2017 ; Zhang et al ., 2020 ). However, the build‐up and secretion of the CCM pathway intermediate PCA remain rate‐limiting for high CCM production (Suastegui et al ., 2016 ; Leavitt et al ., 2017 ; Snoek et al ., 2018 ; Wang et al ., 2020 ). This observation is propelled by the fact that PCA accumulation supposedly is not limited to suboptimal catalytic activity or expression of the downstream CCM pathway enzyme, PCA decarboxylase, as PCA accumulation also remains a persistent issue for microbial biosynthesis of other PCA‐derived chemicals and nutraceuticals without decarboxylase requirements (Strucko et al ., 2015 , 2017 ; D'Ambrosio et al ., 2020 ). Yet, no attempts for directed evolution of the heterologous enzymes towards increased PCA‐derived production has to our knowledge been performed. Here, we demonstrate rapid evolution of PCA decarboxylase subunits using a simple experimental design enabled by OrthoRep and five biosensor‐assisted selection cycles ultimately yielding > 13‐fold higher CCM production compared with the subunits encoded in the wild‐type PCA decarboxylase complex. Fig. 1 Schematic illustration of the in vivo directed evolution workflow. A. Schematic illustration of the 3‐step cis,cis ‐muconic acid pathway, comprising heterologous expression of PaAroZ , KpAroY subunits (B, D, and Ciso), as well as CaCatA and overexpression of Tkl1 (Weber et al ., 2012 ; Curran et al ., 2013 ). B. Schematic illustration of the parental strain (Sc‐105, see Table S5 ) used for in vivo directed evolution of the cis,cis ‐muconic acid pathway enzymes KpAroY. B and KpAroY. Ciso in yeast cells. The strain replicates and expresses the biosensor, all cis,cis ‐muconic acid pathway enzymes except KpAroY. B and KpAroY. Ciso , and the variant error‐prone TP‐DNAP (expressed from AR‐Ec633, see Table S4 ) from the nucleus. All components required for OrthoRep replication and transcription are encoded on p2, whereas, genes encoding KpAroY. B and KpAroY. Ciso are expressed from p1. C. Schematic illustration of the in vivo directed evolution workflow showing the passaging regimes of the parental strain undergoing (i) the five consecutive rounds of OrthoRep coupled with biosensor‐based selection or (ii) fifteen bulk passages to effect drift without biosensor‐based selection.",
"discussion": "Discussion In this study, we demonstrate the successful merger of two synthetic biology tools for the benefit of evolving superior enzymes without the use of labour‐intensive library designs or costly low/semi‐throughput analytical facilities. Importantly, this study show‐cases the evolution of metabolic pathway enzymes without any native growth advantage for the cells, a condition that most previous in vivo directed evolution requires (Esvelt et al ., 2011 ). Furthermore, the merger of OrthoRep and biosensors for directed evolution as demonstrated in this study complements the development of selections associated with growth under (strong) selection pressures (Ravikumar et al ., 2018 ; Zhong et al ., 2020 ). While this study was a successful demonstration of continuous hypermutation of target genes applied for metabolic enzyme evolution, more prospecting and evolution‐guided engineering of the CCM pathway is still warranted. For biobased CCM production, such efforts should not be limited to improving the suboptimal catalytic activity or expression of pathway enzymes downstream of PCA, i.e. PCA decarboxylase and catechol 1,2‐dioxygenase. One strategy to consider further is the need to limit the secretion or passive diffusion, of PCA across the cellular membrane as is often observed in yeast engineered to produce PCA‐derived chemicals and nutraceuticals (Hansen et al ., 2009 ; Weber et al ., 2012 ; Curran et al ., 2013 ; Suastegui et al ., 2016 ; Leavitt et al ., 2017 ). With the evolved PCA decarboxylase subunits identified from this study, a logical next step could be to move from in vivo directed evolution of predefined target genes towards genome‐wide adaptive laboratory evolution. For such purposes, transferring the biosensor read‐out from fluorescence to growth would be beneficial from a technical and scalability point of view (Zhong et al ., 2020 ). For such a purpose, the coupling of CCM to growth using synthetic control circuits founded on BenM, or the recently developed vanillin biosensor, could be useful for genome‐wide searches of nucleotide polymorphisms and chromosomal re‐arrangements limiting PCA efflux and/or further boosting PCA metabolic flux respectively (Ambri et al ., 2020 ; D'Ambrosio et al ., 2020 ). Another aspect to consider is to diversify the selection criteria beyond the stringent one used in this study (Fig. 1C ). Specifically, tuning the selection strength during continuous evolution regimes has previously been demonstrated to enable mutational drifting and adaptation of robust proteins (Bershtein et al ., 2008 ; Steinberg and Ostermeier, 2016 ; Zhong et al ., 2020 ). With the coupling of OrthoRep to FACS‐compatible screens as demonstrated in this study, such tuning should be possible to implement and further explored, in order not to outpace the rate of adaption using the conservative and stringent cut‐off for selection as applied in this study. Ultimately, this could expand both the robustness and catalytic activity of further evolved PCA decarboxylases, but also increase the relatively low number of mutations observed per evolved PCA decarboxylase variant. Furthermore, toggled selection regimes of neutral drifting interrupted by selection (Rix et al ., 2020 ; Zhong et al ., 2020 ), may also increase the hit‐rate of the continuous evolution, and limit the false‐discovery rate observed in this study (> 9/22, > 0.45) (Fig. 3 ). Extending from this, it is also worth considering the use of biosensor variants with operational ranges > 100 mg l −1 CCM (Fig. 3B and C ) or tuning of cultivation time (Skjoedt et al . 2016 ; Snoek et al . 2019 ), to increase the hit‐rate of PCA decarboxylase variants with even higher catalytic activity. In summary, we consider that our study serves as a first demonstration of rapid evolution of metabolic enzymes without any direct fitness advantage using continuous hypermutation, and furthermore moves forward the engineering of S. cerevisiae , and potentially other microbial chassis, for the industrial production of CCM as a precursor for further hydrogenation into adipic acid and nylon‐6,6 for the bioplastics industry."
} | 3,407 |
33057032 | PMC7560658 | pmc | 7,812 | {
"abstract": "The ongoing research on and development of increasingly intelligent artificial systems propels the need for bio inspired pressure sensitive spiking circuits. Here we present an adapting and spiking tactile sensor, based on a neuronal model and a piezoelectric field-effect transistor (PiezoFET). The piezoelectric sensor device consists of a metal-oxide semiconductor field-effect transistor comprising a piezoelectric aluminium-scandium-nitride (Al x Sc 1−x N) layer inside of the gate stack. The so augmented device is sensitive to mechanical stress. In combination with an analogue circuit, this sensor unit is capable of encoding the mechanical quantity into a series of spikes with an ongoing adaptation of the output frequency. This allows for a broad application in the context of robotic and neuromorphic systems, since it enables said systems to receive information from the surrounding environment and provide encoded spike trains for neuromorphic hardware. We present numerical and experimental results on this spiking and adapting tactile sensor.",
"introduction": "Introduction Neural systems and their surroundings are in an ongoing interaction. Humans, mammals and even simple forms of living species like invertebrates are well adapted to permanently changing environments which ensures their survival 1 . Herein, sensation defines the ability to convey information by a chain of transducer stages towards the brain 2 , 3 . It is the somatic sensory system which plays a major role in the transmission of mechanical stimuli from the environment to the brain 2 , 4 . The necessary signal conversion is done by mechanoreceptors underneath the skin which are sensitive to physical distortion and therefore necessary for the perception of size, shape and consistence of objects. Some important classes of mechanoreceptors responsible for a neuronal answer to mechanical stimulation are Meissner’s corpuscles, Merkel cells and Pacinian corpuscles, which can be found in glabrous skin 5 – 7 . Regarding mechanical perception Edgar Douglas Adrian launched a series of seminal papers formulating the firing rate hypothesis based on a number of intriguing experiments as early as 1926. Thereby, he also established the term adaptation in the framework of biological signal processing 8 – 10 . Today, the adaptation to a stimulus is considered as a classical response function, which is not just restricted to the encoding of external signals into internal spiking representations. Besides the adaptation of sensory neurons to a constant stimulus, spike-frequency adaptation is also present in neurons even far from sensory systems and can rise not just through cellular 11 , 12 , but also network induced mechanisms 13 , 14 . These biological findings served as a guideline to develop a tactile sensor based on piezoelectric Al x Sc 1−x N (AlScN) within the fields of neuromorphic engineering and robotics. During the last years, a tremendous effort was made to develop different concepts regarding touch and tactile sensing. Besides traditional piezoelectric materials as for example PbZr x Ti 1−x O 3 (PZT) 15 or BaTiO 3 16 these concepts range from mechanical fluid based 17 sensors to electrical approaches based on graphene 18 – 21 , polymers 22 – 26 , and field-effect transistors coupled capacitively with nanowires 27 , 28 . Moreover, the possible application of pressure sensors for smart prosthetics 29 and their capability of connecting them directly to nerve cells were explored 30 . For a deep and comprehensive overview about recent developments and further possible applications of tactile sensors and eSkin, we refer to the following review articles by Dahiya et al . 31 and Soni and Dahiya 32 . In this context we would like to highlight the piezoelectric oxide semiconductor field effect transistor (POSFET) as a tactile sensor because this device has a few features in common with the here presented PiezoFET 23 , 33 – 35 . The POSFET exploits the organic ferroelectric copolymer PVDF-TrFE in the gate stack. Because ferroelectric materials are piezoelectric, the POSFET is a stress sensitive device, too 36 , 37 . Via the piezoelectric effect, stress induced interfacial charges modulate the channel current of the transistor. This enables the tactile sensor function of the device. Although this fundamental working principle is the same for the POSFET and PiezoFET, we would like to emphasise that the here presented PiezoFET based on AlScN exhibits a few distinguishing features. In contrast to the organic copolymer PVDF-TrFE, AlScN is compatible with Si-CMOS technology 38 – 40 . This allows the development of highly compact and integrated tactile sensor units consisting of PiezoFETs and subsequent neuromorphic circuits. Moreover, the ferroelectric copolymer PVDF-TrFE typically needs an initial wake-up procedure which consists of applying an electric field of about 80–100 V/µm. Such a procedure could be unfavourable for densely packed integrated sensor arrays. The electric field is necessary for the polymer device to function fully 23 , 34 , 41 . Besides the development of single sensor devices for the conversion of stress into an electrical quantity, also a tremendous progress regarding analogue circuit design was made over the last decades 42 . This ranges from the development of biologically inspired circuits like the adaptive exponential I&F neuron 43 , 44 and their integration in spiking deep neural networks 45 to mixed signal circuits which exploit events for processing visual information 46 – 49 . For a review on other neuromorphic systems and circuits, as well as mimicking other senses, the reader can refer to Bartolozzi et al . 50 . In the here presented concept, the piezoelectric thin film inside the gate structure of the PiezoFET responds to a mechanical deformation with a polarisation. For the encoding of the signal into spikes, a Leaky Integrate and Fire (LIF) approach is used. Here, the focus was on the development of a circuitry comprising just few components. Therefore, the LIF neuron model we use does not model the conductance of ion channels explicitly. In order to generate spikes, a negative differential resistance is used which is based on the positive feedback coupling of two transistors. This is a difference to other concepts like the spike based readout of the POSFET 35 . In the following, we present the PiezoFET in more detail. Subsequently, the entire unit including the spike generation circuit is described. This encompasses a short introduction of the used neuron model as well as numerical results on the proposed spiking and adapting circuit. Finally, experimental findings performed with the touch sensor to demonstrate the spike coding of the applied forces as well as the exponential adaptation of the system are presented.",
"discussion": "Conclusion and discussion In the first experiments carried out by Adrian et al., the exact differentiation between individual specialized cells and mechanisms was hardly possible. However, the basal biological mechanism of stimulus adaptation in a sensory system was well observable. Inspired by this seminal work, we developed a system for the encoding of mechanical stress induced through an applied force into a spiking response of a leaky integrate and fire neuron model. The depolarization of the build-up charge of the piezoelectric AlScN layer due to leakage currents, offered the opportunity to incorporate the biological important process of adaption in the sensory unit. Furthermore, the numerical and experimental findings are in good agreement. Interestingly, the presented concept and materials allow integration in Si-technology. Thus, densely packed sensor arrays for artificial spiking neural networks or oscillator-based information processing may be realized. Although, the results presented here are based on the bending of a cantilever which results in a horizontal stress on a piezoelectric material by applying a vertical force, the system could be used in biologically inspired force sensor systems as the theoretical considerations are based on the stress to which the device is subjected. Furthermore, a combination with other effects like piezoresistive modulated transistor responses might be useful and feasible as well 54 , 55 , 63 . Thus, we think that this work is a promising step towards the integration of spiking sensor units into dense interconnected systems as AlScN is a Si-CMOS compatible material 38 – 40 , 51 . Instead of using cantilevers, also planar chips with numerous PiezoFETs are feasible for spatial resolved tactile sensor units. The integration of higher numbers of adapting sensors can for example be used as an input for pattern recognition systems as the local activity changes with a variation in the applied force. The straightforward and technically simple implementation of this biologically essential adaptation process into the sensory unit, offers interesting perspectives as an interface stage for linking neuromorphic computing systems with tactile stimuli from the environment."
} | 2,262 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.