Add files using upload-large-folder tool

05076b04f05611927d832f0da3d4ffa4dcfd58c6fd3c8e741c288ede24fc3e8e/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "All-flexible chronoepifluidic SERS patch.\n(a) Schematic illustration of an all-flexible chronoepifluidic SERS patch (CEP-SERS patch) comprising plasmofluidic channel layer (PCL) and dermal-contact layer (DCL). PCL includes an all-flexible plasmonic SERS substrate and a chronoepifluidic sweat sampler. The flexible SERS substrate features plasmonic nanoislands on ultrathin fluorocarbon-coated PDMS membrane using low-temperature solid-state dewetting of thin silver film. The sweat sampler sequentially collects the sweat via the microfluidic channel with capillary bursting valves, which spatially separates the sweat over time. DCL interconnects the PCL on the skin for stable sweat collection, which contains medical adhesive with a sweat collection port. The CEP-SERS patch provides conformal contact on human skin and label-free sweat profiling of diverse metabolites from sequentially sampled sweat. (b) Schematic illustration of micro and nanofabrication for the CEP-SERS patch. The device fabrication includes fluorocarbon coating (S1), thermal evaporation of Ag (S2), low-temperature solid-state dewetting (S3), and microfluidic encapsulation (S4). The ultrathin fluorocarbon film effectively dewets the thin metal film on the sweat sampler, resulting in plasmonic structures with strong electromagnetic hotspots, which facilitate highly sensitive SERS analysis. Optical images of (c) the PCL (Scale bar: 1 mm) with an inset SEM image of Ag nanoislands (Scale bar: 100 nm) and (d) the CEP-SERS patch on the skin (Scale bar: 10 mm).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "All-flexible SERS Substrate: Ag nanoislands on ultrathin fluorocarbon.\n(a) SEM images (left) and E-field distribution (middle) of Ag nanoislands (Scale bar:\u00a0 100 nm), structural features including diameter and surface coverage (right top), and absorption spectra (right bottom) for 10 nm Ag film on PDMS without fluorocarbon coating (green), single-dewetted 10 nm Ag film (red), and double-dewetted 10 nm Ag film (blue) on fluorocarbon coated PDMS. (b) SERS intensity of 10 \u00b5M R6G at 1365 cm\u207b\u00b9 and SNR depending on the fluorocarbon thickness (tFC). (c) SERS intensity for 1 \u00b5M R6G depending on the Ag film thickness and dewetting repetition (blue: single dewetting, red: repeated dewetting). (d) SERS intensity depends on different concentrations of R6G. (e) Long-term stability of the CEP-SERS patch depending on storage duration measured by SERS peak intensity of 10 \u03bcM R6G solution. (f) Mechanical stability measured by SERS peak intensity variation of benzenethiol at 1070 cm\u22121 after 200 cycles of twisting (red) and bending (blue) (Scale bar: 5 mm). The error bars represent one standard deviation from the mean.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Chrono sample collection and isolation through microfluidic sequential sampler.\n(a) Schematic illustration of sequential sampling (left) and SEM images of capillary bursting valves on the microfluidic sequential sampler (right) (Scale bar: 200 \u03bcm). (b) Optical images of sequential sampling of colored dye (Scale bar: 5 mm). (c) Measured sampling interval depending on flow rate and chamber volume. Each color represents varying chamber volume (blue: 1.5 \u03bcL, green: 0.9 \u03bcL, red: 0.5 \u03bcL). (d) Schematic illustration of on-chip sample isolation utilizing air pocket barrier. (e) The effectiveness of sample isolation is demonstrated by measuring the SERS intensity of diffused R6G at the chamber filled with DI water depending on storage duration with (red) and without (blue) sample isolation. The error bars represent one standard deviation from the mean.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Machine-learned label-free quantification of metabolites.\n(a) SERS peak intensity of uric acid (red, at 635 cm-1), lactate (blue, at 859 cm-1), and tyrosine (green, at 1353 cm-1) depending on the concentration. The error bars represent one standard deviation from the mean. (b) Uric acid profiling in chrono-sampled artificial sweat (left) tracked by SERS intensity at 635 cm-1. Colored bars represent calibrated SERS intensity of input uric acid concentration. Optical images of the sequential sampling over time depending on the chamber volume (blue: 1.5 \u03bcL, red: 0.5 \u03bcL), which visualize sparse and dense sampling intervals (right). The scale bars represent 5 mm. Machine-learned quantification of (c) uric acid, (d) lactate, and (e) tyrosine in the mixture presented by prediction depending on true concentration. (center line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range) Extracted SHAP feature importance (FI, solid line) and SERS spectra at 10 mM (dotted line) of (f) uric acid, (g) lactate, and (h) tyrosine. The bars represent the matched SERS peaks (colored) and notches (gray).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Label-free human sweat profiling of assorted metabolites.\n(a) Schematic illustration of on-body evaluation. Optical images of (b) on-body evaluation (left) and the CEP-SERS patch attached to human skin (right), (c) sequential sweat sampling (top), and sample preservation under different physical distortions (bottom). The scale bar represents 10 mm. Accuracy of label-free quantification for (d) uric acid and (e) lactate in the human sweat. (f) Comparison between tyrosine concentrations depending on participants measured by CMA (light gray) and SERS (dark gray). (g-i) Chronological profiling of each metabolite during exercise (left) measured by SERS (red rectangles) and FMA (blue circles) presented with schematic illustrations of each metabolic pathway (right). The error bars represent one standard deviation from the mean. (j) Measured heart rate (HR, black solid line), RER (orange bar: 0.7 < RER < 0.85), oxygen (VO2, blue dotted line), and carbon dioxide (VCO2, red dotted line) intake during on-body evaluation to monitor exercise intensity. Concentration comparison of (k) uric acid and (l) tyrosine under different physiological conditions. (center line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range; cross, outlier; point, data) (m) Relative metabolic phenotyping of individuals (IN, thin solid line) and averaged result (bold solid line) under fasting conditions (blue) and after purine-rich diet intake (red).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

05076b04f05611927d832f0da3d4ffa4dcfd58c6fd3c8e741c288ede24fc3e8e/preprint/preprint.md

@@ -0,0 +1,182 @@
+# Abstract
+
+Wearable sweat sensors allow non-invasive metabolic profiling for timely intervention in proactive healthcare. However, molecular recognition elements in conventional sensors still hinder a comprehensive understanding of an individual's unique physiological profile. Here we report an all-flexible chronoepifluidic surface-enhanced Raman spectroscopy (SERS) patch (CEP-SERS patch) for label-free sweat profiling. The CEP-SERS patch features the integration of nanoplasmonics and functional microfluidics for precise chronological profiling of metabolites. An ultrathin fluorocarbon film facilitates large-area nanofabrication of plasmonic structures on a functional microfluidic channel via low-temperature solid-state dewetting of a thin silver film. The CEP-SERS patch facilitates conformal contact on human skin and SERS detection of diverse metabolites from sequentially sampled sweat. Machine-learned quantification of metabolites including lactate, uric acid, and tyrosine has successfully profiled SERS detection of sweat during assorted physical activities. This CEP-SERS patch can provide a new strategy for delineating the physiological phenotype of individuals in personalized healthcare.
+
+[Biological sciences/Biotechnology/Nanobiotechnology/Nanofabrication and nanopatterning](/browse?subjectArea=Biological%20sciences%2FBiotechnology%2FNanobiotechnology%2FNanofabrication%20and%20nanopatterning)  
+[Physical sciences/Nanoscience and technology/Nanobiotechnology/Microfluidics](/browse?subjectArea=Physical%20sciences%2FNanoscience%20and%20technology%2FNanobiotechnology%2FMicrofluidics)  
+[Physical sciences/Engineering/Biomedical engineering](/browse?subjectArea=Physical%20sciences%2FEngineering%2FBiomedical%20engineering)  
+[Physical sciences/Materials science/Materials for optics/Nanophotonics and plasmonics](/browse?subjectArea=Physical%20sciences%2FMaterials%20science%2FMaterials%20for%20optics%2FNanophotonics%20and%20plasmonics)
+
+# Introduction
+
+Metabolic phenotyping is pivotal in precision medicine, revealing individual health traits for personalized interventions [1]. Metabotypes surpass genetic predispositions, offering valuable insights into the diversity of individuals influenced by extrinsic determinants such as behavioral patterns, regimens, or gut microbial activity [2,3]. The interrelation of the metabolite reflects precision health status through metabolic pathway alterations induced by physiological changes or disorders. [4–8]. In particular, transient metabolic alterations provide a physiological cue for proactive healthcare. Such cues facilitate the instant recognition of physiological disorders to promote optimal health outcomes. For instance, postprandial metabolic rates provide crucial information on individual responses to dietary intake, aiding in the development of personalized nutrition plans [9]. In addition, exercise or lifecycle routines can be tailored by monitoring daily activity-induced metabolite fluctuations [10]. However, conventional approaches profile static metabolite in blood [11,12] or urine [13–15], which poses challenges in dynamic metabolic profiling.
+
+Recent biosensing wearables profile the metabolic kinetics in biological fluids such as tear fluid [16,17], saliva [18], or sweat [19,20]. Sweat, unlike others, exhibits chemical abundance [21,22] and simple sample collection with less contamination [23], thus allowing in-situ profiling of metabolite alteration in proactive healthcare [24]. For example, analyzing lactate dynamics suggests optimal exercise routines by evaluating the lactate threshold or maximal lactate steady state during physical activities [25–27]. In addition, postprandial changes of branched-chain amino acid or uric acid provide prognostic cues for metabolic syndrome [28–30] or gout [31], respectively. Moreover, cortisol levels [32–34] indicate psychological stress responses during daily life. Furthermore, sweat is easily collected by epidermal interfaces with simple absorbents [35–37], or microfluidic patches [38–40], exhibiting high compatibility with assorted wearable sensors. Such epidermal sweat sensors often utilize electrochemical [26,27,30,31,33,34,36,38,41] or colorimetric approaches [42–45]. However, molecular recognition elements such as antibodies or enzymes on active sensing sites still hinder multiple and unveiled biomarkers detection for a comprehensive understanding of an individual's unique physiological profile.
+
+Surface-enhanced Raman spectroscopy (SERS) allows label-free and quantitative detections of diverse biochemicals [46–48]. Plasmonic nanostructures are integrated into flexible epidermal patches for label-free metabolic profiling through sweat. [48,49,50,51] Furthermore, recent advancements in plasmonic patches combine with epidermal microfluidic channels for not only transient analysis [49–57] but also in-situ collection and chronological profiling of metabolites [58,59,60]. For instance, rigid SERS substrates are frequently integrated into elastomeric microfluidic patches [58,61] to analyze metabolic alterations [58]. Partial rigidity of the SERS substrate still constrains conformal skin contact and mechanical durability in wearable applications, whereas plasmonic paper-based microfluidic sensors feature high flexibility thus overcoming such technical limitations. However, such sensors still lack delicate microfluidic control [48,52,55,62,63], sensitivity, and specificity [59,64], thus restricting the direct quantification of multiple analytes in complex sweat. In particular, such SERS patches encounter compatibility issues with sophisticated functional microfluidics such as capillary bursting valves [65–67], facilitating sequential sampling for sweat profiling. As a result, the all-flexible epifluidic SERS patch still remains a technical challenge in precise and multiplexed sweat profiling.
+
+Here we report an all-flexible chronoepifluidic SERS patch (CEP-SERS patch) for label-free profiling of sweat metabolites. The CEP-SERS patch consists of a plasmofluidic channel layer (PCL) for sweat collection, storage, and SERS analysis, and a dermal contact layer (DCL) for skin attachment (Fig. 1a). PCL includes a flexible plasmonic SERS substrate and a microfluidic sequential sampler. The SERS substrate features plasmonic nanoislands on ultrathin fluorocarbon-coated PDMS membrane, driven by low-temperature solid-state dewetting of thin silver film. The ultrathin fluorocarbon film effectively dewets the thin metal film on the sequential sampler, resulting in plasmonic structures with strong electromagnetic hotspots for highly sensitive SERS analysis. In addition, the sequential sampler serially collects the sweat via the microfluidic channel with capillary bursting valves, which spatially separate the sweat over time. Furthermore, DCL contains medical adhesive with a sweat collection port and interconnects the PCL on the skin for stable sweat collection. Finally, the CEP-SERS patch allows machine-learned label-free quantification of multiple metabolites in chrono-sampled sweat, thus quantitatively profiling sweat over time during physical activities. Such all-flexible feature ensures conformal contact for on-skin chronological sweat collection and label-free quantification of metabolites.
+
+The CEP-SERS patch was fabricated by integrating micro- and nanofabricated PCL and DCL (Fig. 1b). First, an ultrathin fluorocarbon layer was coated on the polydimethylsiloxane (PDMS) chronoepifluidic sweat sampler via atmospheric pressure chemical vapor deposition (AP-CVD) (S1). The fluorocarbon coating significantly reduces the surface energy and increases the surface roughness, thereby facilitating the low-temperature solid-state dewetting of an Ag thin film. Subsequently, a 10 nm-thick Ag thin film was thermally evaporated onto the fluorocarbon-coated sweat sampler (S2). The evaporated Ag thin film was thermally dewetted at 160°C for 30 minutes to form Ag nanoislands (S3). Steps S2 and S3 were repeated to enhance the plasmonic hotspots on the PCL. Finally, an air-venting outlet was punched on the PCL attached to the DCL with a sweat port for encapsulation and skin attachment (S4). Figure 1c and 1d present optical images of the fabricated PCL, including an inset scanning electron microscopy (SEM) image of the Ag nanoislands, and the CEP-SERS patch on the skin. Details and stepwise optical images of the fabrication procedure are described in the Supplementary Materials (Fig. S1).
+
+# Result
+
+## All-flexible SERS Substrate: Ag nanoislands on ultrathin fluorocarbon
+
+The plasmonic properties of Ag nanoislands on the CEP-SERS patch are precisely controlled by the thicknesses of fluorocarbon and Ag thin film. Figure 2a presents SEM images (left) and E-field distribution (middle) of Ag nanoislands, calculated by the finite-difference time-domain method (Lumerical FDTD Solutions) method. The structural features (right top) and absorption properties (right bottom) are shown for 10 nm Ag film on PDMS without fluorocarbon coating (green), single-dewetted 10 nm Ag film (red), and double-dewetted 10 nm Ag film (blue) on fluorocarbon coated PDMS. Unlike bare PDMS, the fluorocarbon intermediate layer provides low surface energy, leading to an enlargement of nanoisland diameters by facilitating lower-temperature solid-state dewetting of thin film (Fig. S2). In addition, two-step dewetting further increases the nanoisland diameters and the packing density up to 40%. These morphological modifications induce a redshift of plasmon resonance (Fig. S3) and enhance E-field intensity for highly-sensitive SERS detection.
+
+The fluorocarbon thickness is determined by signal-to-noise ratio (SNR) in SERS measurements. Figure 2b shows the SERS intensity of 10 µM Rhodamine 6G (R6G) at 1365 cm⁻¹ and SNR depending on the fluorocarbon thickness. A thick fluorocarbon layer enhances the SERS intensity (Fig. S4a) but introduces SERS background noise (Fig. S4b). Therefore, the fluorocarbon thickness was optimized to 2 nm to maximize the SNR, defined as the ratio of the SERS peak intensity at 1365 cm⁻¹ to the standard deviation of the fluorocarbon SERS noise signals. Next, the Ag film thickness is optimized for the maximum SERS intensity. Figure 2c shows SERS intensity for 1 µM R6G at 1365 cm⁻¹ depending on the Ag film thickness and dewetting repetition. A thick Ag film facilitates the formation of large, densely packed nanoislands during the dewetting process (Fig. S5), leading to enhanced SERS intensity (blue bar). Repeated dewetting further enhances the SERS intensity (red bar) by providing abundant plasmonic hotspots. Note that the coalescence of nanoislands at a total thickness of 24 nm leads to a drastic decline in the SERS signal (Fig. S6). As a result, the CEP-SERS patch is fabricated by a 10 nm repeated dewetting process, which provides the maximum SERS intensity (Fig. S7). The SERS intensity of R6G exhibits strong linearity with concentration (Fig. 2d and Fig. S8). The SERS substrate provides an average SERS enhancement factor of 1.8 × 10⁷ with a uniformity of 11.8% (Fig. S9). The CEP-SERS patches are vacuum-sealed for further on-body analysis (Fig. S10) and over 85% of the performance is preserved for up to 25 days (Fig. 2e). The mechanical stability of the CEP-SERS patch is confirmed by measuring the SERS intensity of benzenethiol at 1070 cm⁻¹. The SERS performance remains stable with no notable degradation after 200 cycles of twisting (red) and bending (blue) (Fig. 2f). The CEP-SERS patch remains stable and effective for wearable sweat profiling.
+
+## Microfluidic Sequential Sampler
+
+The sequential sweat sampler allows chronological sample collection through capillary bursting valves (CBVs) in a microfluidic network. Figure 3a illustrates the sampling sequence (left) and SEM images of the microfabricated CBVs (right). The bursting pressure (BP) gradients guide a sample fluid into different chambers for sequential sampling. The BPs are controlled by the width and diverging angle of the CBVs. In particular, the narrow width and the high diverging angle increase the BP in the hydrophobic channel (Fig. S11). The sample fluid enters the first chamber through CBV 1 and fills the chamber. Once the chamber is filled, the fluid then flows into the subsequent chamber through CBV 2. CBV 3 has the highest BP for air ventilation during the sampling. Note that air ventilation allows stable sweat collection without the formation of bubbles. The microfluidic parameters are determined by two-phase fluid dynamic analysis (COMSOL Multiphysics 6.1, Fig. S12). The sequential loading of distinct colored solutions visualizes the stepwise procedure of chrono-sample collection (Fig. 3b). In addition, the sampling interval is determined by the chamber volume and the flow rate (Fig. 3c). Each chamber sequentially stores a collected sample in a separate space, and the small chamber volume allows dense sampling intervals over time (Fig. S13 and Supplementary Video 1). In this experiment, the sweat patch with a microchamber volume of 0.5 µL fulfills each chamber within ~2 minutes at a flow rate of 0.25 µL/min.
+
+The collected samples are preserved without mixing through on-chip isolation and evaporation suppression. The isolation prevents diffusion between fluids, allowing for long-term storage without mixing (Fig. 3d). An air pocket barrier creates trapped microbubbles that separate each sample. The effectiveness of isolation is demonstrated by measuring SERS intensity from the deionized water (DI water) in the chamber after sequentially injecting DI water and R6G solution. The isolated chambers maintain initial concentration for 30 hours (red), while the absence of the air pocket barrier leads to mixing (blue). Low-permeable polyester film, attached to the medical adhesive at the sequential sampler, prevents evaporation. This method retains about 80% of samples after 30 hours. Note that evaporation-induced loss is quantified by measuring fluid and air bubble fractions (Fig. S14). This microfluidic sampler facilitates stable sweat collection on the skin and long-term preservation.
+
+## Machine-learned label-free quantification of metabolites
+
+The CEP-SERS patch allows label-free quantification of assorted metabolites. Figure 4a shows the SERS peak intensities of uric acid (red, at 635 cm⁻¹), lactate (blue, at 859 cm⁻¹), and tyrosine (green, at 1353 cm⁻¹) depending on the different concentrations in the range of physiological level (SERS spectra are presented in Fig. S15). A strong linear relationship between SERS peak intensity and concentration facilitates label-free metabolite quantification. Note that intrinsic Raman background noises from PDMS were initially removed for accurate signal interpretation of metabolites. The metabolic profiling over time is further shown by tracking the changes in the SERS peak intensity of the target molecules within artificial sweat samples (Fig. S16a). The uric acid concentration is sequentially adjusted to 80 µM, 10 µM, and 20 µM over the injection time by using a syringe pump (Fig. S16b). Figure 4b illustrates the variation in SERS peak intensity of uric acid in chrono-sampled artificial sweat over the collection period (left), utilizing the CEP-SERS patch with chamber volumes of 1.5 µL (blue) and 0.5 µL (red) (SERS spectra are presented in Fig. S16c and d). The CEP-SERS patch with a small chamber volume effectively improves the precise quantification of rapid variations in target molecular concentration by providing dense sampling intervals (right).
+
+Machine-learned label-free quantification of metabolites is further conducted by using an autoencoder with a logistic regression model. The autoencoder-based prediction model is trained to minimize the total loss function, including the reconstruction and prediction losses (Fig. S17) [68]. Target metabolites including uric acid, lactate, and tyrosine are mixed with 41 different combinations of concentration, and a total of 1,476 spectra are provided for robust machine-learned quantification, which considers various background states of sweat (Fig. S18 and S19). The SERS spectra mapped onto the two-dimensional (2D) latent space via autoencoder show that the individual classes are diagonally aligned depending on the concentration of each metabolite (Fig. S20). Figure 4c-e shows the predicted concentrations and the corresponding true values for uric acid (R²: 0.71–0.80), lactate (R²: 0.65–0.83), and tyrosine (R²: 0.82–0.92), which are evaluated with 10 times of repeated random sampling cross-validation (Fig. S21). The concentration of each metabolite is well predicted to the true value and the quantification process is explained through the feature extraction calculated by the Shapley additive explanation (SHAP) value (Fig. S22) [69,70]. The SHAP feature importance (solid line) is presented with corresponding SERS spectra at 10 mM concentration (dotted line) for uric acid (Fig. 4f), lactate (Fig. 4g), and tyrosine (Fig. 4h). The extracted features contain the matched characteristics to the SERS spectra including SERS peaks (colored bar) and notches (gray bar). This validation demonstrates the machine-learned quantifications are explainable and established based on the molecule-specific SERS signals.
+
+## Label-free human sweat profiling of assorted metabolites
+
+The CEP-SERS patch collects human sweat on the skin and label-freely profiles the transient alteration of metabolites. The on-body evaluation includes a treadmill warm-up followed by a climb mill exercise, performed on separate days under fasting conditions and after a purine-rich diet intake (Fig. 5a, Fig. S23). The CEP-SERS patches are attached to multiple sites on the forehead and shoulder of participants' skin to collect exercise-induced sweat (Fig. 5b). The sweat samples are collected during the evaluation through the CEP-SERS patch (Fig. 5c). The inset images show the sweat collection using the patch without plasmonic structures to clearly present the sequential sampling. All flexibility of the patch ensures conformal dermal contact, resulting in stable sequential sweat collection (top). In addition, the CEP-SERS patch securely preserves sweat samples under various physical stresses that may occur in a wearable environment such as compressing, twisting, or detaching (bottom). Note that the biophysical signals such as heart rate, respiratory exchange ratio (RER), oxygen, and carbon dioxide intake are simultaneously measured during the evaluation to monitor the exercise intensity.
+
+The CEP-SERS patch performs label-free profiling of chrono-sampled human sweat using machine-learned quantification models for individual metabolites. Sweat samples from four healthy participants are analyzed to assess metabolic differences during exercise with a purine-rich meal versus fasting. The average SERS signals measured from human sweat capture key features of various metabolites in sweat (Fig. S24). SERS signals from chrono-sampled sweat are obtained seven times from each chamber to reduce measurement errors (Fig. S25 - S28). The sweat flow rate for each participant is determined by dividing the total volume of the collected samples by the sampling time (Fig. S29). Label-free uric acid, lactate, and tyrosine quantifications in human sweat are validated using commercial fluorometric assay (FMA) or colorimetric assay (CMA) kits. The sweat is further collected by swiping the microtube to the forehead for FMA and CMA. Machine-learned predictions for uric acid and lactate demonstrate strong agreement with FMA measurements, achieving R² values of 0.96 and 0.86, respectively (Fig. 5d and e). In addition, tyrosine prediction from participant 2–4 shows an error margin of ~5 µM to the CMA kit (Fig. 5f). For Participant 1, contaminants including skin lipids during sample collection significantly affected the absorbance properties, resulting in a substantial difference of ~25 µM compared to the CMA measurement. The metabolites in the collected sweat samples (Fig. 5g-i) are chronologically profiled with biophysical signals (Fig. 5j) through the machine-learned SERS quantification models (Chronological profiling results of metabolites and biophysical signals for four participants under fasting conditions and purine-rich diet intake are presented in Fig. S30 - S33). The CEP-SERS patch captures the metabolic alterations associated with the metabolic pathway of each metabolite and dilution. The metabolites in sweat are often diluted over time during perspiration [24,31,34,71]. In contrast, lactate temporarily increases, which reflects the eccrine sweat gland metabolism [24,26,72] and anaerobic metabolism [73,74] during exercise. The dietary intake increases the overall concentrations of uric acid [31] and tyrosine (amino acid) [30,71,75] in the on-body evaluation (Fig. 5k and 5l). The metabolic alterations under fasting (blue) and purine-rich diet intake conditions (red) show physiological states for four participants. The averaged alteration (bold red line) reveals that lactate levels remain stable due to minimal dietary influence. In contrast, uric acid and tyrosine increase post-ingestion, reflecting physiological changes in purine metabolism and protein digestion, respectively (Fig. 5m). The CEP-SERS patch combined with machine-learned metabolic quantification has successfully captured transient physiological changes by label-free and multiplexed detection of exercise- and intake-induced metabolic alterations.
+
+# Conclusions
+
+Metabolic profiling offers significant insights into individual physiological states, enhancing the scope of digital phenotyping and advancing personalized healthcare. Wearable sweat sensors non-invasively capture the transient biochemical alterations during assorted activities and allow on-body metabolic profiling. Furthermore, SERS holds significant potential to unveil comprehensive physiological information by facilitating label-free universal molecular recognition in sweat. Unlike others (Table S1), our CEP-SERS patch features the integration of nanoplasmonics and functional microfluidics for precise chronological profiling of metabolites. The plasmonic nanostructures are integrated into the microfluidic sampler via large-area fabrication of Ag nanoislands on fluorocarbon-coated PDMS, facilitating both SERS analysis and stable sequential sweat collection. In addition, SERS spectra from chrono-sampled sweat samples are profiled through a robust machine-learned quantification model that accounts for various concentration combinations of the background in sweat. The CEP-SERS patch has successfully captured the exercise- and intake-induced alterations in multiple metabolites. This CEP-SERS patch combined with machine-learned quantification can provide a new strategy for delineating the physiological phenotype of individuals in personalized healthcare.
+
+# Methods
+
+## Numerical analysis
+
+### E-field distribution.
+The electric fields of Ag nanoislands with varying geometries were numerically calculated utilizing a three-dimensional finite-difference time-domain method (Lumerical FDTD Solutions). Geometric parameters of the Ag nanoislands were derived from binary segmented SEM images employing Image J software, with Ag nanoislands modeled as a cylindrical shape.
+
+### Fluid dynamics.
+Fluid dynamics at the capillary bursting valve were numerically calculated by utilizing finite elements methods (FEM, COMSOL Multiphysics 6.1). The calculations were conducted by integrating the two-phase flow and level-set physics modules. In addition, the two-phase flow, level set, and wetted wall phenomena were coupled to derive comprehensive computational results. A time-dependent solver was utilized to compute the temporal evolution of the sequential sampling.
+
+## Materials and reagents
+Rhodamine 6G (R6G, R4127-25G, dye content ~95%), benzenethiol (W361607-SAMPLE-K, ≥98%), sodium L-lactate (L7022-5G, ~98%), urea (U5378-100G), tyrosine (PHR1097-1G, certified reference material), glycine (G7126-100G, ≥99%), L-alanine (A7469-25G, ≥98.5%), L-glutamic acid monosodium salt monohydrate (49621-250G, ≥98.0%), uric acid (U2625-25G, ≥99%), L-ascorbic acid (A5960-25G, ≥99.0%), D-(+)-glucose (G8270-100G, ≥99.5%), creatinine (PHR1462-1G, certified reference material), sodium chloride (S7653-250G, ≥99.5%), and potassium chloride (P5405-250G, ≥99.0%) are purchased from Sigma Aldrich. Medical adhesives were purchased from 3M (Tegaderm 1622W, 1522).
+
+## Fabrication of CEP-SERS patch
+A Si wafer was immersed in buffered oxide etchant for 30 seconds to remove the native oxide layer. Photoresist (SU-8 2000.5, MicroChem Corp.) was then spin-coated to a thickness of 500 nm to form an adhesion layer. An additional photoresist (SU-8 2100, MicroChem Corp.) was applied to fabricate a microfluidic channel mold with a thickness of 200 µm. A 10:1 mixture of PDMS base and curing agent (Sylgard 184, Dow Corning Corp.) was spin-coated onto the microfluidic channel mold at 100 rpm for 30 seconds and then at 160 rpm for 30 seconds. The PDMS was cured on a hotplate at 80°C for 1 hour and 30 minutes. The PDMS layer was peeled off from the mold after curing and washed in isopropyl alcohol for 5 minutes. The fluorocarbon layer was coated on the microfluidic channel via AP-CVD with parameters set to 150 W plasma power, a helium flow rate of 5.0 L/min, and a fluorocarbon flow rate of 2 sccm (IHP-1000, APP Korea). Ag film was thermally deposited at a deposition rate of 1.0 Å/s by using a thermal evaporator (SNTEC Inc., Korea) and dewetted on a hotplate at 160°C for 30 minutes. The Ag deposition and dewetting steps were repeated twice to create strong plasmonic hotspots. The air ventilation outlet and sweat inlet were punched on PCL and medical adhesive film (Tegaderm 1622W, 3M) for DCL, respectively. The two layers were bonded by double-sided medical adhesive (1522, 3M) to complete the CEP-SERS patch fabrication.
+
+## Characterization and measurement
+
+### Absorption measurement.
+Absorption spectra were measured by using a microscopic spectrometer setup comprising an inverted microscope (Axiovert 200M, Carl Zeiss) integrated with a white light LED lamp (MCWHL5-C4, Thorlabs Inc.) and a spectrometer (MicroSpec 2300i) featuring a charge-coupled device (CCD) camera (Model PIXIS: 400BR, Princeton Instruments). Absorption spectra of Ag nanoislands on fluorocarbon-coated PDMS were acquired utilizing a 20× objective lens (NA=0.5).
+
+### SERS measurement.
+A helium-neon laser (HRP050, Thorlabs Inc.) operating at a wavelength of 633 nm was utilized in conjunction with a spectrometer featuring a CCD camera, both integrated with an inverted microscope. Light excitation and collection were performed via a 20× objective lens (NA = 0.5). The excitation laser was operated at a power of 5 mW, while data acquisition times were set to 1 second for characterization and extended to 10 seconds for metabolites and human sweat sample measurements. Note that a single SERS spectrum from human sweat was acquired by averaging 6 times of the measurement to minimize signal variances. All the SERS spectra are measured as solution state, and an equivalent volume of DI water was applied to the sample before detection for dry samples.
+
+## Machine-learned metabolic quantification
+
+### SERS data preparation.
+First, 1,476 spectra with 41 different concentration combinations including target molecules are provided for machine-learned quantification. The measured SERS spectra have a vector length of 1321, which contains the signal intensities for the Raman shift range of 457–1674 cm⁻¹. Each vector was normalized to a range of 0–1.
+
+### Metabolic quantification.
+The model consists of a symmetric encoder and decoder with four equally sized layers of 1321 nodes, maintaining a consistent number of neurons across all layers, with a latent layer of length 2 between the encoder and decoder. The model was trained with the Adam optimizer, and the hyperparameters were heuristically tuned to optimize performance. Initially, the learning rate, weight decay, and batch size were set to 1e⁻⁴, 1e⁻⁵, and 32 respectively. The maximum number of training epochs was set to 150, with early stopping employed. After the 50th epoch, a scheduled weight decay adjustment was applied to promote further model generalization, reducing the learning rate and weight decay to 1e⁻⁵ and 1e⁻⁶, respectively. The model predicts the concentration of each metabolite based on the values along the concentration axis in the latent space. Note that three-quarters of the entire SERS spectra data were used for training, with the remainder reserved for validation. In addition, 10 times of repeated random sampling cross-validation ensures robust model evaluation.
+
+### SERS feature importance calculation.
+SHAP (SHapley Additive exPlanations) values were utilized to identify the significant features for concentration prediction. The SHAP calculations were performed using only the encoder part of the model, focusing on the feature importance independent of the increase or decrease in concentration. For each sample, the SHAP values were squared and then averaged across all samples to obtain the overall importance of each feature, thereby identifying key spectral features contributing to the model's predictions.
+
+## On-body evaluation
+
+### Protocols.
+The participant performed the following exercise protocol on separate days under fasting conditions and after the purine-rich diet intake, which included 125 g to 250 g of sardines. On the days when the purine-rich diet was consumed, the exercise was initiated after a 1.5-hour rest period. The participants wear the CEP-SERS patches attached to the forehead and shoulder, a wearable metabolic system (Cosmed K5, Cosmed) for respiratory gas analysis, and a heart rate sensor (Polar OH1, Polar Electro Oy) for heart rate monitoring. The participant performed a warm-up consisting of two cycles of running on a treadmill at a 7.2 km/hr speed for 4 minutes, followed by a 3-minute rest period. Subsequently, the participant engaged in exercise on a climbmill with progressively increasing intensity until the heart rate reached 80% of the maximum heart rate. The participant completed an additional 5-minute rest period after exercise, concluding the protocol. Note that sweat samples were additionally collected using a microtube swapped on the forehead to measure the actual concentration of sweat using the fluorometric assay or colorimetric assay kit. Sweat was collected once after the treadmill exercise, at 3–5 minute intervals during the climb mill exercise, and once after a 5-minute rest period following the completion of all exercise tasks. All human trials were approved by the Institutional Review Board of Korea Advanced Science and Technology (protocol number: KH2024-085). In addition, we obtained informed consent from all participants before the experiment.
+
+### Fluorometric assays for validation of uric acid and lactate quantification.
+The concentrations of uric acid and lactate in collected sweat were validated by using a commercial uric acid assay kit (MAK077-1KT, Sigma Aldrich) and a lactate assay kit (BM-LAC-100, PicoSens™). For uric acid measurement, 5 µL of sweat sample was mixed with 45 µL of uric acid assay buffer, followed by the addition of 2.17 µL of probe and 2.17 µL of enzyme mix. The mixture was thoroughly mixed by pipetting and incubated for 30 minutes at 37°C. For lactate measurement, sweat samples were diluted 200-fold with deionized water. A 10 µL aliquot of the diluted sweat sample was mixed with 40 µL of lactate assay buffer, followed by the addition of 0.43 µL of probe and 2.17 µL of enzyme mix. The mixture was thoroughly mixed by pipetting and incubated for 30 minutes at room temperature. Fluorescence intensity was measured at excitation and emission wavelengths of 535 nm and 590 nm, respectively. Uric acid and lactate concentrations were calculated using a standard curve of each molecule.
+
+### Colorimetric assays for validation of tyrosine quantification.
+The concentrations of tyrosine in collected sweat were validated by using a commercial tyrosine assay kit (MET-5073, Cell Biolabs Inc.). Sweat samples were prepared by adjusting their volume to 50 µL with standard tyrosine solutions. The mixed samples were chilled on ice for 30 minutes and then centrifuged at 10,000 g for 15 minutes at 4°C. The supernatant was carefully transferred to prevent the pellet from dissolving and subsequently passed through a QIAquick Spin Column (28115) by centrifugation (MICRO 17 TR, Hanil Science Co.) at 5,000 g for 10 minutes at 4 °C. The filtered samples were then used for tyrosine quantification. The 10x enzyme from the kit was diluted to the assay solution, and 15 µL of each prepared sample and standard tyrosine solution was added to individual wells of a transparent 384-well plate, followed by 15 µL of the diluted enzyme solution to make a total volume of 30 µL per well. The plate was incubated at room temperature for 10 minutes on a horizontal shaker, and absorbance was measured at 490 nm using a plate reader (CYTATION 5 imaging reader, Agilent BioTek). Tyrosine concentrations were then calculated using a standard curve.
+
+# References
+
+1. Holmes, E., Wilson, I. D., and Nicholson, J. K. Metabolic phenotyping in health and disease. Cell, 134(5), 714–717 (2008).
+2. Holmes, E., Loo, R. L., Stamler, J., Bictash, M., Yap, I. K., Chan, Q., … and Elliott, P. Human metabolic phenotype diversity and its association with diet and blood pressure. Nature, 453(7193), 396–400 (2008).
+3. Nicholson, J. K., Holmes, E., Kinross, J. M., Darzi, A. W., Takats, Z., and Lindon, J. C. Metabolic phenotyping in clinical and surgical environments. Nature, 491(7424), 384–392 (2012).
+4. Dumas, M. E., Kinross, J., and Nicholson, J. K. Metabolic phenotyping and systems biology approaches to understanding metabolic syndrome and fatty liver disease. Gastroenterology, 146(1), 46–62 (2014).
+5. Holmes, E., Wijeyesekera, A., Taylor-Robinson, S. D., and Nicholson, J. K. The promise of metabolic phenotyping in gastroenterology and hepatology. Nature reviews Gastroenterology & hepatology, 12(8), 458–471 (2015).
+6. Everett, J. R., Holmes, E., Veselkov, K. A., Lindon, J. C., and Nicholson, J. K. A unified conceptual framework for metabolic phenotyping in diagnosis and prognosis. Trends in Pharmacological Sciences, 40(10), 763–773 (2019).
+7. Suhre, K. Metabolic profiling in diabetes. J Endocrinol, 221(3), R75-R85 (2014).
+8. Inoue, K., Tsutsui, H., Akatsu, H., Hashizume, Y., Matsukawa, N., Yamamoto, T., and Toyo'Oka, T. Metabolic profiling of Alzheimer's disease brains. Scientific reports, 3(1), 2364 (2013).
+9. Secor, S. M. Specific dynamic action: a review of the postprandial metabolic response. Journal of Comparative Physiology B, 179, 1–56 (2009).
+10. Emerson, S. R., Kurti, S. P., Emerson, E. M., Cull, B. J., Casey, K., Haub, M. D., and Rosenkranz, S. K. Postprandial metabolic responses differ by age group and physical activity level. The Journal of nutrition, health and aging, 22(1), 145–153 (2018).
+11. Fiehn, O. and Kind, T. Metabolite profiling in blood plasma. Metabolomics: Methods and protocols, 3–17 (2007).
+12. Grauslys, A., Phelan, M. M., Broughton, C., Baines, P. B., Jennings, R., Siner, S., … and Carrol, E. D. NMR-based metabolic profiling provides diagnostic and prognostic information in critically ill children with suspected infection. Scientific Reports, 10(1), 20198 (2020).
+13. Want, E., Wilson, I., Gika, H. et al. Global metabolic profiling procedures for urine using UPLC–MS. Nature Protocols 5, 1005–1018 (2010).
+14. Mathé, E. A., Patterson, A. D., Haznadar, M., Manna, S. K., Krausz, K. W., Bowman, E. D., … and Harris, C. C. Noninvasive urinary metabolomic profiling identifies diagnostic and prognostic markers in lung cancer. Cancer research, 74(12), 3259–3270 (2014).
+15. Luan, H., Liu, L. F., Tang, Z., Zhang, M., Chua, K. K., Song, J. X., … and Cai, Z. Comprehensive urinary metabolomic profiling and identification of potential noninvasive marker for idiopathic Parkinson’s disease. Scientific reports, 5(1), 13888 (2015).
+16. Park, W., Seo, H., Kim, J. et al. In-depth correlation analysis between tear glucose and blood glucose using a wireless smart contact lens. Nature Communications 15, 2828 (2024).
+17. Keum, D. H., Kim, S. K., Koo, J., Lee, G. H., Jeon, C., Mok, J. W., … and Hahn, S. K. Wireless smart contact lens for diabetic diagnosis and therapy. Science Advances, 6(17), eaba3252 (2020).
+18. Kim, J., Imani, S., de Araujo, W. R., Warchall, J., Valdés-Ramírez, G., Paixão, T. R., … and Wang, J. Wearable salivary uric acid mouthguard biosensor with integrated wireless electronics. Biosensors and Bioelectronics, 74, 1061–1068 (2015).
+19. Brunmair, J., Gotsmy, M., Niederstaetter, L., Neuditschko, B., Bileck, A., Slany, A., … and Gerner, C. Finger sweat analysis enables short interval metabolic biomonitoring in humans. Nature Communications, 12(1), 5993 (2021).
+20. Bariya, M., Li, L., Ghattamaneni, R., Ahn, C. H., Nyein, H. Y. Y., Tai, L. C., and Javey, A. Glove-based sensors for multimodal monitoring of natural sweat. Science advances, 6(35), eabb8308 (2020).
+21. Tu, J., Torrente-Rodríguez, R. M., Wang, M. and Gao, W. The era of digital health: a review of portable and wearable affinity biosensors. Advanced Functional Materials, 30, 1906713 (2020).
+22. Heikenfeld, J. et al. Accessing analytes in biofluids for peripheral biochemical monitoring. Nature Biotechnology, 37, 407–419 (2019).
+23. D. S. Yang, R. Ghaffari, and J. A. Rogers Sweat as a diagnostic biofluid. Science, 379(6634), 760–761 (2023).
+24. Min, J., Tu, J., Xu, C., Lukas, H., Shin, S., Yang, Y., … and Gao, W. Skin-interfaced wearable sweat sensors for precision medicine. Chemical reviews, 123(8), 5049–5138 (2023).
+25. Yang, G., Hong, J., and Park, S. B. Wearable device for continuous sweat lactate monitoring in sports: a narrative review. Frontiers in Physiology, 15, 1376801 (2024).
+26. Komkova, M. A., Eliseev, A. A., Poyarkov, A. A., Daboss, E. V., Evdokimov, P. V., Eliseev, A. A., & Karyakin, A. A. Simultaneous monitoring of sweat lactate content and sweat secretion rate by wearable remote biosensors. Biosensors and Bioelectronics, 202, 113970 (2022).
+27. Xuan, X., Chen, C., Molinero-Fernandez, A., Ekelund, E., Cardinale, D., Swarén, M., … and Crespo, G. A. Fully integrated wearable device for continuous sweat lactate monitoring in sports. ACS sensors, 8(6), 2401–2409 (2023).
+28. Lynch, C., Adams, S. Branched-chain amino acids in metabolic signaling and insulin resistance. Nature Reviews Endocrinology 10, 723–736 (2014).
+29. Mayers, J., Wu, C., Clish, C. et al. Elevation of circulating branched-chain amino acids is an early event in human pancreatic adenocarcinoma development. Nature Medicine 20, 1193–1198 (2014).
+30. M. Wang, Y. Yang, J. Min, Y. Song, J. Tu, D. Mukasa, … and W. Gao A wearable electrochemical biosensor for the monitoring of metabolites and nutrients. Nature Biomedical Engineering, 6(11), 1225–1235 (2022).
+31. Yang, Y., Song, Y., Bo, X. et al. A laser-engraved wearable sensor for sensitive detection of uric acid and tyrosine in sweat. Nature Biotechnology 38, 217–224 (2020).
+32. Russell, E., Koren, G., Rieder, M., and Van Uum, S. H. The detection of cortisol in human sweat: implications for measurement of cortisol in hair. Therapeutic drug monitoring, 36(1), 30–34 (2014).
+33. Parlak, O., Keene, S. T., Marais, A., Curto, V. F., and Salleo, A. Molecularly selective nanoporous membrane-based wearable organic electrochemical device for noninvasive cortisol sensing. Science advances, 4(7), eaar2904 (2018).
+34. Tang, W., Yin, L., Sempionatto, J. R., Moon, J. M., Teymourian, H., and Wang, J. Touch-based stressless cortisol sensing. Advanced Materials, 33(18), 2008465 (2021).
+35. Parrilla, M., Guinovart, T., Ferré, J., Blondeau, P., and Andrade, F. J. A wearable paper-based sweat sensor for human perspiration monitoring. Advanced Healthcare Materials, 8(16), 1900342 (2019).
+36. Nagamine, K., Mano, T., Nomura, A., Ichimura, Y., Izawa, R., Furusawa, H., … and Tokito, S. Noninvasive sweat-lactate biosensor emplsoying a hydrogel-based touch pad. Scientific reports, 9(1), 10102 (2019).
+37. Zhao, F. J., Bonmarin, M., Chen, Z. C., Larson, M., Fay, D., Runnoe, D., and Heikenfeld, J. Ultra-simple wearable local sweat volume monitoring patch based on swellable hydrogels. Lab on a Chip, 20(1), 168–174 (2020).
+38. Nyein, H.Y.Y., Bariya, M., Tran, B. et al. A wearable patch for continuous analysis of thermoregulatory sweat at rest. Nature Communications 12, 1823 (2021).
+39. Reeder, J.T., Xue, Y., Franklin, D. et al. Resettable skin interfaced microfluidic sweat collection devices with chemesthetic hydration feedback. Nature Communications, 10, 5513 (2019).
+40. Reeder, J. T., Choi, J., Xue, Y., Gutruf, P., Hanson, J., Liu, M., … and Rogers, J. A. Waterproof, electronics-enabled, epidermal microfluidic devices for sweat collection, biomarker analysis, and thermography in aquatic settings. Science Advances, 5(1), eaau6356 (2019).
+41. Gao, W., Emaminejad, S., Nyein, H. et al. Fully integrated wearable sensor arrays for multiplexed in situ perspiration analysis. Nature 529, 509–514 (2016).
+42. S. Emaminejad, W. Gao, W. Wu, Z. A. Davies, H. Yin Yin Nyein, S. Challa, … and R. W. Davis Autonomous sweat extraction and analysis applied to cystic fibrosis and glucose monitoring using a fully integrated wearable platform. Proceedings of the National Academy of sciences, 114(18), 4625–4630 (2017).
+43. R. T. Ray, M. Ivanovic, P. M. Curtis, D. Franklin, K. Guventurk, W. J. Jeang, … and J. A. Rogers Soft, skin-interfaced sweat stickers for cystic fibrosis diagnosis and management. Science translational medicine, 13(587), eabd8109 (2021).
+44. Koh, A., Kang, D., Xue, Y., Lee, S., Pielak, R. M., Kim, J., … and Rogers, J. A. A soft, wearable microfluidic device for the capture, storage, and colorimetric sensing of sweat. Science translational medicine, 8(366), 366ra165-366ra165 (2016).
+45. Kim, J., Wu, Y., Luan, H., Yang, D. S., Cho, D., Kwak, S. S., … and Rogers, J. A. A skin-interfaced, miniaturized microfluidic analysis and delivery system for colorimetric measurements of nutrients in sweat and supply of vitamins through the skin. Advanced Science, 9(2), 2103331 (2022).
+46. Wang, Y., Zhao, C., Wang, J., Luo, X., Xie, L., Zhan, S., … and Ying, Y. Wearable plasmonic-metasurface sensor for noninvasive and universal molecular fingerprint detection on biointerfaces. Science Advances, 7(4), eabe4553 (2021).
+47. Bi, X., Czajkowsky, D.M., Shao, Z. et al. Digital colloid-enhanced Raman spectroscopy by single-molecule counting. Nature (2024).
+48. Yu, W., Li, Q., Ren, J., Feng, K., Gong, J., Li, Z., … and Yang, L. A sensor platform based on SERS detection/janus textile for sweat glucose and lactate analysis toward portable monitoring of wellness status. Biosensors and Bioelectronics, 263, 116612 (2024).
+49. Ma, H., Cui, Q., Xu, L., Tian, Y., Jiao, A., Wang, C., … and Chen, M. Silk fibroin fibers decorated with urchin-like Au/Ag nanoalloys: a flexible hygroscopic SERS sensor for monitoring of folic acid in human sweat. Optics Express, 29(19), 30892–30904 (2021).
+50. Zhang, X., Wang, X., Ning, M., Wang, P., Wang, W., Zhang, X., … and Li, S. Fast synthesis of Au nanoparticles on metal–phenolic network for sweat SERS analysis. Nanomaterials, 12(17), 2977 (2022).
+51. Koh, E. H., Lee, W. C., Choi, Y. J., Moon, J. I., Jang, J., Park, S. G., … and Jung, H. S. A wearable surface-enhanced Raman scattering sensor for label-free molecular detection. ACS Applied Materials and Interfaces, 13(2), 3024–3032 (2021).
+52. Chen, Z., Wang, W., Tian, H., Yu, W., Niu, Y., Zheng, X., … and Huang, Y. Wearable intelligent sweat platform for SERS-AI diagnosis of gout. Lab on a Chip, 24(7), 1996–2004 (2024).
+53. Ye, H., Chen, X., Huang, X., Li, C., Yin, X., Zhao, W., and Wang, T. Patterned Gold Nanoparticle Superlattice Film for Wearable Sweat Sensors. Nano Letters, 24(35), 11082–11089 (2024).
+54. Xiong, S., Wang, C., Zhu, C., Dong, P., and Wu, X. Dual Detection of Urea and Glucose in Sweat Using a Portable Microfluidic SERS Sensor with Silver Nano-Tripods and 1D-CNN Model Analysis. ACS Applied Materials & Interfaces (2024).
+55. Hu, M., Zhu, K., Wei, J., Yang, K., Wu, L., Zong, S., and Wang, Z. Silk fibroin-based wearable SERS biosensor for simultaneous sweat monitoring of creatinine and uric acid. Biosensors and Bioelectronics, 265, 116662 (2024).
+56. Guan, P. C., Qi, Q. J., Wang, Y. Q., Lin, J. S., Zhang, Y. J., and Li, J. F. Development of a 3D Hydrogel SERS Chip for Noninvasive, Real-Time pH and Glucose Monitoring in Sweat. ACS Applied Materials & Interfaces, 16(36), 48139–48146 (2024).
+57. Zhang, H., Zhang, H., Sikdar, D., Liu, X., Yang, Z., Cheng, W., and Chen, Y. Jellyfish-like Gold Nanowires as FlexoSERS Sensors for Sweat Analysis. Nano Letters, 24(36), 11269–11278 (2024).
+58. He, X., Fan, C., Luo, Y., Xu, T., and Zhang, X. Flexible microfluidic nanoplasmonic sensors for refreshable and portable recognition of sweat biochemical fingerprint. npj Flexible Electronics, 6(1), 60 (2022).
+59. Mogera, U., Guo, H., Namkoong, M., Rahman, M. S., Nguyen, T., and Tian, L. Wearable plasmonic paper–based microfluidics for continuous sweat analysis. Science advances, 8(12), eabn1736 (2022).
+60. Li, G., Zhao, X., Tang, X., Yao, L., Li, W., Wang, J.,and Hao, Q. Wearable Hydrogel SERS Chip Utilizing Plasmonic Trimers for Uric Acid Analysis in Sweat. Nano Letters, 24(42), 13447–13454 (2024).
+61. Xiao, J., Wang, J., Luo, Y., Xu, T., and Zhang, X. Wearable plasmonic sweat biosensor for acetaminophen drug monitoring. ACS sensors, 8(4), 1766–1773 (2023).
+62. Jain, V., Ochoa, M., Jiang, H. et al. A mass-customizable dermal patch with discrete colorimetric indicators for personalized sweat rate quantification. Microsystems and Nanoengineering 5, 29 (2019).
+63. Cheng, Y., Feng, S., Ning, Q. et al. Dual-signal readout paper-based wearable biosensor with a 3D origami structure for multiplexed analyte detection in sweat. Microsystems and Nanoengineering 9, 36 (2023).
+64. Li, Y., Guo, Y., Chen, H., Xiao, X., Long, F., Zhong, H., … and Liu, Z. Flexible Wearable Plasmonic Paper-Based Microfluidics with Expandable Channel and Adjustable Flow Rate for Portable Surface-Enhanced Raman Scattering Sweat Sensing. ACS Photonics (2024).
+65. Choi, J., Kang, D., Han, S., Kim, S. B., and Rogers, J. A. Thin, soft, skin-mounted microfluidic networks with capillary burst valves for chrono‐sampling of sweat. Advanced healthcare materials, 6(5), 1601355 (2017).
+66. Wu, C. H., Ma, H. J. H., Baessler, P., Balanay, R. K., and Ray, T. R. Skin-interfaced microfluidic systems with spatially engineered 3D fluidics for sweat capture and analysis. Science Advances, 9(18), eadg4272 (2023).
+67. Wu, Y., Kanatzidis, E. E., Avila, R., Zhou, M., Bai, Y., Chen, S., … and Rogers, J. A. 3D-printed epidermal sweat microfluidic systems with integrated microcuvettes for precise spectroscopic and fluorometric biochemical assays. Materials Horizons, 10(11), 4992–5003 (2023).
+68. Lee, S., Kim, H., and Lee, D. Linearization autoencoder: an autoencoder-based regression model with latent space linearization. bioRxiv, 2022–06 (2022).
+69. Lundberg, S. M., Erion, G., Chen, H., DeGrave, A., Prutkin, J. M., Nair, B., … and Lee, S. I. From local explanations to global understanding with explainable AI for trees. Nature machine intelligence, 2(1), 56–67 (2020).
+70. Chen, H., Covert, I. C., Lundberg, S. M., & Lee, S. I. Algorithms to estimate Shapley value feature attributions. Nature Machine Intelligence, 5(6), 590–601 (2023).
+71. Zhong, B., Qin, X., Xu, H., Liu, L., Li, L., Li, Z., … and Wang, L. Interindividual-and blood-correlated sweat phenylalanine multimodal analytical biochips for tracking exercise metabolism. Nature Communications, 15(1), 624 (2024).
+72. Derbyshire, P. J., Barr, H., Davis, F., and Higson, S. P. Lactate in human sweat: a critical review of research to the present day. The journal of physiological sciences, 62, 429–440 (2012).
+73. Xuan, X., Perez-Rafols, C., Chen, C., Cuartero, M., and Crespo, G. A. Lactate biosensing for reliable on-body sweat analysis. ACS sensors, 6(7), 2763–2771 (2021).
+74. Karpova, E. V., Laptev, A. I., Andreev, E. A., Karyakina, E. E., and Karyakin, A. A. Relationship between sweat and blood lactate levels during exhaustive physical exercise. ChemElectroChem, 7(1), 191–194 (2020).
+75. Fernstrom, M. H. and Fernstrom, J. D. Protein consumption increases tyrosine concentration and in vivo tyrosine hydroxylation rate in the light-adapted rat retina. Brain research, 401(2), 392–396 (1987).
+
+# Supplementary Files
+
+- [SupplementaryVideo1.mp4](https://assets-eu.researchsquare.com/files/rs-5624954/v1/bcf57a48282f214fb62dfe51.mp4)  
+  Supplementary video 1
+
+- [SupplementaryInformationKHJeongXKAIST.docx](https://assets-eu.researchsquare.com/files/rs-5624954/v1/21e55f8070f7b828cda87db6.docx)  
+  Supplementary information updated

055c635ef5ff7a481cce5739a3e68be679d4efd18c3744431eaf8e18a25f3f77/preprint/images_list.json

@@ -0,0 +1,34 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Validation of pgSIT target genes. Cas9/gRNA-mediated disruption of \u03b2Tub or myo-fem results in (A-C) male (\u2642) sterility or (D-E) female (\u2640) flightlessness, respectively. Schematics of genetic crosses to assess the efficiency of (A) \u03b2Tub or (D) myo-fem disruption in the F1 transheterozygous progeny. (B) Histogram indicating the percent of fertile progeny for each of the various progeny genotypes using gRNA\u03b2Tub#7 line. (C) Imaging of seminal fluid from WT and gRNA\u03b2Tub#7/+;Cas9/+ mosquitoes, showing the difference in spermatid elongation caused by the disruption of \u03b2Tub (Video S1). (E) Histogram showing percent of fertile and flight-capable mosquitoes in each cross using gRNAmyo-fem#1 line. (Video S1-S3). (F) Imaging showing the specific wing posture phenotype induced by the myo-fem disruption in females, but not in males, in which the resting wings were uplifted. Data from both paternal Cas9 crosses (Cas9\u2642 \ud835\uddd1 gRNA\u2640) and maternal Cas9 crosses (Cas9\u2640 \ud835\uddd1 gRNA\u2642) are shown (Fig. S2, S3, Table S3). Bar plots show means \u00b1 one standard deviation (SD), biological replicates, and mean and SD values rounded to a whole number. Statistical significance was estimated using a two-sided Student\u2019s t test with unequal variance. (p \u2265 0.05ns,  p < 0.05*, p < 0.01**, and p < 0.001***). ",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Genetic characterization of pgSIT. (A) The pgSIT cross between transhomozygous gRNA \u2642\u2019s harboring both gRNA\u03b2Tub#7 and gRNAmyo-fem#1 (termed: gRNA\u03b2Tub+myo-fem) and the homozygous Cas9. The pgSIT cross was initiated reciprocally to generate F1 transheterozygous progeny carrying either maternal or paternal Cas9. (B) Histogram comparing the survival and fitness of transheterozygous and heterozygous Cas9 or gRNA progeny to those of WT (Table S4, Video S4).(C) Experimental set-up to determine whether prior matings with pgSIT\u2642\u2019s suppresses WT \u2640 fertility. WT \u2640\u2019s were cohabitated with pgSIT\u2642\u2019s for 2, 6, 12, 24, or 48 hours then WT \u2640\u2019s were transferred to a new cage along with WT \u2642\u2019s and mated for an additional 2 days. The \u2640\u2019s were then blood fed and individually transferred to a vial. Eggs were collected and hatched for fertility determination. Following this, non-fertile \u2640\u2019s were then placed back into cages along WT \u2642\u2019s for another chance to produce progeny. This was repeated for up to five gonotrophic cycles, and the percentage of fertile \u2640\u2019s in each group of 50 \u2640\u2019s was plotted (Table S8). (D) Flight activity of individual mosquitoes including was assessed for 24 hours using a vertical Drosophila Activity Monitoring (DAM) System, which uses an infrared beam to record flight (Table S6, Video S5). (E) To quantify the attractiveness of \u2642\u2019s to \u2640\u2019s for mating, we used a mating-behavior lure of a tone mimicking \u2640 flight. A 10-second 600 Hz sine tone was applied on one side of the cage, and a number of mosquito \u2642\u2019s landing on the mesh around a speaker was scored. Heatmaps were generated using Noldus Ethovision XT. (Table S7, Video S6). Plots show biological replicates and means \u00b1 SDs. Statistical significance was estimated using a two-sided Student\u2019s t test with unequal variance. (p \u2265 0.05ns,  p < 0.05*, p < 0.01**, and p < 0.001***). ",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Multigenerational cage trials demonstrating efficient population suppression. (A) To generate sufficient mosquitoes, three lines were raised separately, including homozygous Cas9, transhomozygous gRNA\u03b2Tub+myo-fem, and WT. To generate pgSIT progeny, virgin Cas9 \u2640\u2019s were genetically crossed to gRNA\u03b2Tub+myo-fem \u2642\u2019s, and eggs were collected. (B) To perform multigenerational population cage trials of pgSIT, two strategies were employed: release of eggs (B, top panel); release mature adults (B, bottom panel).  For both strategies, multiple pgSIT:WT release ratios were tested, including: 1:1, 5:1, 10:1, 20:1, and 40:1. Each generation, total eggs were counted, and 100 eggs were selected randomly to seed the subsequent generation. The remaining eggs were hatched to measure hatching rates and score transgene markers. This procedure was repeated after each generation until each population was eliminated (Table S7). (C) Multigeneration population cage data for each release threshold plotting the proportion of eggs hatched each generation. ",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Model-predicted impact of releases of pgSIT eggs on Ae. aegypti population density and elimination. (A) Releases were simulated on the island of Onetahi, Tetiaroa, French Polynesia, a field site for releases of Wolbachia-infected \u2642 mosquitoes, using the MGDrivE simulation framework (19) and parameters described in Table S17. Human structures are depicted and were modeled as having an equilibrium population of 16 adult Ae. aegypti each. (B) Weekly releases of up to 400 pgSIT eggs per wild Ae. aegypti were simulated in each human structure over 10-24 weeks. The pgSIT construct was conservatively assumed to decrease male mating competitiveness by 25% and adult lifespan by 25%. Elimination probability was calculated as the percentage of 200 stochastic simulations that resulted in local Ae. aegypti elimination for each parameter set. Sample time-series depicting female Ae. aegypti population density are depicted above and below the heatmap. (C) Elimination probability (given 18 weekly releases of 200 pgSIT eggs per wild Ae. aegypti) is depicted for a range of pgSIT \u2642 fitness profiles. Elimination is possible for a wide range of reductions in male mating competitiveness (0-50%) and adult lifespan (0-50%) for an achievable release scheme.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

055c635ef5ff7a481cce5739a3e68be679d4efd18c3744431eaf8e18a25f3f77/preprint/preprint.md

@@ -0,0 +1,228 @@
+# Abstract
+
+The mosquito *Aedes aegypti* is the principal vector for arboviruses including dengue/yellow fever, chikungunya, and zika, infecting hundreds of millions of people annually. Unfortunately, traditional control methodologies are insufficient, so innovative control methods are needed. To complement existing measures, here we develop a molecular genetic control system termed precision guided sterile insect technique (pgSIT) in *Aedes aegypti*. PgSIT uses a simple CRISPR-based approach to generate sterile males that are deployable at any life stage. Supported by mathematical models, we empirically demonstrate that released pgSIT males can compete, suppress, and eliminate mosquitoes in multigenerational population cages. This platform technology could be used in the field, and adapted to many vectors, for controlling wild populations to curtail disease in a safe, confinable, and reversible manner.
+
+Biotechnology and Bioengineering  
+Molecular Genetics  
+pgSIT  
+Aedes aegypti  
+CRISPR/Cas9  
+dengue fever  
+yellow fever  
+zika  
+chikungunya  
+vector control  
+population elimination
+
+# Introduction
+
+Mosquitoes are the world's deadliest animals, killing more humans than any other animal (1) due to transmitting the majority of vector-borne diseases, such as the notorious arboviruses dengue, Zika, yellow fever, and chikungunya transmitted by *Aedes* mosquitoes. The predominating strategy to control these devastating diseases is the use of insecticides, though mosquitoes are evolving and spreading insecticide resistance (2), hampering control efforts. Therefore, there is an urgent demand for innovative mosquito-control technologies that are effective, sustainable, and safe.
+
+Alongside traditional control measures, several genetic-based techniques are being used to combat mosquitoes. These include multiple male (♂) release programs aimed at population suppression, such as the classical radiation-based sterile insect technique (SIT), relying on releasing irradiated sterile ♂’s (3). Alternative approaches include the *Wolbachia*-based incompatible insect technique (IIT), relying on the release of *Wolbachia* infected ♂’s (4, 5), or the antibiotic-based Release of Insects carrying a Dominant Lethal (RIDL) (6). Moreover, emerging CRISPR-based homing gene drives that spread target genes through a population faster than through traditional Mendelian inheritance are presently under development with the aim of safe implementation in the future (7, 8).
+
+As an alternative, a CRISPR-based technology termed precision-guided SIT (pgSIT), was recently developed in flies (9). pgSIT uses a binary approach to simultaneously disrupt genes essential for female (♀) viability and ♂ fertility, resulting in the exclusive survival of sterile ♂’s that can be deployed at any life stage to suppress populations. It requires two breeding strains, one expressing Cas9 and the other expressing guide RNAs (gRNAs). Mating between these strains results in RNA-guided mosaic target gene mutations throughout development, ensuring complete penetrance of desired phenotypes. Compared to alternatives, pgSIT does not require the use of radiation, *Wolbachia*, nor antibiotics, and will not persist in the environment. Unfortunately, this technology is presently only accessible in flies, and for population control techniques, its equivalent needs to be developed for mosquitoes.
+
+To address this, here we systematically engineer pgSIT in *Ae. Aegypti* using a system that simultaneously disrupts genes essential for ♂ fertility and ♀ flight, which is necessary for mating, blood feeding, reproduction, and predator avoidance—meaning survival in general (10). Using our technology, we demonstrate resulting progeny of flightless ♀’s and fit sterile ♂’s that can compete, suppress, and eliminate mosquito populations in multigenerational population cages. Mathematical models suggest that releases of *Ae. aegypti* pgSIT eggs could effectively eliminate a local *Ae. aegypti* population using achievable release schemes. Taken together, this study suggests pgSIT may be an efficient technology for mosquito population control and the first example of one suited for real-world release.
+
+# Results
+
+## Validation of pgSIT target genes
+
+To engineer pgSIT in Ae. aegypti, we first validated target genes by generating transgenic gRNA-expressing lines targeting two conserved genes: β-Tubulin 85D (βTub, AAEL019894), specifically expressed in mosquito testes (11–13) and essential for spermatogenesis and ♂ fertility (14), and myosin heavy chain (myo-fem, AAEL005656), expressed nearly exclusively in ♀ pupae (11, 12) and essential for ♀ flight (15) (Fig. S1, Table S1). To ensure efficient disruption, each gRNA line encoded four U6–promoter-driven (16) gRNAs targeting unique sites in the coding sequence of either βTub (U6-gRNAβTub - marked with 3xP3-GFP) or myo-fem (U6-gRNAmyo−fem - marked with 3xP3-tdTomato) (Fig. S2-4). Multiple independent transgenic lines were generated, and to assess their activity, we conducted bidirectional crosses with Cas9 controlled by a homozygous nuclear pore complex protein (Cas9 - marked with Opie2-CFP) (17) (Table S2, S3, Fig. S2-4). The resulting transheterozygous F1 progeny (gRNA/+, Cas9/+) were assessed and crossed to wildtype (WT) for further evaluation. For the βTub crosses, fertility of the F1 transheterozygous ♂’s ranged from 0–94.9%, with two lines achieving 100% sterility from immotile sperm (14), while F1 transheterozygous ♀’s maintained normal fertility (Fig. 1, S2, Table S3, Video S1). For myo-fem crosses, all F1 transheterozygous ♀’s generated from ⅗ lines were flightless, while F1 transheterozygous ♂’s maintained normal flight (Fig. 1, Table S3, Fig. S3, Video S2,3). As expected, ♀ flightlessness significantly reduced mating ability and blood consumption as many get trapped on the water surface following eclosion, resulting in reduced fecundity, fertility, and survival. Sanger sequencing of genomic DNA revealed expected mutations at the βTub- and myo-fem-targeted loci.
+
+## Development of pgSIT and fitness assessments
+
+To generate a pgSIT strain capable of targeting both βTub and myo-fem simultaneously, we combined two gRNA lines that exclusively produced sterile ♂’s (gRNAβTub #7) or flightless ♀’s (gRNAmyo−fem #1) (Fig. 1, Fig. S2-S3, Table S3) by repeated introgression, generating a trans-homozygous stock (termed gRNAβTub + myo−fem) (Fig. S4). To assess its activity, we bidirectionally crossed gRNAβTub + myo−fem to Cas9. Importantly, these crosses yielded all flightless ♀’s (termed pgSIT♀) and sterile ♂’s (termed pgSIT♂) with normal flight and mating capacity (Fig. 2, S5, Table S4-8, Video S4-6). We next determined transgene integration sites, single copy number per transgene, and confirmed target gene disruptions by both amplicon sequencing (Fig. S6) and Nanopore genome sequencing using transheterozygous pgSIT♂s (Fig. S7-S9, Table S9-10). We also performed transcriptome sequencing of pupae comparing pgSIT♂s and pgSIT♀s to WT to quantify target gene reduction, expression from transgenes, and to assess global expression patterns (Fig. S8-S10, Table S11-S15). As expected, we observed significant target gene disruption in pgSIT individuals, robust expression from our transgenes, and non-target gene misexpression, which would be expected given the significant phenotypes observed (i.e. flightless females and spermless males).
+
+To explore potential fitness effects, we assayed several fitness parameters including ♀ fecundity, fertility, flight activity, ♂ mating capacity, ♂ sound attraction, larva-pupa development time, pupa-adult development time, and longevity (Fig. 2, S5, Table S5-8, Video S5-6). The pgSIT♀’s were flightless with significantly reduced fecundity, fertility, and survival, indicating they would be very unlikely to survive in the wild, let alone transmit pathogens. For pgSIT♂’s, other than slightly delayed larva-pupa development time, we did not detect significant differences in fitness parameters. Previous studies demonstrated that Ae. aegypti ♀’s typically mate only once in their lifetime, a behavior known as monandry (18). To explore whether prior matings with pgSIT♂’s could suppress ♀ fertility, we initiated experiments in which WT ♀’s were first mated with pgSIT♂’s for a period of time (2, 6, 12, 24, or 48 hrs) followed by WT ♂’s (48 hrs). Fertility was measured for up to five gonotrophic cycles. We found that prior exposure to pgSIT♂’s ensured long lasting reductions in ♀ fertility, spanning 5 gonotrophic cycles, with longer exposures (24 and 48 hrs) resulting in near complete suppression of ♀ fertility (Fig. 2, Table S8).
+
+## pgSIT induced population suppression
+
+To assess whether pgSIT ♂’s could compete and suppress populations, we conducted discrete, multi-generational, population cage experiments by repeatedly releasing either eggs or adult ♂’s each generation, using several introduction frequencies (pgSIT:WT − 1:1, 5:1, 10:1, 20:1, and 40:1) (Fig. 3, Table S16). To measure efficacy each generation, we counted the total number of eggs laid and hatched and confirmed the lack of presence of marker genes in hatched larvae indicating released pgSIT ♂’s were indeed sterile. Adult releases at high release thresholds (20:1, 40:1) eliminated all populations by generation 3, and at lower release thresholds (10:1), we saw elimination by generation 6. For the egg releases, elimination was achieved by generation 6 for 4/6 populations at high release thresholds (20:1, 40:1).
+
+## Theoretical performance of pgSIT in a wild population
+
+To explore the potential for pgSIT ♂’s to suppress Ae. aegypti populations in the wild, we simulated releases of pgSIT eggs on the island of Onetahi, Tetiaroa, French Polynesia (Fig. 4), a field site for releases of Wolbachia-infected ♂ mosquitoes, using the MGDrivE simulation framework (19). Weekly releases of up to 400 pgSIT eggs per wild adult were simulated in each human structure over 10–24 weeks. The scale of these releases was chosen considering adult release ratios of 10:1 are common for sterile male mosquito interventions (6) and female Ae. aegypti produce > 30 eggs per day in temperate climates (20). We also assumed 25% reductions in male mating competitiveness and adult lifespan for pgSIT males by default because, although pgSIT fitness effects were not apparent from laboratory experiments, they may become apparent in the field. Results from these simulations suggest that significant population suppression (> 96%) is seen for a wide range of achievable release schemes, including 13 weekly releases of 120 or more pgSIT eggs per wild adult (Fig. 4, Video S7). Population elimination was common for larger yet achievable release schemes, including 18 weekly releases of 200 or more pgSIT eggs per wild adult, and 24 weekly releases of 100 or more pgSIT eggs per wild adult. Results also suggest a wider range of pgSIT fitness profiles (e.g. a 50% reduction in male mating competitiveness and 25% adult lifespan reduction) could lead to population elimination for these release schemes (Fig. 4).
+
+# Discussion
+
+While many technologies for halting the spread of deadly mosquito-borne pathogens exist, none are without significant drawbacks such that additional measures are needed. By disrupting essential genes throughout development, we demonstrate efficient production of short-lived, flightless pgSIT ♀’s and fit sterile pgSIT ♂’s. Importantly, when repeatedly released into caged populations, the pgSIT ♂’s competed with WT ♂’s thereby suppressing, and even eliminating, populations using release ratios that are achievable in the field (4–6). Mathematical models suggest that population elimination could be accomplished in the field through sustained releases of ~100–200 or more pgSIT eggs per wild Ae. aegypti adult, even if fitness costs significantly exceed those measured in contained laboratory experiments.
+
+For pgSIT to be realized in the wild, the two strains will first need to be separately and continuously mass-reared in a facility, without contamination, and crossed to produce sterile ♂’s. While this can be viewed as rate-limiting (21), it offers stability, as the binary CRISPR system will remain inactive until crossed—thereby reducing the evolution of suppressors or mutations that could disrupt the system. Additionally, each sorted ♀ can produce up to 450 eggs in her lifetime (22), which improves scalability. Moreover, once crossed, the resulting progeny are essentially dead-ends (i.e. sterile ♂’s /flightless ♀’s), hatched among high numbers of sterile pgSIT ♂’s, and should not contribute to the gene pool (23). We demonstrate here that the technology is fully penetrant by screening >100K individuals.
+
+pgSIT offers an alternative approach to scalability that should help decrease costs and increase efficiency. For instance, the required genetic cross at scale can be initiated using existing robotic sex sorting devices (www.senecio-robotics.com) or (5). Upon sex sorting and crossing, the resulting pgSIT progeny can be distributed and released at any life stage, mitigating requirements for sex separation at field sites. This strategy will be especially effective for mosquitoes that diapause during the egg stage (e.g. Aedes species) because it will enable long-term egg accumulation. Eggs could be distributed to logistically spaced remote field sites where they can hatch, develop, and compete with wild mosquitoes (Fig. S11). This attractive feature should reduce the costs of developing multiple production facilities requiring on-site sex separation for manual release of fragile adults.
+
+It should be noted that the releases of adult pgSIT ♂’s unexpectedly resulted in faster population suppression as compared to egg releases in multigenerational population cage experiments. We believe this to result from the slightly reduced egg hatching rates of pgSIT ♂’s and their delayed larva-pupa development time, which likely enabled the co-released WT ♂’s first access to WT ♀’s. While this could impact the discrete generation population cage experiments conducted here, it should not be problematic for suppressing continuous populations in the wild.
+
+Finally, notwithstanding its inherently safe nature, pgSIT requires genetic modification, and regulatory use authorizations will need to be granted prior to implementation. While this could be viewed as a limitation (21), we don’t expect obtaining such authorizations to be insurmountable. In fact, we envision pgSIT to be regulated in a similar manner to Oxitec’s RIDL technology, which has been successfully deployed in many locations and recently received experimental use authorizations in the USA.
+
+Overall, the inherent self-limiting nature of pgSIT, offers a controllable safe alternative to technologies that can persist and spread in the environment, such as gene drives (8). Going forward, pgSIT may provide an efficient, safe, scalable, and environmentally friendly alternative next-generation technology for wild population control of mosquitoes resulting in wide-scale prevention of human disease transmission.
+
+## Supplementary materials
+
+### Mosquito rearing and maintenance
+
+Ae. aegypti mosquitoes were derived from the Liverpool strain (wildtype [WT]) previously used to generate the reference genome (24). Mosquitoes were raised in incubators at 27.0°C with 20–40% humidity and a 12-hour light/dark cycle in cages (Bugdorm, 24.5 cm × 24.5 cm × 24.5 cm). Adults were provided 0.3 M aqueous sucrose ad libitum, and ♀’s were blood fed on anesthetized mice for two consecutive days for ~15 minutes at a time. Oviposition substrates were provided ~3 days following the second blood meal. Eggs were collected and aged for ~4 days to allow for embryonic development, then were hatched in deionized H₂O in a vacuum chamber. Roughly ~400 larvae were reared in plastic containers (Sterilite, 34.6 cm × 21 cm × 12.4 cm, USA) with ~3 liters of deionized H₂O, and fed fish food (TetraMin Tropical Flakes, Tetra Werke, Melle, Germany). For genetic crosses, to ensure ♀ virginity, pupae were separated and sexed under the microscope by sex-specific morphological differences in the genital lobe shape (at the end of the pupal abdominal segments just below the paddles) before being released to eclose in cages. These general rearing procedures were followed unless otherwise noted. Mosquitoes were examined, scored, and imaged using the Leica M165FC fluorescent stereo microscope equipped with the Leica DMC2900 camera. For higher resolution images, we used a Leica DM4B upright microscope equipped with a VIEW4K camera enabling time lapse videos. Time lapse videos of caged adult mosquitoes were taken with a mounted Canon EOS 5D Mark IV using a 24–105mm image stabilizer ultrasonic lens.
+
+### Guide RNA design and testing
+
+Two target genes were selected for gRNA design: β-Tubulin 85D (βTub, AAEL019894) and myosin heavy chain (myo-fem, AAEL005656). For each target gene, DNA sequences were first identified using reference genome assembly (24), and genomic target sites were validated using PCR amplification and Sanger sequencing (Table S18 for primer sequences). Gene structures, transcripts, and exon-intron junction boundaries were carefully evaluated using comprehensive developmental transcriptome data (11, 24) loaded into an internal genome browser. Target gRNA sequences were selected to be 20 bp (N20) in length, excluding the PAM (NGG) (25). For in silico gRNA selection, we used either CHOPCHOP V3.0.0 (https://chopchop.cbu.uib.no) or CRISPOR (http://crispor.tefor.net) to minimize potential genomic off-target cleavage events. In total, we designed four gRNAs targeting βTub and four gRNAs targeting myo-fem (Table S18). To confirm gRNA activity in vivo, each gRNA was in vitro synthesized prior to construct design (Synthego, CA, USA). Then 100 ng/ul of gRNA was individually injected into fifty preblastoderm stage embryos (0.5–1 hr old) derived from Exu-Cas9 maternally depositing mothers, per previous embryo-injection protocols (16, 17). The surviving G0 progeny were pooled (2–5 individuals per pool), and genomic DNA was extracted using the DNeasy blood and tissue kit (Qiagen, Cat No./ID: 69506) following the manufacturer's protocols. To molecularly characterize the induced mutations, target loci were PCR amplified from extracted genomic DNA, and the PCR products were gel purified (Zymo Research, Zymoclean Gel DNA Recovery Kit, Cat No./ID: D4007). The purified products were either sent directly for sequencing or subcloned (Invitrogen, TOPO-TA, Cat No./ID: LS450641), wherein single colonies were selected and cultured in Laurel Broth (LB) with ampicillin before plasmid extraction (Zymo Research, Zyppy plasmid miniprep kit, Cat No./ID: D4036) and Sanger sequencing. Mutated alleles were identified in silico by alignment with WT target sequences. All primers used for PCR and sequencing, including gRNA target sequences, are listed in Table S18.
+
+### Construct molecular design and assembly
+
+The Gibson enzymatic assembly method was used to engineer all constructs in this study (26). To generate the Nup50-Cas9 construct marked with CFP, OA-874PA (Addgene #164846), we used our previous plasmid for Cas9 expression (Addgene plasmid #100608) as the backbone (17). The fragments of T2A-eGFP-P10-3’UTR and OpIE2-dsRed-SV40 were removed by cutting with restriction enzyme FseI. Then, the P10-3’UTR fragment was amplified from Addgene plasmid #100608 with primers 874-P10 and 777B. Another fragment, OpIE2-CFP-SV40, was synthesized using gBlocks® Gene Fragment service (Integrated DNA Technologies, Coralville, Iowa). Both fragments were provided for the Gibson assembly into the cut backbone. We designed two constructs, OA-1067A1 (Addgene #164847) and OA-1067K (Addgene #164848), each carrying four different gRNAs targeting either β-Tubulin 85D (βTub, AAEL019894) or myosin heavy chain (myo-fem, AAEL005656) genes.
+
+To engineer these plasmids, four intermediate plasmids, OA-1055A (gRNA βTub1&2), OA-1055B (gRNA βTub3&4), OA-1055W (gRNA myo−fem1&2), and OA-1055X (gRNA myo−fem3&4), each harboring two gRNAs, were generated by cutting a backbone plasmid OA-984 (Addgene plasmid #120363), which contains piggyBac elements and the 3xP3-tdTomato transformation marker, with the restriction enzymes AvrII and AscI. Two gBlocks® Gene Fragments were then cloned in, each containing two gRNAs: one driven by U6b (AAEL017774) and one by U6c (AAEL017763) promoters (17). To assemble the final plasmid OA-1067A1, an intermediate plasmid OA-1067A was generated by linearizing the plasmid OA-1055B with the restriction enzyme BglII and inserting in the fragment of U6b-gRNA βTub1-U6c-gRNA βTub2 amplified with primers 1167.C1 and 1067.C2 from plasmid OA-1055A. Then, the fragment of 3xP3-tdTomato was removed from plasmid OA-1067A using the restriction enzymes AscI and NotI and replaced with the 3xP3-eGFP transformation marker amplified with primers 1067A1.C1 and 1067A1.C2 from the plasmid OA-961B (Addgene plasmid #104967). To assemble the final plasmid OA-1067K, OA-1055W was linearized with the restriction enzyme FseI, and the insertion of U6b-gRNA myo−fem3-U6c-gRNA myo−fem4 was amplified with primers 1167.C5 and 1067.C6 from the plasmid OA-1055X. During each cloning step, single colonies were selected and cultured in LB medium with ampicillin, and then the plasmids were extracted (Zymo Research, Zyppy plasmid miniprep kit, Cat No./ID: D4036) and Sanger sequenced. Final plasmids were maxi-prepped using (Zymo Research, ZymoPURE II Plasmid Maxiprep kit, Cat No./ID: D4202) and Sanger sequenced. All primers are listed in Table S18. Complete plasmid sequences and plasmid DNA are available at www.addgene.com.
+
+### Generation of transgenic lines
+
+Transgenic lines were generated by microinjecting preblastoderm stage embryos (0.5–1 hr old) with a mixture of the piggybac plasmid (200 ng/ul) and a transposase helper plasmid (phsp-Pbac, (200 ng/ul). Embryonic collection and microinjections were performed following previously established procedures (17). After 4 days of development post-microinjection, G0 embryos were hatched in deionized H₂O in a vacuum chamber. Surviving G0 pupae were separated and sexed and divided into separate ♀ or ♂ cages (~20 cages total). The pupae eclosed inside these cages along with added WT ♂ pupae (added into the ♀ cages) or WT ♀ pupae (added into the ♂ cages) at 5:1 ratios (WT:G0). Several days post-eclosion (~4–7), enabling sufficient time for development and mating, a blood meal was provided, and eggs were collected, aged, then hatched. The hatched larvae with positive fluorescent markers were individually isolated using a fluorescent stereo microscope (Leica M165FC). To isolate separate insertion events, selected transformants were individually crossed to WT (5:1 ratios of WT:G1), and separate lines were established (Table S2). These were subjected to many generations of backcrosses to WT to isolate single insertion events. Each of these individual gRNA lines (OA-1067A1: gRNA βTub and OA-1067K: gRNA myo−fem) were maintained as mixtures of homozygotes and heterozygotes with periodic selective elimination of WTs. The Cas9 line (OA-874PA: Nup50-Cas9) was homozygosed by ~10 generations of single-pair sibling matings selecting individuals with the brightest expressing transformation markers. Homozygosity was confirmed genetically by repeated test crosses to WT.
+
+### Genetic testing of established lines
+
+To assess the activity of the transgenic lines generated, we performed a series of genetic crosses by releasing sexed pupae into cages. We first crossed gRNA lines (gRNA ♂ 𝗑 WT ♀) to generate heterozygotes. We next reciprocally crossed heterozygous gRNA βTub/+ (lines #1–10) and the heterozygous gRNA myo−fem/+ (lines #1–5), with homozygous Cas9 (1 ♂ 𝗑 10 ♀). To measure the fecundity, the resulting transheterozygous F₁ progeny (gRNA βTub/+; Cas9/+) or (gRNA myo−fem/+; Cas9/+), were reciprocally crossed to WT’s (50 ♂ 𝗑 50 ♀), keeping track of the grandparents genotypes (Fig. S2, S3, Table S3). Control crosses of: WT ♂ 𝗑 WT ♀; WT ♂ 𝗑 Cas9 ♀; Cas9 ♂ 𝗑 WT ♀; gRNA/+ ♂ 𝗑 Cas9 ♀; gRNA/+ ♀ 𝗑 Cas9 ♂; gRNA/+ ♀ 𝗑 WT ♂; and gRNA/+ ♂ 𝗑 WT ♀ were also set up for comparisons (50 ♂ 𝗑 50 ♀). Adults were allowed to mate in the cage for 4–5 days, then blood meals were provided, and eggs were collected and hatched. The percentage of egg hatching (i.e. fertility) was estimated by dividing the total number of eggs laid by the total number of hatched eggs. Larvae-to-adult survival rates were calculated by dividing the total number of adults that emerged by the total number of larvae. Pupae-adult survival rates were calculated by dividing the number of dead pupae by the total number of pupae. Flight capacity for each sex was calculated by dividing the total number that were flightless (observed by eye) by the total of number of adult mosquitoes of that sex. Blood acquisition rates were calculated by dividing the number of blood-fed ♀’s by the total number of ♀’s. To investigate ♂ internal anatomical features, testes and ♂ accessory glands (n=20) were dissected in 1% PBS buffer for imaging.
+
+### Generation and characterization of gRNA βTub + myo−fem
+
+To generate gRNA βTub + myo−fem, we genetically crossed gRNA βTub#7 (marked with 3xp3-GFP) with gRNA myo−fem#1 (marked with 3xp3-tdTomato). Resulting F1 transheterozygotes gRNA βTub#7/+ ; gRNA myo−fem#1/+ were subjected to multiple generations of single-pair sibling matings, carefully selecting individuals with the brightest expressing transformation markers, to generate a transhomozygous stock (termed: gRNA βTub + myo−fem). Zygosity was confirmed genetically by repeated test crosses to WT. To measure efficacy, we bidirectionally crossed gRNA βTub + myo−fem with Cas9 (50 ♂ 𝗑 50 ♀), generating F1 transheterozygotes gRNA βTub + myo−fem/+ ; Cas9/+. Control crosses were also setup for comparisons: gRNA βTub + myo−fem ♂ 𝗑 gRNA βTub + myo−fem ♀; gRNA βTub + myo−fem ♂ 𝗑 WT ♀; gRNA βTub + myo−fem ♀ 𝗑 WT ♂; Cas9 ♂ 𝗑 Cas9 ♀; Cas9 ♂ 𝗑 WT ♀; and Cas9 ♀ 𝗑 WT ♂; (50 ♂ 𝗑 50 ♀). To determine the fecundity and fertility, resulting transheterozygous F1’s (~3 days old) were bidirectionally crossed to WT’s (50 ♂ 𝗑 50 ♀; 10 replicates each). These were allowed to mate for ~2 days and then blood fed. Afterwards, eggs were collected for up to five consecutive gonotrophic cycles and hatched.
+
+### Determination of transgene integration sites and copy number
+
+To determine the transgene insertion site(s) and copy number(s), we performed Oxford Nanopore DNA sequencing. We extracted genomic DNA using the Blood & Cell Culture DNA Midi Kit (Qiagen, Cat# 13343) from twenty adult transheterozygous pgSIT ♂’s (3 days old) harboring all three transgenes (Cas9/+ ; gRNA βTub#7/+ ; gRNA myo−fem#1/+), following the manufacturer’s protocol. The sequencing library was prepared using the Oxford Nanopore SQK-LSK109 genomic library kit and sequenced on a single MinION flowcell (R9.4.1) for 72 hrs to generate an N50 read length for the set of 4088 bp. Basecalling was performed using ONT Guppy basecalling software version 4.4.1, generating 2.94 million reads above quality threshold Q ≥ 7, which corresponds to 8.68 Gb of sequence data. To determine transgene copy number(s), reads were mapped to the AaegL5.0 reference genome (24) supplemented with transgene sequences (OA-1067A1: gRNA βTub; OA-1067K: gRNA myo−fem; and OA-874PA: Nup50-Cas9) using minimap2 (27). In total, 2,862,171 out of 2,936,275 reads (97.48%) were successfully mapped with a global genome-wide depth of coverage of 5.495. We calculated the mean coverage depth for all contigs in the genome (2310) and the three plasmids (OA-1067A1: gRNA βTub; OA-1067K: gRNA myo−fem; and OA-874PA: Nup50-Cas9) as well as normalized coverage (Table S9-S10). Transgene coverage ranged from 5.1 to 7.6, and normalized coverage ranged from 0.93 to 1.38. As compared to the three chromosomes, the coverages are consistent with the transgenes present at a single copy (Fig. S7).
+
+To identify transgene insertion sites, we inspected reads that aligned to the transgenes in the Interactive Genomics Viewer (IGV) browser. The reads extending beyond the boundaries of the transgenes were then analyzed to determine mapping sites within the genome. For OA-874PA, one read spanned the whole transgene (~11.5 kb) and extended 4 and 3.5 kb on both sides. The extending portions mapped to both sides of the position on NC_035109.1:33,210,105 (chromosome 3), with the nearest gene being AAEL023567, which is ~5 kb away. For OA-1067K, one read covered ~7 kb of the transgene extending ~10 kb off the 3' end, 9 kb of which map to the NC_035108.1:287,686 − 296,810 region (chromosome 2). A few other shorter reads map to the same location. The site is located in the intron of AAEL005206, which is a capon-like protein, and based on the RNA-seq data, it's expression does not appear to be affected in pgSIT animals. For OA-1067A1, the nanopore sequencing was unable to resolve the insertion site, presumably due to its insertion in one of the remaining gaps in the genome. Finally, using nanopore data, we confirmed genomic deletions in both pgSIT target genes - see AEL019894 and AAEL005656 as expected (Fig. S8-S9). The nanopore sequencing data has been deposited to the NCBI sequence read archive (SRA) under BioProject ID is PRJNA699282 with accession number SRR13622000.
+
+### Transcriptional profiling and expression analysis
+
+To quantify target gene reduction and expression from transgenes as well as to assess global expression patterns, we performed Illumina RNA sequencing. We extracted total RNA using miRNeasy Mini Kit (Qiagen, Cat# 217004) from ten sexed pupae: WT ♀, WT♂, transheterozygous pgSIT ♂’s, and pgSIT ♀ harboring all three transgenes Cas9/+; gRNA βTub#7/+; gRNA myo−fem#1/+ with each genotype in biological triplicate (12 samples total), following the manufacturer’s protocol. DNase treatment was conducted using DNase I, RNase-free (ThermoFisher Scientific, Cat# EN0521), following total RNA extraction. RNA integrity was assessed using the RNA 6000 Pico Kit for Bioanalyzer (Agilent Technologies #5067−1513), and mRNA was isolated from ~1 µg of total RNA using NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB #E7490). RNA-seq libraries were constructed using the NEBNext Ultra II RNA Library Prep Kit for Illumina (NEB #E7770) following the manufacturer’s protocols. Briefly, mRNA was fragmented to an average size of 200 nt by incubating at 94°C for 15 min in the first strand buffer. cDNA was then synthesized using random primers and ProtoScript II Reverse Transcriptase followed by second strand synthesis using NEB Second Strand Synthesis Enzyme Mix. Resulting DNA fragments were end-repaired, dA tailed, and ligated to NEBNext hairpin adaptors (NEB #E7335). Following ligation, adaptors were converted to the “Y” shape by treating with USER enzyme, and DNA fragments were size selected using Agencourt AMPure XP beads (Beckman Coulter #A63880) to generate fragment sizes between 250 and 350 bp. Adaptor-ligated DNA was PCR amplified followed by AMPure XP bead clean up. Libraries were quantified using a Qubit dsDNA HS Kit (ThermoFisher Scientific #Q32854), and the size distribution was confirmed using a High Sensitivity DNA Kit for Bioanalyzer (Agilent Technologies #5067−4626). Libraries were sequenced on an Illumina HiSeq2500 in single read mode with the read length of 50 nt and sequencing depth of 20 million reads per library. Base calls were performed with RTA 1.18.64 followed by conversion to FASTQ with bcl2fastq 1.8.4. The reads were mapped to the AaegL5.0 (GCF_002204515.2) genome supplemented with OA-874PA, OA-1067A1, and OA-1067K sequences using STAR. On average, ~97.5% of the reads were mapped (Table S11). Gene expression was then quantified using featureCounts against the annotation release 101 GTF downloaded from NCBI (GCF_002204515.2_AaegL5.0_genomic.gtf). TPM values were calculated from counts produced by featureCounts and combined (Table S12).
+
+PCA and hierarchical clustering of the data show that the samples generally behaved as expected in clustering by sex and genotype (Fig. S10). DESeq2 was then used to perform differential expression analyses between pgSIT vs WT samples within each sex (Fig. S10, Table S13, S14), and a two-factor design consistently showed what changed in response to the genotype in both sexes (Table S15). In a comparison between pgSIT ♀ and WT ♀, 660 genes were upregulated in pgSIT ♀ and 392 were downregulated at an adjusted p-value < 0.05. The target gene, AAEL005656, was significantly downregulated in pgSIT ♀ (Fig. S10C). In a comparison between pgSIT ♂’s and WT♂ (Table S13), 2067 genes were upregulated in pgSIT ♂’s and 2722 were downregulated at an adjusted p-value < 0.05. The target gene, AEL019894, was strongly downregulated in pgSIT ♂ (Fig. S10D). It’s important to note here that the CRISPR/Cas9 pgSIT system disrupts the DNA (not the RNA) so transcription is expected to occur; however, the transcripts produced will encode mutations and should be degraded by nonsense -mediated mRNA decay (NMD) mechanisms. Indeed, these mutant RNA’s can be observed in the IGV (Fig. S8, S9). In the two-factor comparison, 1447 genes were upregulated in pgSIT and 2563 were downregulated at an adjusted p-value < 0.05 (Fig. S10E). For each DESeq2 comparison, gene ontology enrichments were performed on significantly differentially expressed genes, and these are provided as tabs in the corresponding tables (Table S13-S15). All Illumina RNA sequencing data has been deposited to the NCBI sequence read archive (SRA) under BioProject ID is PRJNA699282 with accession numbers SRR13620773-SRR13620784.
+
+### Amplicon Sequencing of Target Loci
+
+To sample a variety of molecular changes at the gRNA target sites (myo-fem and βTub), we used the Amplicon-EZ service by Genewiz® and followed the Genewiz® guidelines for sample preparation. Genomic DNA from 50 WT and 50 pgSIT sexed pupae (25♀ + 25♂) were extracted separately using DNeasy Blood and Tissue Kit (Qiagen, Cat No./ID: 69506) following the manufacturer's protocols. Primers with Illumina adapters (Table S18) were used to PCR amplify the genomic DNA. PCR products were purified using the Zymoclean Gel DNA Recovery Kit (Zymo Research, Cat No./ID: D4007). Roughly 50,000 one-directional reads were generated by Genewiz® and uploaded to Galaxy.org for analysis. Quality control for the reads was performed using FASTQC. Sequence data were then paired and aligned against the myo-fem or βTub sequence using Map with BWA-MEM under “Simple Illumina mode”. Sequence variants were detected using FreeBayes, with parameter selection level set to “simple diploid calling.” The amplicon sequencing data has been provided as File S1.
+
+### Prior mating with pgSIT ♂’s suppress ♀ fertility
+
+To determine whether prior matings with pgSIT ♂ could reduce ♀ fertility, we initiated 15 cages each consisting of 250 mature (4–5 days old) pgSIT ♂ combined with 50 mature (4–5 days old) WT virgin ♀. We allowed the pgSIT ♂’s to mate with these ♀’s for a limited period of time (including: 2, 6, 12, 24, and 48 hrs; 3 replicate cages each). Cages were shaken every 3 minutes for the first half hour to increase mating opportunities. Following these time periods, all ♀’s were removed and transferred to new cages along with 250 WT mature ♂’s, cages were again shaken every 3 minutes for the first half hour to increase mating opportunities and left to mate for an additional 2 days. The ♀’s were then blood fed, and each blood fed ♀ was individually transferred to a single narrow Polystyrene vial (Genesee Scientific Cat# 32–116), and eggs were collected and hatched for fertility determination. Following this, non-fertile ♀’s were then placed back into cages along with the original WT ♂’s, plus an additional 50 mature WT ♂’s, for another chance to produce progeny. This was repeated for up to five gonotrophic cycles. As controls, cages with 250 WT ♂’s and 50 WT ♀’s, or 50 unmated blood fed WT ♀’s with no ♂’s added, or 50 unmated blood fed WT ♀’s with 250 WT ♂ adults were also set up (Table S8).
+
+### Life table parameters
+
+Life table parameters were assessed by comparing WT, homozygous gRNA βTub + myo−fem, homozygous Cas9, and transheterozygous pgSIT (gRNA βTub + myo−fem/+; Cas9/+) generated with Cas9 inherited from either the mother (maternal Cas9) or father (paternal Cas9). Larva/pupae development times were recorded as the number of days from hatched larvae to pupae and then to adults. One hundred larvae from each line were placed in separate larval rearing containers (Sterilite, 34.6 cm × 21 cm × 12.4 cm, USA), each with 3 liters of deionized water, and fed once a day. Larvae were counted twice daily until pupation, and then the date of pupation and emergence were recorded. Larval to pupae development time was calculated for each sex. Pupae were transferred to plastic cups (Karat, C-KC9) with 100 ml of water, and survivors were recorded until adulthood. ANOVA and Tukey post-hoc tests were performed to compare differences in larval and pupal development among all groups.
+
+For measuring ♂ /♀ longevity, we tested the variation in ♂ and ♀ longevity among different lines using two methods: (i) released along with WT of the opposite sex or (ii) without WT of the opposite sex. (i) One hundred WT, homozygous gRNA βTub + myo−fem, homozygous Cas9 newly eclosed adult mosquitoes (fifty ♂’s and fifty ♀’s) were maintained in a cage; fifty newly-eclosed pgSIT ♂’s (maternal cas9) and fifty newly-eclosed pgSIT ♂’s (paternal cas9) were caged with fifty newly-eclosed WT ♀’s; and finally, fifty newly-eclosed pgSIT ♀’s (maternal cas9) and fifty newly-eclosed pgSIT ♀’s (paternal cas9) were caged with fifty newly-eclosed WT ♂’s. (ii) Fifty ♂’s or ♀’s from each line were released into a cage separately without the opposite sex. Adults were provided with 10% sucrose and monitored daily for survival until all mosquitoes had died (3 replicates).
+
+For measuring ♀ fecundity and fertility, ♀’s (n=50) and ♂’s (n=50) three days post-emergence raised under the same standardized larval conditions were placed into a cage and allowed to mate for 2 days. ♀ mosquitoes were blood fed until fully engorged and were individually transferred into plastic vials with oviposition substrate. Eggs were stored in the insectary for 4 days to allow full embryonic development and then were hatched in a vacuum chamber. Fecundity was calculated as the number of eggs laid per ♀, and fertility was calculated as the percentage of eggs hatched per ♀. An analysis of variance (ANOVA) and a Tukey post-hoc test were performed to compare differences in fecundity and fertility among all groups.
+
+♂ mating capacity (how many ♀’s can be mated by one mature ♂ ) was measured as follows. Fifteen mature WT ♀’s were caged with 1 mature ♂ of each genotype for 24 hours (1♂:15♀ ratio). After 24 hours, the single ♂ was removed from all cages. Two days after the single ♂ was removed, 75 WT ♂’s were added to each cage that previously had a pgSIT ♂ (5♂:1♀ ratio). Blood meals were provided, and each blood fed ♀ was individually transferred to a single vial for egg collection. The fecundity and fertility of each ♀ was determined. The mating capacity was calculated as the total number of ♀’s - total number of fertile ♀’s. The mating capacity of WT, homozygous gRNA βTub + myo−fem, and homozygous Cas9 ♂ was equal to the number of fertile ♀’s. All statistical analyses were performed using GraphPad Prism software (GraphPad Software, La Jolla, California, USA). P values > 0.05 were considered not significant.
+
+### Flight activity quantification
+
+Mosquitoes were reared at 28°C, 80% relative humidity under a 12:12 hr light:dark regime, and measurements of flight activity were performed using a Drosophila Activity Monitoring (DAM) System (TriKinetics, LAM25) using large tubes designed for mosquitoes (TriKinetics, PGT 25 x 125 mm Pyrex Glass). Individual 4–7 day-old, non-blood fed virgin ♀ and non-mated ♂ mosquitoes were introduced into the monitoring tubes, which contained 10% sucrose (Sigma, Cat. S0389) at both ends of the tube as the food source. The DAM System was positioned vertically during the assays. Flight activity was measured over a period of 24 hrs by automatically calculating the number of times that mosquitoes passed through the infrared beam in the center of the tubes. The walls of the monitoring tubes were coated with Sigmacote (Sigma, Cat. SL2) to inhibit mosquitoes from walking upward. For preparing the wingless mosquitoes, the animals were anesthetized on ice, and the wings were removed using Vannas Scissor (World Precision Instruments, Cat. 14003). The wingless mosquitoes were allowed to recover for 12 hrs before recording. Mosquitoes were manually checked after flight activity recording to ensure survival. Data acquisition was performed using the DAMSystem (TriKinetics) (Fig. 2D, Video S5, Table S6).
+
+### Sound attraction assay
+
+The sound attraction assay was performed in a chamber with a temperature of 28°C and humidity of 80%. Seven-day old ♂’s were sex separated after the pupae stage. The day before testing, 30–40 ♂’s were transferred by mouth aspiration to a 15-cm³ mesh cage with a 10% sucrose bottle. ♂ mosquitoes were allowed to recover in the cage under a 12 hr:12 hr light:dark regime for 24 hrs. For each trial, a 10-second 600 Hz sine tone was applied on one side of the cage as a mating behavior lure, mimicking ♀ flight tones. The number of mosquitoes landing on the mesh area around the speaker box(10 cm²) was quantified at 5-second intervals throughout the stimulus. The average percent of mosquitoes landing around the speaker area out of the total cage post-sound presentation was calculated (Fig. 2E, Video S6, Table S7). Heatmaps were generated using Noldus Ethovision XT.
+
+### Multigenerational population cage trials
+
+To perform multigenerational population cage trials, two strategies were employed: (i) release of eggs; (ii) release mature adults (Fig. 3, Table S16). Cage trials were carried out using discrete non-overlapping generations. For the first release of eggs strategy (i), WT eggs and pgSIT eggs were hatched together using the following ratios of 1:1 (100:100), 1:5 (100:500), 1:10 (100:1000), 1:20 (100:2000), and 1:40 (100:4000), and three biological replicates for each ratio (15 cages total). All eggs were hatched simultaneously, then separated into multiple plastic containers (Sterilite, 34.6 cm × 21 cm × 12.4 cm, USA). Roughly 400 larvae were reared in each container using standard conditions with 3 liters of deionized water and were allowed to develop into pupae. Pupae were placed in plastic cups (Karat, C-KC9) with ~100 ml of water (~150 pupae per cup) and transferred to large cages (BugDorm, 60cm × 60cm × 60cm) to eclose. All adults were allowed to mate for ~5–7 days. ♀’s were blood fed, and the eggs were collected. Eggs were counted and stored for ~4 days to allow full embryonic development, then 100 eggs were selected randomly and mixed with pgSIT eggs with ratios of 1:1 (100:100), 1:5 (100:500), 1:10 (100:1000), 1:20 (100:2000), and 1:40 (100:4000) to seed for the following generation, and this procedure continued for all subsequent generations. The remaining eggs were hatched to measure hatching rates and to screen for the possible presence of transformation markers. The hatching rate was estimated by dividing the number of hatched eggs by the total number of eggs.
+
+For the release of mature adults strategy (ii), 3–4-days-old mature WT adult ♂’s were released along with mature (3–4 days old) pgSIT adult ♂’s at release ratios: 1:1 (50:50), 1:5 (50:250), 1:10 (50:500), 1:20 (50:1000), and 1:40 (50:2000), with three biological replicates for each release ratio (15 cages total). One hour later, 50 mature (3–4 days old) WT adult ♀’s were released into each cage. All adults were allowed to mate for 2 days. ♀’s were then blood fed and eggs were collected. Eggs were counted and stored for four days to allow full embryonic development. Then, 100 eggs were randomly selected, hatched, and reared to the pupal stage, and the pupae were separated into ♂ and ♀ groups and transferred to separate cages. Three days post eclosion, 50 (1:1), 250 (1:5), 500 (1:10), 1000 (1:20), and 2000 (1:40) age-matched pgSIT mature ♂ adults were caged with these mature ♂’s from 100 selected eggs. One hour later, mature ♀’s from 100 selected eggs were transferred into each cage. All adults were allowed to mate for 2 days. ♀’s were blood fed, and eggs were collected. Eggs were counted and stored for 4 days to allow full embryonic development. The remaining eggs were hatched to measure hatching rates and to screen for the possible presence of transformation markers. The hatching rate was estimated by dividing the number of hatched eggs by the total number of eggs. This procedure continued for all subsequent generations.
+
+### Mathematical modeling
+
+To model the expected performance of pgSIT at suppressing and eliminating local Ae. aegypti populations, we used the MGDrivE simulation framework (19). This framework models the egg, larval, pupal, and adult mosquito life stages with overlapping generations, larval mortality increasing with larval density, and a mating structure in which females retain the genetic material of the adult ♂ with whom they mate for the duration of their adult lifespan. The inheritance pattern of the pgSIT system was modeled within the inheritance module of MGDrivE, along with impacts on adult lifespan, male mating competitiveness, and pupatory success. We distributed Ae. aegypti populations according to human structures sourced from OpenStreetMap on the basis that Ae. aegypti is anthropophilic. Each human structure was assumed to have an equilibrium population of 16 adult Ae. aegypti, producing an equilibrium island population of 992. We implemented the stochastic version of the MGDrivE framework to capture random effects at low population sizes and the potential for population elimination. Weekly releases of up to 400 pgSIT eggs were simulated in all human structures of Onetahi over a period of 10–24 weeks. 200 repetitions were carried out for each parameter set, and mosquito genotype trajectories, along with the proportion of simulations that led to local population elimination, were recorded. Complete model and intervention parameters are listed in Table S17.
+
+# References
+
+1. T. C. Winegard, *The Mosquito: A Human History of Our Deadliest Predator* (Penguin, 2019).
+2. C. L. Moyes, J. Vontas, A. J. Martins, L. C. Ng, S. Y. Koou, I. Dusfour, K. Raghavendra, J. Pinto, V. Corbel, J.-P. David, D. Weetman, Contemporary status of insecticide resistance in the major Aedes vectors of arboviruses infecting humans. *PLoS Negl. Trop. Dis.* **11**, e0005625 (2017).
+3. D. A. Dame, C. F. Curtis, M. Q. Benedict, A. S. Robinson, B. G. J. Knols, Historical applications of induced sterilisation in field populations of mosquitoes. *Malar. J.* **8 Suppl 2**, S2 (2009).
+4. X. Zheng, D. Zhang, Y. Li, C. Yang, Y. Wu, X. Liang, Y. Liang, X. Pan, L. Hu, Q. Sun, X. Wang, Y. Wei, J. Zhu, W. Qian, Z. Yan, A. G. Parker, J. R. L. Gilles, K. Bourtzis, J. Bouyer, M. Tang, B. Zheng, J. Yu, J. Liu, J. Zhuang, Z. Hu, M. Zhang, J.-T. Gong, X.-Y. Hong, Z. Zhang, L. Lin, Q. Liu, Z. Hu, Z. Wu, L. A. Baton, A. A. Hoffmann, Z. Xi, Incompatible and sterile insect techniques combined eliminate mosquitoes. *Nature.* **572**, 56–61 (2019).
+5. J. E. Crawford, D. W. Clarke, V. Criswell, M. Desnoyer, D. Cornel, B. Deegan, K. Gong, K. C. Hopkins, P. Howell, J. S. Hyde, J. Livni, C. Behling, R. Benza, W. Chen, K. L. Dobson, C. Eldershaw, D. Greeley, Y. Han, B. Hughes, E. Kakani, J. Karbowski, A. Kitchell, E. Lee, T. Lin, J. Liu, M. Lozano, W. MacDonald, J. W. Mains, M. Metlitz, S. N. Mitchell, D. Moore, J. R. Ohm, K. Parkes, A. Porshnikoff, C. Robuck, M. Sheridan, R. Sobecki, P. Smith, J. Stevenson, J. Sullivan, B. Wasson, A. M. Weakley, M. Wilhelm, J. Won, A. Yasunaga, W. C. Chan, J. Holeman, N. Snoad, L. Upson, T. Zha, S. L. Dobson, F. S. Mulligan, P. Massaro, B. J. White, Efficient production of male Wolbachia-infected Aedes aegypti mosquitoes enables large-scale suppression of wild populations. *Nat. Biotechnol.* **38**, 482–492 (2020).
+6. D. O. Carvalho, A. R. McKemey, L. Garziera, R. Lacroix, C. A. Donnelly, L. Alphey, A. Malavasi, M. L. Capurro, Suppression of a Field Population of Aedes aegypti in Brazil by Sustained Release of Transgenic Male Mosquitoes. *PLoS Negl. Trop. Dis.* **9**, e0003864 (2015).
+7. K. C. Long, L. Alphey, G. J. Annas, C. S. Bloss, K. J. Campbell, J. Champer, C.-H. Chen, A. Choudhary, G. M. Church, J. P. Collins, K. L. Cooper, J. A. Delborne, O. R. Edwards, C. I. Emerson, K. Esvelt, S. W. Evans, R. M. Friedman, V. M. Gantz, F. Gould, S. Hartley, E. Heitman, J. Hemingway, H. Kanuka, J. Kuzma, J. V. Lavery, Y. Lee, M. Lorenzen, J. E. Lunshof, J. M. Marshall, P. W. Messer, C. Montell, K. A. Oye, M. J. Palmer, P. A. Papathanos, P. N. Paradkar, A. J. Piaggio, J. L. Rasgon, G. Rašić, L. Rudenko, J. R. Saah, M. J. Scott, J. T. Sutton, A. E. Vorsino, O. S. Akbari, Core commitments for field trials of gene drive organisms. *Science* **370**, 1417–1419 (2020).
+8. J. Champer, A. Buchman, O. S. Akbari, Cheating evolution: engineering gene drives to manipulate the fate of wild populations. *Nat. Rev. Genet.* **17**, 146–159 (2016).
+9. N. P. Kandul, J. Liu, H. M. Sanchez C, S. L. Wu, J. M. Marshall, O. S. Akbari, Transforming insect population control with precision guided sterile males with demonstration in flies. *Nat. Commun.* **10**, 84 (2019).
+10. D. Navarro-Payá, I. Flis, M. A. E. Anderson, P. Hawes, M. Li, O. S. Akbari, S. Basu, L. Alphey, Targeting female flight for genetic control of mosquitoes. *PLoS Negl. Trop. Dis.* **14**, e0008876 (2020).
+11. O. S. Akbari, I. Antoshechkin, H. Amrhein, B. Williams, R. Diloreto, J. Sandler, B. A. Hay, The developmental transcriptome of the mosquito Aedes aegypti, an invasive species and major arbovirus vector. *G3* **3**, 1493–1509 (2013).
+12. S. Gamez, I. Antoshechkin, S. C. Mendez-Sanchez, O. S. Akbari, The Developmental Transcriptome of Aedes albopictus, a Major Worldwide Human Disease Vector. *G3* **10**, 1051–1062 (2020).
+13. E. C. Degner, Y. H. Ahmed-Braimah, K. Borziak, M. F. Wolfner, L. C. Harrington, S. Dorus, Proteins, Transcripts, and Genetic Architecture of Seminal Fluid and Sperm in the Mosquito Aedes aegypti. *Mol. Cell. Proteomics.* **18**, S6–S22 (2019).
+14. J. Chen, J. Luo, Y. Wang, A. S. Gurav, M. Li, O. S. Akbari, C. and Montell, Suppression of female fertility in Aedes aegypti with a CRISPR-targeted male-sterile mutation. *Under Revision at PNAS* (2021).
+15. S. O’Leary, Z. N. Adelman, CRISPR/Cas9 knockout of female-biased genes AeAct-4 or myo-fem in Ae. aegypti results in a flightless phenotype in female, but not male mosquitoes. *PLOS Neglected Tropical Diseases* **14**, e0008971 (2020).
+16. M. Li, T. Yang, N. P. Kandul, M. Bui, S. Gamez, R. Raban, J. Bennett, H. M. Sánchez C, G. C. Lanzaro, H. Schmidt, Y. Lee, J. M. Marshall, O. S. Akbari, Development of a confinable gene drive system in the human disease vector. *Elife* **9**, doi: 10.7554/eLife.51701 (2020).
+17. M. Li, M. Bui, T. Yang, C. S. Bowman, B. J. White, O. S. Akbari, Germline Cas9 expression yields highly efficient genome engineering in a major worldwide disease vector, Aedes aegypti. *Proc. Natl. Acad. Sci. U. S. A.* **114**, E10540–E10549 (2017).
+18. E. C. Degner, L. C. Harrington, Polyandry Depends on Postmating Time Interval in the Dengue Vector Aedes aegypti. *Am. J. Trop. Med. Hyg.* **94**, 780–785 (2016).
+19. H. M. C. Sánchez, S. L. Wu, J. B. Bennett, J. M. Marshall, MGDrivE: A modular simulation framework for the spread of gene drives through spatially explicit mosquito populations. *Methods Ecol. Evol.*, 229–239 (2019).
+20. E. A. Mordecai, J. M. Cohen, M. V. Evans, P. Gudapati, L. R. Johnson, C. A. Lippi, K. Miazgowicz, C. C. Murdock, J. R. Rohr, S. J. Ryan, V. Savage, M. S. Shocket, A. Stewart Ibarra, M. B. Thomas, D. P. Weikel, Detecting the impact of temperature on transmission of Zika, dengue, and chikungunya using mechanistic models. *PLoS Negl. Trop. Dis.* **11**, e0005568 (2017).
+21. J. Bouyer, M. J. B. Vreysen, Concerns about the feasibility of using “precision guided sterile males” to control insects. *Nature Communications* **10**, doi: 10.1038/s41467-019-11616-9 (2019).
+22. H. Briegel, M. Hefti, E. DiMarco, Lipid metabolism during sequential gonotrophic cycles in large and small female Aedes aegypti. *J. Insect Physiol.* **48**, 547–554 (2002).
+23. N. P. Kandul, J. Liu, H. M. Sanchez C, S. L. Wu, J. M. Marshall, O. S. Akbari, Reply to “Concerns about the feasibility of using ‘precision guided sterile males’ to control insects.” *Nature Communications* **10**, doi: 10.1038/s41467-019-11617-8 (2019).
+24. B. J. Matthews, O. Dudchenko, S. B. Kingan, S. Koren, I. Antoshechkin, J. E. Crawford, W. J. Glassford, M. Herre, S. N. Redmond, N. H. Rose, G. D. Weedall, Y. Wu, S. S. Batra, C. A. Brito-Sierra, S. D. Buckingham, C. L. Campbell, S. Chan, E. Cox, B. R. Evans, T. Fansiri, I. Filipović, A. Fontaine, A. Gloria-Soria, R. Hall, V. S. Joardar, A. K. Jones, R. G. G. Kay, V. K. Kodali, J. Lee, G. J. Lycett, S. N. Mitchell, J. Muehling, M. R. Murphy, A. D. Omer, F. A. Partridge, P. Peluso, A. P. Aiden, V. Ramasamy, G. Rašić, S. Roy, K. Saavedra-Rodriguez, S. Sharan, A. Sharma, M. L. Smith, J. Turner, A. M. Weakley, Z. Zhao, O. S. Akbari, W. C. Black 4th, H. Cao, A. C. Darby, C. A. Hill, J. S. Johnston, T. D. Murphy, A. S. Raikhel, D. B. Sattelle, I. V. Sharakhov, B. J. White, L. Zhao, E. L. Aiden, R. S. Mann, L. Lambrechts, J. R. Powell, M. V. Sharakhova, Z. Tu, H. M. Robertson, C. S. McBride, A. R. Hastie, J. Korlach, D. E. Neafsey, A. M. Phillippy, L. B. Vosshall, Improved reference genome of Aedes aegypti informs arbovirus vector control. *Nature* **563**, 501–507 (2018).
+25. M. Li, L. Y. C. Au, D. Douglah, A. Chong, B. J. White, P. M. Ferree, O. S. Akbari, Generation of heritable germline mutations in the jewel wasp Nasonia vitripennis using CRISPR/Cas9. *Sci. Rep.* **7**, 901 (2017).
+26. D. G. Gibson, L. Young, R.-Y. Chuang, J. C. Venter, C. A. Hutchison 3rd, H. O. Smith, Enzymatic assembly of DNA molecules up to several hundred kilobases. *Nat. Methods* **6**, 343–345 (2009).
+27. H. Li, Minimap2: pairwise alignment for nucleotide sequences. *Bioinformatics* **34**, 3094–3100 (2018).
+28. M. U. G. Kraemer, M. E. Sinka, K. A. Duda, A. Q. N. Mylne, F. M. Shearer, C. M. Barker, C. G. Moore, R. G. Carvalho, G. E. Coelho, W. Van Bortel, G. Hendrickx, F. Schaffner, I. R. F. Elyazar, H.-J. Teng, O. J. Brady, J. P. Messina, D. M. Pigott, T. W. Scott, D. L. Smith, G. R. W. Wint, N. Golding, S. I. Hay, The global distribution of the arbovirus vectors Aedes aegypti and Ae. albopictus. *Elife* **4**, e08347 (2015).
+
+# Supplementary Files
+
+- [GUIDEDOA2100081SupplementaryTables.zip](https://assets-eu.researchsquare.com/files/rs-367110/v1/15aa6ab78b5b0a2f16b521d3.zip)  
+  Supplemental Tables  
+  Table S1. myo-fem and βTub developmental gene expression data. Ae. albopictus and Ae. aegypti myo-fem and βTub TPM gene expression across different developmental timepoints  
+  Table S2. Embryo microinjection and transgenic line generation  
+  Table S3. Single gene disruption  
+  Table S4. pgSIT cross data  
+  Table S5. Life Parameters. Comparisons of life parameters of different mosquito lines. To evaluate the potential fitness costs associated with the pgSIT components, several life-table parameters such as fecundity, larval development time, ♂ insemination capacity, mating competitiveness, and adult survival rate were measured among WT, homozygous gRNAβTub+myo-fem, homozygous Cas9 and heterozygous pgSIT lines (gRNAβTub+myo-fem,+/Cas9,+). Compared to WT, homozygous gRNAβTub+myo-fem, homozygous Cas9, and pgSIT♂’s that produced from maternal Cas9 or paternal Cas9 had no significant differences in larval and pupal survival rate, larval and pupal development time, ♂ insemination ability, and adult survival (p < 0.05). Maternal Cas9 refers to paternal gRNAβTub/+; gRNAmyo-fem/+; maternal Cas9/+, and paternal Cas9 corresponds to maternal gRNAβTub/+; gRNAmyo-fem/+; paternal Cas9/+. Each strain is labeled with a superscript (a-e) and is used to indicate where the significance lies. For example, an ANOVA with a post hoc Tukey’s analysis for ♀ fecundity indicates differences are between wildtype vs gRNA, wildtype vs Cas9, and wildtype vs transheterozygotes.  
+  Table S6. pgSIT flight capacity assay. Flight activities were monitored over a 24-hour period using the DAM system. The counts are the number of times the mosquitoes passed the infrared beam.  
+  Table S7. Sound attraction assay  
+  Table S8. pgSIT♂ sterilize WT ♀’s  
+  Table S9. The mean coverage depth from the Nanopore DNA sequencing for all contigs in the genome (2310) and the three plasmids (OA-1067A1: gRNAβTub; OA-1067K: gRNAmyo-fem; and OA-874PA: Nup50-Cas9) as well as normalized coverage based on the global number (Table S12). Transgene coverage ranged from 5.1 to 7.6 and normalized coverage ranged from 0.93 to 1.38.  
+  Table S10. Nanopore coverage means  
+  Table S11. Mapping Stats for RNA sequencing  
+  Table S12. RNAseq expression data  
+  Table S13. deseq2_liverpool_males_pgSIT_males.annotations  
+  Table S14. deseq2_liverpool_females_pgSIT_females.annotations.xlsx  
+  Table S15. deseq2_liverpool_pgSIT.annotations  
+  Table S16. Multigenerational population cage data  
+  Table S17. Parameters used in Aedes aegypti population suppression model.  
+  Table S18. Primer and gRNA sequences
+
+- [FileS1ampliconEZseqdata.xlsx](https://assets-eu.researchsquare.com/files/rs-367110/v1/61e3706a7e8424f4340f94f0.xlsx)  
+  Supplemental Dataset  
+  File S1. Amplicon EZ sequencing data.
+
+- [SupplementalVideo1.TimelapseofTubmutantandWTtestesandsperm.mp4](https://assets-eu.researchsquare.com/files/rs-367110/v1/929ebe64f1baa40111bb99e4.mp4)  
+  Video S1. Timelapse of βTub mutant and WT testes and sperm. βTub mutant and WT testes were imaged at 10X and 63X.
+
+- [SupplementalVideo2myofemfemaleseclosingfromcup.mp4](https://assets-eu.researchsquare.com/files/rs-367110/v1/000315cb62bc18b5d0237cea.mp4)  
+  Video S2. myo-fem mutant ♀’s eclosing. Flightless myo-fem mutant ♀’s have abnormal wing postures restricting their escape from rearing cups following eclosion, which reduces survival.
+
+- [SupplementalVideo3.Timelapseofmhcmyofemmutantflight.mp4](https://assets-eu.researchsquare.com/files/rs-367110/v1/2e3330ba5bc7098d305d3513.mp4)  
+  Video S3. Timelapse of myo-fem mutant flight. Cages consisting of myo-fem mutant ♀’s, myo-fem mutant ♂’s, WT ♀’s, and WT ♂’s were recorded over 5.5 minutes. The cages were occasionally tapped to stimulate movement.
+
+- [SupplementalVideo4TimelapseofpgSITandWTmosquitoes.mp4](https://assets-eu.researchsquare.com/files/rs-367110/v1/7a610233597a47e053a24add.mp4)  
+  WT ♂’s were recorded for 5.5 minutes. The cages were occasionally tapped to stimulate movement/flight.
+
+- [SupplementalVideo5DAMassayvideo.mp4](https://assets-eu.researchsquare.com/files/rs-367110/v1/bb195384128c54c48797ca78.mp4)  
+  Video S5. DAM assay video. Short clip of the DAM assay’s monitor tube in action with a WT ♀ passing the infrared beam during flight.
+
+- [SupplementalVideo6.MaleCourtshipAssay.mov](https://assets-eu.researchsquare.com/files/rs-367110/v1/1e63a54b367e872d452eab51.mov)  
+  Video S6. Male courtship assay. pgSIT males are strongly attracted to the female flight tone indicating strong mating behavior.
+
+- [SupplementlVideo7.mp4](https://assets-eu.researchsquare.com/files/rs-367110/v1/4cb4bbe51eb2b71bdbc9d2b5.mp4)  
+  Video S7. Model-predicted impact of releases of pgSIT eggs in Onetahi, Tetiaroa, French Polynesia. Time-series for female Ae. aegypti population density and elimination probability are depicted for four sample release schemes depicted in Figure 4.
+
+- [SupplementalFigures.pdf](https://assets-eu.researchsquare.com/files/rs-367110/v1/24d25732d9dcc7b8c41125e6.pdf)  
+  Supplemental Figures  
+  Figure S1. Conservation of target genes in Diptera.  
+  Figure S2. Assessment of independent gRNAβTub lines.  
+  Figure S3. Assessment of independent gRNAmyo-fem lines.  
+  Figure S4. Transmitted light and fluorescent images of mosquito life stages of strains used in this study.  
+  Figure S5. Fitness of transheterozygous pgSIT mosquitoes in comparison with WT and parental lines.  
+  Figure S6. Illumina NGS-based amplicon sequencing results representing myo-fem and βTub knockout in pgSIT mosquitoes.  
+  Figure S7. Determination of transgene copy number using Oxford Nanopore genome sequencing.  
+  Figure S8. Integrated genome browser snapshot depicting pgSIT sequencing results for myo-fem.  
+  Figure S9. Integrated genome browser snapshot depicting pgSIT sequencing results for βTub.  
+  Figure S10. Transcriptional profiling and expression analysis.  
+  Figure S11. Scaling pgSIT to control populations of mosquitoes and molecular mechanisms.

07212abf54a75e3870791e75d8bb51a41460b19b6906e8342e0a5c26aa7ffe91/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Site locations and mean modelled forest cover (MFC) calculated from past forest recontructions48 within the LBK distribution based on reference 7 (see Methods). All sites are LBK, except for Alf\u00f6ld Linear Pottery (ALP, SI Table 1). N=number of sites. 1. Cuiry-l\u00e8s-Chaudardes; 2. Maastricht-Cannerberg; 3. Maastricht-Klinkers; 4. Geleen-Janskamperveld; 5. Erkelenz-K\u00fcckhoven; 6. Konigshoven 14 (FR 5); 7. Langweiler 842; 8. Ensisheim-Ratfeld6,8; 9. Colmar6; 10. Sierentz6; 11. Bischoffsheim6,8,50; 12. Herxheim45 13. Kilianst\u00e4dten; 14. Vaihingen an der Enz27; 15. Heilbronn-Neckargartach8; 16. Dillingen-Steinheim; 17. Derenburg Meerenstieg II44; 18. Halberstadt Sonntagsfeld44; 19. Karsdorf44; 20. Altscherbitz; 21. Brodau42; 22. Aiterhofen8; 23. Lerchenhaid8; 24. Stephansposching; 25. Rutzig/Haid8; 26. P\u0142onia 2; 27. Brzezin 7; 28. Karwowo 1; 29. \u017bal\u0119cino; 30. \u017buk\u00f3w; 31. \u010cern\u00fd V\u016fl28,47,50; 32. Bylany39,41; 33. Chot\u011bbudice28,47,50\u00a034. Stroegen; 35. T\u011b\u0161etice-Kyjovice8,50; 36. Brunn am Gebirge 37. Gnadendorf8; 38. Vedrovice-S\u00eddli\u0161t\u011b8; 39. Asparn a. d. Zaya/Schletz8; 40. Blatn\u00e98; 41. Chabsko 24; 42. \u017begotki; 43. Bo\u017cejewice 22/23; 44. Ro\u017cniaty 2; 45. Radojewice 29; 46. Kuczkowo 5; 47. Siniarzewo 1; 48. Kopyd\u0142owo 619,35; 49. Ludwinowo 733,40,46,50, 50. Bodzia 1; 51. Kruszyn 13; 52. Modlnica 5; 53. Vr\u00e1ble-Ve\u013ek\u00e9 Lehemby43; 54. Balatonsz\u00e1rsz\u00f3-Kis-erdei-d\u0171l\u01518,50; 55. \u0160t\u00farovo; 56. Tolna-M\u00f6zs-K\u00f6zs\u00e9gi-Cs\u00e1d\u00e9s-f\u00f6ldek; 57. Apc-Berekalja I50; 58. F\u00fczesabony-Gubak\u00fat (ALP)8,50; 59. Polg\u00e1r-Ferenci-h\u00e1t (ALP)8; 61. Fels\u0151vad\u00e1sz-V\u00e1rdomb (ALP); 60. Garadna- Elker\u00fcl\u0151 \u00fat (ALP).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "A. Histogram of the cattle/dairy ruminant diet \u03b413C values based on unpublished and published \u03b413C16:0 values from dairy lipids6,33,39-42 recovered from pottery vessels (blue), collagen \u03b413C values8,27,43-45 (green) and, highest and lowest \u03b413C values from the sequential bioapatite analysis of cattle teeth19,46 (red) (SI Table 2-4);\u00a0B. Biplot of longitude and cattle diet values from the three datasets, (colour legend as in A). The dotted line with error margins in A and B represents the diet value of \u221227.7 \u00b1\u00a01\u2030 for a forest dwelling ruminant based on the mean \u03b413C value of published8,27,44,45and unpublished contemporaneous deer bone collagen samples (SI Table 6); C. Interpolation map for cattle diet inferred from bone collagen \u03b413C values; D. Interpolation map for dairy ruminant diet inferred from dairy lipid \u03b413C values (colour legend as in C).\u00a0",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Boxplot of diet \u03b413C values based on bone collagen, bioapatite (max and min) and dairy lipids for sites within each river basin, ordered according to the median MFC for the basin (in brackets).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "A. Mid-range, highest and lowest \u03b413C values per tooth sampled from Bischoffsheim (BIS), Chot\u011bbudice (CHO)28,47, \u010cern\u00fd V\u016fl28 (CER), T\u011b\u0161etice-Kyjovice (TES), Ludwinowo 7 (LUD)46, Apc-Berekalja I (APC) and Balatonsz\u00e1rsz\u00f3 (BAL). The grey box marks the corrected bioapatite value of \u221213.2\u2030 for a forest dwelling ruminant based on deer bone collagen data; B. Range in dietary \u03b2 (\u039415NGlx-Phe) values for each sampled cattle tooth, with the grey box marking the upper limit of the reference for woody plants (\u22129.3\u00a0\u00b1\u00a01.6\u203038); C. Amplitude in \u03b413C values\u00a0(max \u03b413Cbioap\u00a0\u2212\u00a0min \u03b413Cbioap ) from individual sampled teeth in comparison to mean modelled forest cover (MFC); D. Amplitude in \u03b2 values per cattle tooth sampled from Bischoffsheim (BIS), Ludwinowo 7 (LUD), Apc-Berekalja I (APC) and Balatonsz\u00e1rsz\u00f3-Kis-erdei-d\u0171l\u0151 (BAL) in comparison to MFC.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "5A-F: Combined stable isotopic results of incremental analysis of bioapatite (\u03b413C (grey diamonds) and \u03b418O (white diamonds)) and dentine (dietary \u03b2 values are black filled squares) from cattle teeth samples: A. BIS3; B. BIS4; C. LUD1; D. APC1; E.\u00a0BAL3; F. BAL5. The black dotted line is the upper limit of both dietary \u03b2 values for woody plants and hypothetical \u03b413C value for forest based on contemporary deer collagen samples (\u221213.2\u2030, based on \u221227.7\u2030 adjusted for \u0394bioap-diet by \u221214.5\u203053).\u00a0",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

07212abf54a75e3870791e75d8bb51a41460b19b6906e8342e0a5c26aa7ffe91/preprint/preprint.md

@@ -0,0 +1,246 @@
+# Abstract
+
+Stable isotope signatures of domesticates found on archaeology sites provide information about past human behaviour, such as the evolution and adaptation of husbandry strategies. A dynamic phase in cattle husbandry evolution is during the 6th millennium BCE, where the first cattle herders of central Europe spread rapidly through diverse forested ecological niches, where little is known about pasturing strategies. Here we investigate cattle pasturing and foddering practices using a multi-regional dataset of stable isotope values (δ13C and δ18O; compound-specific stable isotopic analysis δ15N-amino acids and δ13C-dairy fats) measured from cattle bone and teeth, and pottery residues from early farming contexts, and palaeoenvironmental information. Our analysis reveals that farmers practiced different pasturing strategies with the intensive use of forested ecosystems in some areas for both graze and seasonal forage. We propose that the diversity of strategies is related to the adaptation of herding to new environments, which had a positive impact on cattle breeding and milk availability for human consumption.
+
+**pasture ecology** **Linearbandkeramik** **bioapatite** **collagen** **multi-isotope approach** **compound-specific stable isotopes** **seasonality**
+
+# Introduction
+
+Early husbandry practices were an important step in the evolution of domesticated animals, either via conscious or unconscious selection, shaping animal morphology, phenotype, and genotypes. Subsistence practices, such as husbandry, are central components of human cultural activities<sup>1-3</sup>. A dynamic phase in the evolution of cattle husbandry in Europe was during the rapid introduction of farming represented by the *Linearbandkeramik* (LBK) culture. These communities settled across a wide geographical range, initially in Hungary and eastern Austria between 5545–5360 cal BCE<sup>4</sup> and then expanded into other regions to the north, east and west between 5380 to 5315 cal BCE<sup>5–8</sup>, during a period time characterised by numerous short-term and significant fluctuations in precipitation<sup>9</sup>. The traditional image of these communities is one of homogeneity<sup>8</sup>, partly reinforced by ancient DNA analysis where large-scale genomic studies suggest that LBK communities were the outcome of monolithic demic diffusion<sup>10,11</sup>. However, regionally focused studies demonstrate that complex interactions between local hunter-gatherers and farmers existed<sup>12</sup> as well as subtle differences in animal husbandry strategies with greater exploitation of pigs in certain areas, such as Alsace<sup>6,13,14</sup>. Moreover, detailed palaeodietary studies indicate significant differences in human diet between regional populations<sup>8</sup>.
+
+Cattle (*Bos taurus*) were a central part of LBK economies and contributed to the success and rapid expansion of these communities<sup>3,15</sup>, providing an important source of both milk and meat<sup>15,16</sup>. In comparison to other domesticates, cattle are slow to gain maximum weight with traditional land races reaching sexual maturity around 2–3 years with first parturition occurring between 3 and 4 years<sup>15</sup>. Maintaining herds of sub-adults and adults represents an investment in terms of pasture and fodder resources<sup>1</sup>, particularly during winter and early spring when grazing may be restricted by snow or flooding<sup>17</sup>. The LBK communities inhabited diverse range of environments from the seasonally flooded marshes of the Carpathian Basin<sup>18</sup> to the semi-open woodlands of the Polish lowlands<sup>19</sup> and the forested steppe in northern and central Bohemia, and southern Moravia<sup>20</sup> to the closed mixed oak forests of the Rhine valley<sup>21</sup>. European woodlands depending on their canopy structure and composition have always been important in traditional pasture systems<sup>17,22,23</sup>. Branches and leaves (leafy hay), a fodder rich in nutrients and minerals<sup>24</sup>, has been used since the Late Neolithic<sup>25</sup> as an alternative animal feed<sup>17,22</sup>. Given the relative diversity in forested environments within LBK cultural distribution, husbandry practices, particularly foddering and pasturing strategies may have been shaped in part by these local ecosystems.
+
+Without direct physical evidence of forest fodder consumption, carbon isotope ratios (δ<sup>13</sup>C) of animal bone and teeth can potentially identify the use of forest environments for pasture and forage<sup>26–30</sup>. This is based on the well-known ‘canopy effect’ principle<sup>31</sup> (*see Methods*), where relative negative δ<sup>13</sup>C values of bone collagen, dentine and bioapatite samples may indicate animals feeding under a dense forest canopy. Within the context of our study, an initial investigation observed a strong relationship between δ<sup>13</sup>C values from LBK cattle bone collagen samples and geographical/paleoenvironmental proxies, with values become increasingly more negative moving westwards, not evident in sheep/goat samples<sup>30</sup>. Another means to determine domesticated ruminant diet is via δ<sup>13</sup>C values of dairy lipids trapped within the fabric of pottery. These are biosynthesised from dietary carbohydrates<sup>32</sup>, which in herbivores is directly related to consumed fodder. Dairy lipids are routinely detected using compound-specific stable isotope analysis (CSIA) of fatty acids (FA), providing direct evidence of milking<sup>33,34</sup> and changes in pasture strategies over time<sup>35</sup>. Deciduous forest canopies by their nature change in density reaching their maximum during late summer<sup>36</sup>. These changes in the canopy impact foliar δ<sup>13</sup>C values, and thus seasonally collected fodder can be identified using δ<sup>13</sup>C/δ<sup>18</sup>O values of sequential samples of the enamel following the tooth growth axis. This methodology has detected cattle winter foddering in central Europe: one individual at Chotěbudice (LBK, Czechia<sup>28</sup>) and in both sheep and cattle from Bercy (Middle Neolithic, Paris Basin<sup>26</sup>). While these results are promising, carbon isotope compositions cannot distinguish between plants growing within open canopy forests or forest edges, to those growing in open landscapes<sup>37</sup> nor can they determine the plant type. The recent development of a proxy based on CSIA of collagen/dentine amino acid (AA) δ<sup>15</sup>N values provides an independent means of directly identifying the fodder source i.e., woody vs. herbaceous plants<sup>38</sup>. Combining this proxy with δ<sup>13</sup>C/δ<sup>18</sup>O values, within a single tooth, provides a powerful tool to assess the seasonal use of forest resources.
+
+Stable isotope signatures of domesticated animals found on archaeology sites can be used to track the adaptation of husbandry strategies by early farmers to local ecosystems. Herein, we analyse δ<sup>13</sup>C, δ<sup>18</sup>O and δ<sup>15</sup>N values from bone and dentine collagen and bioapatite from cattle, and pottery dairy lipids from unpublished and published studies<sup>6,8,19,27,28,33,35,39–47</sup> of early farming contexts characterised by a material culture with numerous shared elements across a wide geographical range (SI Table 1–5, 1,541 isotopic measurements, 61 sites; Figure 1). Together with site-specific paleoenvironmental data, we characterise pasture and forage strategies adopted by pioneer farming communities in central Europe. The paleoenvironmental proxies include mean forest cover (MFC), mean precipitation summer (PMS) and winter (PMW); mean temperature summer (TMS) and winter (TMW) based on palynological data<sup>48,49</sup>, and modern river information (river drainage systems (RDS); Strahler stream order (SSO); distance to nearest river (DNR)). The combination of these datasets offers unique insights into how local forested ecosystems influenced early cattle pasture management and seasonal foddering practices in central Europe.
+
+# Results
+
+The global distribution of the δ¹³C values for LBK cattle diet inferred from collagen and dental bioapatite, and for ruminant diet based on dairy lipids δ¹³C values ranged from −31.2 to −19.8‰. There was no significant difference between diet δ¹³C values estimated from three datasets (ANOVA, df = 2, p = 0.1; Figure 2A: where each histogram represents a specific dataset). A significant correlation was found between diet δ¹³C values and longitude (Pearson’s correlation r = 0.6, p < 0.001, Figure 2B; SI Table 7). We compared the datasets to an average diet value for deer, a mainly forest-dwelling ruminant, based on published and unpublished bone collagen samples from LBK sites (SI Table 6; −27.7 ± 1‰, N = 33). The general trend observed for all datasets was values from sites in the west were within or below our forest dwelling ruminant baseline in contrast to those in the east (Figure 2B). This is further highlighted by the interpolated maps of diet δ¹³C values, where values from collagen (Figure 2C) and dairy lipids (Figure 2D) become progressively more depleted in ¹³C moving east to west.
+
+Correlation analysis between all values and paleoenvironmental proxies found only PMW to have a correlation coefficient greater than 0.5 (SI Table 7; r = −0.54, p < 0.0001). Within the dairy lipid dataset, other proxies were found to be significantly correlated with correlation coefficient greater than 0.5 (Dairy lipids diet δ¹³C values ~ TMW: r = −0.63, p < 0.0001; ~MFC: r = 0.5, p < 0.0001). However, in both the collagen and bioapatite datasets, PMW and TMW were found to be significant but with r values < 0.4 which explains less than 16% of the variation in these datasets. Correlation between two variables is not causation, especially within datasets where other sources of variation exist and cannot be explored due to lack of information, which is often the case for archaeological data. We used Partial Least Square (PLS) regression analysis to further explore the relationship between diet-inferred δ��³C values and paleoenvironmental proxies (SI 1). For all datasets, the r² value for the PLS regression models were < 0.4. For both datasets, the variables TMW and PMW were within the top three components ranked for importance. However, the optimal models to explain diet δ¹³C values based on bone collagen and dairy lipids used a single component (longitude) based on the RMSE scores of 1.24 and 3.15 respectively. For bioapatite diet δ¹³C values, all three variables (longitude, TMW and PMW) were included in the optimal PLS model.
+
+For cattle herders, it is important to be close to water sources since cattle are obligate drinkers. Most sites are situated near to streams with no tributaries (SSO 1) with the average distance of 1.5 km to closest water source. The diet δ¹³C values for each tissue from the major river drainage systems (RDS; Seine, Meuse, Rhine, Elbe, Danube, Vistula, and Oder; Figure 3) decreased from east to west. Furthermore, samples from sites situated in the Seine, Meuse, and Middle/Upper Rhine River basins where the mean MFC ranged from 63.9 to 75.2% had lower δ¹³C values in comparison to areas with higher MFC (>75%). This contrasts with the canopy effect, where we would expect negative values where the MFC was greatest.
+
+## Individual analysis
+
+Incremental bioapatite samples of the third molar from 46 cattle exhibited δ¹³C values between −12‰ to −10‰, except at Bischoffsheim where all values were below −12‰ (Figure 4A, SI Table 4 and SI Figure 1). Individuals from Bischoffsheim, Balatonszárszó, Chotěbudice²⁸ and Apc-Berekalja I exhibited low δ¹³C values that occur when the δ¹⁸O values are low i.e. during cold months⁵¹ (SI Figure 1). Within individual teeth, the amplitude between the highest and lowest values ranged from 0.3‰ to 3.0‰ (SI Table 4; Figure 4C) and was negatively correlated with MFC (r = −0.7, p < 0.05). At sites where MFC was <75%, such as Bischoffsheim and Balatonszárszó, inter-tooth variation in amplitude was greatest. The diet β values based on CSIA-AA δ¹⁵N values of sequential dentine samples from 12 cattle sampled here for δ¹³C/δ¹⁸O analysis, ranged from −8.7‰ to −3.6‰. Using modern references³⁸, we can identify the plant source i.e., herbaceous (−5.4 ± 2.1‰) or woody (−9.3 ± 1.6‰). The lowest values were found at Bischoffsheim while a single tooth from Balatonszárszó also exhibited values below the upper limit for diet based on woody plants (BAL 2, Figure 4B). Balatonszárszó individuals exhibited the greatest range in values (−7.7 to −3.6‰; SI Table 6) while largest intra-tooth variation between the highest and lowest β values was at Bischoffsheim (Figure 4D).
+
+We integrated bioapatite (δ¹⁸O/δ¹³C) and dentine (CSIA-β) datasets (Figure 5A–F). For interpretation, a delay of around 6 months is taken into account between the formation of dentine and the mineralisation of enamel based on experimental analysis⁵². At Bischoffsheim, the δ¹³C values fell within the range expected for the forest dwelling herbivore, with the lowest δ¹³C values recorded when the δ¹⁸O values were lowest i.e. winter. In the same teeth, β values showed variation over the annual cycle, with the lowest values indicating of woody plants in the diet (Figure 5A/B). These values occur, considering the delay in enamel mineralisation, when the δ¹⁸O values were low indicating consumption of woody plants in winter. This is also observed in BAL3 (Figure 5E) where low δ¹³C values and δ¹⁸O values coincided with a decreasing trend in β values, suggesting leafy hay foddering or pasturing within forested environments during the winter months. Individuals LUD1, APC1 and BAL5 exhibit β values between −6.3 and −3.5‰ indicating a solely herbaceous plant diet (Figure 5C/D/F).
+
+# Discussion
+
+The holistic analysis of δ¹³C values from cattle teeth and bone, and ruminant dairy lipids provides a unique insight into early herding pasturing and foddering strategies. Our analysis shows the earliest evidence for seasonal winter foddering with leafy hay for cattle in central Europe and intensive use of forested areas for cattle and domesticated ruminant pasture west of the Rhine. We discuss the impact of climate and other factors on these results, present an overall synthesis of early cattle land use and explore the potential long-term impact of forest pasture on early human societies and their cattle herds.
+
+## Impact of climate and other factors on carbon stable isotopes
+
+Carbon stable isotope ratios of C₃ plants are sensitive to changes in their growing environments<sup>37</sup>, these values are passed onto grazing herbivores and subsequently human consumers. Forested environments also impact foliar δ¹³C values<sup>31,37</sup> due to the ‘canopy’ effect, a central principal of our study. We observed in areas where MFC was low that diet δ¹³C values were also low, which was unexpected. However, while the proxy MFC provides a unique measure for paleo-forest cover, it cannot characterise woodland composition. This is important for characterising canopy structure and density as it is a mosaic of different tree species<sup>36</sup>. Moving from the paleoenvironment to paleoclimatic conditions, climate has an important impact on foliar δ¹³C values where plants growing in warm and dry environments exhibiting higher values than those growing in cool humid conditions, which is driven in principle by stomatal closure and its impact on ¹³C discrimination<sup>54</sup>. Our analysis demonstrated a correlation between diet δ¹³C values, and winter precipitation (PMW) and temperatures (TMW), chiming with pervious large-scale palaeodietary analysis albeit using modern mean summer temperatures<sup>55</sup>. However correlation analysis of sheep/goat bone collagen found no significant result between paleoclimate and δ¹³C values<sup>30</sup>. It is difficult to assess the impact of local environments and climatic conditions on ancient plant foliar δ¹³C values and those of herbivores partly due to the poor temporal resolution as well as choice of paleoenvironmental proxies to accurately reflect past ecosystems.
+
+The δ¹³C values of dairy FA, such as palmitic acid (C₁₆:₀) reflect a dairy ruminant’s diet at the time of lactation as they are largely biosynthesised *de novo* from dietary carbohydrates or incorporated from dietary FA<sup>32</sup>. Our results showed a clear decrease in dairy lipid δ¹³C values moving east to west. Pottery vessels can be used for multiple purposes, where mixtures of animal products within a single vessel can affect the dairy C₁₆:₀ FA δ¹³C values. Relatively high δ¹³C values (ca. −26.5‰, equivalent to diet value of −25‰) could be explained by the mixing of animal products in vessels. For example, processing non-ruminant products, such as pig fat, with δ¹³C values ca. 3‰ more enriched than ruminant fats<sup>56</sup>, would produce the observed isotopic shift. Pigs are more frequent in faunal assemblages to the west of the Rhine during the LBK<sup>57</sup>, and pottery lipids<sup>6</sup> and yet we still observe low dairy lipids δ¹³C values. Processing of dairy and freshwater aquatic products in the same vessel would lead to negative shifts in δ¹³C values. However, there is currently little evidence for the exploitation of freshwater resources in the LBK<sup>57</sup>. Therefore, the east to west trend observed within dairy lipid δ¹³C values appears to mainly reflect ruminant fodder resources. The large variation observed in values in some regions (Paris Basin, Alsace, Carpathian Basin) may reflect changes in forage during the lactation period for example, provision of leafy hay.
+
+## Integrated perspectives of cattle pasturing practices and local ecological contexts
+
+The lack of significant difference between δ¹³C values from cattle skeletal material and dairy lipids, as well as the previous analysis of sheep/goat bone collagen<sup>25</sup> supports the hypothesis that cattle were the primary source of dairy lipids<sup>15,33</sup>. The subsequent overview provides a synthesis of our results where we explore the evolution of cattle herding management strategies within forested ecosystems. Beginning in the Carpathian Basin, the region during the 6th millennium BC, was characterised by extensive marshland environments and gallery forests (MFC mean, 70.5%) strongly influenced by the Danube and its tributaries. Sites from this region represent some of the oldest in our dataset, Apc-Berekalja I and Balatonszárszó (5470–4950 cal BCE and 5335–4900 cal BCE respectively). In general, β and δ¹³C values indicate the diet of cattle consisted of herbaceous plants from open environments. Cattle ranging between different ecological niches (marsh, forest, open steppe) may cause the high variation observed in bioapatite δ¹³C values (Figure 4A). In addition, there are some indications of the use of woodland resources for forage. At Apc-Berekalja I, dairy lipids are depleted in ¹³C (−26.0 ± 0.5‰ (N =9)) in comparison to bioapatite samples (−24.7 ± 2.4‰ Nteeth =7), which may be a result of leafy hay provisioned during milking. The β values from Balatonszárszó also suggest some contribution of woody plants in the cattle diet during winter. Overall, in this region we propose that cattle were managed extensively moving between different pastures with their fodder supplemented at specific times, for example, during lactation or when snow cover restricted pasture access.
+
+In areas, such as Poland and Czechia, the ratio of open areas to forests was low (MFC, 80–85%). The Polish samples overall exhibit relatively higher diet δ¹³C values than the reference for a forest-dwelling herbivore. CSIA and bulk values from teeth, bones and pottery suggest that cattle grazed on herbaceous plants from open or open canopy forest environments. The low variation in δ¹³C values of bioapatite, collagen and dairy lipids from Polish sites suggested limited diversity in pasture types. Interestingly in subsequent TRB phase at Kopydłowo, dairy lipids and collagen indicate herders made greater use of densely forested areas<sup>19,35</sup>. Moving south-west, where palynological investigations indicate the presence of open canopy steppe forest<sup>58</sup>, one individual at Chotěbudice (Northern Czechia) had δ¹³C values occurring in winter indicative of leafy hay foddering (CHO9, range −27.3‰ to −25.7‰)<sup>28</sup>. Other cattle teeth from here, Černý Vůl and Těšetice-Kyjovice exhibit values that fall within the upper range of dietary values expected for a forest dwelling herbivore. We cannot rule out the use of forests in these regions for forage resources, but the absence of low δ¹³C values may indicate cattle grazing in open areas surrounding the settlements, for example, river terraces and forest fringes.
+
+In the western part of our study region, the MFC was less than 75% and forests typically consisted of mixed deciduous species. Focusing on the Alsace regions (Bas-Rhin and Haut-Rhin) bordering the Rhine, halophilic species, such as hazel were more frequent in the Haut-Rhin suggesting open canopy forested environments<sup>10</sup>. To the north, all our isotope datasets at Bischoffsheim (Bas-Rhin) support the use of forests as pasture with clear evidence of winter provision of collected leafy hay. The mean cattle diet values are found to be more enriched in ¹³C at sites in the Haut-Rhin region, for example, Ensisheim-Ratfeld (δ¹³C lipid: −26.1 ± 0.9‰, N =19; δ¹³C coll: −26.8 ± 1‰, N =14). The variation in canopy structure and species composition would explain the observed difference between north and south, while the use of forest pasture in the Bas-Rhin region may have been related to the increased presence of pigs<sup>6,13,14</sup>, which encourage the development of undergrowth via rooting<sup>23</sup>.
+
+The Paris Basin was more open landscape (MFC: 65%). However, the average δ¹³C values at Cuiry-lès-Chaudardes, from both cattle bone collagen (−28.1 ± 0.5‰, N =8) and ruminant dairy lipids (−27.7 ± 1.2‰ (N =49)) reflect forage depleted in ¹³C, such as forest pasture and fodder. Further, comparison with Bischoffsheim (Student t-test, bone collagen t=0.3 p=0.7), supports this hypothesis. The significant presence of dairy lipids at the site suggests a strong emphasis on milk production and processing. Dairy lipids δ¹³C values from Cuiry-lès-Chaudardes are not significantly different to cattle bone collagen samples from Bischoffsheim (T-test, bone collagen t=1.1 p=0.7). Cattle as the main dairy species should be approached with caution and various scenarios examined, because the slaughter profile analysis shows an orientation towards meat production<sup>16</sup>. Alternatively, the use of forest resources for cattle fodder and pasture may have fuelled dairy production while providing a by-product in the form of fatten male calves for meat production. It is also possible that dairy sheep/goat were pastured in woodland environments alongside cattle, winter provisioning of forest resources of both species is evident at more recent sites within the Paris Basin<sup>26</sup>.
+
+## The impact of the use of forest resources on human societies and their herds
+
+A recent study on cattle birth periodicity, predicted from bioapatite δ¹⁸O values including individuals from our dataset, demonstrated that there was an increased frequency of out-of-season births in LBK<sup>50</sup> in comparison to other early Neolithic groups. The oestrous cycle in cattle is not governed by seasonal variation in temperatures as in sheep/goat<sup>59</sup>. Provision of leafy hay, a rich source of minerals and nutrients<sup>24</sup>, would have increased winter survival of females as well as improved their body condition, fertility and survival of offspring<sup>59</sup>. A greater proportion in out-of-season births would have resulted in overlapping lactations and thus extended period of milk availability for human populations. Provision of feed, such as leafy hay, during milking increases the milk let-down<sup>60</sup> and certain tree species also improve milk quality<sup>24</sup>. Processing milk into a variety of storable products, such as milk casein balls or hard cheeses, would have helped support them during lean months and food crises<sup>61</sup>. Pasturing animals within forests would have increased contact with local hunter-gatherer-fisher groups, with surplus dairy products as a medium of exchange<sup>62</sup>. In the Paris Basin, Middle Neolithic farming populations of the region exhibit both cultural and biological links with these communities<sup>7,12</sup>. The forest and forest pasturing, which began in the LBK, may have acted as a catalyst stimulating interaction and exchange between groups.
+
+Cattle were an important commodity to the first farmers of central Europe and this study reveals complex regional strategies that evolved in response to different forested ecosystems and resource availability, and potentially lead to benefits to both humans and cattle populations. Supplementing cattle diets with leafy hay would have improved cattle health and influenced birth seasonality providing a direct benefit to communities with increased availability in milk. Pasturing and foddering of animals within forests contributed to the opening-up of these environments, facilitating the spread of prehistoric farming societies across Europe. The arrival of farmers and their herds to central Europe disrupted woodland environments marking an important point in the emergence of man-made ecosystems.
+
+# Methods
+
+## Stable isotope principles (Rosalind E. Gillis/ Iain P. Kendall)
+
+The canopy effect is a result of atmospheric CO₂ under the canopy being ¹³C-depleted relative to the atmosphere due to the uptake of recycled CO₂ respired by decomposition of ¹³C-depleted organic matter. This is coupled with decreased light intensity that reduces photosynthetic efficiency, discriminating against the transfer of ¹³C (ref. 37,54). These relatively depleted values are passed on to animals pastured and foddered on plants found under the canopy, and thus their milk and body tissues, such as carcass fats, collagen, and enamel, are expected to display relatively low δ¹³C values (ref. 63). Stable carbon and oxygen isotope values from sequential bioapatite samples can be used to investigate seasonal changes in diet. Oxygen isotopes of bioapatite precipitate in equilibrium with body water, influenced by the seasonal variability in stable isotopic composition of water (ref. 51). Thus, δ¹⁸O values provide a seasonal reference for δ¹³C values (summer: high δ¹⁸O values; winter: low δ¹⁸O values).
+
+Direct evidence of the type of plants (woody/herbaceous) consumed can be determined using the dietary β values based on δ¹⁵N CSIA of AA from incremental samples of dentine from cattle molars. These values represent the Δ¹⁵N Glx-Phe values of the plants at the base of the food web, using a known trophic offset of −4.0‰ between cattle and their diet (ref. 49). The dietary β values can then be compared with established ranges of Δ¹⁵N Glx-Phe values expected for herbaceous (−5.4 ± 2.1‰) and woody plants (−9.3 ± 1.6‰), based on modern references (ref. 38). This difference in values is likely due to the involvement of Phe in the phenylpropanoid pathway, by which lignin is produced, leading to isotopic fractionation and enrichment of the remaining Phe pool available for protein biosynthesis. This results in the more negative Δ¹⁵N Glx-Phe values observed in woody plants relative to herbaceous plants, as the former are assumed to produce more lignin.
+
+Diet values were calculated using the following enrichment values: Δ lipids-diet is +1.5‰ based on Δ lipids-collagen = 6.6 ‰ (ref. 63) and Δ collagen-diet = −5.1 ‰ (ref. 64); Δ bioap-diet = −14.5‰ (ref. 53). The spacing between diet-inferred δ¹³C values and herbivore bone collagen δ¹³C values has been proposed to be between 5.1 to 5.3‰ (ref. 64). Here we use 5.1‰ as to be comparable with previous stable isotope studies of LBK faunal material (ref. 28). The enrichment of bioapatite in ¹³C varies between species depending on the difference in physiology and size of the species (ref. 48). We have used an enrichment factor of 14.5‰ based on a recent synthesis of the spacing between diet, CO₂ breath and bioapatite in animals of different digestive systems (ref. 53). The spacing between collagen and fat δ¹³C values has been proposed to be −6.6‰ for consumers of terrestrial C₃ diets (ref. 63). The diet-inferred δ¹³C values are thus calculated by adding 1.5‰ to the δ¹³C values of the C₁₆:0 fatty acid.
+
+## Lipid residue analysis of pottery vessels and determination of δ¹³C values from dairy lipids (Mélanie Roffet-Salque)
+
+Lipid residue analyses and interpretations were based on established protocols (ref. 65). Briefly, 1 to 3 g samples were taken from potsherds and their surfaces cleaned with a modelling drill to remove exogenous lipids (e.g. soil or finger lipids arising from handling). The sherds were ground to a powder in a glass pestle with a mortar. The powdered sherd was transferred to a glass culture tube, internal standard added for quantification (n-tetratriacontane, 20 μg) and acidified methanol solution (H₂SO₄/MeOH, 4 v/v, 5 mL, 70 °C, 1 h) added. The lipids were then extracted from the aqueous phase with n-hexane (4 x 3 mL). The solvent was evaporated under a gentle stream of nitrogen to obtain the total lipid extract (TLE). Aliquots of the TLE were trimethylsilylated using N,O-bis(trimethylsilyl)trifluoroacetamide containing 1 % trimethylsilyl chloride (20 μL, 70 °C, 1 h) and re-dissolved in n-hexane for analysis by gas chromatography (GC) and GC-combustion-isotope ratio mass spectrometry (GC-C-IRMS).
+
+All GC analyses were performed on a Hewlett Packard 5890 series II chromatograph. Helium was used as carrier gas at constant flow (2 mL min⁻¹) and a flame ionization detector (FID) was used to monitor column effluent. Trimethylsilylated total lipid extracts (1 μL) were injected through an on-column injector, in track-oven mode onto a fused silica capillary column (50 m x 0.32 mm i.d.) coated with a dimethylpolysiloxane stationary phase (J&W Scientific, CP-Sil 5 CB, 0.1 μm film thickness). The oven temperature was programmed, after an isothermal hold at 50 °C for 2 min, to 300 °C at 10 °C min⁻¹, followed by a second isothermal hold at 300 °C for 10 min. Peaks were identified by comparison of retention times with those of an external standard and quantification was achieved by the internal standard method. Data acquisition and processing were carried out by the Clarity software.
+
+GC-MS analyses of trimethylsilylated aliquots were performed using a Finnigan Trace MS quadrupole MS coupled to a Trace GC. Diluted samples were introduced using a PTV injector in the splitless mode onto a 50 m x 0.32 mm i.d. fused silica capillary column coated with a HP-1 stationary phase (100 % polymethylpolysiloxane, 0.17 μm film thickness; Agilent Technologies). The initial injection port temperature was 50 °C with an evaporation phase of 1 min, followed by a transfer phase from 50 °C to 300 °C at 14.5 °C s⁻¹, followed by an isothermal hold at 300 °C. The GC oven temperature was programmed as for the GC analyses. The MS was operated in the electron ionisation (EI) mode (70 eV) with a GC interface temperature of 300 °C and a source temperature of 200 °C. The emission current was 150 μA and the MS set to acquire in the range of m/z 50-650 Daltons at 8.3 scans per s. Data acquisition and processing were carried out using the XCalibur 1.2 software. Peaks were identified based on their mass spectra, GC retention times and by comparison with the NIST mass spectral library (version 2.0a).
+
+Compound-specific δ¹³C values of FA were determined using an Isoprime 100 GC-C-IRMS system. The same GC conditions were used as for the GC analyses (HP-1 column, 100 % dimethylpolysiloxane, 50 m x 0.32 mm x 0.17 μm, Agilent Technologies). Each sample was run at least in duplicate. Instrument stability was monitored by running a fatty acid methyl ester standard mixture every 2 or 4 runs. Results were calibrated against a CO₂ reference gas injected directly in the ion source as two pulses at the beginning of each run. Instrumental precision was 0.3 ‰.
+
+Animal fats were identified as dairy lipids when their Δ¹³C (= δ¹³C₁₈:₀ – δ¹³C₁₆:₀) values were ≤ −3.1‰ as proposed by Dunne et al. (ref. 34). We examined δ¹³C values of C₁₆:₀ fatty acids from a total of 352 extracts identified as originating from animal dairy lipids (of which 135 are published (ref. 6,33,35,39-42)) from 44 sites. The species-specific identification of dairy species (cattle/sheep/goats) is not obtainable through the molecular or isotopic composition of the extracts and thus the dairy lipids from this study can come from any of these species.
+
+## δ¹³C and δ¹⁸O analysis of bioapatite samples (Rosalind E. Gillis)
+
+Cattle third molars (M3) were selected for stable isotopic analysis: 1) because the archaeozoological material was highly fragmented, making it difficult to distinguish between M1 and M2, and 2) to avoid effects from suckling and weaning. Each tooth sampled represents an individual except BIS3 and 4, which appear to come from the same individual. A minimum of eight M3s with early stages of occlusal wear were sampled from each site except at Těšetice-Kyjovice where only two teeth were sampled. Tooth surfaces were cleaned using an abrasive tungsten drill bit to remove dental calculus, cementum, and sediments. Enamel samples were removed by drilling with a diamond bit on the buccal side of the proximal lobe perpendicular to the crown growth axis. The purification protocol for stable isotopic analysis followed Balasse et al. (ref. 66). Purified enamel samples weighing between 551-650 μg were analysed on a Kiel IV device interfaced to a Delta V Advantage IRMS at the Service de Spectrométrie de Masse Isotopique du MNHN (SSMIM, Paris). The accuracy and precision of the measurements were verified using an internal laboratory calcium carbonate standard (Marbre LM normalized to NBS 19). Over the period of analysis, an average of six LM samples were analysed per run. These gave a mean δ¹³C value of 2.13 ± 0.03 ‰ (1σ) (theoretical value normalized to NBS 19 = 2.13‰) and a mean δ¹⁸O value of −1.66 ± 0.15 ‰ (1σ) (theoretical value = −1.83‰). Results are expressed relative to the VPDB standard.
+
+## Compound-specific stable isotope analysis of AAs from cattle dentine (Iain P. Kendall)
+
+Cattle third molar (M3) teeth were sequentially sampled at six points along the growth axis of each tooth. Dentine was collected as a powder, using a modelling drill with a diamond abrasive drill bit. Once formed, dentine in teeth is not remodelled, and therefore the collagen preserves the isotopic composition of the period of formation. For each sample, the AA norleucine was added as an internal standard to ca. 15 mg of dentine. Demineralisation of the inorganic fraction and hydrolysis of the collagen was achieved in one step by heating with acid (6 M HCl, 5 mL; 100°C, 24 h), and the solution blown to dryness under nitrogen. AA purification and derivatisation to N-acetyl isopropyl (NAIP) ester derivatives were prepared according to established protocols (ref. 67,68).
+
+AAs were identified by GC-FID by comparison with AA standards and quantified by comparison with a known amount of norleucine internal standard. Their δ¹⁵N values were determined by GC-C-IRMS as described in Styring, et al. (ref. 68) with a modified GC method, using DB-35 capillary column (30 m × 0.32 mm internal diameter; 0.5 µm film thickness; Agilent Technologies, UK) and the oven temperature of the GC held at 40°C for 5 min before programming at 15°C min⁻¹ to 120°C, then 3°C min⁻¹ to 180°C, then 1.5°C min⁻¹ to 210°C and finally 5°C min⁻¹ to 270°C and held for 1 min. A Nafion drier removed water and a cryogenic trap removed CO₂ from the oxidised and reduced sample. Isotopic compositions are expressed using the delta scale as follows: δ¹⁵N = Rsample/Rstandard - 1, where R is the ¹⁵N/¹⁴N ratio, and the standard is atmospheric N₂ (AIR). All δ¹⁵N values are reported relative to reference N₂ of known isotopic composition, introduced directly into the ion source in four pulses at the start and end of each run. Each reported δ¹⁵N value is the mean of triplicate determinations. A standard mixture of AAs of known δ¹⁵N values was analysed every three runs to ensure acceptable instrument performance.
+
+## Paleoenvironmental variables (Marco Zanon)
+
+The identification of past forest composition is hampered by the location of pollen cores as well as modelling uncertainties. Localised exploitation of forest resources may be underrepresented in traditional paleoecological investigations due to difficulties in capturing small-scale landscape dynamics. These difficulties may variously stem from a lack of targeted investigations, or from an absence of suitable archives. To sidestep these issues, and to proceed with a complete comparison of faunal and land cover data, we make use of interpolated reconstructions covering the whole study area. Such large-scale interpolated reconstructions may still be unable to fully resolve local dynamics, yet their use allows us to initiate a comparison between geographically spread-out datasets. The MFC data were generated from the interpolated Holocene reconstructions by Zanon et al. (ref. 48). We chose to use MFC values (%) sampled from the 7500, 7250, and 7000 cal BP (i.e. 5550, 5300, and 5050 cal BC) time slices at the location of every site in the faunal data set, and subsequently averaged.
+
+Paleoclimate information (summer and winter temperature and precipitation) is based on the modelled values presented in Sánchez Goñi, et al. (ref. 49) and available for the time slice ~7100 ± 100 years cal. BP (ca. 5150 ± 100 BC). We applied inverse distance weighted interpolation to all data points using the R package gstat 2.0-6 (ref. 69). The optimal power value for each variable was selected via leave-one-out cross-validation, using the root-mean-square error as a metric to assess the model performance. We then sampled the interpolated climate values at the location of every site within the data set.
+
+## Statistical analysis (Marco Zanon/Rosalind E. Gillis)
+
+The interpolated diet-inferred value maps (fig. 2C-D) were produced as follows: the median δ¹³C values for each site were interpolated via Inverse Distance Weighted interpolation through the R package gstat 2.0-6 (ref. 69). The optimal power value for each variable was selected via leave-one-out cross-validation, using the root-mean-square error as a metric to assess the model performance. The size of the “bullseye” depends partly on purely graphical choices (number and width of the colour intervals) and partly on the parameters of the interpolation algorithm. Statistical analysis and graphic production were carried out using the free platform R program (ref. 70) (SI 1).
+
+# References
+
+1. Bogucki, P. Animal traction and household economies in Neolithic Europe. *Antiquity* **67**, 492–503 (1993).
+
+2. Marciniak, A. *Placing animals in the Neolithic: Social zooarchaeology of prehistoric farming communities.*, 279 (University College London Press, 2005).
+
+3. Bogucki, P. *Forest farmers and stockholders. Early agriculture and its consequences in North-Central Europe.*, (Cambridge press, 1988).
+
+4. Jakucs J, Bánffy E, Oross K, Voicsek V, Ramsey C B, Dunbar E, Kromer B, Bayliss A, Hofmann D, Marshall P, Whittle A. Between the Vinča and *Linearbandkeramik* Worlds: The Diversity of Practices and Identities in the 54th–53rd Centuries cal BC in Southwest Hungary and Beyond. *Journal of World Prehistory* **29**, 267–336 (2016).
+
+5. Pavlů, I. The Neolithisation of Central Europe. *Archeologické rozhledy* **LVII**, 293–302 (2005).
+
+6. Casanova, E. et al. Spatial and temporal disparities in human subsistence in the Neolithic Rhineland gateway. *Journal of Archaeological Science* **122**, 105215, doi:10.1016/j.jas.2020.105215 (2020).
+
+7. Gronenborn, D., Antunes, N. & Rivollat, M. in *Neolithisation and the Origins of Complexity. DAI Research Cluster 1: From sedentarization to the complex society: settlement, economy, environment, cult. Proceedings of the workshops in Tehran, Lisbon and Lima. Berlin* Vol. 17 (eds Markus Reindel et al.) (Menschen - Kulturen - Traditionen / People - Cultures -Traditions, 2021).
+
+8. Whittle, A. W. R. & Bickle, P. *The first farmers of central Europe: diversity in LBK lifeways.*, (Oxbow Books, 2013).
+
+9. Pechtl, J. & Land, A. Tree rings as a proxy for seasonal precipitation variability and Early Neolithic settlement dynamics in Bavaria, Germany. *PLoS One* **14**, e0210438, doi:10.1371/journal.pone.0210438 (2019).
+
+10. Lipson, M. et al. Parallel palaeogenomic transects reveal complex genetic history of early European farmers. *Nature* **551**, 368–372, doi:10.1038/nature24476 (2017).
+
+11. Haak, W. et al. Ancient DNA from European early neolithic farmers reveals their near eastern affinities. *PLoS Biol* **8**, e1000536, doi:10.1371/journal.pbio.1000536 (2010).
+
+12. Rivollat, M. et al. Ancient genome-wide DNA from France highlights the complexity of interactions between Mesolithic hunter-gatherers and Neolithic farmers. *Sci Adv* **6**, eaaz5344, doi:10.1126/sciadv.aaz5344 (2020).
+
+13. Arbogast, R. M. & Jeunesse, C. in *The origins and spread of domestic anaimals in southwest Asia and Europe* (eds S. Colledge et al.) 271–282 (Left coast press, 2013).
+
+14. Johnson, E. V., Timpson, A., Thomas, M. G. & Outram, A. K. Reduced intensity of bone fat exploitation correlates with increased potential access to dairy fats in early Neolithic Europe. *Journal of Archaeological Science* **94**, 60–69, doi:10.1016/j.jas.2018.04.001 (2018).
+
+15. Gillis, R. E. et al. The evolution of dual meat and milk cattle husbandry in Linearbandkeramik societies. *Proceedings of the Royal Society B: Biological Sciences* **284**, 20170905 (2017).
+
+16. Hachem, L. *Le site néolithique de Cuiry-lès-Chaudardes-I. De L'analyse de la faune á la structuration sociale.*, Vol. 120 (VML, 2011).
+
+17. Austad, I. in *The cultural landscape- Past, Present and Future* (eds H. H. Birks, H. J.B. Birks, P. Emil Kaland, & D. Moe) 11–30 (Cambridge University Press, 1988).
+
+18. Gardner, A. The ecology of Neolithic environmental impacts - re-evaluation of existing theory using case studies from Hungary & Slovenia. *Documenta Praehistorica* **XXVI**, 163–184 (1999).
+
+19. Marciniak, A. et al. Animal husbandry in the Early and Middle Neolithic settlement at Kopydłowo in the Polish lowlands. A multi-isotope perspective. *Archaeological and Anthropological Sciences* **9**, 1461–1479, doi:10.1007/s12520-017-0485-6 (2017).
+
+20. Vostrovská, I., Bíšková, J., Lukšíková, H., Kočár, P. & Kočárová, R. The Environment and Subsistence of the Early Neolithic Settlement Area at Těšetice-Kyjovice, Czech Republic. *Environmental Archaeology* **24**, 248–262, doi:10.1080/14614103.2018.1424981 (2018).
+
+21. Sudhaus, D. & Friedmann, A. Holocene Vegetation and Land Use History in the Northern Vosges (France). *Quaternary Science Journal* **64**, 55–66, doi:10.3285/eg.64.2.01 (2015).
+
+22. Halstead, P., Tierney, J., Butler, S. & Mulder, Y. Leafy Hay: an Ethnoarchaeological Study in NW Greece. *Environmental Archaeology* **1**, 71–80, doi:10.1179/env.1996.1.1.71 (1996).
+
+23. Kreuz, A. Closed forest or open woodland as natural vegetation in the surroundings of Linearbandkeramik settlements? *Vegetation History and Archaeobotany* **17**, 51–64, doi:10.1007/s00334-007-0110-1 (2007).
+
+24. Hejcmanová, P., Stejskalová, M. & Hejcman, M. Forage quality of leaf-fodder from the main broad-leaved woody species and its possible consequences for the Holocene development of forest vegetation in Central Europe. *Vegetation History and Archaeobotany* **23**, 607–613, doi:10.1007/s00334-013-0414-2 (2013).
+
+25. Rasmussen, P. Analysis of goat/sheep faeces from Egolzwil 3, Switzerland: evidence for branch and twig foddering of livestock in the Neolithic. *Journal of Archaeological Science* **20** (1993).
+
+26. Balasse, M., Boury, L., Ughetto-Monfrin, J. & Tresset, A. Stable isotope insights (δ18O,δ13C) into cattle and sheep husbandry at Bercy (Paris, France, 4th millennium BC): birth seasonality and winter leaf foddering. *Environmental Archaeology* **17**, 29–44, doi:10.1179/1461410312z.0000000003 (2013).
+
+27. Fraser, R. A., Bogaard, A., Schäfer, M., Arbogast, R. & Heaton, T. H. E. Integrating botanical, faunal and human stable carbon and nitrogen isotope values to reconstruct land use and palaeodiet at LBK Vaihingen an der Enz, Baden-Württemberg. *World Archaeology* **45**, 492–517, doi:10.1080/00438243.2013.820649 (2013).
+
+28. Berthon, R., Kovačiková, L., Tresset, A. & Balasse, M. Integration of Linearbandkeramik cattle husbandry in the forested landscape of the mid-Holocene climate optimum: Seasonal-scale investigations in Bohemia. *Journal of Anthropological Archaeology* **51**, 16–27, doi:10.1016/j.jaa.2018.05.002 (2018).
+
+29. Drucker, D. G., Bridault, A., Hobson, K. A., Szuma, E. & Bocherens, H. Can carbon-13 in large herbivores reflect the canopy effect in temperate and boreal ecosystems? Evidence from modern and ancient ungulates. *Palaeogeography, Palaeoclimatology, Palaeoecology* **266**, 69–82, doi:10.1016/j.palaeo.2008.03.020 (2008).
+
+30. Gillis, R. E. & Zanon, M. in *Biodiversités, environnements et sociétés depuis la Préhistoire : nouveaux marqueurs et approches intégrées / Biodiversities, Environments and Societies since Prehistory: New Markers and Integrated Approaches. Proceedings of the 41th Rencontres internationales d'archéologie et d'histoire* (eds E. Nicoud, M. Balasse, E. Desclaux, & I. Théry-Parisot) (APDCA – CEPAM, Nice Côte d'Azur, 2021).
+
+31. Van der Merwe, N. J. & Medina, E. The Canopy Effect, Carbon Isotope Ratios and Foodwebs in Amazonia. *Journal of Archaeological Science* **18**, 249–259 (1991).
+
+32. Palmquist, D. L. & Mattos, W. Turnover of lipoproteins and transfer to milk fat of dietary (1-Carbon-14) linoleic acid in lactating cows. *Journal of Dairy science* **61**, 561–565 (1978).
+
+33. Salque, M. et al. Earliest evidence for cheese making in the sixth millennium BC in northern Europe. *Nature* **493**, 522–525 (2013).
+
+34. Dunne, J. et al. First dairying in green Saharan Africa in the fifth millennium BC. *Nature* **486**, 390–394 (2012).
+
+35. Roffet-Salque, M. & Evershed, R. P. in *Kopydłowo, stanowisko 6. Osady neolityczne z pogranicza Kujaw i Wielkopolski* (eds A. Marciniak, I. Sobkowiak-Tabaka, Marta Bartkowiak, & Mikołaj Lisowski) 133–142 (Wydawnictwo Profil-Archeo, 2016).
+
+36. Leuschner, C. & Meier, I. C. The ecology of Central European tree species: Trait spectra, functional trade-offs, and ecological classification of adult trees. *Perspectives in Plant Ecology, Evolution and Systematics* **33**, 89–103 (2018).
+
+37. Tieszen, L. T. Natural variations in the carbon isotope values of plants: Implications for Archaeology, Ecology and Paleoecology. *Journal of Archaeological Science* **18**, 227–248 (1991).
+
+38. Kendall, I. P. et al. Compound-specific δ¹⁵N values express differences in amino acid metabolism in plants of varying lignin content. *Phytochemistry* **161**, 130–138, doi:10.1016/j.phytochem.2019.01.012 (2019).
+
+39. Brychova, V. et al. Animal exploitation and pottery use during the early LBK phases of the Neolithic site of Bylany (Czech Republic) tracked through lipid residue analysis. *Quaternary International* **574**, 91–101, doi:10.1016/j.quaint.2020.10.045 (2021).
+
+40. Roffet-Salque, M. et al. in *Ludwinowo 7 - Neolithic settlement in Kuyavia* (ed J. Pyzel) 301–316 (Profil-Archeo, 2019).
+
+41. Matlova, V. et al. Defining pottery use and animal management at the Neolithic site of Bylany (Czech Republic). *Journal of Archaeological Science: Reports* **14**, 262–274 (2017).
+
+42. Salque, M. et al. New insights into the early Neolithic economy and management of animals in Southern and Central Europe revealed using lipid residue analyses of pottery vessels. *Anthropozoologica* **47**, 45–61 (2012).
+
+43. Gillis, R. E. et al. Stable isotopic insights into crop cultivation, animal husbandry, and land use at the Linearbandkeramik site of Vráble-Veľké Lehemby (Slovakia). *Archaeological and Anthropological Sciences* **12**, 256, doi:10.1007/s12520-020-01210-2 (2020).
+
+44. Oelze, V. M. et al. Early Neolithic diet and animal husbandry: stable isotope evidence from three Linearbandkeramik (LBK) sites in Central Germany. *Journal of Archaeological Science* **38**, 270–279, doi:10.1016/j.jas.2010.08.027 (2011).
+
+45. Dürrwächter, C., Craig, O. E., Collins, M. J., Burger, J. & Alt, K. W. Beyond the grave: variability in Neolithic diets in Southern Germany? *Journal of Archaeological Science* **33**, 39–48, doi:10.1016/j.jas.2005.06.009 (2006).
+
+46. Kendall, I. P., Gillis, R. E., Balasse, M. & Evershed, R. P. in *Ludwinowo, stanowisko 7. Osada neolityczna na Kujawach / Ludwinowo, site 7. Neolithic settlement in Kuyavia Ocalone Dziedzictwo Archeologiczne* (ed J. Pyzel) 277–288 (PROFIL-ARCHEO, 2019).
+
+47. Kovačiková, L., Bréhard, S., Šumberová, R., Balasse, M. & Tresset, A. The new insights into the subsistence and early farming from Neolithic settlements in Central Europe: the archaeozoological evidence from the Czech Republic. *Archaeofauna* **21**, 71–79 (2012).
+
+48. Zanon, M., Davis, B. A. S., Marquer, L., Brewer, S. & Kaplan, J. O. European Forest Cover During the Past 12,000 Years: A Palynological Reconstruction Based on Modern Analogs and Remote Sensing. *Front Plant Sci* **9**, 253, doi:10.3389/fpls.2018.00253 (2018).
+
+49. Sánchez Goñi, M. F. et al. The expansion of Central and Northern European Neolithic populations was associated with a multi-century warm winter and wetter climate. *The Holocene* **26**, 1188–1199, doi:10.1177/0959683616638435 (2016).
+
+50. Balasse, M. et al. Seasonal calving in European Prehistoric cattle and its impacts on milk availability and cheese-making. *Sci Rep* **11**, 8185, doi:10.1038/s41598-021-87674-1 (2021).
+
+51. Gat, J. R. Oxygen and hydrogen isotopes in the hydrologic cycle. *Annual Revue Earth Planet Science* **24**, 225–262 (1996).
+
+52. Balasse, M. Potential biases in sampling design and interpretation of intra-tooth isotope analysis. *International Journal of Osteoarchaeology* **13**, 3–10, doi:10.1002/oa.656 (2003).
+
+53. Cerling, T. E. et al. CH₄/CO₂ Ratios and carbon isotope enrichment between diet and breath in herbivorous mammals. *Frontiers in Ecology and Evolution* **9**, doi:10.3389/fevo.2021.638568 (2021).
+
+54. Farquhar, G. D., Ehleringer, J. & Hubick, K. T. Carbon Isotope Discrimination And Photosynthesis. *Annual Review of Plant Biology* **40**, 503–537 (1989).
+
+55. Hedges, R. et al. in *The first farmers of central Europe: diversity in LBK lifeways* (eds A. W. R Whittle & P Bickle) Ch. 9, 343–384 (Oxbow, 2013).
+
+56. Copley, M. S. et al. Direct chemical evidence for widespread dairying in prehistoric Britain. *PNAS* **100**, 1524–1529 (2003).
+
+57. Manning, K. et al. in *The origins and spread of domestic animals in Southwest Asia and Europe* (eds S. Colledge et al.) Ch. 12, 237–252 (Left coast press, 2013).
+
+58. Pokorný, P. et al. Mid-Holocene bottleneck for central European dry grasslands: Did steppe survive the forest optimum in northern Bohemia, Czech Republic? *The Holocene* **25**, 716–726, doi:10.1177/0959683614566218 (2015).
+
+59. Hall, S. J. & Hall, J. G. Inbreeding and population dynamics of the Chillingham cattle (*Bos taurus*). *Journal of Zoology, London* **216**, 479–493 (1988).
+
+60. Labussière, J. in *Biology of Lactation* (eds J. Martinet, L.-M. Houdebine, & H.H. Head) 307–343 (INRA, 1999).
+
+61. Halstead, P. & O’Shea, J. in *Bad Year Economics. Cultural Responses to Risk and Uncertainty* (eds P. Halstead & J. O´Shea) 1–7 (Cambridge University Press, 1989).
+
+62. Ingold, T. *Hunter, Pastoralists and Ranchers Reindeer Economics and thier transformations.*, (Cambridge University Press, 1988).
+
+63. Tieszen, L. T. & Fagre, T. in *Prehistoric human bone: Archaeology at the molecular level* (eds J. B. Lambert & G. Grupe) (Springer-Verlag, 1993).
+
+64. Schoeninger, M. J. & DeNiro, M. J. Nitrogen and Carbon Isotopic composition of bone collagen from marine and terrestrial animals. *Geochimica et Cosmochimica Acta* **48**, 625–639 (1984).
+
+65. Correa-Ascencio, M. & Evershed, R. P. High throughput screening of organic residues in archaeological potsherds using direct methanolic acid extraction. *Analytical Methods* **6**, 1330–1340 (2014).
+
+66. Balasse, M. Reconstructing dietary and environmental history from enamel isotopic analysis: time resolution of intra-tooth sequential sampling. *International Journal of Osteoarchaeology* **12**, 155–165, doi:10.1002/oa.601 (2002).
+
+67. Corr, L. T., Berstan, R. & Evershed, R. P. Optimisation of derivatisation procedures for the determination of δ¹³C values of amino acids by gas chromatography/combustion/isotope ratio mass spectrometry. *Rapid Commun Mass Spectrom* **21**, 3759–3771, doi:10.1002/rcm.3252 (2007).
+
+68. Styring, A. K. et al. Practical considerations in the determination of compound-specific amino acid delta15N values in animal and plant tissues by gas chromatography-combustion-isotope ratio mass spectrometry, following derivatisation to their N-acetylisopropyl esters. *Rapid Commun Mass Spectrom* **26**, 2328–2334, doi:10.1002/rcm.6322 (2012).
+
+69. Gräler, B., Pebesma, E. & Heuvelink, G. Spatio-Temporal Interpolation using gstat. *The R Journal* **8**, 204–218 (2016).
+
+70. R: A language and environment for statistical computing. R Foundation for Statistical Computing, (URL https://www.R-project.org/. Vienna, Austria, 2017).
+
+# Supplementary Files
+
+- [SI1Statisticalinformation.pdf](https://assets-eu.researchsquare.com/files/rs-1419935/v1/4a6f66353764f48554b5beda.pdf)
+- [SITABLE17.xlsx](https://assets-eu.researchsquare.com/files/rs-1419935/v1/27f2ba2383de893b2a453405.xlsx)  
+  SI Dataset 1 to 7
+- [SIFigure1.pdf](https://assets-eu.researchsquare.com/files/rs-1419935/v1/66b61e9777f1c9b3d12dd322.pdf)

0e1902d9b362bf87da2f4c01091362768c84791f9a3091efa963dfe66addc35b/preprint/images_list.json

@@ -0,0 +1,34 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Dependence of motility and its cost on the expression of E. coli flagellar genes in nutrient-rich medium. a, Schematic representation of the flhDC operon in strain MG1655, with native (MG1655 WT) or inducible (Ptac) regulation of expression. The native regulatory region of the flhDC operon, including the upstream IS1H insertion element, was replaced in the Ptac strain with the tac promoter inducible by isopropyl \u03b2-d-1-thiogalactopyranoside (IPTG); an additional copy of the lacI gene (Lac repressor) was inserted upstream of the tac promoter to reduce the basal expression. b, Flow cytometry measurement of PfliC-GFP reporter activity in mid-exponential cultures of MG1655 WT or its Ptac derivative grown in tryptone broth (TB) medium. Flagellar gene expression in the Ptac strain was induced with the indicated concentrations of IPTG (in \u00b5M). Flow cytometry histograms of three biological replicates (n = 3) are shown as violin plots in different hues (AU \u2013 arbitrary units). c, PfliC reporter activity determined either as the median GFP intensity at mid-exponential growth phase in flow cytometry (FC) measurements (black symbols) or as the peak of GFP expression normalized by OD600 in plate reader (PR) cultures (red symbols). Both data sets were aligned by MG1655 WT expression (horizontal dashed line). Points are the mean values (n = 3) and error bars are the standard deviations (mean\u2009\u00b1\u2009s.d.). d, Dependence of the average cell swimming velocity in cultures of the indicated E. coli strains on the activity of the PfliC reporter as determined by flow cytometry. The average swimming velocity was calculated as the product of the swimming fraction and the swimming velocity of motile cells (see Extended Data Fig. 1c,d for individual values). Motility and reporter expression were determined separately for each replicate culture (indicated by individual symbols). e, The growth fitness cost of flagellar gene expression. Fitness cost was determined as the percentage of cells (in %) of either the MG1655 WT or Ptac strain induced by different concentrations of IPTG in co-cultures with the non-flagellated \u0394flhC strain after 24 h of growth with shaking (200 rpm) in TB medium. The strains were initially co-inoculated in a 1:1 ratio. PfliC activity measured in the plate reader was used to plot the data; mean\u2009\u00b1\u2009s.d. (n = 3) is shown for both parameters.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Limitation of E. coli motility at high expression of flagellar genes. a-c, Changes in E. coli flagellation with varying expression of flagellar genes. Fluorescence microscopy images of MG1655 WT or Ptac cells grown either without (Ptac0) or with 50 \u03bcM IPTG (Ptac50), stained with amino-specific fluorescent dye to visualize flagella (a). Corresponding quantification of the number (b, N flagella) and length (c, in \u00b5m) of flagella as a function of PfliC activity measured by flow cytometry (FC, n = 3 biological replicates, mean\u2009\u00b1\u2009s.d.). Data from the same experiments were used to quantify both the number and length of flagella; n = 100 cells from different fields of view (b) and n = 35-50 flagellar filaments in 10-20 cells (c). See Extended Data Fig. 3 for value distributions and significance analysis. d, Dependence of swimming velocity on the number of flagellar filaments, predicted by the RFT physical model of the multi-flagellated microswimmer (see Supplementary Note 2 for details). Our RFT model takes into account that cells with a higher number of flagella also have longer filaments, as observed experimentally.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Motility of E. coli as a function of gene expression in minimal medium. a, Flow cytometry measurements of the PfliC-GFP reporter of MG1655 WT or Ptac strains grown to mid-exponential phase in M9 minimal medium, with either glucose (left) or succinate (right) as the sole carbon source. Labels are as in Fig. 1b. Flow cytometry histograms of three biological replicates are shown as violin plots in different hues (AU \u2013 arbitrary units). b, Dependence of average swimming velocity on the median PfliC reporter activity (flow cytometry, FC) for the indicated carbon sources and strains. Each dot represents an independent culture (biological replicate) for which both expression (PfliC reporter activity) and swimming were determined. c, Percentage of GFP-positive cells within the population of the MG1655 WT, Ptac and \u0394ydiV (lacks YdiV, the negative regulator of FlhDC; open symbols) strains as a function of median PfliC reporter activity, both measured by flow cytometry as in (a). Each symbol represents an independent culture. The inset describes different conditions used for the starting culture: the overnight culture pre-grown in TB (TB) or M9 glucose (M9+glu) was diluted to the fresh TB or M9 media (1:100 and 1:1000 indicate the dilution).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Motility of natural E. coli isolates. a, Relation between flagellar regulon activity and motility for representative ECOR strains (indicated here and throughout by their number in the collection) compared to MG1655 WT and Ptac strains; corresponding inducer concentrations (IPTG, \u00b5M) used for the Ptac strain are indicated by numbers in red. All E. coli cultures were grown in a liquid TB medium (indicated by L). The same mid-exponential cell culture was used to measure the PfliC reporter activity in the plate reader (GFP fluorescence normalized to OD600) and average swimming velocity (see Methods for details). Each point represents the mean value for both parameters (n = 3), with error bars indicating the standard deviations. b, Diameters of spreading zones formed by MG1655 WT, Ptac and ECOR strains in porous 0.27% TB agar, measured after 4-5 h incubation at 34\u00b0C (n = 3; mean\u2009\u00b1\u2009s.d.). c, Correlation between PfliC reporter activity of in E. coli strains grown in liquid (L) or semi-solid (indicated by S) medium (0.5% TB agar) (n = 3; mean\u2009\u00b1\u2009s.d.). d, Dependence of swimming velocity on\u00a0 PfliC activity for ECOR, MG1655 WT and Ptac strains grown on semi-solid (S) medium (n = 3, mean\u2009\u00b1\u2009s.d.). Data for other ECOR strains are shown in Extended Data Fig. 10. \u00a0",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

0e1902d9b362bf87da2f4c01091362768c84791f9a3091efa963dfe66addc35b/preprint/preprint.md

@@ -0,0 +1,251 @@
+# Abstract
+
+Regulatory strategies that allow microorganisms to balance their investment of limited resources in different physiological functions remain poorly understood, particularly for numerous cellular functions that are not directly required for growth. Here, we investigate the allocation of resources to flagellar swimming, the most prominent and costly behavior in bacteria that is not directly required for growth. We show that the dependence of motile behavior on gene expression is determined by the hydrodynamics of propulsion, which limits the ability of bacteria to increase their swimming by synthesizing more than a critical number of flagellar filaments. Together with the fitness cost of flagellar biosynthesis, this defines the physiologically relevant range of investment in motility. Gene expression in all *E. coli* isolates tested falls within this range, with many strains maximizing motility under nutrient-rich conditions, particularly when grown on a porous medium. The hydrodynamics of swimming may further explain the bet-hedging behavior observed at low levels of motility gene expression.
+
+[Biological sciences/Microbiology/Bacteria/Bacterial systems biology](/browse?subjectArea=Biological%20sciences%2FMicrobiology%2FBacteria%2FBacterial%20systems%20biology) [Biological sciences/Microbiology/Bacteria/Bacterial physiology](/browse?subjectArea=Biological%20sciences%2FMicrobiology%2FBacteria%2FBacterial%20physiology) [Biological sciences/Biophysics/Motility/Cellular motility](/browse?subjectArea=Biological%20sciences%2FBiophysics%2FMotility%2FCellular%20motility)
+
+# Introduction
+
+Microorganisms, like all living systems, must achieve multiple physiological objectives that may change when encountering new environments. To perform successfully, microorganisms have therefore evolved numerous regulatory mechanisms responsible for allocating limited resources to specific physiological functions<sup>1, 2</sup>. Bacteria, including *Escherichia coli*, have become convenient models to address this fundamental resource allocation problem<sup>3</sup>, with a primary focus on proteome partitioning<sup>4–7</sup>. To allocate their proteomic resources into protein biosynthesis as a function of growth rate, bacteria appear to obey linear rules known as growth laws<sup>4, 5</sup>: the fraction of the proteome responsible for biomass production expands with growth rate, whereas the fraction responsible for nutrient uptake and catabolism decreases with growth rate. This leads to the negative linear relation between the expression of carbon catabolic genes and growth rate, known as the C-line<sup>5</sup>, which has been proposed to maximize growth. However, although growth maximization is an important research allocation strategy<sup>4–6, 8</sup>, it is not always the case<sup>9, 10</sup> and cells may instead prioritize other targets such as energy yield or stress response<sup>11, 12</sup>. Furthermore, while previous studies have mostly focused on the optimized expression of catabolic<sup>4, 5, 8, 10</sup>, anabolic<sup>5, 6</sup>, or ribosomal<sup>2, 5, 6</sup> genes, microbial strategies for resource allocation to multiple functions not directly required for growth remain unclear<sup>13</sup>.
+
+The most prominent example of such a costly physiological function is swimming motility. Motile bacteria are propelled by the rotation of long helical flagellar filaments powered by a motor that is typically proton-driven<sup>14</sup>. Motility enables bacteria to follow spatial gradients of nutrients or harmful chemicals sensed by the chemotaxis signaling pathway<sup>15, 16</sup>. Motility consumes several percent of total cellular resources in *E. coli* and other bacteria<sup>17–19</sup>, primarily due to the protein budget required for the biosynthesis of flagella<sup>20, 21</sup>. Consistent with this high cost, several studies have observed a trade-off between growth and motility in *E. coli*<sup>21–24</sup>. However, the exact dependence of this trade-off on the absolute level of resource allocation to swimming motility remains uninvestigated.
+
+Interestingly, the flagellar regulon in *E. coli* is controlled by catabolite repression<sup>25</sup>, such that flagellar gene expression increases in minimal medium during growth on poor carbon sources in accordance with the C-line<sup>7, 21</sup>. The physiological relevance of such an investment strategy remains debated. One proposed explanation is that it ensures an anticipatory allocation of resources towards motility, in proportion to the potential benefit of finding additional nutrient sources via chemotaxis, which is higher in nutrient-poor environments<sup>21</sup>. Alternatively, it has been suggested that the number of flagella is tuned to match growth rate-dependent changes in cell size<sup>23</sup>.
+
+In this study, we quantified the relation between the expression of motility genes and motile behavior, as well as the impact of motility on the growth fitness of *E. coli*. We demonstrate that major limitations on resource investment in motility, at both high and low levels of gene expression, arise from hydrodynamic constraints on bacterial swimming. Together with the fitness cost of flagellar synthesis and operation, this creates the physiologically relevant range within which the expression level of motility genes can vary depending on the conditions. We observe that within this range, *E. coli* follows different strategies of resource allocation towards motility depending on the medium, growth rate and isolate.
+
+# Results
+
+Native regulation of motility genes in nutrient-rich medium maximizes swimming while limiting the cost of expression
+
+To investigate how motility and growth depend on the expression of flagellar genes, we engineered a derivative of *E. coli* K-12 strain MG1655 with titratable expression of the *flhDC* operon that encodes the master activator of the entire flagellar regulon (Fig. 1a, Supplementary Table 1 and Methods). Expression of the flagellar regulon at different levels of *Ptac*-flhDC induction was quantified using a fluorescent reporter for flagellin (*fliC* gene) promoter activity (P<sub><em>fliC</em></sub>), which was previously shown to efficiently report the production of flagella in *E. coli* <sup>20, 21, 26</sup>. Reporter activity was measured using either a plate reader to follow changes in in the mean expression over time (Extended Data Fig. 1a, b), or flow cytometry to determine the distribution of single-cell expression levels within the cell population at a defined time point in mid-exponential phase (Fig. 1b). We confirmed that both readouts yielded similar results for *E. coli* cultures grown in nutrient-rich tryptone broth (TB) medium, with native MG1655 (wild-type; MG1655 WT) expression falling at an intermediate level within the range covered by the inducible *Ptac* strain (Fig. 1c and Extended Data Fig. 1b).
+
+To understand how motility changes as a function of gene expression, we characterized swimming behavior in populations of MG1655 WT and *Ptac* cells using differential dynamic microscopy <sup>27</sup> (see Supplementary Note 1 and Extended Data Fig. 2). We observed that population-averaged cell swimming velocity initially increased with expression at low levels of induction, but saturated at high levels of expression (Fig. 1d). Notably, this saturation occurred around the level of motility gene expression seen in the wild-type strain. A similar pattern was observed when the fraction of well-swimming cells within the population, as determined by our motility assay, and the swimming velocity of only these cells were plotted individually (Extended Data Fig. 1c, d). The cell swimming velocity at the highest expression level was even slightly reduced (Fig. 1d and Extended Data Fig. 1c). Two other derivatives of *E. coli* K-12, W3110 <sup>28</sup> and RP437 (the latter is commonly studied as a wild type for *E. coli* chemotaxis <sup>29</sup>), both showed a similar relation between flagellar gene expression and motility, but were slightly less motile than MG1655 WT (Fig. 1d). The poorer swimming performance of RP437 may be a consequence of its extensive mutagenization <sup>29</sup>, and a previous study showed that the motility of this strain can be improved by experimental evolution <sup>20</sup>.
+
+We further investigated the effect of motility on fitness by co-culturing CFP-labeled MG1655 WT or *Ptac* strains with a non-flagellated YFP-labeled ΔflhC strain. The fitness cost of flagellar regulon activity over a culture passage was determined as the reduction in relative cell number of the tested strain in the co-culture from the initial 50% at inoculation <sup>20, 21</sup>. This cumulative fitness cost gradually increased with the level of motility genes expression over the entire range of induction tested (Fig. 1e). Thus, expression of motility genes beyond the native level in *E. coli* K-12 strains does not appear to provide any additional benefit, but nevertheless imposes an increasing fitness cost.
+
+## Hydrodynamic constraints limit cell velocity at high levels of flagellar production
+
+The saturation of *E. coli* motility at high levels of flagellar gene expression could be due either to some bottleneck in the biogenesis of functional flagella or to limits in the physical propulsion by multiple flagella. To distinguish between these two possibilities, we first determined how the activity of the flagellar regulon corresponds to changes in flagellation. Staining flagella with an amino-specific fluorescent dye <sup>30</sup> revealed a clear dependence of the number and length of flagella on the expression of the flagellar regulon (Fig. 2a). The average number of flagellar filaments per cell showed an approximately linear increase with the activity of the P<sub><em>fliC</em></sub> reporter (Fig. 2b, Extended Data Fig. 3a). The length of flagellar filaments also showed a moderate increase followed by an apparent saturation (Fig. 2c, Extended Data Fig. 3b). These results were consistent with increased amounts of intra- and extracellular flagellin, determined by immunoblotting (Extended Data Fig. 4). Thus, *E. coli* cells can synthesize more flagella at levels of motility gene expression that exceed those of wild-type cells, but this increase does not translate into higher swimming velocity.
+
+Alternatively, this saturation of swimming with flagella number could be explained by the physics of *E. coli* motility. The hydrodynamics of flagella-propelled bacterial swimming is well understood and can be captured by relatively simple mathematical models such as resistive force theory (RFT) <sup>31, 32</sup>. We therefore used RFT to describe the swimming of a multi-flagellated bacterium, where multiple flagella form a tight bundle that rotates to propel the cell (Supplementary Note 2 and Extended Data Fig. 5a). Based on our experimental measurements (Extended Data Fig. 5b,c), we assume that the flagellar motors operate at a constant speed that does not depend on the number of flagella, which may be the maximum speed of the motor torque-speed relationship. Indeed, the load per motor is low and decreases as the number of flagella increases (Extended Data Fig. 5g), because now multiple motors share the torque generation necessary for bundle rotation and cell propulsion. Our model predicts that swimming velocity should initially increase with motility gene expression and then saturate, in agreement with the experimental data (Fig. 2d and Extended Data Fig. 5d-f). The initial increase stems from the increase of flagellar length and the increased thickness of the bundle formed by more flagella. Saturation then occurs in the RFT model at high number of filaments because the viscous drag of the cell body becomes negligible compared to the drag of the flagella themselves. As a consequence, any increase in thrust resulting from adding more flagella is offset by an equal increase in viscous drag, since the two have identical dependencies on flagellar length and bundle thickness. Although our model is clearly simplified, in particular, does not capture all the complexity of flagella bundle hydrodynamics <sup>33</sup>, it strongly indicates that the ability of *E. coli* to increase its swimming velocity by increasing the number and length of flagella is indeed limited by the hydrodynamics and mechanics of flagellar propulsion in viscous media.
+
+## Motility gene expression follows the potential benefit of chemotaxis under carbon-limited conditions
+
+Since expression of the flagellar regulon is under catabolite repression during carbon-limited growth in minimal media, we asked whether this regulation serves to maximize swimming, as observed in nutrient-rich medium, or whether it optimizes an alternative target. Consistent with its C-line-dependent regulation <sup>7, 21, 25</sup>, the expression of motility genes in the MG1655 WT strain grown in the minimal medium was much lower in the presence of a good (glucose) than a poor (succinate) carbon source (Fig. 3a). Expression in the *Ptac* strain at a given induction was also lower during growth on glucose, but this dependence was weaker, as expected for promoters that are not catabolite repressed <sup>34</sup>. Despite these differences, both swimming velocity (Fig. 3b) and growth fitness cost (Extended Data Fig. 6) in the *Ptac* strain showed the same dependence on motility gene expression for both carbon sources. MG1655 WT levels also fit to this curve, but unlike growth in nutrient-rich medium, the native activity of the flagellar regulon clearly does not maximize swimming velocity in this case.
+
+Instead, we hypothesized that native gene expression under carbon-limited growth might correlate with the potential benefit that could be achieved in a given carbon source by performing chemotaxis towards sources of additional nutrients, as proposed before <sup>21</sup>. Following this previous study, we measured the benefit of chemotaxis by providing localized sources of amino acids in co-culture between the *Ptac* strain (labeled with CFP) and its motile but non-chemotactic ΔcheY derivative (labeled with YFP) for different levels of motility gene induction (Extended Data Fig. 7). While the benefit of chemotaxis saturated at high levels of motility gene expression in both carbon sources, saturation occurred at much lower expression in the presence of glucose, with the point of saturation close to the native level of expression in the respective carbon source.
+
+Another notable finding was the appearance of two distinct subpopulations, with almost negative and strongly positive expression, at low average levels of reporter activity in the *Ptac* strain (Fig. 3a and Extended Data Fig. 8). Interestingly, this separation appeared to be a function of the average reporter activity and did not depend on the carbon source (Fig. 3a and Fig. 3c). In this low expression range, the proportion of positive cells in the population increased up to a critical level of expression, after which the distribution became unimodal and it was rather the mean of the positive peak that increased with induction. Motility gene expression in MG1655 WT cells was above the critical level where bimodal behavior becomes apparent, even in culture grown on glucose. To investigate whether native regulation could also exhibit bimodality, we further reduced motility gene expression in wild-type cells by prolonged growth under catabolite repression in glucose, either by using a higher dilution of the TB-grown overnight culture or by pre-growing the overnight culture in glucose (Fig. 3c, Extended Data Fig. 9a). Indeed, both conditions reduced P<sub><em>fliC</em></sub> activity in the MG1655 WT cell population and revealed a bimodal pattern similar to that observed in the *Ptac* strain. Bimodality was also observed for a non-induced *Ptac* strain grown in TB (Fig. 1c and Fig. 3c). Thus, bimodality appears to depend solely on the expression level and not on the details of transcriptional regulation of the flhDC operon or on the growth medium.
+
+Motility gene expression in *E. coli* has previously been shown to be pulsatile <sup>26, 35</sup> and this may be the cause of the observed bimodality. In the closely related species *Salmonella enterica*, motility genes are also known to exhibit bistable expression <sup>36</sup>. Both bistability (in *S. enterica*) and pulsatility (in *E. coli*) of expression were attributed to negative regulation of FlhDC activity by YdiV (RflP) <sup>37</sup>, with organism-specific differences in the topology of the YdiV regulatory circuit <sup>35, 37</sup>. We therefore tested whether regulation by YdiV could be responsible for the emergence of bimodality in our experiments. As expected, the expression level of motility genes in a ΔydiV strain was elevated, and it was above the bimodality threshold in glucose even when the culture was inoculated from TB at a 1:1000 dilution (Fig. 3c and Extended Data Fig. 9b). However, when the expression level was sufficiently lowered by pre-growth in glucose, two distinct subpopulations could be clearly observed in the ΔydiV strain, suggesting that negative regulation by YdiV is not sufficient to explain the bimodal activation of the P<sub><em>fliC</em></sub> reporter.
+
+## Activity of the flagellar regulon in natural isolates of *E. coli*
+
+Finally, to investigate how investment in motility varies among *E. coli* strains that may have adapted to different ecological niches, we used the ECOR collection, which contains 72 isolates from different hosts and geographical regions <sup>38</sup>. From this collection, we first selected 61 strains that were sensitive to kanamycin and thus transformable with the P<sub><em>fliC</em></sub> reporter plasmid, and then discarded 23 non-swimming isolates that did not spread in porous (0.27%) TB agar. From the remaining 38 spreading isolates, a subset of 24 strains with moderate and good spreading abilities was chosen for further investigation (Supplementary Table 2).
+
+Although the activity of the P<sub><em>fliC</em></sub> reporter varied widely among the TB-grown ECOR strains, it was consistently below or similar to that of the MG1655 WT strain (Fig. 4a and Extended Data Fig. 10a), indicating that the investment in motility by natural *E. coli* isolates is under similar limitation as in the K-12 strains. However, the swimming velocity of the majority of ECOR strains grown in liquid TB medium was lower than that of MG1655 WT and *Ptac* strains at similar levels of P<sub><em>fliC</em></sub> reporter activity (Fig. 4a and Extended Data Fig. 10a). Since previous studies showed that the motility of several pathogenic *E. coli* strains <sup>39</sup> and other bacteria <sup>40</sup> can be activated when cells are grown on a surface or in a porous medium, we measured the ability of ECOR strains to spread in porous 0.27% TB agar. Indeed, the spreading of most ECOR strains, including those that were poorly motile when grown in liquid, was comparable to that of MG1655 WT and *Ptac* (Fig. 4b).
+
+A possible explanation for this difference could be increased expression of motility genes in cells grown in porous media or on a semi-solid agar surface, where flagella rotate under high load <sup>39, 41–43</sup>. We therefore measured the activity of the P<sub><em>fliC</em></sub> reporter in cultures grown on 0.5% TB agar plates. In this case, expression in individual strains correlated well with their spreading (Extended Data Fig. 10b). While we indeed observed an upregulation of reporter activity in such surface-grown compared to liquid-grown cultures for a few isolates (e.g. ECOR-72), this was not the case for the majority of ECOR strains (Fig. 4c, Extended Data Fig. 10c and Supplementary Table 2). However, when the motility of cells grown on an agar surface was subsequently analyzed in motility buffer (see Methods for details), the average cell swimming velocity was indeed higher for many ECOR strains compared to liquid-grown cultures, now showing a dependence of swimming velocity on expression similar to the MG1655 WT and *Ptac* strains (Fig. 4d, Extended Data Fig. 10d and Supplementary Table 2). Thus, the observed poor motility of many ECOR isolates grown in liquid medium cannot be generally explained by low activity of the flagellar regulon but rather indicates some deficiency in flagellar assembly or function in liquid-grown cell. Notably, however, both motility gene expression and swimming of all ECOR strains were always below or comparable to that of MG1655 WT, further supporting the fundamental nature of limitation imposed on *E. coli* motility by hydrodynamics.
+
+# Discussion
+
+How microorganisms regulate the allocation of their limited cellular resources under varying environmental conditions remains an open question. Although optimality theory<sup>50</sup> predicts that gene expression levels should have been evolutionarily tuned to maximize an organism’s fitness, such optimization is a multifactorial problem with mostly uncharacterized constraints and trade-offs between conflicting optimization goals. Particularly challenging to understand are microbial strategies for allocating resources to costly functions that do not directly benefit growth or are not used under certain conditions, which can account for up to half of cellular protein resources<sup>13, 44, 45</sup>.
+
+Here, we investigated resource allocation to flagellar motility, the most prominent of such non-growth related cellular functions in bacteria, by titrating the expression of the flagellar gene regulon and quantifying its impact on *E. coli* motility. We observed that the biogenesis of the motility apparatus, i.e., the number of flagella and their length, shows a dependence on gene expression over a wide range, demonstrating that *E. coli* can increase its flagellation beyond the level observed in wild-type strains with the native regulation of gene expression. The effect on growth fitness increases proportionally with resource investment, too, consistent with flagella biosynthesis being the major component of motility costs<sup>20, 21</sup>. In contrast, cell swimming velocity increases as a function of motility gene expression until the number of flagella reaches ~ 5, but saturates above this level. This dependence of swimming velocity on the number and length of filaments was well captured by a mathematical model describing the swimming of a multi-flagellated bacterium using the resistive force theory, suggesting that the observed saturation of cell velocity is the consequence of hydrodynamic constraints on *E. coli* motility. Further supporting the general nature of this relation, not only the K-12 strains, but also the majority of motile natural isolates of *E. coli* mapped to the same unique expression-swimming relation under conditions that favored their motility.
+
+Strikingly, although the activity of the flagellar regulon differed among the wild-type *E. coli* strains tested and between conditions, it was invariably confined to the sub-saturating part of the expression-swimming relation. In a fraction of the strains, including K-12 derivatives and several natural isolates, motility gene expression in the nutrient-rich medium was most likely selected to maximize swimming velocity. This could indicate a high importance of swimming, e.g., for colonization of the environment<sup>19, 46</sup>. However, even in these strains, expression levels remain bounded by the critical level at which swimming velocity saturates, indicating that cells avoid unnecessary resource expenditures that provide no additional benefit. Expression levels in other *E. coli* isolates map to different points on the expression-swimming curve, covering the range below saturation of motility. Such heterogeneity could be due to different selection pressures on motility in the ecological niches occupied by these isolates, which is consistent with findings that differences in motility allow coexistence and niche segregation between *E. coli* strains, both *in vitro*<sup>25</sup> and in an animal host<sup>47</sup>.
+
+While many *E. coli* strains, including the K-12 derivatives and some natural isolates, swim similarly well when grown in either liquid or porous media, we observed that most natural isolates showed good motility only when grown in porous or semi-solid media, possibly reflecting conditions in the animal gut. The mechanism underlying this effect needs to be further characterized, but it does not seem to be explained by a previously reported mechanosensing-based upregulation of the entire flagellar gene regulon in porous media<sup>39</sup>. Many *E. coli* isolates swim poorly when grown in liquid despite having comparatively high activity of the flagellar regulon, and only achieve the motility expected based on their gene expression when grown on semi-solid medium. For these isolates, growth in liquid may result in the assembly of poorly functional motors or flagella. A potential mechanism for such flagellar motor remodeling in *E. coli* could be the previously described recruitment of additional force-generating units under load<sup>41, 43</sup>, but it remains to be seen whether this recruitment is sufficiently long-lasting to account for these isolates retaining high motility even after transfer to a liquid environment.
+
+When grown under carbon limitation, *E. coli* cells exhibited similar expression-swimming and expression-cost relations in both good and poor carbon sources, despite expected growth-dependent changes in cell size<sup>23</sup>. However, under these conditions, native expression of *E. coli* motility genes clearly does not maximize swimming. Instead, it correlates well with saturation of the benefit that *E. coli* could derive from chemotaxis-dependent accumulation to sources of additional nutrients, consistent with the strategy of anticipatory investment in motility<sup>21</sup>.
+
+The reduced activity of the flagellar regulon under carbon-limited growth revealed another prominent feature of its regulation in *E. coli*, namely the appearance of two distinct subpopulations of cells below a certain threshold of average P<sub><em>fliC</em></sub> reporter activity. This bimodality may be related to the recently described pulsatile activation of flagellar genes in *E. coli* at intermediate expression levels of the master regulator FlhDC<sup>26, 35</sup>. However, whereas this previous work concluded that pulsatility of expression is caused by the negative regulation of FlhDC by YdiV<sup>26</sup>, this regulation was not sufficient to explain the bimodality in our experiments. Furthermore, based on the established quantitative relation between gene expression and swimming motility, we could speculate on possible physiological reasons for such differentiation into distinct subpopulations. The bimodality of gene expression in microorganisms is commonly interpreted as stochastic bet-hedging behavior, which may be a better strategy in an unpredictable environment than a single adaptive phenotype<sup>48–50</sup>. While similar arguments were used to rationalize the differentiation of a bacterial population into motile and non-motile phenotypes<sup>26, 35, 36</sup>, here we propose a different, though not mutually exclusive, explanation. We noticed that the bimodality in our experiments occurs at the average expression that is below the level that would correspond to approximately two flagella per cell. Given that swimming with fewer than two flagella becomes inefficient, we argue that the observed bifurcation serves to avoid this “average”, poorly motile phenotype, which is unable to benefit from motility but still pays the fitness cost. Such “enforced” bet hedging may provide an alternative explanation for evolutionarily selected bimodality of gene expression, which is likely to apply not only to bacterial motility, but also to other cases where an intermediate phenotype is less fit than either of the extreme phenotypes. Thus, the hydrodynamics of flagella-mediated motility may not only determine the upper limit of swimming velocity at high levels of motility gene expression, but may also explain its bimodality at low levels of expression.
+
+# Methods
+
+## Strains and growth conditions
+
+All *E. coli* strains, including natural isolates from the *E. coli* Reference Collection (ECOR) <sup>38</sup> and plasmids used in this study are described in Supplementary Tables 1 and 2. The strain with inducer-dependent expression of *flhDC* operon (*Ptac*) was constructed previously <sup>21</sup> by replacing the native regulatory region of the *flhDC* operon, including the upstream *IS1H* insertion element, in the MG1655 *Δflu* background with the *tac* promoter inducible by isopropyl β-d-1-thiogalactopyranoside (IPTG). To reduce the basal expression of the *flhDC* operon, the *lacI* gene encoding the Lac repressor was additionally inserted upstream of the *tac* promoter. Deletion of the *ydiV* gene in MG1655 *Δflu* and its *Ptac* derivative was performed by P1 transduction from the KEIO collection <sup>51</sup> followed by curation of the resistance cassette by FLP recombination <sup>52</sup>. Deletion of the *flu* gene encoding the major *E. coli* adhesin, antigen 43, in the MG1655 group strains was used to prevent autoaggregation of motile planktonic cells <sup>53</sup> and thus facilitate subsequent characterization of motility <sup>21</sup>.
+
+To evaluate the activity of the flagellar regulon, strains were transformed with the plasmid carrying the GFP reporter for *fliC* promoter (P<sub><em>fliC</em></sub>) as described previously <sup>21</sup>. For pairwise growth competition experiments, performed as before <sup>21</sup>, the strains were labeled by expression of either cyan or yellow fluorescent proteins (CFP or YFP) from the pTrc99a vector under the control of the IPTG-inducible synthetic P<sub><em>trc</em></sub> promoter <sup>54</sup>. Since pTrc99a carries an extra copy of *lacI*, which reduces the leaky expression from the genomic P<sub><em>tac</em></sub> promoter and thus the inducibility of expression in the *Ptac* strain, an empty pTrc99a vector was transformed into *Ptac* and other *E. coli* K-12 strains for comparability.
+
+*E. coli* strains were grown in either lysogeny broth (LB; 10 g l<sup>−1</sup> of tryptone, 5 g l<sup>−1</sup> of yeast extract, 5 g l<sup>−1</sup> of NaCl), tryptone broth (TB; 10 g l<sup>−1</sup> of tryptone, 5 g l<sup>−1</sup> of NaCl), and either M9 (5× stock made with 64 g l<sup>−1</sup> of Na<sub>2</sub>HPO<sub>4</sub>-7H<sub>2</sub>O, 15 g l<sup>−1</sup> of KH<sub>2</sub>PO<sub>4</sub>, 2.5 g l<sup>−1</sup> of NaCl, 5.0 g l<sup>−1</sup> of NH<sub>4</sub>Cl, 2 mM MgSO<sub>4</sub>, 0.1 mM CaCl<sub>2</sub>, 1µM FeSO<sub>4</sub>, and 1µM ZnCl<sub>2</sub>) or Tanaka (34 mM Na<sub>2</sub>HPO<sub>4</sub>, 0.3 mM MgSO<sub>4</sub>, 64 mM KH<sub>2</sub>PO<sub>4</sub>, 10 µM CaCl<sub>2</sub>, 1µM FeSO<sub>4</sub>, and 1µM ZnCl<sub>2</sub>) <sup>55</sup> minimal media supplemented with 0.4% glucose or 15 mM succinate as the sole carbon source. Ampicillin (100 µg ml<sup>−1</sup>) and/or kanamycin (100 µg ml<sup>−1</sup>), and isopropyl β-d-1 thiogalactopyranoside (IPTG) were added to the media when necessary.
+
+## Reporter activity measurements
+
+P<sub><em>fliC</em></sub> reporter activity was measured by either flow cytometry or plate reader assay. Unless otherwise stated, for flow cytometry, overnight cultures grown in TB (37°C, 200 rpm) were diluted 1:100 in 10 ml of the respective target medium. When minimal medium was used, cells were washed three times in medium without carbon source before inoculation. Cultures were incubated at 34°C with shaking (270 rpm) and harvested at mid-exponential phase (OD<sub>600</sub> = 0.4–0.6 for TB or 0.3–0.5 for M9). Cultures were diluted ~ 50-fold in tethering buffer (6.15 mM K<sub>2</sub>HPO<sub>4</sub>, 3.85 mM KH<sub>2</sub>PO<sub>4</sub>, 0.1 mM EDTA, 1 µM methionine, 10 mM sodium lactate, pH 7.0) and fluorescence was detected using a 488 nm laser (100 mW) and a 510/20 nm bandpass filter for GFP on a BD LSRFortessa SORP cell analyzer (BD Biosciences, Germany). 30,000 individual events were analyzed in each experimental run. Gating was first performed on an FSC-A/SSC-A plot and on an SSC-W over SSC-H plot to exclude doublets. Events in the samples with fluorescence intensities higher than the background signal from the MG1655 *WT* or *Ptac* strain without the reporter plasmid were considered ‘positive’. The proportion of ‘positive’ events per sample and summary statistics (mean, median fluorescence values) of both the ‘positive’ and the ‘whole’ population were assessed during the measurements using BD FACSDiva™ Software v8.0.1 during measurements. Data were collected in FCS 3.0 file format and analyzed using the flowCore package in R v. 4.2.2.
+
+For growth and expression measurements in the BioTek Synergy H1 plate reader, cultures were inoculated into the 96-well plates (Greiner Bio-One) at a dilution of 1:1000 and grown at 34°C with double orbital shaking at a frequency of 548 cycles per minute (CPM) and a shaking amplitude of 2mm for 24 h (TB) or for 48–64 h (M9). GFP fluorescence was quantified using a monochromator-based filter set (excitation 485 nm, emission 530 nm, with a bandpass ≤ 18 nm for detection). Fluorescence and optical density (OD<sub>600</sub>) were measured every 10 min. For experiments shown in Extended Data Fig. 7, the TECAN Infinite M1000 PRO plate reader was used instead for consistency with the previous study <sup>21</sup>.
+
+Reporter activity in ECOR isolates was measured after growth in liquid TB medium or on the surface of semi-solid TB agar (0.5%). For the liquid medium setup, day cultures were prepared in the same manner as for flow cytometry. For the semi-solid condition, 20 µL of the same overnight culture was spread on the surface of TB agar using glass beads. After drying for 15–20 min, the plates were incubated at 34°C for the same time as the strain grew in liquid medium until OD<sub>600</sub> = 0.4–0.6 (i.e., 2.5-4h). Cells were gently washed from the plates with 2 ml of motility buffer (6.15 mM K<sub>2</sub>HPO<sub>4</sub>, 3.85 mM KH<sub>2</sub>PO<sub>4</sub>, 0.1 mM EDTA, 67 mM NaCl, pH 7.0) and adjusted if necessary to final OD<sub>600</sub> = 0.5, and 1 ml of a liquid-grown culture was also washed once in motility buffer. After another washing step, the cells were resuspended in 1 ml motility buffer supplemented with 1% glucose and 0.001% Tween-80. GFP fluorescence was measured in a TECAN Infinite 200 PRO plate reader at 480 nm wavelength, 9 nm bandwidth for excitation and 510 nm wavelength, 20 nm bandwidth for emission.
+
+## Analysis of swimming velocity and flagella rotation
+
+Bacterial cell motility was analyzed as previously described <sup>21, 56</sup>. Briefly, 1 ml of the same cell culture as prepared for flow cytometry was gently centrifuged (4000 rpm, 5 min), washed twice in motility buffer, and resuspended in 1 ml motility buffer supplemented with 1% glucose and 0.001% Tween-80. 3–5 µL of this cell suspension was introduced into a custom-made chamber between two coverslips, and motility was imaged by phase-contrast video-microscopy (Nikon TI Eclipse, 10x objective with NA = 0.3, Phase 1 ring, CMOS camera EoSens 4CXP), with 10,000 frames being recorded at a rate of 100 frames per second (fps). Motility parameters, in particular the fraction of swimming cells and the swimming velocity of the swimmers, are extracted from the movies using differential dynamic microscopy (DDM) <sup>55</sup> (see Supplementary Note 1).
+
+To determine the frequency of flagella rotation, samples were prepared in the same manner as described for swimming velocity analysis. A 10,000-frame movie with a field of view of 512 x 512 px <sup>2</sup> (1 px = 0.7 µm) was acquired far from the sample surfaces under dark field illumination (Nikon TI Eclipse, 10x objective with NA = 0.3, CMOS camera EoSens 4CXP) at a rate of 800 fps. Dark field illumination is obtained by combining an aligned Ph3 condenser ring with the 10x objective on the Nikon TI Eclipse microscope. All data were analyzed using the dark field flicker microscopy (DFFM) method <sup>57</sup> (see Supplementary Note 1) implemented in ImageJ (https://imagej.nih.gov/ij/) with custom-written plugins. Briefly, DFFM uses the flickering that results from changes in the direction in which light is scattered by anisotropic objects as they rotate to measure the rotation speeds of the cell body and flagella.
+
+## Motility assay in soft agar
+
+Motility driven spreading of *E. coli* in 0.27% TB soft agar was analyzed as previously described <sup>39</sup>. Briefly, 2 µl of overnight cultures grown in TB (37°C, 200 rpm) were transferred to the soft agar plates, and the diameters of the spreading zones were measured after 4–5 h of incubation at 34°C by capturing images with an iPad camera and quantifying the diameter of the spreading zone using ImageJ.
+
+## Pairwise growth competition
+
+Growth competition assays were performed as previously described <sup>21</sup>. Briefly, the overnight cultures of the MG1655 *WT* or *Ptac* strain expressing CFP and the *ΔflhC* strain expressing YFP, grown individually in TB (37°C, 200 rpm), were mixed in a 1:1 ratio to final OD<sub>600</sub> = 0.0025 in 2.5 mL of fresh media and cultured for 24 h (TB) or 48–72 h (M9 minimal medium) at 34°C and 200 rpm. The expression of YFP and CFP was induced with 10 µM IPTG for the co-culture containing the MG1655 *WT* strain or by the corresponding IPTG concentrations used for induction of the chromosomal *Ptac* promoter. For the chemotactic benefit assay, differentially labeled non-chemotactic *ΔcheY* strain and MG1655 *WT* or *Ptac* strains were grown in Tanaka minimal medium for 72 h without shaking in the presence of nutrient gradients generated by 40 µL large agarose beads (2% agarose) containing 12% casein hydrolysate as described previously <sup>21</sup>. The initial and final proportions of CFP- and YFP-labeled cells were measured by flow cytometry on the BD LSRFortessa SORP cell analyzer (BD Biosciences). The sample was excited with lasers at 447 nm (75 mW), 514 nm (100 mW), and 488 nm (20 mW), with the latter used to identify all cells. CFP and YFP emission signals were detected at 470/15 nm and 542/27 nm, respectively. The fraction of CFP/YFP-‘positive’ events per sample was assessed during the measurements using BD FACSDiva™ Software v8.0.1. Summary statistics were collected in csv file format and analyzed in R v. 4.2.2.
+
+## Measurements of flagellar length and number
+
+For flagella staining, 1 ml of the mid-exponential cell culture grown in TB as described above was centrifuged (3000g, 3 min) and gently washed three times in Buffer A (10 mM KPO<sub>4</sub> buffer, 0.1 mM EDTA dipotassium salt, 67 mM NaCl, 0.001% Tween-80, pH 7.0). The cell pellet was resuspended in 400 µL of Buffer B (same as Buffer A but adjusted to pH 7.8 with NaHCO<sub>3</sub>), and 8 µl of 10 µg ml<sup>−1</sup> Alexa Fluor 594 succinimidyl ester dye dissolved in DMSO was added to the mixture. Samples were incubated at 30°C in the dark with gentle shaking (100 rpm) for 90 min, washed three times in Buffer A and diluted fivefold in Buffer A. 3–5 µl of cell suspension was applied to a 1% agarose pad (in tethering buffer) and transferred to a 2-well µ-Slide (ibidi, Germany).
+
+Fluorescence widefield images were acquired using a Zeiss Elyra 7 inverted microscope with a 63x oil/1.46 oil objective and a further 1.6X magnification. The sample was excited with a 561 nm 500 mW laser (1% power) using a quadruple band dichroic and emission filter. The fluorescence emission of the succinimidyl ester was detected at 595/50 nm interval with a PCO 4.2 Edge sCMOS camera, the exposure time was 100 ms. The number of flagella was quantified for randomly selected 100 cells in multiple fields of view, including both flagellated and non-flagellated cells. The length of flagellar filaments (35–50 filaments per condition) was measured using segmented line tool of ImageJ.
+
+## Immunoblot analysis of intra- and extracellular flagellin
+
+To shear flagellar filaments, a 1 ml aliquot of the mid-exponential cell culture was passed through a 1 ml syringe with the 26G needle 20 times, and centrifuged at 2500 g for 10 min. The supernatant and cell pellet, resuspended in 333 µL of TB medium, were further analyzed by immunoblot. To transfer the samples to the membrane after SDS-PAGE, a PerfectBlue Semi-Dry Electroblotter (Peqlab, VWR, Germany) was used at constant amperage for 1 h (150 mA for 8*6 cm membrane and 1.5 mm thick gel). After transfer, the membrane was stained with Revert™ 700 Total Protein Stain for Western Blot Normalization (LI-COR Biosciences, Germany) and, after blocking, incubated overnight (4°C, orbital shaking) with the primary anti-flagellin antibody (Antikoerper, Germany) diluted 1:10000 followed by the secondary IRDye 800CW anti-rabbit IgG antibody (LI-COR Biosciences, Germany) antibody at a dilution of 1:10000. Fluorescence was measured using an Odyssey Clx Infrared Imaging System (LI-COR Biosciences, Germany) in two channels (700 and 800 nm). Images were analyzed and processed using ImageJ.
+
+## The model of flagellum-mediated bacterial swimming
+
+The model for multiflagellated propulsion extends the classical force balance analysis for uniflagellated propulsion <sup>31, 58</sup> and accounts for our measurements of swimming speed, cell body rotation speed, and flagellar rotation speed, as well as flagellar length, flagellar number, and cell size. The model is described in detail in Supplementary Note 2. Briefly, we assume that the *N* flagella form a single tight bundle, described in the framework of resistive force theory <sup>31, 59–61</sup> as a helix of larger thickness for a higher number of flagella, which is justified considering several macroscopic experiments at low Reynolds number with multiple helices <sup>62, 63</sup>. We account for the increase in both flagellar length and flagellar number with increasing *flhDC* induction. The cell body is described as a counter-rotating rod <sup>64, 65</sup> of fixed size, consistently with our observation. The flagellar motor speed is assumed to be constant, in agreement with our measurements of the flagella and cell body rotation speeds. The balance of forces and torques acting on the cell body and the flagellar bundle provides predictions of the swimming speed and the rotation frequencies.
+
+# References
+
+1. Molenaar, D., van Berlo, R., de Ridder, D. & Teusink, B. Shifts in growth strategies reflect tradeoffs in cellular economics. *Mol. Syst. Biol.* **5**, 323 (2009).
+
+2. Basan, M. Resource allocation and metabolism: the search for governing principles. *Curr. Opin. Microbiol.* **45**, 77-83 (2018).
+
+3. Goelzer, A. & Fromion, V. Resource allocation in living organisms. *Biochem. Soc. Trans.* **45**, 945-952 (2017).
+
+4. Scott, M., Gunderson, C.W., Mateescu, E.M., Zhang, Z. & Hwa, T. Interdependence of Cell Growth and Gene Expression: Origins and Consequences. *Science* **330**, 1099-1102 (2010).
+
+5. You, C. *et al.* Coordination of bacterial proteome with metabolism by cyclic AMP signalling. *Nature* **500**, 301-6 (2013).
+
+6. Li, G.W., Burkhardt, D., Gross, C. & Weissman, J.S. Quantifying absolute protein synthesis rates reveals principles underlying allocation of cellular resources. *Cell* **157**, 624-35 (2014).
+
+7. Hui, S. *et al.* Quantitative proteomic analysis reveals a simple strategy of global resource allocation in bacteria. *Mol. Syst. Biol.* **11**, 784 (2015).
+
+8. Dekel, E. & Alon, U. Optimality and evolutionary tuning of the expression level of a protein. *Nature* **436**, 588-92 (2005).
+
+9. Price, M.N. *et al.* Indirect and suboptimal control of gene expression is widespread in bacteria. *Mol. Syst. Biol.* **9**, 660 (2013).
+
+10. Towbin, B.D. *et al.* Optimality and sub-optimality in a bacterial growth law. *Nat. Commun.* **8**, 14123 (2017).
+
+11. Radzikowski, J.L. *et al.* Bacterial persistence is an active σ<sup>S</sup> stress response to metabolic flux limitation. *Mol. Syst. Biol.* **12**, 882 (2016).
+
+12. Zhu, M. *et al.* A fitness trade-off between growth and survival governed by Spo0A-mediated proteome allocation constraints in *Bacillus subtilis*. *Sci. Adv.* **9**, eadg9733 (2023).
+
+13. Balakrishnan, R. & Cremer, J. Conditionally unutilized proteins and their profound effects on growth and adaptation across microbial species. *Curr. Opin. Microbiol.* **75**, 102366 (2023).
+
+14. Berg, H.C. The rotary motor of bacterial flagella. *Ann. Rev. Biochem.* **72**, 19-54 (2003).
+
+15. Wadhams, G.H. & Armitage, J.P. Making sense of it all: bacterial chemotaxis. *Nat. Rev. Mol. Cell. Biol.* **5**, 1024-37 (2004).
+
+16. Colin, R., Ni, B., Laganenka, L. & Sourjik, V. Multiple functions of flagellar motility and chemotaxis in bacterial physiology. *FEMS Microbiol. Rev.* **45** (2021).
+
+17. Milo, R., Jorgensen, P., Moran, U., Weber, G. & Springer, M. BioNumbers--the database of key numbers in molecular and cell biology. *Nucleic Acids Res.* **38**, D750-3 (2010).
+
+18. Colin, R. & Sourjik, V. Emergent properties of bacterial chemotaxis pathway. *Curr. Opin. Microbiol.* **39**, 24-33 (2017).
+
+19. Keegstra, J.M., Carrara, F. & Stocker, R. The ecological roles of bacterial chemotaxis. *Nat. Rev. Microbiol.* **20**, 491-504 (2022).
+
+20. Ni, B. *et al.* Evolutionary Remodeling of Bacterial Motility Checkpoint Control. *Cell. Rep.* **18**, 866-877 (2017).
+
+21. Ni, B., Colin, R., Link, H., Endres, R.G. & Sourjik, V. Growth-rate dependent resource investment in bacterial motile behavior quantitatively follows potential benefit of chemotaxis. *Proc. Natl. Acad. Sci. USA* **117**, 595-601 (2020).
+
+22. Gude, S. *et al.* Bacterial coexistence driven by motility and spatial competition. *Nature* **578**, 588-592 (2020).
+
+23. Honda, T. *et al.* Coordination of gene expression with cell size enables *Escherichia coli* to efficiently maintain motility across conditions. *Proc. Natl. Acad. Sci. USA* **119**, e2110342119 (2022).
+
+24. Fraebel, D.T. *et al.* Environment determines evolutionary trajectory in a constrained phenotypic space. *eLife* **6**, e24669 (2017).
+
+25. Liu, M. *et al.* Global transcriptional programs reveal a carbon source foraging strategy by *Escherichia coli*. *J. Biol. Chem.* **280**, 15921-7 (2005).
+
+26. Kim, J.M., Garcia-Alcala, M., Balleza, E. & Cluzel, P. Stochastic transcriptional pulses orchestrate flagellar biosynthesis in *Escherichia coli*. *Sci. Adv.* **6**, eaax0947 (2020).
+
+27. Wilson, L.G. *et al.* Differential Dynamic Microscopy of Bacterial Motility. *Phys. Rev. Lett.* **106**, 018101 (2011).
+
+28. Hayashi, K. *et al.* Highly accurate genome sequences of *Escherichia coli* K-12 strains MG1655 and W3110. *Mol. Syst. Biol.* **2**, 2006.0007 (2006).
+
+29. Parkinson, J.S. Complementation analysis and deletion mapping of *Escherichia coli* mutants defective in chemotaxis. *J. Bacteriol.* **135**, 45-53 (1978).
+
+30. Turner, L., Ping, L., Neubauer, M. & Berg, H.C. Visualizing Flagella while Tracking Bacteria. *Biophys. J.* **111**, 630-639 (2016).
+
+31. Chattopadhyay, S., Moldovan, R., Yeung, C. & Wu, X.L. Swimming efficiency of bacterium *Escherichia coli*. *Proc. Natl. Acad. Sci. USA* **103**, 13712-13717 (2006).
+
+32. Elgeti, J., Winkler, R.G. & Gompper, G. Physics of microswimmers--single particle motion and collective behavior: a review. *Rep. Prog. Phys.* **78**, 056601 (2015).
+
+33. Kamdar, S. *et al.* Multiflagellarity leads to the size-independent swimming speed of peritrichous bacteria. *Proc. Natl. Acad. Sci. USA* **120**, e2310952120 (2023).
+
+34. Klumpp, S., Zhang, Z. & Hwa, T. Growth Rate-Dependent Global Effects on Gene Expression in Bacteria. *Cell* **139**, 1366-1375 (2009).
+
+35. Sassi, A.S., Garcia-Alcala, M., Kim, M.J., Cluzel, P. & Tu, Y. Filtering input fluctuations in intensity and in time underlies stochastic transcriptional pulses without feedback. *Proc. Natl. Acad. Sci. USA* **117**, 26608-26615 (2020).
+
+36. Koirala, S. *et al.* A Nutrient-Tunable Bistable Switch Controls Motility in *Salmonella enterica* Serovar Typhimurium. *mBio* **5**, e01611-14 (2014).
+
+37. Spöring, I. *et al.* Regulation of Flagellum Biosynthesis in Response to Cell Envelope Stress in *Salmonella enterica* Serovar Typhimurium. *mBio* **9**, e00736-17 (2018).
+
+38. Ochman, H. & Selander, R.K. Standard reference strains of *Escherichia coli* from natural populations. *J. Bacteriol.* **157**, 690-693 (1984).
+
+39. Laganenka, L., Lopez, M.E., Colin, R. & Sourjik, V. Flagellum-Mediated Mechanosensing and RflP Control Motility State of Pathogenic *Escherichia coli*. *mBio* **11**, e02269-19 (2020).
+
+40. Cairns, L.S., Marlow, V.L., Bissett, E., Ostrowski, A. & Stanley-Wall, N.R. A mechanical signal transmitted by the flagellum controls signalling in *Bacillus subtilis*. *Mol. Microbiol.* **90**, 6-21 (2013).
+
+41. Lele, P.P., Hosu, B.G. & Berg, H.C. Dynamics of mechanosensing in the bacterial flagellar motor. *Proc. Natl. Acad. Sci. USA* **110**, 11839-44 (2013).
+
+42. Tipping, M.J., Delalez, N.J., Lim, R., Berry, R.M. & Armitage, J.P. Load-dependent assembly of the bacterial flagellar motor. *mBio* **4**, e00551-13 (2013).
+
+43. Nord, A.L. *et al.* Catch bond drives stator mechanosensitivity in the bacterial flagellar motor. *Proc. Natl. Acad. Sci. USA* **114**, 12952-12957 (2017).
+
+44. O'Brien, E.J., Utrilla, J. & Palsson, B.O. Quantification and Classification of E. coli Proteome Utilization and Unused Protein Costs across Environments. *PLoS Comput. Biol.* **12**, e1004998 (2016).
+
+45. Balakrishnan, R., de Silva, R.T., Hwa, T. & Cremer, J. Suboptimal resource allocation in changing environments constrains response and growth in bacteria. *Mol. Syst. Biol.* **17**, e10597 (2021).
+
+46. Cremer, J. *et al.* Chemotaxis as a navigation strategy to boost range expansion. *Nature* **575**, 658-663 (2019).
+
+47. Laganenka, L. *et al.* Chemotaxis and autoinducer-2 signalling mediate colonization and contribute to co-existence of *Escherichia coli* strains in the murine gut. *Nat. Microbiol.* **8**, 204-217 (2023).
+
+48. Kollmann, M., Løvdok, L., Bartholomé, K., Timmer, J. & Sourjik, V. Design principles of a bacterial signalling network. *Nature* **438**, 504-507 (2005).
+
+49. Veening, J.-W., Smits, W.K. & Kuipers, O.P. Bistability, Epigenetics, and Bet-Hedging in Bacteria. *Ann. Rev. Microbiol.* **62**, 193-210 (2008).
+
+50. Norman, T.M., Lord, N.D., Paulsson, J. & Losick, R. Stochastic Switching of Cell Fate in Microbes. *Ann. Rev. Microbiol.* **69**, 381-403 (2015).
+
+51. Baba, T. *et al.* Construction of *Escherichia coli* K-12 in-frame, single-gene knockout mutants: the Keio collection. *Mol. Syst. Biol.* **2**, 2006.0008 (2006).
+
+52. Link, A.J., Phillips, D. & Church, G.M. Methods for generating precise deletions and insertions in the genome of wild-type *Escherichia coli*: application to open reading frame characterization. *J. Bacteriol.* **179**, 6228-6237 (1997).
+
+53. Ulett, G.C., Webb, R.I. & Schembri, M.A. Antigen-43-mediated autoaggregation impairs motility in *Escherichia coli*. *Microbiology (Reading)* **152**, 2101-2110 (2006).
+
+54. Press, M.O. *et al.* Genome-scale Co-evolutionary Inference Identifies Functions and Clients of Bacterial Hsp90. *PLoS Gen.* **9**, e1003631 (2013).
+
+55. Tanaka, S., Lerner, S.A. & Lin, E.C.C. Replacement of a Phosphoenolpyruvate-dependent Phosphotransferase by a Nicotinamide Adenine Dinucleotide-linked Dehydrogenase for the Utilization of Mannitol. *J. Bacteriol.* **93**, 642-648 (1967).
+
+56. Colin, R., Zhang, R. & Wilson, L.G. Fast, high-throughput measurement of collective behaviour in a bacterial population. *J. R. Soc. Interface* **11**, 20140486 (2014).
+
+57. Martinez, V.A. *et al.* Flagellated bacterial motility in polymer solutions. *Proc. Natl. Acad. Sci. USA* **111**, 17771-17776 (2014).
+
+58. Purcell, E.M. The efficiency of propulsion by a rotating flagellum. *Proc. Natl. Acad. Sci. USA* **94**, 11307-11311 (1997).
+
+59. Gray, J. & Hancock, G.J. The Propulsion of Sea-Urchin Spermatozoa. *J. Exp. Biol.* **32**, 802-814 (1955).
+
+60. Lighthill, J. Flagellar Hydrodynamics. *SIAM Review* **18**, 161-230 (1976).
+
+61. Johnson, R.E. & Brokaw, C.J. Flagellar hydrodynamics. A comparison between resistive-force theory and slender-body theory. *Biophys. J.* **25**, 113-127 (1979).
+
+62. Kim, M.J. *et al.* Particle image velocimetry experiments on a macro-scale model for bacterial flagellar bundling. *Exp. Fluids* **37**, 782-788 (2004).
+
+63. Danis, U. *et al.* Thrust and Hydrodynamic Efficiency of the Bundled Flagella. *Micromachines* **10**, 449 (2019).
+
+64. Tirado, M.M. & de la Torre, J.G. Rotational dynamics of rigid, symmetric top macromolecules. Application to circular cylinders. *J. Chem. Phys.* **73**, 1986-1993 (2008).
+
+65. Tirado, M.M., Martínez, C.L. & de la Torre, J.G. Comparison of theories for the translational and rotational diffusion coefficients of rod-like macromolecules. Application to short DNA fragments. *J. Chem. Phys.* **81**, 2047-2052 (1984).
+
+# Supplementary Files
+
+- [SupplementarytablesLisevich.xlsx](https://assets-eu.researchsquare.com/files/rs-4044856/v1/e32df38d9e5d9a3855049da3.xlsx)  
+  Supplementary Tables 1 and 2
+
+- [SupplementaryNotesLisevich.docx](https://assets-eu.researchsquare.com/files/rs-4044856/v1/851b67dc5e815f30619846db.docx)  
+  Supplementary Notes 1 and 2
+
+- [ExtendedDataFigures.docx](https://assets-eu.researchsquare.com/files/rs-4044856/v1/0c495b5d934c3f87d2059909.docx)

0ebe1024cda95d94795111845d7994a9cff18fe03d3b90b80d825560ab7a095a/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Cryo-EM structures of human OGT and OGT\u2013OGA complex.\u00a0a, The dynamic protein O-GlcNAcylation cycle mediated by OGT and OGA. \u00a0b, Domains and motifs of human OGT and OGA. \u00a0TPR, tetratricopeptide repeat. c, Human 293 cells with the endogenous OGT loci tagged with the HaloTag were treated with the HaloPROTAC3 compound for different durations. Total cell lysates were blotted with the indicated antibodies. d, Cryo-EM map of human OGT dimer (left panel) and the structure of the OGT dimer fitted into the EM map (right panel). \u00a0GTD, glycosyltransferase family B domain. e, Close-up views of the OGT dimerization interface. \u00a0f, O-GlcNAcylation of recombinant TAB1 by OGT WT or its monomeric 4A mutant in the presence of UDP-GlcNAc. \u00a0g, Cryo-EM map of human OGT\u2013OGA complex (left panel) and the structure of the OGT\u2013OGA complex fitted into the EM map. \u00a0GHD, glycoside hydrolase domain; IDR, intrinsically disordered region. h, Superimposition of the structures of the OGT\u2013OGA complex and the OGT dimer (colored in gray and with only one protomer shown).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Substrate recognition by human OGT.\u00a0a, Extensive interactions between human OGT and OGA. \u00a0b-e Close-up views of the OGT\u2013OGA interfaces, with key interacting residues, GlcNAc, and UDP shown as sticks and labeled. f, O-GlcNAcylation assay of the catalytically inactive OGA mutant (D175N or D175N/S405A) by OGT wild type (WT) and the indicated mutants in the presence or absence of UDP-GlcNAc. g, Binding between GST-OGA (residues 371\u201344) and OGT WT and the indicated mutants. The input proteins and proteins bound to GST beads were analyzed by SDS-PAGE and Coomassie staining.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Inhibition of OGT by OGA. a, EM density of an OGA IDR segment (colored in yellow) that interacts with the asparagine ladder of the OGT TPR domain (colored in purple). b,c, Close-up views of an OGA segment interacting with the intervening domain (ID), the UDP-binding pocket, and the active site of OGT. d, O-GlcNAcylation assay of TAB1 by the indicated OGT and OGA proteins in the presence or absence of UDP-GlcNAc.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Inhibition of OGA by OGT. a, Cartoon drawing of the crystal structure of the human OGA dimer bound to a TAB1 peptide (PDB code: 5VVU). b, Superimposition of the structures of the OGA molecules in the OGA dimer (as in a) and the OGT\u2013OGA complex, showing the steric clashes between OGT and the second OGA molecule in the OGA dimer. c, Close-up view of the superimposed structures in b. d,e, Close-up views of the interactions between OGT and the GHD of OGA. f, Removal of TAB1 O-GlcNAcylation by OGA and its inhibition by OGT. O-GlcNAcylated TAB1 (G-TAB1) was incubated with OGA or the catalytically inactive OGA D175N mutant in the absence or presence of the indicated OGT proteins. The reaction mixtures were analyzed by SDS-PAGE, stained with Coomassie brilliant blue (CBB), and blotted with the anti-O-GlcNAc antibody.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpg",
+        "caption": "Substrate recognition by OGT and homeostatic control of O-GlcNAcylation.\u00a0a, Model of substrate recognition by OGT. For clarity, only one OGT monomer is shown. OGT binds to intrinsically disordered regions (IDRs) of substrates through its TPR domain. The superhelical turns of the TPR domain have to partially unwind to allow the IDR to access its lumen. The superhelical turns then reform and wrap around the IDR in a topological embrace, which lengthens the lifetime of the OGT\u2013substrate complex and increases the processivity of O-GlcNAcylation.\u00a0b, Homeostatic control of O-GlcNAcylation by the mutual inhibition between OGT and OGA. Large pools of OGT and OGA molecules in the cell are bound to each other and mutually inhibited. This ensures low levels of O-GlcNAcylation and de-GlcNAcylation. Under conditions that favor O-GlcNAcylation (e.g. elevated levels of UDP-GlcNAc), OGT substrates compete effectively with OGA for OGT binding and modification. The dislodged OGA forms active OGA dimers, which remove substrate O-GlcNAcylation to keep the overall O-GlcNAcylation at a new steady state. As UDP-GlcNAc levels fall, OGA rebinds OGT and the system returns to the basal state.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

0ebe1024cda95d94795111845d7994a9cff18fe03d3b90b80d825560ab7a095a/preprint/preprint.md

@@ -0,0 +1,222 @@
+# Abstract
+
+O-GlcNAcylation is a conserved post-translational modification that attaches N-acetyl glucosamine (GlcNAc) to myriad cellular proteins<sup>1–4</sup>. In response to nutritional and hormonal signals, O-GlcNAcylation regulates diverse cellular processes by modulating the stability, structure, and function of target proteins. Misregulation of O-GlcNAcylation is implicated in cancer, diabetes, and neurodegeneration<sup>5–7</sup>. A single pair of enzymes, the O-GlcNAc transferase (OGT) and the O-GlcNAcase (OGA), catalyzes the addition and removal of O-GlcNAc on over 3,000 proteins in the human proteome<sup>8,9</sup>. How OGT selects its native substrate(s) and maintains the homeostatic control of O-GlcNAcylation of so many substrates against OGA are not understood. Here we show that chemically induced degradation of OGT co-depletes OGA in human cells, suggesting the existence of a stable OGT–OGA complex<em>in vivo</em>. The cryo-electron microscopy (cryo-EM) structures of human OGT and the OGT–OGA complex reveal that OGT forms a functionally important scissor-shaped dimer. A long flexible OGA segment occupies the extended substrate-binding groove of OGT and positions a serine for O-GlcNAcylation, thus preventing OGT from modifying other substrates. Conversely, OGT disrupts the functional dimerization of OGA and occludes its active site, resulting in the blocking of access by other substrates. This unexpected but direct mutual inhibition between OGT and OGA limits futile O-GlcNAcylation cycles and maintains O-GlcNAc homeostasis.
+
+**Biological sciences/Biochemistry/Glycobiology**  
+**Biological sciences/Biochemistry/Enzyme mechanisms**  
+**Biological sciences/Biochemistry/Structural biology/Electron microscopy/Cryoelectron microscopy**
+
+# Introduction
+
+O-GlcNAcylation entails the covalent attachment of O-GlcNAc to the hydroxyl group of serines or threonines of target proteins<sup>1–4</sup>. It modifies thousands of human proteins that have functions in diverse cellular processes and are localized to major cellular compartments, including the nucleus, the cytoplasm, and the mitochondria<sup>8,9</sup>. Protein O-GlcNAcylation is a dynamic and reversible process, which is mediated by a single pair of enzymes, OGT and OGA (Fig. <span class="InternalRef" refid="Fig1">1</span> a,b)<sup>1,2</sup>. OGT contains an N-terminal tetratricopeptide repeat (TPR) domain required for substrate recognition and a C-terminal glycosyltransferase family B domain (GTD) that catalyzes the modification of substrates with uridine diphosphate GlcNAc (UDP-GlcNAc) as the cofactor<sup>10–12</sup>. OGA consists of an N-terminal glycoside hydrolase domain (GHD) that removes O-GlcNAc from substrates, a stalk domain, and a C-terminal histone acetyltransferase (HAT)-like domain<sup>10,13−15</sup>. OGA forms a functional homodimer in which the catalytic domain from one monomer is capped by the stalk domain of the other to generate the substrate-binding cleft<sup>10,13−16</sup>.
+
+The OGT cofactor UDP-GlcNAc is the product of the hexosamine biosynthetic pathway that links the metabolism of glucose, amino acids, fatty acids, and nucleotides<sup>1,2,4,17</sup>. Its production is sensitive to changes in glycolysis, amino acid synthesis, fatty acid levels, and nucleotide levels. Protein O-GlcNAcylation is thus an important nutrient sensor. In response to nutrient availability and various stress conditions, O-GlcNAcylation modulates the structure, stability, localization, and function of target proteins, thereby regulating the cell cycle, signal transduction, gene transcription, and protein translation, among other cellular processes<sup>2,17,18</sup>. Well-characterized O-GlcNAcylation substrates include the tumor suppressor TP53<sup>19</sup>, TGFβ activated kinase 1 binding protein 1 (TAB1)<sup>20</sup>, histones<sup>21</sup>, the histone methyltransferase EZH2<sup>22</sup>, the microtubule-binding protein TAU<sup>23</sup>, and α-synuclein<sup>24</sup>. Aberrant O-GlcNAcylation has been implicated in human metabolic syndromes and chronic diseases, including cancer, diabetes, cardiovascular diseases, and neurodegenerative diseases<sup>5–7,25</sup>.
+
+Biochemical and structural studies on human OGT and OGA have provided critical insights into the mechanistic underlying their enzymatic activities towards peptide substrates. However, how OGT enforces selectivity towards its native (but not peptide) substrate remains a long-standing question in the field. In addition, it is still unknown how OGT and OGA coordinate their enzymatic activities towards diverse substrates to achieve optimal O-GlcNAcylation levels in response to the nutrient status<sup>2,9,10</sup>. Paradoxically, human OGA has been reported to directly interact with its opposing enzyme OGT<sup>26</sup>. The nature and functional importance of this interaction remain unexplored, however. Here, we show that human OGA requires OGT for stability in human cells, suggesting that a large pool of OGA forms a complex with OGT. We have determined the cryo-EM structure of the full-length human OGT–OGA complex, revealing that neither OGT nor OGA in the complex can act on other substrates. The unexpected direct mutual inhibition between this enzyme pair limits futile O-GlcNAc cycling and maintains steady-state levels of O-GlcNAcylation.
+
+# Results
+
+## Cryo-EM structure of human OGT dimer
+
+Using CRISPR-Cas9 genome editing technology, we knocked in the HaloTag into the endogenous OGT loci in HEK293 cells. Addition of the HaloPROTAC compound induced the efficient and rapid degradation of the resulting OGT-HaloTag fusion protein (Fig. 1c). Expectedly, the overall cellular O-GlcNAc levels were greatly reduced upon OGT depletion, confirming the essential role of OGT in cellular O-GlcNAcylation.
+
+We expressed and purified full-length human OGT (Extended Data Fig. 1a). Recombinant wild-type OGT, but not its catalytically deficient H508A mutant, catalyzed efficient O-GlcNAcylation of the full-length human TAB1 in a UDP-GlcNAc-dependent manner, indicating that the recombinant OGT protein was active (Extended Data Fig. 1b). We next determined the structure of OGT an overall resolution of 3.69 Å using single-particle cryo-EM (Fig. 1d, Extended Data Fig. 2, and Extended Data Fig. 3). Consistent with a recent report, OGT formed a scissor-shaped dimer (Fig. 1d). The TPR domain of each monomer consists of 27 antiparallel α-helices, which stack together in a right-handed superhelical conformation with a lumen of 22 Å in diameter (Fig. 1d). The conserved asparagine residues line this lumen and align in a ladder-like configuration to engage substrates. Consistent with earlier studies, TPR6 and TPR7 of OGT mediate its dimerization (Fig. 1e). Mutations of four conserved residues at the dimer interface to alanine (OGT 4A, W208A/L209A/I211A/H212A) disrupted OGT dimerization, based on size exclusion chromatography and cryo-EM analysis (Extended Data Fig. 4). OGT 4A was less active in catalyzing O-GlcNAcylation of TAB1 in vitro (Fig. 1f), indicating that OGT functions as a homodimer.
+
+## Cryo-EM structure of the OGT–OGA complex
+
+OGT and OGA have been shown to directly interact with each other. HaloPROTOC-mediated depletion of OGT co-depleted endogenous OGA in HEK293 cells (Fig. 1c). Depletion of OGA could similarly co-deplete OGT in human cells. These findings suggest that large pools of OGT and OGA might form stable complexes in human cells and depend on each other for stability. Indeed, recombinant human OGT and OGA formed a stable complex and co-fractionated on size exclusion chromatography (Extended Data Fig. 5a). We then determined the cryo-EM structure of the OGT–OGA complex with an overall resolution of 3.92 Å (Fig. 1g, Extended Data Fig. 5b-d, and Extended Data Fig. 6).
+
+OGT in the complex has a scissor-like dimeric architecture similar to that of OGT alone (Fig. 1g and Extended Movie 1). Compared to OGT alone, the distance between the two active sites in the dimer (located at the handles of the scissor) becomes shorter in the complex: from ~90 Å in OGT alone to ~80 Å in the OGT–OGA complex (Fig. 1d,g). OGT TPR1–5 become more rigid and compact upon OGA binding in the complex (Fig. 1h).
+
+Only the N-terminal GHD of OGA and two segments of the intrinsically disordered region (IDR) following this domain were observed in cryo-EM maps and could be modeled in the complex (Fig. 2a). The stalk and HAT-like domains of OGA were absent in the cryo-EM maps, presumably due to conformational flexibility. Crystal structures of the N-terminal fragment of OGA have revealed that the catalytic GHD forms a functional homodimer. Interestingly, only one GHD monomer was observed in the OGT–OGA complex. It docks on the convex surface of TPR11–13 in one OGT protomer (Fig. 2a). Residues 377–393 of the OGA IDR form a helix, which packs against the GHD. A long segment of the OGA IDR (residues 394–442) occupies the active site of the catalytic GTD and the entire lumen of the TPR domain of OGT (Fig. 2b-e). While no density was observed for a second OGA GHD at the equivalent site on the other OGT monomer, there was an obvious density belonging to the OGA IDR in the lumen of the TPR domain of this OGT molecule (Extended Data Fig. 5d and Extended Movie 2), suggesting that both OGT molecules are bound by OGA.
+
+## Substrate recognition by human OGT
+
+OGA is a known native substrate of OGT. In the OGT–OGA complex, residues 396–407 of OGA interact with the GTD of OGT. S405 of OGA lies in proximity of UDP-GlcNAc bound at the OGT active site. The amide group of S405 forms a hydrogen bond with the α-phosphate of UDP, a critical interaction required for the transfer of GlcNAc to the residue to be modified. The GlcNAc moiety is bound by active site residues H508, A664, and H930 (Fig. 2e), as observed in crystal structures of OGT bound to peptide substrates. The strong electron density connecting the hydroxyl group of S405 and GlcNAc suggests that OGA S405 is glycosylated (Extended Data Fig. 6c). Indeed, wild-type OGA, but not its S405A mutant, was modified by OGT in vitro, indicating that OGA S405 can be O-GlcNAcylated by OGT (Fig. 2f). Thus, our cryo-EM structure likely captures the post-catalytic OGT–OGA complex, in which the GlcNAc moiety is attached to the hydroxyl group of S405.
+
+Residues 408–442 of OGA adopt an extended conformation, occupy the entire substrate-binding lumen of the TPR domain, and develop extensive hydrophobic interactions and hydrogen bonds (Fig. 2a-e). Notably, the backbone amide or carbonyl groups of OGA form hydrogen bonds or favorable polar interactions with many residues in the asparagine ladder of OGT, including N94 and N97 in TPR3, N128 in TPR4, N165 in TPR5, N196 in TPR6, N230 in TPR7, N264 in TPR8, N298 in TPR9, N332 in TPR10, N366 in TPR11, N403 in TPR12, and N434 in TPR13 (Extended Data Fig. 7a-d). This mode of sequence-independent backbone recognition by OGT ensures that it can interact with a wide range of substrates, provided that these substrates have a long IDR. The asparagine ladder residues in the TPRs proximal to the GTD, such as N400 and N403 in TPR12 and N468 in TPR13.5, interact with the sidechains of D413 and S410 of OGA through hydrogen bonds (Fig. 2d and Extended Data Fig. 7d). These and other OGT–OGA interactions involving specific sidechains of OGA confer the substrate specificity of OGT.
+
+Mutations of certain OGT asparagine ladder residues at the observed OGT–OGA interface, including N94 and N97 in TPR3, N196 in TPR6, and N230 and N233 in TPR7, abolished binding between OGT and the OGT-binding segment of OGA (residues 371–440) in vitro (Fig. 2g and Extended Data Fig. 8a,b). The OGA-binding-deficient OGT mutants were also less efficient in catalyzing the O-GlcNAcylation of OGA in vitro (Fig. 2f) and in supporting the overall cellular O-GlcNAc levels in HEK293 cells depleted of the endogenous OGT (Extended Data Fig. 8c,d). These results validate the functional importance of the OGT–OGA interactions observed in our structure. The substrate recognition mode revealed by the OGT–OGA complex structure is applicable to other OGT substrates in human cells.
+
+Mutations of some asparagine ladder residues, such as N332 in TPR10, N366 in TPR11, and N434 in TPR13, had no effect on OGA binding (Extended Data Fig. 8b). Thus, not all asparagine ladder residues of OGT contribute to OGA binding equally. Binding hotspots in TPR3, TPR6, and TPR7 are more critical for binding between OGT and OGA. Mutation of N196 in TPR6 diminished the O-GlcNAcylation of TAB1 whereas mutations of asparagine residues in TPR3 and TPR7 had little effect (Extended Data Fig. 9), indicating that OGT uses different binding hotspots for recognizing different substrates.
+
+## Competitive inhibition of human OGT by OGA
+
+Enzyme-substrate interactions are typically transient, and modified substrates (i.e., products) are released to enable additional rounds of catalysis. Despite being modified by OGT, the O-GlcNAcylated OGA remains bound to OGT. A segment of its IDR occupies the entire substrate-binding lumen of the TPR domain of OGT and engages the asparagine ladder residues (Fig. 3a). Furthermore, OGA residues surrounding O-GlcNAcylated S405 maintain hydrophobic and hydrogen bonding interactions with the GTD and the intervening domain (ID) of OGT (Fig. 3b-3c). Specifically, OGA W387 and Q396 directly contact T811 and Q812 in OGT ID. Q396, S399, and R400 of OGA also engage the UDP-binding loop of OGT (residues 905–907). V402, A403, and H404 from OGA contact the UDP moiety (Fig. 3c). Thus, by occupying both the substrate-binding groove of the TPR domain and the active site of OGT, OGA acts as a competitive inhibitor of OGT to prevent the binding and modification of other substrates.
+
+Indeed, compared to OGT alone, the OGT–OGA complex was less efficient in catalyzing the O-GlcNAcylation of TAB1 (Fig. 3d). Importantly, OGT in complex with the catalytically inactive OGA D175N mutant was also deficient in modifying TAB1, ruling out the possibility that OGA in the complex actively reverses TAB1 O-GlcNAcylation. These results indicate that OGT in the OGT–OGA complex is incapable of modifying other substrates.
+
+## Structural basis of the inhibition of OGA by OGT
+
+Human OGA forms a domain-swapped homodimer, in which the C-terminal helix (residues 676–694) from one monomer docks on the stalk domain of the other monomer (Fig. 4a). In each monomer, the GHD packs against its own stalk domain, with the intervening IDR being absent from the structure. The glycopeptide substrate binds at a pocket formed by the GHD of one monomer and the stalk domain of the other monomer. Thus, the dimerization of OGA is required for substrate recognition and subsequent removal of GlcNAc.
+
+OGT binding triggers a dramatic conformational change of OGA. Instead of the domain-swapped dimer, OGA in the OGT–OGA complex is monomeric. Only its GHD and a segment of the IDR are visible in the complex. The GHD docks onto the convex surface of TPR11 and TPR12 of OGT whereas the IDR binds to the lumen of the TPR domain of OGT (Fig. 2a). Superimposing the structures of the OGA dimer and the OGT–OGA complex reveals that OGT in the complex develops serious steric clashes with the stalk domain in the same OGA monomer and with the other OGA monomer (Fig. 4b,c). Thus, OGT binding is incompatible with OGA dimerization. In addition, TPR11 and TPR12 of OGT completely occlude the catalytic residues of OGA (Fig. 4d, e). Therefore, OGT inhibits the O-GlcNAcase activity of OGA by disrupting its dimerization and by shielding its active site.
+
+OGA efficiently removed O-GlcNAcylation of TAB1 in vitro, as evidenced by the weaker signals on the anti-O-GlcNAc blot and by the downshift of glycosylated TAB1 bands on Coomassie-stained gel (Fig. 4f). Compared to OGA alone, the OGT–OGA complex had much weaker O-GlcNAcase activity towards TAB1. OGA bound to the catalytically dead H508A mutant of OGT also exhibited weaker O-GlcNAcase activity, suggesting that O-GlcNAcylation of OGA by OGT is not required for OGA inhibition.
+
+# Discussion
+
+The single pair of enzymes OGT and OGA regulates dynamic cycling of O-GlcNAcylation on thousands of cellular proteins with diverse functions. Our structural and functional analyses of the OGT–OGA complex provide key insight into how OGT recognizes its substrates. The GTD and ID of OGT interact with 9–10 residues N-terminal to the serine/threonine residue to be O-GlcNAcylated. The TPR domain of OGT wraps around an extended 40-residue segment C-terminal to the modification site. Thus, remarkably, OGT can simultaneously engage a total of 50 residues in an extended conformation in each substrate. This interaction is reminiscent of the interactions between karyopherins and nuclear localization signals (NLS) <sup>34</sup>, although the NLS segments are much shorter. Many asparagines in the asparagine ladder of OGT interact with the peptide backbone in a sequence-independent manner. Different subsets of TPRs are required for the recognition of different substrates, suggesting that sparse binding hotspots involving sidechain interactions determine the substrate specificity of OGT. This type of backbone-dominated binding mode enables the recognition of a wide variety of substrates by OGT.
+
+OGT prefers to bind to and act on a long segment (at least 50 residues) of intrinsically disordered regions (IDRs) in substrates. Because the OGT TPR domain forms an α-solenoid with two complete superhelical turns, it needs to undergo conformational changes for substrates to gain access to the lumen of all the TPRs, as seen in the OGT–OGA complex (Extended Movie 1 and Extended Movie 2). We propose that the TPR domain of OGT can transiently adopt an open conformation in which the superhelical turns of the α-solenoid are partially unwound (Fig. <span class="InternalRef" refid="Fig5">5</span> a). An IDR segment of the substrate first binds to a subset of TPRs. Other TPRs then form right-handed twists to wrap around a much larger IDR segment of the substrate. Once the substrate occupies the entire lumen of the TPR domain, OGT might scan for optimal sequences within the IDR through passive diffusion and locate specific serines/threonines in this region for modification. If the IDR is located in between folded domains, the substrate cannot escape from the TPR lumen of OGT without the straightening of the superhelical α-solenoid. As such, the topological enclosure of substrates by the OGT TPR domain prolongs the lifetime of the enzyme-substrate complex and enables processive modification of multiple serines/threonines in one binding-release cycle.
+
+The fact that the TPR domain of OGT becomes more rigid and compact upon OGA binding is in general agreement with this model. The most relaxed conformations of OGT are unlikely to be captured by our cryo-EM analyses. We hypothesize that the dimerization of OGT stabilizes the mid-sections of the two α-selonoids in the dimer, allowing each selonoid to transiently reach its more straightened conformation. This might explain why OGT dimerization is required for the optimal O-GlcNAcylation of cellular substrates. Future biophysical experiments and molecular dynamics simulations are needed to test this hypothesis.
+
+Being an important nutrient sensing mechanism, O-GlcNAcylation integrates signals from several metabolic pathways <sup>2,4</sup>. Nutrient conditions that elevate UDP-GlcNAc levels generally increase global O-GlcNAcylation. Under a specific nutrient condition, the functions of OGT and OGA need to be coordinated to maintain O-GlcNAc homeostasis. There is mutual regulation between OGT and OGA at transcriptional and post-translational levels. Our discovery that OGT and OGA inhibit each other’s enzymatic activity establishes a direct mechanism of mutual regulation (Fig. <span class="InternalRef" refid="Fig5">5</span> b).
+
+OGT and OGA physically interact and depend on each other for stability in human cells, suggesting that a large pool of OGT molecules is bound to OGA, and vice versa. Being a competitive inhibitor of OGT, OGA acts as a gatekeeper and reduces the accidental O-GlcNAcylation of sub-optimal substrates. Under proper conditions (e.g. high UDP-GlcNAc concentrations), only substrates that outcompete OGA for OGT binding are effectively O-GlcNAcylated. Binding of strong substrates dislodges the inhibited monomeric OGA from OGT. The released OGA is free to form active dimers, which remove O-GlcNAc from substrates and return the O-GlcNAcylation levels to the ground state. The mutual inhibition between OGT and OGA thus ensures the fidelity of O-GlcNAcylation, limits futile O-GlcNAc cycling, and maintains O-GlcNAc homeostasis.
+
+# References
+
+1. Bond, M. R. & Hanover, J. A. A little sugar goes a long way: the cell biology of O-GlcNAc. *J. Cell Biol.* **208**, 869–880 (2015).
+
+2. Yang, X. & Qian, K. Protein O-GlcNAcylation: emerging mechanisms and functions. *Nat. Rev. Mol. Cell Biol.* **18**, 452–465 (2017).
+
+3. Chatham, J. C., Zhang, J. & Wende, A. R. Role of O-Linked N-Acetylglucosamine Protein Modification in Cellular (Patho)Physiology. *Physiol. Rev.* **101**, 427–493 (2021).
+
+4. Zhu, Y. & Hart, G. W. Nutrient regulation of the flow of genetic information by O-GlcNAcylation. *Biochem. Soc. Trans.* **49**, 867–880 (2021).
+
+5. Ferrer, C. M., Sodi, V. L. & Reginato, M. J. O-GlcNAcylation in Cancer Biology: Linking Metabolism and Signaling. *J. Mol. Biol.* **428**, 3282–3294 (2016).
+
+6. Gonzalez-Rellan, M. J., Fondevila, M. F., Dieguez, C. & Nogueiras, R. O-GlcNAcylation: A Sweet Hub in the Regulation of Glucose Metabolism in Health and Disease. *Front. Endocrinol. (Lausanne)* **13**, 873513 (2022).
+
+7. Lee, B. E., Suh, P. G. & Kim, J. I. O-GlcNAcylation in health and neurodegenerative diseases. *Exp. Mol. Med.* **53**, 1674–1682 (2021).
+
+8. Ma, J., Hou, C. & Wu, C. Demystifying the O-GlcNAc Code: A Systems View. *Chem. Rev.* **122**, 15822–15864 (2022).
+
+9. Fehl, C. & Hanover, J. A. Tools, tactics and objectives to interrogate cellular roles of O-GlcNAc in disease. *Nat. Chem. Biol.* **18**, 8–17 (2022).
+
+10. Joiner, C. M., Li, H., Jiang, J. & Walker, S. Structural characterization of the O-GlcNAc cycling enzymes: insights into substrate recognition and catalytic mechanisms. *Curr. Opin. Struct. Biol.* **56**, 97–106 (2019).
+
+11. Lazarus, M. B., Nam, Y., Jiang, J., Sliz, P. & Walker, S. Structure of human O-GlcNAc transferase and its complex with a peptide substrate. *Nature* **469**, 564–567 (2011).
+
+12. Jinek, M. *et al.* The superhelical TPR-repeat domain of O-linked GlcNAc transferase exhibits structural similarities to importin alpha. *Nat. Struct. Mol. Biol.* **11**, 1001–1007 (2004).
+
+13. Roth, C. *et al.* Structural and functional insight into human O-GlcNAcase. *Nat. Chem. Biol.* **13**, 610–612 (2017).
+
+14. Li, B., Li, H., Hu, C. W. & Jiang, J. Structural insights into the substrate binding adaptability and specificity of human O-GlcNAcase. *Nat. Commun.* **8**, 666 (2017).
+
+15. Li, B., Li, H., Lu, L. & Jiang, J. Structures of human O-GlcNAcase and its complexes reveal a new substrate recognition mode. *Nat. Struct. Mol. Biol.* **24**, 362–369 (2017).
+
+16. Elsen, N. L. *et al.* Insights into activity and inhibition from the crystal structure of human O-GlcNAcase. *Nat. Chem. Biol.* **13**, 613–615 (2017).
+
+17. Ong, Q., Han, W. & Yang, X. O-GlcNAc as an Integrator of Signaling Pathways. *Front. Endocrinol. (Lausanne)* **9**, 599 (2018).
+
+18. Stephen, H. M., Adams, T. M. & Wells, L. Regulating the Regulators: Mechanisms of Substrate Selection of the O-GlcNAc Cycling Enzymes OGT and OGA. *Glycobiology* **31**, 724–733 (2021).
+
+19. Yang, W. H. *et al.* Modification of p53 with O-linked N-acetylglucosamine regulates p53 activity and stability. *Nat. Cell Biol.* **8**, 1074–1083 (2006).
+
+20. Pathak, S. *et al.* O-GlcNAcylation of TAB1 modulates TAK1-mediated cytokine release. *EMBO J.* **31**, 1394–1404 (2012).
+
+21. Sakabe, K., Wang, Z. & Hart, G. W. Beta-N-acetylglucosamine (O-GlcNAc) is part of the histone code. *Proc. Natl. Acad. Sci. U. S. A.* **107**, 19915–19920 (2010).
+
+22. Lo, P. W. *et al.* O-GlcNAcylation regulates the stability and enzymatic activity of the histone methyltransferase EZH2. *Proc. Natl. Acad. Sci. U. S. A.* **115**, 7302–7307 (2018).
+
+23. Liu, F., Iqbal, K., Grundke-Iqbal, I., Hart, G. W. & Gong, C. X. O-GlcNAcylation regulates phosphorylation of tau: a mechanism involved in Alzheimer's disease. *Proc. Natl. Acad. Sci. U. S. A.* **101**, 10804–10809 (2004).
+
+24. Marotta, N. P. *et al.* O-GlcNAc modification blocks the aggregation and toxicity of the protein alpha-synuclein associated with Parkinson's disease. *Nat. Chem.* **7**, 913–920 (2015).
+
+25. Bolanle, I. O., Riches-Suman, K., Williamson, R. & Palmer, T. M. Emerging roles of protein O-GlcNAcylation in cardiovascular diseases: Insights and novel therapeutic targets. *Pharmacol. Res.* **165**, 105467 (2021).
+
+26. Whisenhunt, T. R. *et al.* Disrupting the enzyme complex regulating O-GlcNAcylation blocks signaling and development. *Glycobiology* **16**, 551–563 (2006).
+
+27. Buckley, D. L. *et al.* HaloPROTACS: Use of Small Molecule PROTACs to Induce Degradation of HaloTag Fusion Proteins. *ACS Chem. Biol.* **10**, 1831–1837 (2015).
+
+28. Meek, R. W. *et al.* Cryo-EM structure provides insights into the dimer arrangement of the O-linked beta-N-acetylglucosamine transferase OGT. *Nat. Commun.* **12**, 6508 (2021).
+
+29. Kositzke, A. *et al.* Elucidating the protein substrate recognition of O-GlcNAc transferase (OGT) toward O-GlcNAcase (OGA) using a GlcNAc electrophilic probe. *Int J Biol Macromol* **169**, 51–59 (2021).
+
+30. Stephen, H. M., Praissman, J. L. & Wells, L. Generation of an Interactome for the Tetratricopeptide Repeat Domain of O-GlcNAc Transferase Indicates a Role for the Enzyme in Intellectual Disability. *J. Proteome Res.* **20**, 1229–1242 (2021).
+
+31. Gorelik, A. & Ferenbach, A. T. CRISPR-Cas9-mediated depletion of O-GlcNAc hydrolase and transferase for functional dissection of O-GlcNAcylation in human cells. *bioRxiv*, doi: https://doi.org/10.1101/2020.1108.1119.258079 (2020).
+
+32. Lin, C. H., Liao, C. C., Chen, M. Y. & Chou, T. Y. Feedback Regulation of O-GlcNAc Transferase through Translation Control to Maintain Intracellular O-GlcNAc Homeostasis. *Int. J. Mol. Sci.* **22** (2021).
+
+33. Khidekel, N. *et al.* Probing the dynamics of O-GlcNAc glycosylation in the brain using quantitative proteomics. *Nat. Chem. Biol.* **3**, 339–348 (2007).
+
+34. Wing, C. E., Fung, H. Y. J. & Chook, Y. M. Karyopherin-mediated nucleocytoplasmic transport. *Nat. Rev. Mol. Cell Biol.* **23**, 307–328 (2022).
+
+# Methods
+
+## Protein expression and purification
+
+The cDNAs encoding the full-length of human OGT (UniProt: O15294), OGA (UniProt: O60502), and TAB1 (UniProt: Q15750) were cloned from a human cDNA library or synthesized by the Tsingke Biotech Company. Human OGT was cloned into the pET21b vector (Novagen) to produce human OGT with a C-terminal His<sub>6</sub> affinity tag. Human OGA and TAB1 were individually cloned into the pRSF vector to produce proteins with an N-terminal His<sub>6</sub> tag followed by the PreScission protease site. The fragments of OGA (residues 370–440 and 441–511) were cloned into the pGEX-6P-1 vector. Site-directed mutagenesis of human OGT and OGA was performed using the Q5® Site-Directed Mutagenesis Kit (New England Biolabs). OGT and its mutants were also cloned into the pCS2 expression vector with N-terminal Myc tags for expression in human cells. OGA and TAB1 were cloned into the pCS2 expression vector with N-terminal Flag tags. All constructs were verified by DNA sequencing.
+
+BL21 (DE3) *Escherichia coli* cells containing the desired plasmids were grown in LB medium with ampicillin (100 µg/l for the pET21b and pGEX-6p-1 plasmids) or kanamycin (50 µg/l for the pRSF plasmid) with shaking at 37˚C to OD<sub>600</sub> of 1.0. The cell cultures were cooled to 18˚C, induced with 0.5 mM IPTG for 12–15 h, and harvested by centrifugation. The cell pellets were suspended in the lysis buffer (25 mM Tris-HCl pH 8.0, 150 mM NaCl, 5 mM β-mercaptoethanol, 1 mM PMSF, and with 5 mM imidazole added for His<sub>6</sub>-tagged proteins). The cells were lysed by sonication or an UltraHigh Pressure Homogenizer and centrifuged at 40,000g for 50 min at 4˚C. The supernatants were incubated with pre-equilibrated Ni<sup>2+</sup>-NTA agarose beads for His<sub>6</sub>-tagged proteins (Qiagen) or Glutathione Sepharose 4B beads for GST-tagged proteins (GE Healthcare) for 2 h at 4˚C. The beads were washed with 20 column volumes (CV) of the wash buffer (25 mM Tris pH 8.0, 150 mM NaCl, 5 mM β-mercaptoethanol, with 20 mM imidazole added for His<sub>6</sub>-tagged proteins). The His<sub>6</sub>-tagged proteins were eluted with 15 ml elution buffer (25 mM Tris-HCl pH 8.0, 150 mM NaCl, 5 mM β-mercaptoethanol, and 250 mM imidazole). The GST-tagged proteins were eluted with 5 ml elution buffer (25 mM Tris-HCl pH 8.0, 150 mM NaCl, 5 mM β-mercaptoethanol, and 15 mM reduced GSH) or incubated with the HRV 3C protease overnight. The proteins were loaded onto a Resource Q column and fractionated by the AKTA Pure system (GE Healthcare) equilibrated with Buffer A (25 mM Tris pH 8.0, 10 mM NaCl, 5 mM β-mercaptoethanol). The proteins were eluted with a linear 5–50% gradient of Buffer B (25 mM Tris pH 8.0, 1 M NaCl, 5 mM β-mercaptoethanol) over 13 CVs. The pooled peak fractions were concentrated and loaded onto a Superose 6 10/300 Increasing Column equilibrated with Buffer C (25 mM HEPES pH 7.4, 100 mM NaCl, 2 mM DTT, 0.05% NP40). The peak fractions were collected, analyzed by 10% SDS-PAGE, aliquoted, and stored at -80˚C.
+
+For producing the human OGT–OGA complex, OGT and OGA were co-expressed in BL21 (DE3) and purified using the same protocol as described above. For producing O-GlcNAcylated TAB1 (G-TAB1), TAB1 and OGT were co-expressed in BL21 (DE3). G-TAB1 was purified using the same protocol as described above. TAB1 O-GlcNAcylation was verified by Western blotting using the anti-O-GlcNAc antibody (RL2) (Abcam, ab2739). The peak fractions containing G-TAB1 were collected and stored.
+
+## Pull-down assays
+
+For GST pull-down assays, the concentrations of all proteins used in the assays were adjusted to 1 mg/ml. GST or GST-OGA proteins were used as baits. His<sub>6</sub>-tagged OGT and its mutant proteins were used as preys. Each 10 µl bait samples were incubated with 10 µl Glutathione Sepharose 4B beads for 1 h at 4˚C. The beads were washed 3 times with the binding buffer (25 mM Tris pH 8.0, 150 mM NaCl, 2 mM DTT, 0.05% NP40). Then 10 µl prey proteins were added and incubated with beads in a 100-µl binding reaction for 2 h at 4˚C. The beads were washed 5 times with the binding buffer and boiled with the SDS sample buffer. The samples were analyzed by SDS-PAGE, stained with Coomassie blue, and imaged using the LI-COR Odyssey system.
+
+## O-GlcNAcylation and O-GlcNAcase assays
+
+For the O-GlcNAcylation assays, OGT WT and mutants or the OGT–OGA complex and its mutants were incubated with human TAB1 for 90 min at 37˚C or human OGA proteins (WT, D175N, or D175N/S405A) overnight at 37˚C. The assays were performed in 20 µl volumes with 2 µg enzymes, 2 mM UDP-GlcNAc, 1 µg TAB1 or OGA in the reaction buffer (25 mM Tris pH 8.0, 100 mM NaCl, 5 mM MgCl<sub>2</sub>, 1 mM DTT). The reactions were stopped by the addition of the SDS sample buffer and then analyzed by SDS-PAGE followed by Western blotting with the anti-O-GlcNAc antibody (RL2).
+
+For the O-GlcNAcase assay, human O-GlcNAcylated TAB1 (G-TAB1) was used as the substrate. The assays were performed in 20 µl volumes with 2 µg OGA and 1 µg G-TAB1 in the reaction buffer (25 mM Tris pH 8.0, 100 mM NaCl, 5 mM MgCl<sub>2</sub>, 1 mM DTT) for 90 min at 37˚C. The reaction mixtures were analyzed by SDS-PAGE and Western blotting with the anti-O-GlcNAc antibody (RL2).
+
+## Mammalian cell culture and cellular O-GlcNAcylation assays
+
+HEK293FT cells were cultured in 10-cm or 6-well plates in a 37˚C incubator with 5% CO<sub>2</sub> in the GIBCO™ DMEM (Fisher Scientific) medium supplemented with 10% fetal bovine serum (Sigma Aldrich) and 1% penicillin-streptomycin-glutamine (Invitrogen). The sgRNA targeting the C terminal region of human OGT (5’-agcataaataaagactgcac-3’) was ligated into the pSpCas9(BB)-2A-Puro (PX459) V2.0 vector. The homology-directed repair (HDR) template containing the 5’ homology arm (~ 500 bp), the HaloTag9-3X Flag tag-P2A-Hygro (hygromycin B resistance gene) cassette, and the 3’ homology arm (~ 500 bp) was cloned into the pUC19 vector and co-transfected with the Cas9 plasmid into human 293FT cells using Lipofectamine 3000 (Thermo Fisher Scientific). At 6 h after transfection, the media were replaced with the fresh media containing 5 µM farrerol<sup>35</sup>. At 24 h after transfection, the media were changed to fresh DMEM complete media. After several rounds of hygromycin B (200 µg/ml) selection, single clones were picked and seeded into new 6-well plates. The clones were screened by PCR sequencing and Western blotting with the anti-OGT and anti-Flag antibodies for the integration of the HaloTag9-Flag tag cassette into the endogenous *OGT* locus. Depletion of the resulting OGT-HaloTag9 fusion protein in 293FT cells were induced with the addition of HaloPROTAC3 (2 µM). The cell samples were collected at different timepoints and analyzed by Western blotting.
+
+The 293FT OGT-HaloTag9 knock-in cells were transfected with Myc-OGT WT and mutant plasmids using Lipofectamine 2000 (Thermo Fisher Scientific). The cells were then treated with HaloPROTAC3 (2 µM) in the culture media for 24 h. The cells were harvested, re-suspended in 2X SDS sample buffer, boiled for 5 min at 95˚C, and analyzed by SDS-PAGE followed by Western blotting.
+
+The primary antibodies used in Western blotting included: rabbit anti-OGT (Abcam, ab177941), mouse anti-OGA (Abcam, ab68522), rabbit anti-OGA (Proteintech, 14711-1-AP), mouse anti-Myc (Sigma-Aldrich, M4439), anti-Flag (Sigma-Aldrich, F1804), mouse anti-O-GlcNAc (Cell Signaling, CTD110.6, 9875s), mouse anti-O-GlcNAc (RL2) (Abcam, ab2739), and mouse anti-GAPDH (Proteintech, 60004-1-Ig).
+
+## Cryo-EM data collection and image processing
+
+For cryo-EM grid preparation, 3 µl samples (~ 5 mg/ml) were applied onto glow discharged holey carbon grids (Quantifoil Cu R1.2/1.3, 300 mesh), blotted with a Vitrobot Marker IV (Thermo Fisher Scientific) for 3 s under 100% humidity at 4˚C, and subjected to plunge freezing into liquid ethane. All cryo-EM data were collected using an FEI Titan Krios microscope at 300 kV equipped with a Gatan K3 Summit direct electron detector (super-resolution mode, at a nominal magnification of 105,000) and a GIF-quantum energy filter. Defocus values were set from – 1.8 to -2.3 um. Each stack of 32 frames was exposed for 2.13 s, with a total electron dose of 50 e<sup>–</sup>/Å<sup>2</sup>. AutoEMation was used for fully automated data collection<sup>36</sup>.
+
+All micrograph stacks were motion corrected with MotionCor2<sup>37</sup> with a binning factor of 2, resulting in a pixel size of 0.861 Å. Contrast transfer function (CTF) parameters were estimated using Gctf<sup>38</sup>. Most steps of image processing were performed using cryoSPARC<sup>39</sup>. For 3D processing of the OGT data, a total of 6,266,880 particles were automatically picked from 5,539 micrographs using Gautomatch (developed by Kai Zhang, MRC-LMB). Particles were extracted with a pixel size of 3.444 Å and subjected to several rounds of reference-free 2D classification. 1,264,325 particles were kept after the exclusion of obvious ice contamination and junk particles. Then, *ab initio* models were generated and subsequently used for heterogeneous 3D refinement. The best class of 493,491 particles were reextracted without binning. After the last round of 3D classification, 187,680 particles were used for further 3D refinement, including homologous refinement, heterogeneous refinement, non-uniform refinement, and local refinement. The global resolution of the OGT homodimer is 3.69 Å based on the Fourier Shell Correlation (FSC) 0.143 criterion.
+
+For data processing of the OGT–OGA complex, 5,316,274 particles were picked using Gautomatch from 5,744 micrographs. After particle extraction with a binning factor of 4, 458,373 particles were reextracted without binning after the last round of 3D classification. Upon several rounds of 2D and 3D classification combined with different subsets, three major conformations of the OGT–OGA complex emerged: conformation I (85,916 particles), conformation II (114,543 particles), and conformation III (54,789 particles). After the final round of 3D refinement, the global resolutions of the three conformations were determined to be 5.68 Å, 3.92 Å, and 5.86 Å, respectively, using FSC 0.143 criterion. Finally, cryo-EM density maps were sharpened using the negative B-factor reported by cryoSPARC<sup>39</sup>. Conformation II of the OGT–OGA complex had the highest resolution and was used for further structural analysis.
+
+For data processing of human OGT monomeric W208A/L209A/I211A/H212A (4A) mutant and human OGA protein, 135,924 and 271,000 particles were used for the final round of 2D classification, respectively.
+
+## Model building and refinement
+
+The X-ray structures of human OGT (PDB: 1W3B and 3PE3) or OGA (PDB: 5UN9) were used as the starting models and docked into the final EM maps with UCSF Chimera<sup>40</sup>. The models were manually adjusted and iteratively built in COOT<sup>41</sup> and then refined against summed maps using phenix.real_space_refine implemented in PHENIX<sup>42</sup> until the validation data were reasonable. FSC values were calculated between the resulting models and the two half-maps, as well as the averaged map of the two half-maps. The quality of the models was evaluated with MolProbity<sup>43</sup> and EMRinger<sup>44</sup>. The structure validation statistics were listed in Extended Data Table 1. All structural figures were prepared with PyMOL<sup>45</sup>, Chimera<sup>40</sup> or Chimera X<sup>46</sup>.
+
+## Statistical analysis
+
+No statistical methods were used to predetermine sample size or applied to data analysis. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.
+
+## Data availability
+
+The cryo-EM density maps of the OGT dimer and the OGT–OGA complex have been deposited to the Electron Microscopy Data Bank under the accession numbers EMD-33768 (OGT dimer), EMD-33767 (OGT–OGA conformer I), EMD-33773 (OGT–OGA conformer II), and EMD-33769 (OGT–OGA conformer III). Atomic coordinates have been deposited to the RCSB Protein Data Bank under the accession numbers 7YEA (OGT dimer) and 7YEH (OGT-OGA conformer II).
+
+35. Zhang, W. *et al.* A high-throughput small molecule screen identifies farrerol as a potentiator of CRISPR/Cas9-mediated genome editing. *eLife* **9** (2020).
+
+36. Lei, J. & Frank, J. Automated acquisition of cryo-electron micrographs for single particle reconstruction on an FEI Tecnai electron microscope. *J. Struct. Biol.* **150**, 69-80 (2005).
+
+37. Zheng, S. Q. *et al.* MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. *Nat. Methods* **14**, 331-332 (2017).
+
+38. Zhang, K. Gctf: Real-time CTF determination and correction. *J. Struct. Biol.* **193**, 1-12 (2016).
+
+39. Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. *Nat. Methods* **14**, 290-296 (2017).
+
+40. Pettersen, E. F. *et al.* UCSF Chimera--a visualization system for exploratory research and analysis. *J. Comput. Chem.* **25**, 1605-1612 (2004).
+
+41. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. *Acta Crystallogr. D Biol. Crystallogr.* **66**, 486-501 (2010).
+
+42. Afonine, P. V. *et al.* Real-space refinement in PHENIX for cryo-EM and crystallography. *Acta Crystallogr. D Struct. Biol.* **74**, 531-544 (2018).
+
+43. Chen, V. B. *et al.* MolProbity: all-atom structure validation for macromolecular crystallography. *Acta Crystallogr. D Biol. Crystallogr.* **66**, 12-21 (2010).
+
+44. Barad, B. A. *et al.* EMRinger: side chain-directed model and map validation for 3D cryo-electron microscopy. *Nat. Methods* **12**, 943-946 (2015).
+
+45. Alexander, N., Woetzel, N. & Meiler, J. bcl::Cluster : A method for clustering biological molecules coupled with visualization in the Pymol Molecular Graphics System. *IEEE Int. Conf. Comput. Adv. Bio. Med. Sci.* **2011**, 13-18 (2011).
+
+# Supplementary Files
+
+- [ExtendedDataFiguresTableNCB20221202.docx](https://assets-eu.researchsquare.com/files/rs-2275302/v1/c561a8b9a3d24f7f7d3887fa.docx)
+
+- [movie1.mp4](https://assets-eu.researchsquare.com/files/rs-2275302/v1/9ad1dd154c35c46fc8358b16.mp4)  
+  Cryo-EM maps and structures of human OGT homodimer and OGT-OGA complex
+
+- [movie2.mp4](https://assets-eu.researchsquare.com/files/rs-2275302/v1/ed845ab646899222cfaf4b32.mp4)  
+  The conformational changes of OGT upon OGA binding
+
+- [ValidationReports.zip](https://assets-eu.researchsquare.com/files/rs-2275302/v1/330959f9de3c269cbd8e8c20.zip)  
+  Validation reports

145662a0fee26f65d0087d05ecbd2312d6f95a29c1fecf9ac3e31ceeb5a368a1/preprint/images_list.json

@@ -0,0 +1,58 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Scheme of in vitro co-transcriptional translation assay and quantification.\na, Schematic of the co-transcriptional translation assay. The DNA is constructed with a T7 promoter (yellow), E coli. ribosome binding site (orange) and GFP gene. b, c Example curves of transcription and translation data. The blue curve exemplifies the transcription readout in real-time via a Cy3 molecular beacon that fluoresces when bound to mRNA. The green curve represents translation readout as the fluorescence of sfGFP. The initial linear phases of curves are used to quantify the transcription and translation initiation rate, respectively. The calculation is described in \u201cMethod.\u201d",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "PQS orientation and presence of RNA G4 increases translation efficiency.\na, Schematic of RNA G4 insertion in the 5\u2019-UTR. PQS in non-template (NT) strand leads to the formation of rG4 in the mRNA, while insertion in template (T) strand and control does not. b, c, f, Real-time intensities of transcription, translation and G4 formation assays. The constructs, non-template (NT), control (C), and template (T) are colored in blue, black, and purple, respectively. b, Transcription assay is quantified by Cy3 probe intensity. d, Translation is measured as sfGFP intensity.f, RNA G4 formation is quantified by real-time NMM signal. NT shows NMM signal while C and T have no signal. c, Transcription rates are calculated from the early linear part of the curve in b and normalized to the transcription rate of the control sequence. The transcription rate of NT is 30% higher than T. e, Translation efficiencies are calculated from the translation rates obtained from the early linear part of the curve in dand the normalized transcription rates in c. The translation efficiency was normalized to the control sequence. NT shows enhanced translation. For cand e, data are presented as mean \u00b1 SEM of independent experiments (n > 6). NS: nonsignificant, *P < 0.05, ***P < 0.0005 (two-sided unpaired t-test). g, Halftime to saturation of NT. The halftimes of transcription (in b), translation (in d), and NMM RG4 formation (in f) are 32 \u00b1 2.8, 47 \u00b1 7.5, and 71.7 \u00b1 4.6 min, respectively. Data are presented as mean \u00b1 SEM of independent experiments (n = 3).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Bulkiness of RG4 drives translational enhancement.\na, PQS candidates with varying loop lengths. The core domain of guanine triplets (black) remains constant, and loop sequences (highlighted in red) are varied. The sequences are arranged from shortest to longest loop length, indicating the increase of bulkiness. b, Real-time GFP signal measurements of individual PQS candidates by plate reader. NT, T, and control are colored in blue, purple, and gray, respectively. c, Normalized translation efficiencies are calculated from the early linear part of curve in b. Data are presented as mean \u00b1 SEM of n = 3-5 independent experiments. The numbers on the x-axis represent the PQS ordered as shown in a. NT constructs with longer loop lengths, representing bulkier RG4 structures, resulted in higher translation efficiencies. Shown in c only represents the significance between template and non-template, where **P < 0.005 (two-sided unpaired t-test). d, The correlation between translation efficiency and total loop length. The loop length is the sum of the uridine bases in a. The correlation coefficients are 0.89 and 0.36 for NT and T, indicating a strong correlation between translation efficiency and RG4 for NT. Data are represented as mean \u00b1 SEM of n = 3-5 and 3 for NT and T, respectively.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Hairpin and RG4 structures synergistically promote translation.\na, Schematic of the 5\u2019-UTR. Domain 1 is upstream of the 5\u2019-UTR and the position for hairpin insertion. Domain 2 is the PQS location. Domain 3 is downstream of RG4. Domain 4 begins at the RBS and ends at +9 of translation codons. The predicted mRNA folding energies are provided in Supplementary Table 1. b, Translation efficiencies of NT constructs containing different hairpins and RG4 in the 5\u2019-UTR. The constructs are grouped in order of hairpin stem from short (left) to long (right). The x-axis of each group indicates the RG4 sequence in domain 2, arranged in increasing total loop length. All the translation efficiencies are normalized to 10 hp-control. Data are represented as mean \u00b1 SEM of n = 3-10. c, The correlation between translation efficiency and total loop length. The data are grouped in order of hairpin stem, and each data point is normalized to the control within the group. Strong correlations observed between translation efficiency and loop length for all hairpin structures indicate that dependence of domain 2 RG4 remains regardless of domain 1. d, Heat map of the translation efficiencies from b. The x-axis is RG4 arranged in order of longer loop lengths, and the y-axis is the length of hairpin stem. The color represents the mean translation efficiency and is scaled from 0 (white) to 12 (dark red). The color trend demonstrates the hairpin and RG4 size synergistical dependence on translation efficiencies.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "RG4 does not recruit ribosome or increase RBS accessibility.\na-d Hypothesis one: RG4 increases the affinity to ribosome. a, Schematic of RG4 recruiting the ribosome. b, Competitor RNA used in translation reaction. The RG4 is PQS cMyc (Fig. 3a), and the hairpin is 10 hp (Fig. 4a). RNA sequences are provided in Supplementary Table 2. c, Transcription rate of PQS-NT with addition of competitor RNA. Data are presented as mean \u00b1 SEM of independent experiments (n = 3). Shown in c represents the significance between PQS-NT and addition of competitor, where *P < 0.05 (two-sided paired t-test). d, Normalized GFP intensity. Additions of RNA 1-4 show no significant difference from PQS-NT while RNA containing the RBS (5 and 6) reduces the GFP production, suggesting successful competition requires RBS not RG4. e-g Hypothesis two: RG4 enhances the accessibility of RBS to ribosome. e, Cartoon depicting the release of RBS from other secondary structure by RG4. f, Schematic for examining the accessibility of the RBS. The mRNA is either heated to remove all secondary structure or not heated. A Cy5-labled molecular beacon complementary to the RBS is applied to determine accessibility. g, Binding curves of the RBS molecular beacon to RNAs. The data points are collected by titrating RNA concentration in a 2-fold series dilution from 1.75 \u00b5M to 24 nM and represents the average number from independent experiments (n = 5). The dissociation constants, Kd, are 367 (\u00b1 25), 364 (\u00b1 23), 370 (\u00b1 21) nM for NT (blue rectangle), Control (gray circle), and T (purple diamond), respectively, showing a negligible difference in binding affinity (two-sided paired t-test). h, Fluorescence intensities of molecular beacon with and without heating. There is no significant difference among the control, T, and NT with or without heating. Data are presented as mean \u00b1 SEM of independent experiments (n = 3). NS: nonsignificant (two-sided paired t-test).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "RG4 does not increase mRNA lifetime but may stabilize ribosome bound to mRNA.\na, b, Hypothesis 3: RG4 increases the mRNA lifetime. a, Illustration of how RG4 could prevent the mRNA from degradation. b, RT-PCR is performed with two primers, proximal and distal to the 5\u2019 end. No significant difference is observed among the control, T, and NT transcript levels with either primer, indicating that RG4 does not affect mRNA lifetime. Shown is a representative result from three independent experiments. c-g, Hypothesis 4: RG4 stabilizes the ribosome-bound state. c, A potential mechanism of RG4 stabilizing the ribosome bound to the mRNA by preventing ribosomes from dislodging off the mRNA. d, RG4 specific helicase, DHX36, is used to remove RG4 from mRNA during translation. e, Translation readout as real-time GFP intensity. The titration of helicase demonstrates a dose-dependence such that the more addition of helicase decreases the translation level further. f, Normalized transcription rate of NT construct with DHX36 titration. g, Normalized translation efficiencies are calculated from the early linear part of curve in e, and normalized to transcription level in f. It reveals a dose dependent decrease in translation, indicating that RG4 is essential for translation enhancement. For f and g, data are presented as mean \u00b1 SEM of independent experiments (n = 3). **P < 0.005, ***P < 0.0005 and NS: nonsignificant (two-sided paired t-test).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_7.png",
+        "caption": "Translation enhancement pattern is observed in E. coli.\na, Schematic of the dual-color fluorescence reporter system. The two reporters, mCherry and GFP, have the same T7 promoter (yellow) and RBS (orange). The 5\u2019-UTR of the GFP contains the 10 hp hairpin and PQS (red) that generates RG4. b, Fluorescence imaging of GFP expression in E. coli. c, d, Real-time GFP and mCherry intensities. The curves represent example traces of cMyc, where non-template (NT), control, and template (T) are colored in blue, gray, and purple, respectively. The data is collected after IPTG induction by plate reader. c, The orientation-dependence is observed in GFP. d, The three curves show no difference in mCherry. e,Normalized translation efficiencies are defined by the ratio of GFP and mCherry signal at 210 min after induction. The calculation is described in \u201cMethod\u201d and Supplementary Fig. 4. Data are presented as mean \u00b1 SEM of n = 3 independent experiments. Shown in e only represents the significance between template and non-template, where *P < 0.05 (two-sided paired t-test). The PQS-NT constructs showed higher translation than T and the translation increased with bulky PQS.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

145662a0fee26f65d0087d05ecbd2312d6f95a29c1fecf9ac3e31ceeb5a368a1/preprint/preprint.md

@@ -0,0 +1,175 @@
+# Abstract
+
+Translation initiation in bacteria is frequently regulated by various structures in the 5’ untranslated region (5’UTR). Previously, we demonstrated that G-quadruplex (G4) formation in non-template DNA enhances transcription. In this study, we aimed to explore how G4 formation in mRNA (RG4) at 5’UTR impacts translation using a T7-based in vitro translation system and in *E. coli*. We showed that RG4 strongly promotes translation efficiency in a size-dependent manner. Additionally, inserting a hairpin upstream of the RG4 further enhances translation efficiency, reaching up to a 12-fold increase. We found that the RG4-dependent effect is not due to increased ribosome affinity, ribosome binding site accessibility, or mRNA stability. We proposed a physical barrier model in which bulky structures in 5’UTR prevent ribosome dislodging and thereby increase the translation output. This study provides biophysical insights into the regulatory role of 5’UTR structures in bacterial translation, highlighting their potential applications in tuning gene expression.
+
+[Biological sciences/Biophysics/Molecular biophysics](/browse?subjectArea=Biological%20sciences%2FBiophysics%2FMolecular%20biophysics) [Biological sciences/Biochemistry/RNA](/browse?subjectArea=Biological%20sciences%2FBiochemistry%2FRNA)
+
+# Introduction
+
+Gene expression is a tightly regulated process to ensure efficient utilization of resources and adaptation to changing environments. This regulation occurs at various levels, including transcription, translation, and the level of mRNA and protein<sup>1, 2, 3, 4, 5, 6, 7</sup>. In bacteria, the absence of a nuclear membrane necessitates rapid post-transcriptional regulations to enable quick responses to the environmental stimuli<sup>8, 9, 10, 11</sup>. Untranslated regions (UTR) of RNA have emerged as key players in regulating translation initiation by presenting noncanonical structures to translational machinery or by recruiting proteins and enzymes that recognize RNA sequences, modifications, or structures<sup>12, 13</sup>.
+
+The 5’ untranslated region (5’UTR) of bacterial mRNA serves multiple functions that are critical for gene regulation and protein synthesis. First, it typically contains a conserved AG-rich Shine-Dalgarno (SD) sequence, located a few nucleotides upstream of the translation start site (TSS). The SD sequence base-pairs with the 16S ribosomal RNA (rRNA) to guide the binding of the small ribosomal subunit, providing a well-defined mechanism for initiating translation<sup>14</sup>. Second, bacterial 5’UTRs often harbor cis-acting regulatory elements, such as upstream open reading frames (uORFs), which stall the ribosome and control the access to downstream TSS. Additionally, bacterial 5’UTRs can serve as a platform for RNA-binding proteins and small RNAs that regulate translation efficiency. For example, small RNA coupled with an RNA binding protein Hfq can bind a 5’UTR to stimulate translation initiation or trigger mRNA degradation<sup>10, 15, 16</sup>.
+
+Furthermore, secondary structures within 5’UTR play a critical role in regulating RNA stability and translation efficiency. Specifically, co-transcriptionally folded structures can influence translation initiation and the rate of translation. Previous research on bacterial 5’UTR primarily focused on the ribosome binding site (RBS), which includes the SD sequence and a short range (10–20 nt) upstream and downstream of the SD region<sup>14, 17</sup>. Several studies showed that secondary structures, such as pseudoknots and hairpin stem-loop that form across the SD sequence can inhibit translation by preventing ribosome binding<sup>18, 19</sup>. Notably, temperature sensitive hairpins in the 5’UTR of *E. coli* can regulate translation by masking or unmasking RBS or start codon (AUG) to turn on or off translation initiation, respectively<sup>20</sup>. Similarly, riboswitches control translation through changes in mRNA conformation upon ligand binding, enabling rapid responses to environmental cues<sup>21, 22</sup>. Specific sequence elements, such as purine-rich regions or G-quadruplexes, can also affect translation in a context-dependent manner. Previous studies demonstrated that an RNA G-quadruplex (RG4) structure located near the SD sequence inhibits the base pairing between 16S rRNA and the mRNA<sup>23, 24</sup>. However, it remains unclear how and to what extent the 5’UTR structure impacts bacterial gene regulation.
+
+Our previous research has uncovered the role of potential G-quadruplex sequence (PQS) in non-template DNA in promoting transcription through co-transcriptional formation of R-loop and G4 structure<sup>25</sup>. In this context, the transcribed mRNA bears a G4 structure at the 5’ end, which prompted us to investigate whether such G4 structures in RNA modulates translation outcome. G-quadruplexes form in single stranded DNA or RNA that harbors repetitive runs of guanines interspersed with non-guanine, loop sequences. Four guanine bases come together in a coplanar arrangement to form a tetrad, which stacks in multiple layers. The size and stability of the structure depend on the composition and the length of the loops<sup>26</sup>. Increasing evidence suggests that RG4 structures are involved in translation regulation in eukaryotes, often blocking translation initiation when present in the 5’UTR<sup>27, 28, 29, 30, 31</sup>. Some RG4 structures have also been shown to function as Internal Ribosome Entry Sites (IRES), stimulating translation independent of a start site<sup>32, 33, 34</sup>. Furthermore, computational studies have identified numerous PQS across prokaryotic species positioned non-randomly in non-coding RNA segment that precedes mRNA, indicating an evolutionarily conserved function of 5’UTR RG4 in bacterial genome<sup>35, 36, 37</sup>. While some studies demonstrate reduced translation by the 5’UTR RG4, it is not clear if different sequence composition and hence structure of RG4 would drive a different type and level of translational regulation.
+
+In this study, we investigated the role of 5’UTR RG4 structures in *E. coli* translation. We inserted a series of PQSs in non-template DNA, upstream of a GFP reporter gene to allow for the formation of RG4 at 5’UTR. Using the T7 expression system, we measured *in vitro* transcription and subsequent translation in real-time, which were used to calculate the translation efficiency. We found that the presence of RG4 in the 5’UTR led to enhanced translation both *in vitro* and in *E. coli*. Longer loops within RG4 resulted in higher translation yield. Moreover, insertion of a hairpin upstream of an RG4 further increased translation. Taken together, we demonstrate that the translation enhancement scales with the size of the 5’UTR structures. We propose a mechanism by which the 5’UTR structures act as a physical barrier that may prevent dissociation of ribosome from the mRNA and thereby promote translation.
+
+# Result
+
+## RNA G4 increases translation efficiency
+
+To quantify the translation efficiency, we set up an *in vitro* translation assay which contains reagents for the T7 RNAP for transcription and *E. coli* translation system (Fig. 1a) <sup>25, 38</sup>. We prepared DNA construct with T7 promoter followed by the ribosome binding site (RBS), a fluorescent reporter which encodes *superfold* GFP (*sf* GFP), and a transcription terminator sequence. We used our previously established protocol to measure the real-time transcription; DNA molecular beacon becomes fluorescent when annealed to a transcribed RNA which bears a complementary sequence (Fig. 1b) <sup>39</sup>. In parallel, the intensity of *sf* GFP was obtained and plotted as a translation read-out (Fig. 1c). Hence, the real-time transcription and translation activities can be simultaneously measured by collecting intensities of the molecular beacon and GFP over time using a plate reader. The GFP signal is expected to rise after that of the molecular beacon because of the time delay between transcription and translation and the maturation time required for the *sf* GFP folding <sup>40</sup>. Based on the simultaneous measurement, we can calculate the translational efficiency for each reaction by normalizing the translation signal (*sf*GFP) by the transcription signal (molecular beacon).
+
+To examine the effect of 5’UTR RNA G-quadruplex (RG4) on translation, we inserted a potential G-quadruplex forming sequences (PQS) in between the T7 promoter and the RBS such that the PQS is 44 bp downstream of T7 promoter and 43 bp upstream from the RBS. The PQS was inserted into either a template (T) or a non-template (NT) strand for comparison. Hence, the PQS insertion in NT is expected to produce the G4-bearing transcript which can fold into RG4 while the PQS in T, and the scrambled control (C) sequence will not fold into RG4 (Fig. 2a). In agreement with our previous study, the PQS-NT led to approximately 30% higher transcription efficiency than in PQS-T (Fig. 2b) <sup>25</sup>. The initiation rate of transcription was quantified by taking the linear increase of fluorescence intensity in the initial phase of each curve (Fig. 2c). Surprisingly, the translation reporter, *sf* GFP signal revealed that the PQS-NT induced over four-fold higher protein product compared to the control (Fig. 2d). The translation efficiency (TE) was calculated by dividing the initiation rate of *sf* GFP signal by the transcription initiation rate for each condition. (Fig. 2e). The 4-fold difference observed in translation cannot be explained by the 30% difference in transcription between the NT and T, suggesting an additional mechanism that promotes translation post-transcriptionally. Due to the PQS orientations, we expect that the mRNA from NT, but not T or control forms an RG4 structure at 5’UTR. Thus, we tested for the RG4 formation by applying N-methyl mesoporphyrin IX (NMM) in the transcription reaction <sup>41</sup>. NMM is a G4 ligand that exhibits induced fluorescence upon binding G4 <sup>42</sup>. As expected, the NMM fluorescence displayed a prominent increase over the transcription time in NT, but not in T and C (Fig. 2f), indicating a progressive and robust formation of RG4 exclusively in NT condition. The selective NMM fluorescence for NT suggests that RG4 is likely responsible for the enhanced translation since the three constructs differ only by the PQS region. In addition, the half-life obtained from the fluorescence increase reflected the order of events i.e., transcription signal increased first, followed by the NMM intensity reflecting the RG4 formation, and the GFP intensity (Fig. 2g). This further supports that the RG4 is responsible for the enhanced translation.
+
+### Bulkiness of RG4 drives translational enhancement.
+
+PQS can vary in its sequence composition, which gives rise to diverse conformations of varying stability, and bulkiness <sup>41, 43</sup>. Based on the result obtained above, we asked if different PQS sequences produce various levels of translational enhancement. To focus on the effect of G4, we removed the sequence upstream of PQS and inserted a series of PQS with varying loop length without changing the guanine triplets (Fig. 3a). Since RNA G4 primarily folds in parallel conformation in which all the guanine strands run in the same orientation <sup>44</sup>, we envision that the loop sequences will protrude out from the central tetrad core. The NMM based assay revealed that transcription of all PQS-NT sequences produced RG4 (Supplementary Fig. 1a-c). Despite varying levels of NMM signal acquired for different RG4s, single molecule FRET assay displayed that both small (short looped) and large (long looped) RG4 form a stable G4 structure without structural dynamics (Supplementary Fig. 1d, e). All PQS-NT which result in RG4 containing RNA consistently led to higher translation product than its counterpart PQS-T construct (Fig. 3b). Surprisingly, PQS-NT sequences led to 4-fold higher translation than the control (Fig. 3c), strongly reflecting the role of RG4 in promoting translation. Next, to examine the relationship between the RG4 sequence and the translation level, we plotted the translation efficiency against the total loop length of each RG4. Strikingly, the loop length of RG4 is highly correlated to the translation efficiency with a correlation coefficient of r = 0.89, indicating that the longer loop length which likely represents higher bulkiness of individual RG4 structure drives translation enhancement (Fig. 3d). Overall, our data demonstrate that all RG4 structures elevate translation and the bulkiness of RG4 accentuates the translational enhancement.
+
+### Hairpin and RG4 structures synergistically promote translation.
+
+By comparing the results presented in Fig. 3C and Fig. 2, we noticed that despite the same PQS sequence, the translation enhancement was higher in Fig. 2. Upon close examination, we recognized that there is a 10 bp hairpin forming sequence located upstream of PQS in the construct used in Fig. 2, but not in Fig. 3C. This observation led us to test if an additional 5’UTR structure can further enhance translation. To investigate the effect of two tandem structures, we divided the 5’UTR into four segments (Fig. 4a and Supplementary table 1); upstream of RG4 (1), RG4 (2), downstream of RG4 (3), and RBS to start codon (4). We applied an RNA structure prediction tool (UNAfolds <sup>45</sup>) to calculate the folding energies of all positions except position 2, because RG4 folding cannot be accurately predicted by currently available tools. The folding energy, ΔG estimated for the positions 3, 4, and 3 + 4 were −5.6, 0.9, and −11.8 kcal/mol, respectively, indicating weakly folded state of the downstream sequence (Supplementary table 1). To avoid interference with the ribosome binding, we decided to vary sequence located upstream of PQS. We note that the position 1 sequence used in Fig. 3 has a low folding energy (ΔG = -3.1 kcal/mol) (see Supplementary table 1), which most likely stays unfolded at 37 ℃; we named the construct 0 hp henceforth. To study the impact of hairpin in modulating translation, we introduced hairpins of 4 bp, 8 bp, 10 bp and 12 bp stem length which are expected to have folding energy (ΔG) of -5.9, -10.4, -15.9, and −21.4 kcal/mol, respectively (Fig. 4a). We inserted each hairpin at position 1, upstream of the RG4 sequences. Following the trend seen in 0 hp cases (Fig. 3), the translational efficiency for each hairpin group exhibited an RG4 size dependence (Fig. 4b and c), indicating that the RG4 structures remained in different hairpin constructs. In addition, the translation was further enhanced as a function of the hairpin length (Fig. 4b). For example, the translation of cMyc increased from 1.4 to 5 folds for 0 hp and 12 hp, respectively and that of 199 increased from 3.3 to 8.5 folds for 0 hp and 12 hp, respectively. However, none of the hairpins enhanced translation by itself in the absence of RG4 (the control of each hairpin), suggesting that single hairpin structure cannot drive the translation enhancement independently. Next, we tested tandem hairpins by placing 6, 10, 14 and 17 hp (folding energies ΔG: -6.5, -21.0, -31.3, -42.1 kcal/mol, respectively) in addition to the 10 hp (Supplementary Fig. 2). We found that the two hairpins enhance translation only to a level of a single RG4 (cMyc) regardless of the folding energy (Supplementary table 1). This suggests that the enhancement was induced by the structure rather than the folding energy and that the RG4 is more potent than the hairpins in promoting translation. Indeed, the RG4 loop length-dependent translational efficiency is exhibited for all hairpin variants (Fig. 4c). We compiled and projected all the results to a 2-D heatmap which presents a distinct trend that translation strength is highly correlated with both the hairpin stem size (vertical axis) and G4 loop lengths (horizontal axis) (Fig. 4d). Taken together, hairpin contributes to enhanced translation only when present with RG4. This finding raises an intriguing question about the underlying mechanism by which 5’UTR structure upstream of RBS enhances translation.
+
+What is the mechanism that enables 5’UTR structures to elevate translation in a size dependent manner? We reasoned that the 5’UTR structure may have an impact on either the ribosome or the mRNA. To define the mechanism, we set out to test four hypotheses: i) RG4 increases the ribosome binding affinity (Fig. 5a); ii) RG4 improves the accessibility of RBS to ribosome (Fig. 5e); iii) RG4 increases the mRNA lifetime (Fig. 6a); iv) RG4 stabilizes the ribosome-bound state (Fig. 6c).
+
+## RG4 does not attract ribosome
+
+We tested the hypothesis that RG4 enhances translation by increasing the affinity to ribosomes, perhaps by acting like an IRES (internal ribosome entry site) in viral translation <sup>46</sup>. This hypothesis posits that the mRNA containing RG4 structure (based on PQS-NT) will recruit more ribosomes than the mRNA without RG4 (C, PQS-T) (Fig. 5a), resulting in higher translation. To test this hypothesis, we performed a competition assay in which RG4 bearing competitor RNA was applied to the translation reaction in molar excess. The competitor RNA constructs included a negative control, polyU 40 nt (Fig. 5b-1), an RG4 alone (Fig. 5b-2), a single strand (ss) RNA with a hairpin but without RG4 (Fig. 5b-3), an RG4 flanked by the neighboring sequence found in mRNA (Fig. 5b-4), a ssRNA with RBS (Fig. 5b-5) and a ssRNA with both RG4 and RBS (Fig. 5b-6). If our hypothesis is correct, we expect to see reduced translation in conditions 2, 4, and 6, all of which contain RG4. We confirmed that there was no significant difference in transcription rate among the six conditions, indicating that the competitor RNAs did not affect the overall transcription (Fig. 5c). The translation result revealed that only the RBS containing RNA (5 and 6) lowered the translation significantly while the other conditions, (1–4) did not (Fig. 5d), suggesting that the RBS containing competitor, but not the RG4 bearing strands competed for the ribosome binding, thus lowering the translation. Therefore, we show that the RG4 does not increase affinity toward ribosome.
+
+## RG4 does not increase RBS accessibility
+
+Next, we hypothesized that the formation of RG4 increases RBS accessibility to ribosome by preventing RBS from folding into an inaccessible secondary structure (Fig. 5e). To examine the accessibility of the RBS, we applied a molar excess of a molecular beacon that bears sequence complementary to RBS (Fig. 5f). The beacon fluoresces upon hybridizing to the RBS; hence the intensity of the beacon represents the accessibility of the RBS. We tested the accessibility in two ways. First, we performed titration of purified transcript of NT, C and T to a fixed concentration of molecular beacon (400 nM). The dissociation constant, K<sub>d</sub> was 367 (±25), 364 (±23), 370 (±21) nM for NT, C and T respectively, indicating negligible difference in RBS accessibility among the three constructs (Fig. 5g). Second, we applied the molecular beacon before and after heating up the transcript from NT, C and T to test if the heat induced unfolding will increase the RBS accessibility. Despite the overall increase, the similar intensities of molecular beacon among NT, C and T indicated that RBS is accessible regardless of RG4 (Fig. 5h). Hence, both assays corroborate to reflect that RBS is fully accessible in all three constructs. Therefore, the RG4 unlikely acts via making the RBS accessible to ribosome loading.
+
+## RG4 does not increase mRNA lifetime
+
+Next, we hypothesized that RG4 increases the lifetime of the mRNA by stabilizing the mRNA (Fig. 6a) since secondary structures on RNA can increase RNA lifetime by preventing RNA degradation <sup>13</sup>. We performed RT-PCR to compare the mRNA from NT, C and T after three hours of translation. The highly similar mRNA levels tested by two different sets of primers in NT, C and T provide robust evidence that RG4 does not play a role in stabilizing mRNA (Fig. 6b).
+
+## RG4 may stabilize ribosome bound to mRNA
+
+The negative results obtained in the first three hypotheses strongly suggest that the effect of RG4 in translational regulation must occur after the ribosome loads on the mRNA. This raises a possibility that RG4 may promote translation by stabilizing the ribosome bound to mRNA, perhaps by preventing the ribosome from dislodging (Fig. 6c). To test the hypothesis, we applied an RNA helicase, DHX36 (or RHAU) to unfold the RG4 structure during the translation reaction (Fig. 6d). DHX36 is a well-studied RG4-specific helicase which should effectively remove the RG4 structure formed in mRNA <sup>47, 48, 49, 50, 51</sup>. Previously, we reported an ATP dependent repetitive unwinding mechanism by which DHX36 unfolds RG4 using single-molecule FRET <sup>52</sup>. In agreement with our previous finding, DHX36 displayed a strong affinity to RG4 and unfolded the structure even at sub nanomolar concentration (Supplementary Fig. 3). Strikingly, when applied to the translation reaction, DHX36 reduced translation in a dose-dependent manner (Fig. 6e) without impacting the transcription (Fig. 6f and 6g), strongly suggesting that RG4 structure is responsible for the increased translation. We propose that the RG4 acts as a physical blockade which stabilizes the ribosome bound state by preventing ribosome from dislodging from the mRNA.
+
+### Translation enhancement pattern is observed in *E. coli*.
+
+Next, we asked if the hairpin and RG4 mediated translation enhancement also operates in *E. coli*. Unlike the cell-free translation system which only contains essential reagents for transcription and translation, cellular environment is enriched with other proteins, including helicases, RNA binding proteins and RNases that can modulate the gene expression process and thus change the translation enhancement effect by the 5’UTR structures <sup>53, 54</sup>. To test this, we prepared a dual-color fluorescence plasmid reporter. The T7 promoter-GFP was built with the 5’UTR structures for an experimental readout whereas the T7 promoter-mCherry was constructed without 5’UTR elements to serve as an internal control (Fig. 7a). The GFP expression was normalized against the mCherry expression to obtain the relative translation yield for various 5’UTR sequences (Supplementary Fig. 4a-c). The GFP expression was confirmed and visualized by fluorescence imaging (Fig. 7b). Later, we quantified the GFP and mCherry expression by acquiring real-time fluorescence and absorbance (A<sub>600</sub>) which were recorded simultaneously by the plate reader. The GFP intensity displayed the same orientation dependence of NT > C > T as we observed *in vitro* (Fig. 7c), while the mCherry intensity remained similar (Fig. 7d), suggesting that the same RG4 dependent translation enhancement occurs in *E. coli* cells. By using RT-PCR analysis, we confirmed that the difference is not based on the mRNA expression (supplementary Fig. 4d). We cloned six sets of plasmids with varying PQS sequences inserted either in NT or T and quantified the translation efficiency. The result reflects the same pattern as before; PQS-NT produces higher GFP signal than PQS-T and the translation further increases when bulkier PQS is inserted in NT (Fig. 7e). Taken together, our data indicates that the 5’UTR structure dependent translational enhancement exists both *in vitro* and in cells.
+
+# Discussion
+
+Here we applied a real-time transcription-translation coupled assay (Fig. 1) to demonstrate the impact of RG4 at 5’UTR in promoting translation (Fig. 2). This enhancement is highly correlated with the total loop length i.e the size of the RG4 (Fig. 3) and such effect is further accentuated when a hairpin structure is added in tandem (Fig. 4). We demonstrate that the RG4 mediated translation enhancement is not due to elevated affinity to ribosome, increased accessibility of RBS (Fig. 5), or improved stability of the mRNA (Fig. 6 top). The RG4 structure is the key to promoting the translation as the helicase induced unwinding completely abolished the effect (Fig. 6 bottom). We propose that RG4 serves as a physical blockade that prevents the ribosome from falling off the mRNA and thereby directing it toward the protein synthesis. In addition, we demonstrate that the same mechanism operates in *E. coli* cells (Fig. 7), suggesting its potential application for controllable gene expression in *E. coli*.
+
+We find that RG4 with longer loops induces higher translational enhancement (Fig. 3). While longer loops can enlarge the overall size of the RG4, they can weaken the folded state of the G4, based on the studies done for DNA G4 i.e. longer loop lengths lead to less stable folding of DNA-G4 due to lower folding energy <sup>41, 43</sup>. We tested whether the folded state of RG4 varies between the short loop (111) and the longer loop (199) RG4 via smFRET assay. Surprisingly, both RNAs showed a steady high FRET state, indicating a stably folded G4 structure (Supplementary Fig. 1d, e). This evidence suggests that the G4 folding in RNA is inherently more stable than the G4 in DNA. To further weaken the RG4, we mutated the middle guanine to adenine to destabilize the core of RG4. Strikingly, the mutation constructs still showed a similar enhancement effect as the original RG4 (Supplementary Fig. 5), reflecting that the mutated RNA can still fold into a structure that can increase translational output. Taken the data together with the dual-hairpin result (Supplementary Fig. 2), translational enhancement up to 4–6 folds increase occurs regardless of the structure, but the strongest effect requires both stable folding and large size of the structures, for example, 177, 555 and 199 with a hairpin with a long stem.
+
+The effect of RG4 depends on its location. We found that translation was abolished when PQS was positioned at 10 bp upstream of the SD sequence (data not shown), likely by blocking ribosome subunit association to the RBS. This agrees with Holder and Hartig’s previous work which demonstrated that inserting G-quadruplex 20 bp upstream of the start codon decreased translation efficiency without changing transcription. Again, G4 likely inhibits the interaction of the 16S ribosomal RNA with the SD region <sup>23, 24</sup>. Therefore, we inserted G4 at 46 bp upstream to SD to prevent such structural inhibition. Whether the level of enhancement relies on the distance between RG4 and RBS, or RG4 and other structures warrants future study.
+
+Our observation may be partially explained by the "standby model" <sup>55, 56</sup>, in which an upstream hairpin provides a temporary position for the ribosomal 30S subunit to stay, waiting for the unfolding of the SD sequence. According to this model, the rate-determining step becomes the recruitment of ribosome subunit, which is slower than the waiting time for unfolding. In our case, however the extended sequence adjacent to SD (position 3 in Fig. 4 a) is predicted to have a folding energy of -5.6 kcal/mol, which falls within the energy proposed by the standby model (less than − 10 kcal/mol) <sup>55</sup>. Nevertheless, the current version of the standby model only accounts for a hairpin structure near the SD sequence, which cannot be extended to the effect of the RG4 structure positioned far from the RBS. Therefore, we propose that the RG4 structure with or without the hairpin may play a role of a blockade in preventing ribosome from falling off the mRNA.
+
+We also considered that the translation enhancement may result from the coupled transcription and translation in *E. coli* where RNA synthesis by RNAP facilitates the recruitment of ribosomal subunits and initiate translation before the transcription is terminated i.e. co-transcriptional translation (CTT) <sup>57, 58</sup>. However, in our system, CTT is unlikely because T7 RNAP transcription rate (220 ~ 230 nt/s) <sup>59, 60</sup> is significantly higher than *E. coli* ribosome translation rate (42–51 nt/s) <sup>61</sup>. That is, ribosomes will be loaded onto a nascent RNA post RNA synthesis rather than being coupled to transcription. In order to examine the effect in the presence of CTT, we also cloned the same 5’UTR sequence and GFP gene to an *E. coli* promoter P<sub>L−LacO</sub> system. Surprisingly, we still observed a huge increase of GFP signal in non-template construct than in template (Supplementary Fig. 6), suggesting the structural effect may still function in regular *E. coli* gene under CTT. However, the mechanism in a pure *E. coli* system should be studied more systematically in the future.
+
+To summarize, we have demonstrated a size-dependent 5’UTR structure effect in translation enhancement in T7 and *E. coli* system. We focused primarily on RNA G-quadruplex and hairpin structures. Although we examined varying sizes of both the RG4 and hairpin, our study opens a wide window of opportunity for future studies, for example, investigating the role of a pseudoknot structure positioned either upstream or downstream of RG4 or in between RG4 and RBS. Furthermore, we want to note that in all our experiments, PQS insertion into template (PQS-T) which produces C-rich transcript, strongly suppressed the translation efficiency both *in vitro* and *in vivo* experiments. This suggests an opposite function of C-rich RNA in down-regulating the gene expression. In conclusion, our study provides a new mechanism and function of 5’UTR mRNA in bacterial translation and provides a novel cloning scheme for tunable gene expression system.
+
+# Methods
+
+DNA Preparation  
+All the DNA samples were started from the plasmid construction with a single *GFP* reporter. The details were described in an earlier publication <sup>25</sup>. DNA sequences were listed in Supplementary Table 2. The recombinant plasmids were transformed into NEB-*5α* for DNA extraction and BL21-DE3 for *E. coli* expression assay. For dual-color reporter system, a *mCherry* gene was cloned from pET mCherry vector (Addgene, plasmid #29722) into the GFP plasmid by NEB HiFi DNA Assembly kit. Linear DNA samples for *in vitro* translation were PCR-amplified from GFP plasmid. The T7 promoter forward primer and T7 terminator reverse primers (see Supplementary Table 2) were designed and ordered from Integrated DNA Technologies (IDT) for amplification. The amplified linear DNA held the T7 promoter, 5’UTR with PQS, GFP gene, and T7 terminator and was purified by gel electrophoresis and Gel Extraction Kit (QIAquick).
+
+RNA Preparation  
+RNA samples were prepared by HiScribe T7 Quick High Yield RNA Synthesis kit (NEB) at 37℃ overnight. Each reaction (20 µL) had 1 µg of DNA, NTP mix, T7 RNA Polymerase, and RNase free water. The overnight product was firstly digested by DNase I (0.1 U/µL) in DNase reaction buffer (10 mM Tris-HCl pH 7.6, 2.5 mM MgCl<sub>2</sub>, 0.5 mM CaCl<sub>2</sub>) at 37℃ for 30 minutes. Later, the reaction was quenched by adding 1 µL 0.5 M EDTA, followed by inactivating at 75℃ for 10 minutes. RNA was purified by Monarch RNA CleanUp kit (NEB).
+
+*In Vitro* Translation Assay  
+Ensemble *in vitro* translation assay was conducted by PURExpress *In Vitro* Translation kit (NEB) and performed by TECAN Spart plate reader at 37℃. Each reaction (25 µL) was premixed with 55 ng linear DNA (4 nM), 400 nM molecular beacon (see Supplementary Table 2), RNase inhibitor murine (0.8 unit/µL), and 10 µL Solution A. To measure the RG4 formation, NMM was added in premixed solution at final concentration of 1 µM. The reaction was initiated by adding 7.5 µL Solution B and loaded on a 384-well plate (white and transparent bottom, Thermo Scientific). The Cy3, GFP, and NMM were excited at λ<sub>ex</sub> 545, 485, and 393 nm and detected at λ<sub>em</sub> 570, 510, and 610 nm, respectively. Both excitation and emission were assigned with 10 nm slit size. The initiation transcription rate was quantified from the linear part (10 to 25 minutes) of the Cy3 intensity curve. The translation rate was quantified from the linear part (25 to 50 minutes) of the GFP intensity curve. Each rate was normalized to the transcription rate and translation rate of 10 hp control DNA construct, respectively. The normalized translation efficiency was calculated by dividing the normalized translation rate to the normalized transcription rate. The half-life was defined by the time that the intensity reached 50 percent of the plateau.
+
+*In Vitro* Transcription NMM Assay  
+Ensemble in vitro transcription for real-time NMM measurement was performed by TECAN Spark plate reader at 37℃. Each sample was prepared with 1 nM linear DNA template in transcription buffer (40 mM Tris-HCl pH 8.3, 50 mM KCl, 6 mM MgCl<sub>2</sub>, 2 mM spermidine, 1 mM dithiothreitol), RNase inhibitor murine (0.4 unit/µL), T7 RNA polymerase (1.25 unit/µL), and 1 mM NMM. The reaction was initiated by adding NTP mix for a final concentration of 1 mM. Each reaction (100 µL) was loaded on 96-well transparent plate (Thermo Scientific). The data was collected at λ<sub>ex</sub> 393 nm (slit size 10 nm) and λ<sub>em</sub> 610 nm (slit size 10 nm). For emission spectrum, the data was collected at λ<sub>ex</sub> 393 nm (slit size 10 nm) and λ<sub>em</sub> 580–650 nm (slit size 10 nm).
+
+RNA G4 Competition Assay  
+RNA competition assay was modified from *in vitro* translation assay by adding RNA competitors. PolyU 40 and cMyc RG4 were ordered from IDT. Other RNAs were synthesized from the 5’UTR of 10 hp control and 10hp cMyc DNA. The DNA templates were PCR-amplified by T7 promoter primer, short-length primer 1, and long-length primer 2 (see Supplementary Table 2). RNA purification protocol was described in *RNA Preparation*. The reaction was performed with 55 ng of 10 hp cMyc-NT DNA and 5 µM of competitor RNA by plate reader at 37℃.
+
+RBS Probe Binding Assay and Accessibility Assay  
+A molecular beacon complementary to the ribosome binding site (RBS) was ordered from IDT and labeled by Cy5 and quencher at each end (see Supplementary Table 2). The Cy5 intensity of beacon was measured at λ<sub>ex</sub> 640nm and λ<sub>em</sub> 665nm with 10 nm slit size. The binding assay was conducted by incubating 400 nM beacon and RNAs of 10hp control, T, and NT at 37℃. RNA purification protocol was described in *RNA Preparation*. The data points were collected by titrating RNA concentration in a 2-fold series dilution from 1.75 µM to 24 nM. The K<sub>d</sub> was fitted to the binding curve by OriginPro. For accessibility assay, each sample (10 µg of RNA) was mixed with 400 nM of the RBS beacon and incubated with or without heating treatment. The heated samples were incubated at 80℃ for 5 minutes and cooled by 10℃ intervals for 5 minutes until 37℃. After incubation, intensities of all the samples were measured by TECAN plate reader at 37℃.
+
+RT-qPCR  
+For *in vitro* translation, post-translation samples (20 µL) were treated with DNase I (0.1 U/µL) in DNase reaction buffer (10 mM Tris-HCl pH 7.6, 2.5 mM MgCl<sub>2</sub>, 0.5 mM CaCl<sub>2</sub>) at 37℃ for 30 minutes. The reaction was quenched by 1 µL of 0.5 M EDTA, followed by heat inactivation at 75℃ for 10 minutes. The DNA-free samples were diluted 10-fold, and 2 µL of each diluted sample was used to synthesize cDNA by ProtoScript II cDNA Synthesis Kit (NEB). The reaction was incubated at 25℃ for 5 minutes, followed by 42℃ for 1 hour and 80℃ for 5 minutes. The cDNA was diluted 100-fold before qPCR measurement. The qPCR sample (20 µL) contained 1 µL 100-fold diluted cDNA, 10 µL SYBR Green Supermix (NEB), and 250 nM primers (see Supplementary Table 2). The data points of cycles were plotted in Fig. 6B.
+
+DHX36 purification and titration experiment  
+The *E. coli* strand with DHX36 plasmid was made in the lab and described in a previous publication <sup>52</sup>. The *E. coli* was inoculated in TB medium and grew overnight. Next day, the culture was diluted to OD<sub>600</sub> of 0.01 and grew till OD<sub>600</sub> of 0.6 at 37℃ and induced protein expression with 1 mM IPTG at 14℃ for overnight. The purification protocol followed previous publications <sup>52</sup>. Protein concentration was quantified by standard BSA (NEB) calibration curve by SDS-PAGE, and the aliquots of protein samples were stored in -80℃. For DHX36 titration assay, the protein was diluted by TNM buffer (10 mM Tris-HCl, pH 8.0; 50 mM NaCl; 5 mM MgCl<sub>2</sub>) to avoid the change of reaction volume and buffer condition. 1 µL of each titrated DHX36 and additional 1 mM ATP were premixed with solution A, and the reaction was initiated by adding solution B. The experimental protocol is described in *the In Vitro Translation Assay*.
+
+*E. Coli* dual fluorescence assay  
+PQS-contained dual fluorescence plasmids, described in *DNA Preparation*, were transformed into BL21(DE3) E. coli, and grew in LB medium at 37°C overnight. The cultures were diluted to OD<sub>600</sub> of 0.01 and grew at 37°C by TECAN plate reader with a 24-well transparent plate (1 mL for each culture). OD<sub>600</sub> measurements were performed every 10 min, followed by an orbital shaking mode (215 rpm) for a 10 min interval. The gene expression was induced by 1 mM IPTG at OD<sub>600</sub> of 0.4. After the induction, the protein expression was monitored by an auto-loop measurement of 10 min shaking, OD<sub>600</sub>, GFP (λ<sub>ex</sub> 485/ λ<sub>em</sub> 510, slit 10 nm), and mCherry (λ<sub>ex</sub> 585/ λ<sub>em</sub> 610, slit 10 nm) for 10 hours. The translation efficiency was defined by the ratio of GFP to mCherry and normalized to real-time OD<sub>600</sub> (Supplementary Fig. 4). The ratio of 210 min after induction represented the maximal efficiency of each strain. The cultures after 210-min induction were collected and extracted mRNA for RT-qPCR. The pellet was treated with 20 mg/mL lysozyme, and the mRNA was extracted by using Qiagen RNeasy Kits. The primers of qPCR were listed in Supplementary Table 2.
+
+smFRET Assay  
+The PQS RNA oligonucleotides (see Supplementary Table 2) for smFRET were purchased from IDT with terminal amine modification for Cy3 labeling. The 18-mer RNA primer for immobilizing PQS RNA on slides was purchased from IDT and later labeled Cy5. The labeling protocol was described in a previous publication. RNA samples were annealed in TE buffer (10 mM Tris-HCl and 1 mM EDTA, pH 8) at the ratio 1:1. The mixtures were heated at 80°C for 5 min and slowly cooled to room temperature (1°C per min). The single molecule assays were performed by using a home-built prism-type total internal reflection fluorescence microscope (TIRFM) at room temperature (23.0 ± 1.0°C) <sup>25, 62</sup>. RNA sample (10 nM) was diluted to 25 pM and immobilized on a PEG-coated quartz slide by neutravidin (0.05 mg/mL). The reaction buffer (50 mM Tris-HCl, pH 7.5; 50 mM NaCl; 50 mM KCl; 5 mM MgCl<sub>2</sub>; 5% glycerol; 80 units RNase inhibitor murine) with an oxygen scavenging system (1 mg/mL glucose oxidase, 0.8% v/v glucose, ~ 10 mM Trolox, and 0.03 mg/mL catalase). The smFRET experiment was performed by two solid-state lasers, 532 nm and 640 nm lasers. Each measurement was recorded with a 100 ms time resolution by smCamera software and analyzed with Interactive Data Language (IDL). The outputs were processed with custom MATLAB script to generate trajectories and FRET histograms. The details of smFRET data process were described in previous publications <sup>62, 63</sup>. For DHX36 assay, protein (1 nM) or ATP (1 mM) was premixed into reaction buffer, and a flow system was applied to study real-time binding and unwinding events <sup>63</sup>.
+
+Statistics Analysis  
+Data shown in Fig. 2–7 were obtained from individual and independent experiments. All the numbers were calculated and presented in value ± SEM. The statistics tests were calculated by two-sided paired or unpaired t-test, depending on the data. The average numbers, SEM, and statistics *P*-values were reported in Supplementary Table 3.
+
+# References
+
+1. Buccitelli C, Selbach M. mRNAs, proteins and the emerging principles of gene expression control. *Nat Rev Genet* **21**, 630-644 (2020).
+2. Hershey JWB, Sonenberg N, Mathews MB. Principles of Translational Control. *Cold Spring Harb Perspect Biol* **11**, (2019).
+3. Gebauer F, Hentze MW. Molecular mechanisms of translational control. *Nat Rev Mol Cell Biol* **5**, 827-835 (2004).
+4. Soutourina J. Transcription regulation by the Mediator complex. *Nat Rev Mol Cell Biol* **19**, 262-274 (2018).
+5. Bentley DL. Coupling mRNA processing with transcription in time and space. *Nat Rev Genet* **15**, 163-175 (2014).
+6. Tahmasebi S, Khoutorsky A, Mathews MB, Sonenberg N. Translation deregulation in human disease. *Nat Rev Mol Cell Biol* **19**, 791-807 (2018).
+7. Song P, Yang F, Jin H, Wang X. The regulation of protein translation and its implications for cancer. *Signal Transduct Target Ther* **6**, 68 (2021).
+8. Mejia-Almonte C, et al. Redefining fundamental concepts of transcription initiation in bacteria. *Nat Rev Genet* **21**, 699-714 (2020).
+9. Rodnina MV. Translation in Prokaryotes. *Cold Spring Harb Perspect Biol* **10**, (2018).
+10. Van Assche E, Van Puyvelde S, Vanderleyden J, Steenackers HP. RNA-binding proteins involved in post-transcriptional regulation in bacteria. *Front Microbiol* **6**, 141 (2015).
+11. Martinez LC, Vadyvaloo V. Mechanisms of post-transcriptional gene regulation in bacterial biofilms. *Front Cell Infect Microbiol* **4**, 38 (2014).
+12. Leppek K, Das R, Barna M. Functional 5' UTR mRNA structures in eukaryotic translation regulation and how to find them. *Nat Rev Mol Cell Biol* **19**, 158-174 (2018).
+13. Chen F, Cocaign-Bousquet M, Girbal L, Nouaille S. 5'UTR sequences influence protein levels in Escherichia coli by regulating translation initiation and mRNA stability. *Front Microbiol* **13**, 1088941 (2022).
+14. Wen JD, Kuo ST, Chou HD. The diversity of Shine-Dalgarno sequences sheds light on the evolution of translation initiation. *RNA Biol* **18**, 1489-1500 (2021).
+15. Wassarman KM. Small RNAs in bacteria: diverse regulators of gene expression in response to environmental changes. *Cell* **109**, 141-144 (2002).
+16. Vogel J, Luisi BF. Hfq and its constellation of RNA. *Nat Rev Microbiol* **9**, 578-589 (2011).
+17. Chen H, Bjerknes M, Kumar R, Jay E. Determination of the optimal aligned spacing between the Shine-Dalgarno sequence and the translation initiation codon of Escherichia coli mRNAs. *Nucleic Acids Res* **22**, 4953-4957 (1994).
+18. Endoh T, Sugimoto N. Conformational Dynamics of mRNA in Gene Expression as New Pharmaceutical Target. *Chem Rec* **17**, 817-832 (2017).
+19. Xiao J, et al. Facilitating Protein Expression with Portable 5'-UTR Secondary Structures in Bacillus licheniformis. *ACS Synth Biol* **9**, 1051-1058 (2020).
+20. Sadler FW, Dodevski I, Sarkar CA. RNA Thermometers for the PURExpress System. *ACS Synth Biol* **7**, 292-296 (2018).
+21. Ariza-Mateos A, Nuthanakanti A, Serganov A. Riboswitch Mechanisms: New Tricks for an Old Dog. *Biochemistry (Mosc)* **86**, 962-975 (2021).
+22. Abduljalil JM. Bacterial riboswitches and RNA thermometers: Nature and contributions to pathogenesis. *Noncoding RNA Res* **3**, 54-63 (2018).
+23. Holder IT, Hartig JS. A matter of location: influence of G-quadruplexes on Escherichia coli gene expression. *Chem Biol* **21**, 1511-1521 (2014).
+24. Wieland M, Hartig JS. RNA quadruplex-based modulation of gene expression. *Chem Biol* **14**, 757-763 (2007).
+25. Lee CY, McNerney C, Ma K, Zhao W, Wang A, Myong S. R-loop induced G-quadruplex in non-template promotes transcription by successive R-loop formation. *Nat Commun* **11**, 3392 (2020).
+26. Varshney D, Spiegel J, Zyner K, Tannahill D, Balasubramanian S. The regulation and functions of DNA and RNA G-quadruplexes. *Nat Rev Mol Cell Biol* **21**, 459-474 (2020).
+27. Kumari S, Bugaut A, Huppert JL, Balasubramanian S. An RNA G-quadruplex in the 5' UTR of the NRAS proto-oncogene modulates translation. *Nat Chem Biol* **3**, 218-221 (2007).
+28. Arora A, Dutkiewicz M, Scaria V, Hariharan M, Maiti S, Kurreck J. Inhibition of translation in living eukaryotic cells by an RNA G-quadruplex motif. *RNA* **14**, 1290-1296 (2008).
+29. Agarwala P, Pandey S, Maiti S. Role of G-quadruplex located at 5' end of mRNAs. *Biochim Biophys Acta* **1840**, 3503-3510 (2014).
+30. Agarwala P, Pandey S, Ekka MK, Chakraborty D, Maiti S. Combinatorial role of two G-quadruplexes in 5' UTR of transforming growth factor beta2 (TGFbeta2). *Biochim Biophys Acta Gen Subj* **1863**, 129416 (2019).
+31. Endoh T, Sugimoto N. Conformational Dynamics of the RNA G-Quadruplex and its Effect on Translation Efficiency. *Molecules* **24**, (2019).
+32. Cammas A, et al. Stabilization of the G-quadruplex at the VEGF IRES represses cap-independent translation. *RNA Biol* **12**, 320-329 (2015).
+33. Bhattacharyya D, Diamond P, Basu S. An Independently folding RNA G-quadruplex domain directly recruits the 40S ribosomal subunit. *Biochemistry* **54**, 1879-1885 (2015).
+34. Al-Zeer MA, Dutkiewicz M, von Hacht A, Kreuzmann D, Rohrs V, Kurreck J. Alternatively spliced variants of the 5'-UTR of the ARPC2 mRNA regulate translation by an internal ribosome entry site (IRES) harboring a guanine-quadruplex motif. *RNA Biol* **16**, 1622-1632 (2019).
+35. Rawal P, et al. Genome-wide prediction of G4 DNA as regulatory motifs: role in Escherichia coli global regulation. *Genome Res* **16**, 644-655 (2006).
+36. Kaplan OI, Berber B, Hekim N, Doluca O. G-quadruplex prediction in E. coli genome reveals a conserved putative G-quadruplex-Hairpin-Duplex switch. *Nucleic Acids Res* **44**, 9083-9095 (2016).
+37. Bartas M, et al. The Presence and Localization of G-Quadruplex Forming Sequences in the Domain of Bacteria. *Molecules* **24**, (2019).
+38. Jackson K, Kanamori T, Ueda T, Fan ZH. Protein synthesis yield increased 72 times in the cell-free PURE system. *Integr Biol (Camb)* **6**, 781-788 (2014).
+39. Zhang Z, Revyakin A, Grimm JB, Lavis LD, Tjian R. Single-molecule tracking of the transcription cycle by sub-second RNA detection. *Elife* **3**, e01775 (2014).
+40. Balleza E, Kim JM, Cluzel P. Systematic characterization of maturation time of fluorescent proteins in living cells. *Nat Methods* **15**, 47-51 (2018).
+41. Kim M, et al. Quantitative analysis and prediction of G-quadruplex forming sequences in double-stranded DNA. *Nucleic Acids Res* **44**, 4807-4817 (2016).
+42. Kreig A, Calvert J, Sanoica J, Cullum E, Tipanna R, Myong S. G-quadruplex formation in double strand DNA probed by NMM and CV fluorescence. *Nucleic Acids Res* **43**, 7961-7970 (2015).
+43. Tippana R, Xiao W, Myong S. G-quadruplex conformation and dynamics are determined by loop length and sequence. *Nucleic Acids Res* **42**, 8106-8114 (2014).
+44. Zhang DH, Fujimoto T, Saxena S, Yu HQ, Miyoshi D, Sugimoto N. Monomorphic RNA G-quadruplex and polymorphic DNA G-quadruplex structures responding to cellular environmental factors. *Biochemistry* **49**, 4554-4563 (2010).
+45. Zuker M. Mfold web server for nucleic acid folding and hybridization prediction. *Nucleic Acids Res* **31**, 3406-3415 (2003).
+46. Kieft JS. Viral IRES RNA structures and ribosome interactions. *Trends Biochem Sci* **33**, 274-283 (2008).
+47. Linder P, Jankowsky E. From unwinding to clamping - the DEAD box RNA helicase family. *Nat Rev Mol Cell Biol* **12**, 505-516 (2011).
+48. Sauer M, Paeschke K. G-quadruplex unwinding helicases and their function in vivo. *Biochem Soc Trans* **45**, 1173-1182 (2017).
+49. Fay MM, Lyons SM, Ivanov P. RNA G-Quadruplexes in Biology: Principles and Molecular Mechanisms. *J Mol Biol* **429**, 2127-2147 (2017).
+50. Chen MC, et al. Structural basis of G-quadruplex unfolding by the DEAH/RHA helicase DHX36. *Nature* **558**, 465-469 (2018).
+51. Sauer M, et al. DHX36 prevents the accumulation of translationally inactive mRNAs with G4-structures in untranslated regions. *Nat Commun* **10**, 2421 (2019).
+52. Tippana R, Chen MC, Demeshkina NA, Ferre-D'Amare AR, Myong S. RNA G-quadruplex is resolved by repetitive and ATP-dependent mechanism of DHX36. *Nat Commun* **10**, 1855 (2019).
+53. Khemici V, Linder P. RNA helicases in bacteria. *Curr Opin Microbiol* **30**, 58-66 (2016).
+54. Cetnar DP, Salis HM. Systematic Quantification of Sequence and Structural Determinants Controlling mRNA stability in Bacterial Operons. *ACS Synth Biol* **10**, 318-332 (2021).
+55. Espah Borujeni A, Channarasappa AS, Salis HM. Translation rate is controlled by coupled trade-offs between site accessibility, selective RNA unfolding and sliding at upstream standby sites. *Nucleic Acids Res* **42**, 2646-2659 (2014).
+56. Andreeva I, Belardinelli R, Rodnina MV. Translation initiation in bacterial polysomes through ribosome loading on a standby site on a highly translated mRNA. *Proc Natl Acad Sci U S A* **115**, 4411-4416 (2018).
+57. Irastortza-Olaziregi M, Amster-Choder O. Coupled Transcription-Translation in Prokaryotes: An Old Couple With New Surprises. *Front Microbiol* **11**, 624830 (2020).
+58. Chatterjee S, Chauvier A, Dandpat SS, Artsimovitch I, Walter NG. A translational riboswitch coordinates nascent transcription-translation coupling. *Proc Natl Acad Sci U S A* **118**, (2021).
+59. Golomb M, Chamberlin M. Characterization of T7-specific ribonucleic acid polymerase. IV. Resolution of the major in vitro transcripts by gel electrophoresis. *J Biol Chem* **249**, 2858-2863 (1974).
+60. Anand VS, Patel SS. Transient state kinetics of transcription elongation by T7 RNA polymerase. *J Biol Chem* **281**, 35677-35685 (2006).
+61. Proshkin S, Rahmouni AR, Mironov A, Nudler E. Cooperation between translating ribosomes and RNA polymerase in transcription elongation. *Science* **328**, 504-508 (2010).
+62. Roy R, Hohng S, Ha T. A practical guide to single-molecule FRET. *Nat Methods* **5**, 507-516 (2008).
+63. Lee CY, McNerney C, Myong S. G-Quadruplex and Protein Binding by Single-Molecule FRET Microscopy. *Methods Mol Biol* **2035**, 309-322 (2019).
+
+# Supplementary Files
+
+- [SUPPLEMENTARY.docx](https://assets-eu.researchsquare.com/files/rs-3352233/v1/e93c43beea94666745181260.docx)  
+  5’UTR G-quadruplex structure enhances translation in size dependent manner

177c0f6bb16179cb55f8108ab5d6acf2540fb1f623db75f36294e762ae5c1093/preprint/images_list.json

@@ -0,0 +1,66 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Train of thought of soggy hydrogel scaffold and sketch of the preparing/using method. Upper panels show the properties of the proposed soggy hydrogel scaffold inspired by soggy and shrunk puffed food. Lower panels (I-VI) show the usage of soggy hydrogel scaffold.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Dampening process analysis. a) Mechanism of dampening and shrunk process of fd-pcGelMA. b) Volume changing of fd-pcGelMA during dampening treatment. c) Water content changing of fd-pcGelMA during dampening treatment. d) Morphology changing during dampening treatment. e) Simulation unit and model for dampening process analysis. f) Simulation of dampening-derived shrinking process with CMVTs. g) Simulation of size changing. h) Simulation of bonding H2O changing.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Mechanical changing during L&D treatment. a) Mechanism of mechanical changing during L&D treatment. b) Morphology of fd-pcGelMA with different L&D treatment durations. c) Compressive curves of soggy-pcGelMA with different L&D treatment durations. d) Compressive modulus of soggy-pcGelMA with different L&D treatment durations. e) Simulation of Vons Mises stress distribution. f) Simulation of compressive modulus.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Analysis during re-swelling process. a) Mechanism of soggy-pcGelMA volume enlargement during re-swelling treatment. b) Images of contact angle testing on the soggy-pcGelMA surface. c) Droplet contact angles changing during wetting process. d) Droplet absorbing speed with different dampening treatment. e) Morphology changing of soggy-pcGelMA during wetting process. f) Volume changing during re-swelling process. g) Mass changing during re-swelling process.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Physiochemical properties of the BioBullet. a) Preparation method of BioBullet. b) SEM morphology of lyophilized scaffold. c) SEM morphology of lyophilized scaffold. d) Images of BioBullet for adipose defect. e) BioBullet volume during different treatment process. f) Images of BioBullet re-swelling. g) Mass/Volume changing of BioBullet during re-swelling process. h) BioBullet degradation.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "Biocompatibility of the BioBullet to adipose-derived stem cells (ADSCs). a) Confocal morphology of ADSCs (with Cherry) seeded on BioBullet. b) LIVE/DEAD of ADSCs seeded on BioBullet. c) Cell area proportion on the BioBullet. d) Cell viability on the BioBullet. e) Cell proliferation on the BioBullet. f) Protein expression on the BioBullet.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_7.png",
+        "caption": "BioGun design and usage. a) Sketch of BioGun. b) Assembly method of BioGun. c) Usage of BioGun.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_8.png",
+        "caption": "Minimally invasive repair of adipose tissue defects. a) B-ultrasoundimages of BioBullet minimally invasive injection process with BioGun. b) Micro-CT images of in-vivo re-swelling process of BioBullet. c) Re-swelled BioBullet size in vivo. d) BioBullet with regeneration adipose tissue. e) Confocal morphology of cells on BioBullet in 40-day culturing. f) H&E staining of cells on BioBullets. g) Vessel regeneration on BioBullet in 40-day culturing. h) BioBullet size in 40-day culturing. i) Lipid area on BioBullet in 40-day culturing.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

177c0f6bb16179cb55f8108ab5d6acf2540fb1f623db75f36294e762ae5c1093/preprint/preprint.md

@@ -0,0 +1,267 @@
+# Abstract
+
+Minimally invasive injection of tissue engineering scaffolds has been gaining attention due to featuring several advantages of requiring a small incision, entailing a simple operation procedure, and involving a quick recovery. However, it remains challenge because larger scaffolds must pass through an extremely thin injection needle. Herein, inspired by the phenomenon that puffed food becomes soggy and shrinks when exposed to air, a novel scaffold treatment method is proposed; namely, lyophilization & dampening (L&D) treatment, which reduces the printed hydrogel scaffold volume by around 90%. Lyophilization treatment can remove water inside the scaffolds; Dampening treatment, that is, placing the freeze-dried scaffolds into a vapor atmosphere, can make scaffolds shrunk to a suitable size for minimally invasive injection. Moreover, unlike existing air-dried technique that feature high mechanical modulus, the soggy scaffolds developed herein feature low mechanical modulus (2.88 kPa), thus minimizing the foreign body sensation after implantation. Furthermore, the injected soggy scaffolds can rapidly swell into their original size and act as tissue regeneration media. Accordingly, a specialized tool namely “BioGun” is designed for the minimally invasive injection of soggy scaffolds ("BioBullet”). This novel strategy would potentially overcome existing technical bottlenecks limiting the clinical soft tissue defect repair.
+
+**Biological sciences/Biotechnology/Tissue engineering**  
+**Physical sciences/Engineering/Mechanical engineering**  
+**Minimally invasive treatment**  
+**Soft tissue**  
+**Projection-based bioprinting (PBP)**  
+**Gelatin methacryloyl (GelMA)**  
+**Hydrogel scaffold**
+
+# Introduction
+
+The question of how to achieve the effective repair of soft tissue defects has been a significant topic in the clinical regeneration field. Traditionally, autologous tissue transplantation has been a common method used in clinical soft tissue repair. Such transplantation sources include fat, skin, etc<sup>1, 2, 3</sup>. During the process, a section of healthy soft tissue is cut or extracted from the patients and then transplanted to their defect area by either surgical suture or injection. Because the autologous tissues used in this type of operation are extracted by the patients themselves, no immune reaction occurs. However, autologous tissue transplantation is often accompanied by problems in clinical practice. Take flap transplantation<sup>4, 5</sup> for example, because the transplanted flap is taken from the patient's own healthy skin, in order to ensure the viability of the flap and reduce complications, it is necessary to ensure that the flap tissue has sufficient blood supply, which increases the required volume taken from the donor site, forms a large area of new trauma, and increases the patient's pain. Moreover, flap tissue transplantation is difficult to perform, such that only specialists can conduct the procedure. Besides, the success rate of the operation cannot be guaranteed, and flap necrosis caused by blood circulation disorders can cause serious complications.
+
+With the development of tissue engineering<sup>6, 7, 8</sup> and three-dimensional (3D) bioprinting<sup>9, 10, 11</sup>, the transplantation of bioprinted hydrogel scaffolds has gradually become an alternative strategy for soft tissue repair. For this approach, hydrogel<sup>12, 13, 14</sup> is formed into specific structures by 3D bioprinting before being transplanted into the soft tissue defect area. The excellent biocompatibility supports the cell growth to facilitate tissue regeneration, thus effectively achieving soft tissue reconstruction. Zahra et al.<sup>15</sup> aims to combine the fast gelation properties of alginate with the mechanical properties of silk nanofibrils (SNFs) to develop hybrid ink containing alginate–SNFs for soft tissue engineering. Peng et al.<sup>16</sup> constructed a novel self-adaptive all-in-one delivery chip that combines therapeutic protein release, gene delivery, and electrical conduction in a single microfluidic chip by 3D coaxial bioprinting, which was verified to successfully organize recruitment and neuronal regeneration cues along with bioelectrical signal in one degradable chip for accelerated skin nerve regeneration. However, although such scaffolds have been demonstrated as effective tools for promoting soft tissue repair, the implantation process is always accompanied by extensive surgical trauma, especially when conducted on large soft tissue defects.
+
+Recently, an approach involving the minimally invasive injection of hydrogel scaffolds has been getting attention among biomedical researchers due to featuring the advantages of requiring only a small incision, entailing a simple operation, and involving a quick recovery<sup>17, 18</sup>. So far, owing to their tiny size, microscale hydrogel structures, such as hydrogel microspheres<sup>19, 20, 21</sup>, have become one of the most important structural units used in minimally invasive injections to repair soft tissue defects. However, in terms of repairing large soft tissue defects, large hydrogel scaffolds are always more reliable than microscale ones due to their high structural integrity. Unfortunately, such scaffolds are also hindered by their large size, which prevents them from passing through a thin needle. Because of this, their compatibility with the minimally invasive injection technique has remained a challenge, or else been considered an unachievable goal altogether.
+
+To overcome this limitation and widen their application prospects, inspired by the phenomenon that puffed food gets soggy and shrinks when exposed to the air, we herein propose a pretreatment method to shrink gelatin methacryloyl (GelMA) hydrogel<sup>22, 23, 24</sup> scaffolds prepared by projection-based 3D bioprinting (PBP)<sup>25, 26, 27</sup>, namely lyophilization&dampening (L&D), thus enabling them to pass through a minimally invasive injection needle, which is followed by their rapid re-expansion to their original size, thereby filling up the target defect space *in vivo* (Fig. <span class="InternalRef" refid="Fig1">1</span>). Lyophilization treatment is employed to remove water and form a microscale internal pore network inside the hydrogel scaffolds; Dampening treatment, that is, placing the freeze-dried hydrogel scaffolds into a vapor atmosphere, is used to promote the diffusion of water molecules into the internal pore network to bond to the hydrophilic groups on the pore walls, resulting in the scaffold shrinking and reaching a suitable size for its minimally invasive injection. Once the soggy scaffolds are immersed into an aqueous solution at the soft tissue defect site, they can rapidly re-swell and expand into their original size to fill up the defect space and act as tissue regeneration media. Accordingly, a specialized minimally invasive injection tool, namely “BioGun”, is designed, so that the soggy scaffold termed “BioBullet” can be shot out by it. This novel strategy will potentially overcome existing technical bottlenecks that limit the clinical repair of soft tissue defects by minimally invasive injection in future.
+
+# Results
+
+## Miniaturization benefitted from volume shrinking
+
+Having a scaffold size that is small enough is a key factor determining the success of scaffolds purposed for injection into the defect site of a patient, as a small size is what allows the scaffold to pass through a thin injection needle. Herein, photocrosslinked GelMA structures (pcGelMAs) (10.0 mm × 10.0 mm × 1.5 mm) printed by a PBP device were treated into freeze-dried GelMA structures (fd-pcGelMAs) with a vacuum lyophilizer at -80°C. During this process, the water molecules inside the hydrogel network gradually turn into ice crystals and completely sublimate. As a result, there remains an ocean of pores inside the hydrogel network where the ice crystals were previously situated. Then, the fd-pcGelMAs were placed into a dampening device, which provided thermostatically-controlled vapor (Supplementary Note 1, Fig. S1) and turned the fd-pcGelMAs into soggy GelMA structures (soggy-pcGelMAs). Volume shrinking was the most apparent phenomenon observed during the dampening treatment. The equivalent volume rapidly decreased from 129.91 ± 6.44 mm³ within 0.5 hours and tended to be stable after 2 hours at 11.81 ± 0.45 mm³, which was approximately 1/11th of the original volume. Later, the equivalent volume began to increase slightly (Fig. 2b, d). The water content (Fig. 2c) linearly increased inside the soggy-pcGelMAs during the first 4 hours, after which, the increasing rate plummeted. After 22 hours of the dampening treatment, the water content inside the soggy-pcGelMAs reached only 47.1%, which demonstrated that the quantity of water absorbed during the treatment was very limited compared to that inside the GelMA structures with swelling balance. In addition, the water content only reached 22.2% within the first 0.5 hours, during which time, the soggy-pcGelMAs rapidly shrunk. This indicated that the absorbed water molecules prioritized causing a shrinking effect in the soggy-pcGelMAs, rather than re-swelling effect.
+
+The fd-pcGelMA dampening process is similar to the phenomenon whereby puffed food (bread, cookies, potato chips, etc.) becomes soggy after absorbing moisture from the air. The freeze-dried structure dampening treatment is composed of three steps (Fig. 2a), described as follows: (i) water molecules diffuse into the porous structure; (ii) the water molecules replace air molecules; (iii) the water molecules bond with hydrophilic groups on the inner walls of the fd-pcGelMA, causing inner wall shrinkage. The freeze-drying process can maintain the hydrogel network in the pcGelMA and leave a high quantity of inner pores when the water is removed. During the dampening process, compared to the direct immersion process (Supplementary Note 2, Fig. S2, Fig. S3, Supplementary Video 8), water molecules within the vapor can smoothly enter the fd-pcGelMAs through the pore network and continuously drive out air molecules within them, thereby avoiding the generation of bubbles that prevent water molecules from entering. Once the water molecules reach the inner pores, they rapidly bond to the hydrophilic groups on the inner pore walls of the fd-pcGelMA. Simultaneously, the water molecules break up the initial molecular potential energy system. With hydrogen bonding, one water molecule can act as a medium and bond to several hydrophilic groups on the fd-pcGelMA inner pore walls, causing them to fold. As a result, the fd-pcGelMA gradually shrinks on the macroscale and further removes inner air. Once the bonded water volume is equal to the inner pore volume, that is, once the inner air is totally replaced by water, the shrinking stage is finished. It subsequently enters the re-swelling stage, although the swelling speed can be extremely low in a vapor atmosphere. This is why the equivalent volume rapidly decreases at first before slowly increasing.
+
+To further describe the fd-pcGelMA shrinking behavior during the dampening process, a finite element simulation model was established (Fig. 2e) and parametric sweep analysis was carried out (Supplementary Note 4, Table. S1). Each pore within the pore wall inside the fd-pcGelMA was simplified into a “constant material-volume truss (CMVT)” structure with hydrophilic bonds on the “rods”. The simulation was carried out with the “truss” unit. With the bonding of water molecules onto the “rods”, the rods shortened along the axial direction and the section area was enlarged, keeping the rod volume unchanged. The calculation was set up to stop when the total water volume reached the total pore volume. A crossed shape that was composed of 1 center “truss” and 4 branch “trusses” in each direction was established to simulate the different layers of the pores inside the fd-pcGelMA. Because the water molecules enter from the external environment and can bond to the outer layer pores in advance when they diffuse into the fd-pcGelMA through the pore network, there can be a water molecule quantity decreasing gradient from the outer to inner areas of the fd-pcGelMA. As a result, the water molecule bonding rate decreases from the outer to the inner fd-pcGelMA. From the simulation results (Fig. 2f, g, h, Supplementary Video 1), every “truss” unit gradually shrunk and the quantity of bonded water molecules on the “rods” gradually increased. Firstly, the pores in the five layers shrunk as the bonded water content increased and the total volume rapidly decreased. Then, the pores in the outermost layer reached a balance, such that the pore volume was equal to the bonding water volume. The volume shrinking rate began to decline. At last, all pores in all layers gradually reached the balance described above and the volume remained unchanged. The simulation trends of the volume change and water content change were extremely similar to the trends observed within the experiment. These results and analysis verified that the soggy-pcGelMAs could be treated to reach a suitable size, which would subsequently allow them to pass through a thin injection needle for scaffold injection.
+
+## Potential foreign body sensation according to mechanical properties
+
+The effect of causing a foreign body sensation in the treated patient is another key factor used to judge the feasibility of injectable scaffolds. The injected scaffold should strictly avoid causing a foreign body sensation to protect patients from secondary damage. Thus, such scaffolds should feature a relatively low mechanical modulus. Here, the compression property of the soggy-pcGelMAs after 0, 0.5, 1, 2, 4, and 6 hours of dampening treatment was tested, respectively. From the post-compressing images (Fig. 3b), the compressed soggy-pcGelMAs in the earlier stage of the soggy treatment (0.5–1 h) showed plastic formation, rather than a snapback effect to some height, indicating that they were still in a freeze-dried state. However, with a further increase in the dampening treatment duration, the soggy-pcGelMAs absorbed more water and started to show the elastic formation of a hydrogel to some extent. Those treated for 6 h displayed the recovery of half of their original height when the compression force was removed. These phenomena could potentially be explained from three aspects (Fig. 3a): (i) original cracks; (ii) inner pore wall shrinkage; (iii) persistent hydrogelation.
+
+**(i) Original cracks.** During the freeze-drying treatment, the water turned into ice crystals. It is known that the water crystallization process can cause volume enlargement (the density of ice is 0.9 times that of liquid water). Thus, rather than forming a perfect hydrogel network, tiny cracks can form within the fc-pcGelMAs after the freeze-drying treatment, which may cause structural fluctuation during compression testing. Furthermore, high friction exists among the crack interfaces in the fd-pcGelMAs, which strengthens their mechanical properties to some extent.
+
+**(ii) Inner pore wall shrinkage.** The inner pore walls continually become shorter and thicker during the dampening treatment. This can potentially affect the compression strength and modulus of the structure. Here, the same CMVT structure introduced above was selected as the mechanical property simulation unit in order to demonstrate the influence of dampening shrinkage on the structural compression modulus (Supplementary Note 5, Table. S2). The simulation results (Fig. 3e, f, Supplementary Video 2) show that with decreasing truss length, the compression modulus of the CMVT unit continually increased. Thus, regardless of the effect caused by the bonded water, dampening shrinkage leads to the strengthening of the mechanical properties.
+
+**(iii) Persistent hydrogelation.** The essence of the dampening process is the dissolution of the fd-pcGelMA in the water (vapor). With the water molecules continually entering the fd-pcGelMA, the material system gradually forms a hydrogel state from the dried solid state. Therefore, once the bonded water quantity passes a threshold, the material system tends to display hydrogel properties, such as elastic formation, viscous creep, etc. Importantly, due to the existence of the original cracks discussed above, the accumulated bonded water can act as lubricant among the cracks to sharply weaken the interface friction, leading to substantially lower mechanical strength.
+
+Based on the above three aspects, the phenomenon observed in the compressing curve (Fig. 3c) and the analyzed compression modulus (Fig. 3d) can be easily explained. In the earlier dampening stage (0–0.5 h), the soggy-pcGelMAs exhibited fluctuation, which was caused by the original cracks and the increased compression modulus derived from the shrinkage of the inner pore walls. Then, in the middle dampening stage (0.5–1 h), the soggy-pcGelMAs were in a transitional stage between the dried solid state and the hydrogel state. Thus, the extent of fluctuation was lower and the compression modulus began to decrease due to the lubricating effect of the increase in the bonded water. At last, in the later dampening stage (1–6 h), with further hydrogelation, the material system completely entered a hydrogel state. Hydrogel properties; such as elastic formation and viscous creep; had appeared. Furthermore, the compression modulus reached an extremely low level compared to the dried solid state. Moreover, the compression curve of pcGelMA treated with another common drying/shrinking method, namely air-drying treatment, was tested. It could be seen that although the volume and water content of pcGelMA could be effectively reduced with this method, the compressive modulus of the air-dried pcGelMA was much higher than the one of soggy-pcGelMA (Supplementary Note 3, Fig. S4, Fig. S5, Fig. S6, Fig. S9), which would cause obvious foreign body sensation after injection. These results and analysis verified that the soggy-pcGelMAs could be treated in such a way as to transform them into an ideal mechanical state, thus minimizing the chance of causing a foreign body sensation after scaffold injection.
+
+## Defect-filling capability of the re-swelling process
+
+After a soggy-GelMA is injected into a damaged site on an organ, it is expected to re-swell into a larger volume by absorbing the surrounding bodily fluids, subsequently filling the subcutaneous cavity. It is known that gelatin has a high solubility in water due to containing an abundance of hydrophilic groups on its molecule chains. During the dampening process, vapor enters the inner pores of the fd-pcGelMA. These water molecules then bond to hydrophilic groups, causing the shrinkage of the fd-GelMA. However, under practical circumstances, water molecules bond to hydrophilic groups randomly, so that some hydrophilic groups do not attach to the water molecules and are covered by the wrinkled inner pore walls or have not reached a great enough number (Fig. 4a). Thus, the soggy-GelMA can continue to absorb more water once it is immersed, which is referred to as the re-swelling process. Unlike the vapor used in the dampening treatment, liquid water can provide a sufficient quantity of water molecules and greater outer pressure, which enables it to more rapidly enter the soggy-GelMA and bond to hydrophilic groups within it. Subsequently, the walls of the shrunk soggy-GelMA quickly unfold and re-swell into a larger size.
+
+The surface hydrophilia of the soggy-pcGelMAs that underwent different dampening treatment durations were tested (Fig. 4b, Supplementary Video 3). The results showed that the initial contact angel decreased (Fig. 4c), that is, the surface hydrophilia was enhanced, with the increase in the duration of the dampening treatment. This means that the surface was wetter (contained more water molecules) once a water droplet was dropped onto the surface of the soggy-pcGelMAs. Importantly, the fd-GelMA, namely, soggy-pcGelMAs that did not undergo the dampening treatment, did not allow water molecules to enter smoothly because the presence of inner air bubbles prevented them from doing so (Supplementary Note 3, Fig. S7). With droplets continually wetting the surface, the local surface hydrophilia continually increased, according to the contact angle change, demonstrating that the continuous surface wetting promoted the re-swelling process. Moreover, longer dampening treatment duration led to more rapid droplet absorption (Fig. 4d). This phenomenon was likely due to the higher surface hydrophilia and the presence of fewer inner residual bubbles. The volumes and masses of the soggy-pcGelMAs produced with different re-swelling durations were tested. Volume and mass increased with the increase in the re-swelling process duration (Fig. 4e, f, g, Supplementary Video 4). The re-swelling speed was extremely fast. Within only 5 minutes, the soggy-pcGelMAs reached their original sizes (10.0 mm × 10.0 mm × 1.5 mm). Furthermore, the re-swelling process of air-dried pcGelMAs was also recorded (Supplementary Video 7, Supplementary Note 3, Fig. S8). It could be seen that although the air-dried pcGelMAs could be re-swelled, the re-swelling speed is much lower than that of soggy-pcGelMA, which further verified that the air-drying method was not suitable for preparation of hydrogel scaffold for minimally invasive injection. These results and analysis verified that the soggy-pcGelMAs could successfully re-swell to fill the site of organ damage following scaffold injection.
+
+## Physiochemical properties of BioBullets for adipose defects
+
+Considering the convenience and mass production possibility, soggy-pcGelMAs were designed as “BioBullet”, that is, the soggy-pcGelMAs were loaded in glass tube and further compressed inside, resulting in lower volume (Fig. 5a). Here, we took adipose defect repair as an application scene to demonstrate the feasibility of the proposed strategy herein. Considering that the mammary gland tissue of mice is small and flat-lobed, when designing the scaffold structure, the external dimensions of the BioBullets employed in adipose defect repair were designed as a tablet shape (φ5 mm × 2 mm). Moreover, in order to promote cell migration and nutrient diffusion, a macroscopical grid (600 µm) was designed on the tablet. Then, the printed pcGelMAs underwent the lyophilization and dampening (2 hours) treatment, thereby turning them into soggy-pcGelMAs. At last, the soggy-pcGelMAs were forced into a sterilized glass tube one by one according to the adipose defect size.
+
+From the scanning electronic microscope (SEM) images (Fig. 5b, c), pores could be observed at two scales in the fd-pcGelMAs and soggy-pcGelMAs, namely millimeter-scale pores built by the computer-aided design (CAD) model and micro-scale pores generated by the hydrogel molecules and the ice crystals after the lyophilization treatment. Regarding the micro-scale pores specifically, the porosity results again verified that dampening treatment shrunk the inner pore walls and decreased the inner pore size and porosity (44.76% decreased to 28.01%) inside of the fd-pcGelMAs and soggy-pcGelMAs. Furthermore, due to the introduction of the grid on the structure, the soggy-pcGelMAs could be further flattened (Fig. 5d), rendering them more readily injected into the tissue defect. The volume results (Fig. 5e) showed that after the dampening treatment, the volume of the structure was reduced from 53.94 mm³ to 6.69 mm³; a reduction of 87.60%. Moreover, with further flattening, the volume of the structure decreased to 1.82 mm³, resulting in a final volume reduction of 96.63%.
+
+The structure re-swelling process of BioBullet was further analyzed (Fig. 5f, Supplementary Video 5). The results (Fig. 5g) displayed that the average volume of the BioBullet expanded by 40 times, from 1.82 mm³ to 75.36 mm³. Remarkably, in the first 0.5 minute of re-swelling, due to the exposed area of the scaffold and water was small, the re-swelling curve was flat. After 1 minute, the BioBullet gradually unfolded and the water absorption rate increased. In addition to these tests, we assessed the enzymatic degradation performance of the BioBullet, which is a typical evaluation method used to verify whether a scaffold can be metabolized after implantation. Here, the re-swelled BioBullet and the printed structures that did not undergo any treatment were degraded in collagenase II solution. The degradation results of the two kinds of structures did not show remarkable difference (Fig. 5h), indicating that L&D treatment would not alter the degradation performance of the GelMA scaffolds.
+
+## Biocompatibility of BioBullets to adipose-derived stem cells (ADSCs)
+
+Once the BioBullets are “shot” into the adipose defect, they rapidly re-swell into much larger scaffolds to fill up the defect space. Then, cells surrounding the defect area can migrate to the re-swelled scaffolds to rapidly proliferate and differentiate on them. Among these cell types, adipose-derived stem cells (ADSCs) are one of the most significant in adipose defect repair. Under suitable induction conditions, ADSCs can differentiate into adipose cells, which contain a fat reservoir comprised of triglycerides, and these cells gradually populate the tissue defect area. Therefore, ensuring the proper growth of ADSCs on the re-swelled BioBullets is necessary to guarantee the clinical treatment effect.
+
+The ADSCs were seeded onto the re-swelled BioBullets. After 3 days of culture, the ADSCs were generally attached to the surface of the scaffold in a stretched state (Fig. 6a, c). After 7 days of culture, the ADSCs had migrated rapidly, attached evenly to the entire surface of the BioBullets, and continued to grow downward along the grid pores in the scaffold. The cells showed clear spreading morphology in the absence of adipogenic induction. The LIVE/DEAD staining results (Fig. 6b, d) showed that the cell viability remained above 97% during the 7 days of culturing, indicating that the proposed BioBullets had negligible biological toxicity to the ADSCs. The MTT testing (Fig. 6e) results demonstrated that in the first 3 days, the proliferation rate of the ADSCs on the BioBullets was essentially the same as those cultured in a 96-well plate. However, on the 5th day, the proliferation rate of the cells seeded on the BioBullets was significantly higher than that of those seeded in the 96-well plate. This result was attributed to the greater specific area of the former enabling a greater proportion of cells to attach.
+
+Importantly, previous studies have demonstrated that ADSCs can express the stem cell marker CD44, but not hematopoietic system and epithelial markers, such as CD19. In order to verify the cellular functions of the seeded ADSCs on the BioBullets, Western blotting (WB) was used to detect their specific protein expression levels. The results (Fig. 6f) showed that CD44 was positive in the 7 days of culture and was stably expressed on both the BioBullets and petri cells, which confirmed the cell stemness of the ADSCs during the culture process. In addition, CD19, a marker of epithelial cells, was not expressed in the ADSCs, indicating that they did not differentiate during the culture process. These results verified the excellent cell compatibility and adipogenic induction effect of the BioBullets and the potential of their use in the clinical repair of soft tissue defects.
+
+## A portable minimally invasive therapy tool: BioGun
+
+Based on the previous systematic research and discussion of the BioBullets, their physicochemical properties and biocompatibility confirmed their feasibility as an efficient vehicle for the minimally invasive injection of scaffolds to repair soft tissue defects. However, there was no corresponding surgical tool to deliver these BioBullets. Hence, by referring to the structure of the gun, a portable device called BioGun was also designed to realize the minimally invasive injection of the BioBullets.
+
+The constructed BioGun was composed of pushrod, trigger, barrel, chamber, syringe with water, BioBullet magazine, BioBullet outlet and water outlet (Fig. 7a). Before the surgery, BioBullets were loaded inside of the magazine and the syringe-loaded PBS was inserted into the chamber (Fig. 7b). The BioBullet/water outlet of the BioGun was then aimed at the soft tissue defect area so that the tip was located in the middle of the defect cavity. The pushrod was then pulled to push the BioBullets out of the magazine and inject them into the defect area. Finally, the trigger of the syringe was pushed to inject the PBS into the defect area to speed up the re-swelling process.
+
+In order to verify the feasibility of the BioGun, the assembled BioGun was then applied to carry out a simulation of a minimally invasive injection surgery conducted on a pork adipose tissue defect (Fig. 7c). A cavity with a diameter of 5 mm was artificially formed in the fat area. Then, a BioBullet was injected into the defect site by the BioGun, followed by injecting 0.5 mL PBS. The results showed that only 3 minutes were required to re-swell the BioBullet and completely fill up the entire defect space, confirming the feasibility of the BioGun in performing the minimally invasive injection.
+
+## Minimally invasive repair of adipose tissue defects
+
+In applications for the repair of adipose tissue defects, the desired therapeutic goals are achieving the restoration of the appearance and the regeneration of adipose tissue. To assess this, 8-week-old female mice obtained from the Institute of Cancer Research (ICR) were used. The subcutaneous fat in the inguinal region of the mice was selected from the surgical area. The minimally invasive implantation process with BioBullet and BioGun was completed under the guidance of small animal B-ultrasound (Fig. 8a, Supplementary Video 6). The metal tip of the BioGun was dipped into the skin of each mouse to locate the adipose tissue. When located, the BioBullet was then injected along with some PBS into the adipose tissue. Images captured by a small animal imager (Fig. 8b, c) showed that the BioBullet rapidly expanded within 10 minutes in vivo and its diameter became twice that measured in the wet state, which confirmed that the BioBullet could fill the entire defect space.
+
+From a clinical point of view, the shape in which the defect site recovers and the maintenance of this shape during repair are important evaluation indexes for the repair of soft tissue defects, such as those in adipose tissue. The injected BioBullets were taken out at 14 days, 28 days, and 40 days, respectively. During the 40-day repair process, the BioBullets maintained their morphology without exhibiting spatial collapse and structural fragmentation (Fig. 8d, h). After 14 days of repair, the pores and periphery of the scaffold were tightly wrapped by the adipose tissue, the scaffold was translucent, and the adipose tissue was milky-white. The H&E results (Fig. 8f, i) showed that on the 14th day, a great number of cells grew only around the scaffold, leaving many cavities inside the scaffold that contained no cells. On the 28th day, the scaffold was essentially filled with cells, but the content of balloon-like adipocytes was only approximately 30%, which mainly appeared in the peripheral area of the BioBullets. On the 40th day, the inside of the scaffold was filled with a large amount of vacuolated adipose tissue, with the cell area accounting for 70%. To demonstrate that the cells grown in the scaffolds were indeed adipocytes, we performed Nile red staining, which employs a fat-specific probe to sensitively capture and stain intracellular lipid droplets. The results (Fig. 8e) verified that during the 40-day implantation period, the adipocytes gradually matured, the lipid droplet size increased, and the cell density increased. This indicated that the implanted BioBullets had good compatibility with the adipocytes and could promote their rapid proliferation and growth.
+
+A major clinical difficulty in the repair of large soft tissue is the regeneration of vascular structures within the tissue. In the experiment, we observed the formation of blood vessel cells near the newly grown adipose tissue inside the scaffold. In order to characterize the formation of the vascular structure inside the scaffold during the repair process and to explore its relationship with adipose tissue regeneration, CD31 staining and immunohistochemistry were performed on the tissue sections. Obtaining a positive signal not only confirms the identity of vascular endothelial cells, but also reflects that the cells are already exercising the function of nutrient exchange within the endothelial cells. The results (Fig. 8g) showed that on the 14th day, a large number of CD31-positive cells had aggregated around the BioBullet, indicating that a high degree of endothelial angiogenesis had occurred by this time. On the 28th day, the CD31-positive cells continued to aggregate around the scaffold and generate more voids inside the BioBullet by degradation. This was accompanied by the formation of large pieces of adipose tissue. On the 40th day, a large amount of adipose tissue had grown into the inside of the BioBullet and endothelial cells had formed large blood vessels around the vacuolated adipocytes. This network provides a channel for nutrient loosening and material exchange for the new growth of adipose tissue.
+
+# Discussion
+
+Confronted with the application difficulties of 3D bioprinted hydrogel scaffold in clinical minimally invasive repair of soft tissue defect, which has been restricted by the scaffold size, mechanical properties and biocompatibility, in this work, the soggy puffed food inspired treatment method developed herein successfully achieve the large hydrogel scaffold in minimally invasive injection, which is a bran-new development in this field. This strategy realizes scaffold size shrinking for their subsequent application by minimally invasive injection, avoiding causing a foreign body sensation, re-swelling in the defect space, and promoting tissue regeneration in the soft tissue defect area. We believe that this novel strategy will potentially break through existing technical bottlenecks that limit the clinical repair of soft tissue defects by administering a minimally invasive injection.
+
+For therapy optimization, we encourage researchers in relevant research fields to further develop this strategy from three aspects:
+
+1. Morphology matching. In this paper, the shape of the BioBullet was directly designed as a grid tablet scaffold. Regarding the morphology of complex soft tissue defects, the column structure cannot perfectly fill up the defect space. In the future, doctors can first scan the 3D morphology of the soft defect using CT technology before re-building the 3D model with CAD. With this approach, the injected BioBullet will perfectly fill up not only the main body space, but also the tiny gaps in the soft tissue defect area.
+
+2. Functional additives. Herein, the scaffold was only composed of GelMA hydrogel, without any other functional additives. In its future optimization, researchers may add regeneration drugs or biological factors to speed up the regeneration process. Moreover, by introducing hydrogels with the property of specific controlled release, the releasing behavior can be strictly manipulated to achieve a better repair effect.
+
+3. Tool development. Better tools produce better outcomes. In this work, we have designed the modular BioGun to perform the minimally invasive injection of the BioBullet and auxiliary fluid. In the future, the magazine of BioGun can be re-designed to automatically load a new group of BioBullets once the previous have been “shot” out, thereby better resembling the structure of a real rifle, so that a greater quantity of BioBullets or those with different functions can be loaded simultaneously.
+
+# Methods
+
+**Preparation of GelMA bioink.** The purchased GelMA (EFL-GM-100, purity > 99.9%) powder was removed from a -80°C refrigerator and then dissolved with deionized water to the concentration of 15% w/v. Photoinitiator lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP, purity > 99.8%) was then added into the solution to reach a concentration of 0.5% w/v. The resulting solution was sufficiently dissolved using a 50°C water bath. Then, the light absorber Tartrazine (purity > 95%) was added into the GelMA bioink to reach a concentration of 0.03% w/v. The prepared GelMA bioink was then stored in a 4°C refrigerator.
+
+**Printing of pcGelMAs by PBP.** The pcGelMAs were printed by a desktop PBP printer (EFL-BP8600). First, the printer was powered on and the preheating temperature required for printing the material was set. Both the deposition platform and bioink pool of the printer were then removed. The inside of the bioink pool and the formation area of the deposition platform were each wiped down with a non-woven cloth dipped in 75% alcohol. They were then dried and placed back on the printer. The designed 3D digital model was exported to the ‘.STL’ file format and imported into the printer's host computer software to adjust its size, position, and layout. The light intensity, curing time, and printing temperature were set. After the printer had preheated, the prepared bioink was aspirated a certain amount with a pipette gun and then added to the bioink pool. The amount of bioink added was evaluated according to the actual volume of the structure to be printed. The protective cover was placed and then ‘Start Printing’ was selected. After printing, the printing deposition platform was carefully removed. The hydrogel structure of the printing molding was also carefully removed with a spatula and then immediately placed into the isotonic solvent to wash away the remaining uncrossed bioink on its surface. The washed hydrogel holder was removed from the isotonic aqueous solution, transferred to 75% medical alcohol for immersion, and then placed in a 4°C refrigerator for long-term storage.
+
+**Lyophilization treatment of the pcGelMAs.** The photocrosslinked GelMA hydrogels that had reached swelling equilibrium were transferred into well plates. The well plates were then placed in a -80°C refrigerator for 1 hour to fix the hydrogel morphology, during which time, the vacuum drier was turned on and the cooling temperature was set to -80°C. Then, the well plates containing the hydrogels were transferred into the vacuum drier and the pump was turned on to remove all the gas inside. After 24 hours, the freeze-dried and vacuum-sealed GelMA hydrogels were stored.
+
+**Dampening treatment of the fd-pcGelMAs.** The heater temperature of the dampening treatment device was set to 50°C and the treatment conditions were 25°C and 95%. The fd-pcGelMAs were transferred to a 24-well plate, which was followed by being placed on the metal netting of the dampening treatment device. The treated fd-pcGelMAs (soggy-pcGelMAs) were removed from the device at different time points and restored in a plastic bag using a food vacuum packaging machine.
+
+**Volume measurement of the soggy-pcGelMAs.** From the soggy-GelMA volume observations made during the dampening treatment, soggy-GelMA cuboids fabricated in the same batch were treated by the dampening treatment device simultaneously before being removed for volume analysis at 0.5, 1, 2, 4, 6, and 22 hours (n = 3). Random deformation would occur during both the freeze-drying process of the photocrosslinked GelMA hydrogels and the dampening process of the fd-GelMA. Therefore, obtaining direct measurements with the use of a ruler was impossible. Here, ImageJ was applied to analyze the upper area  
+$${S}_{upper}\left(t\right)$$  
+(regarded as a standard square) of the samples according to photos obtained by a camera (Nikon SLR) and the use of a ruler. Importantly, the molecular network inside the photocrosslinked GelMA hydrogel was roughly isotropic and the sizes in three orthogonal directions could change in equal proportions  
+$$\epsilon$$  
+during the dampening process. Therefore, the volume of the soggy-pcGelMA  
+$${V}_{soggy-GelMA}\left(t\right)$$  
+at a time point  
+$$t$$  
+could be calculated as below:
+
+$${V}_{soggy-pcGelMA}\left(t\right)={V}_{design}\times {\left(\frac{{S}_{upper}\left(t\right)}{{S}_{upper-design}}\right)}^{\frac{3}{2}}$$
+
+where  
+$${V}_{design}$$  
+and  
+$${S}_{upper-design}$$  
+were the volume and the upper area of the designed photocrosslinked GelMA hydrogel (mold size), respectively.
+
+**Water content measurement of the soggy-GelMA.** fd-GelMA cuboids fabricated in the same batch were treated by the dampening treatment device simultaneously and then removed for water content analysis at 0.5, 1, 2, 4, 6, and 22 hours (n = 3). The fd-GelMA and soggy-GelMA were lightly wiped with a non-woven fabric to remove the water on their surface. Then, the masses of the fd-GelMA  
+$${m}_{fd-GelMA}$$  
+and soggy-GelMA  
+$${m}_{soggy-GelMA}\left(t\right)$$  
+were respectively measured with an electronic balance. The water content of the soggy-GelMA  
+$${\psi }_{soggy-GelMA}\left(t\right)$$  
+at a time point  
+$$t$$  
+could be calculated as below:
+
+$${\psi }_{soggy-GelMA}\left(t\right)=\frac{{m}_{soggy-GelMA}\left(t\right)-{m}_{fd-GelMA}}{{m}_{fd-GelMA}}\times 100\%$$
+
+**Compression testing of the soggy-pcGelMAs.** Bioink was printed by a PBP printer into φ4 mm × 4 mm and then immersed in PBS for 24 h to reach swelling equilibrium. Then, L&D treatment was carried out on the pcGelMAs (n = 3). A universal testing machine (UTM-2102) was used to test the compression modulus using a 100 N mechanical sensor, a bar compression test module, an inlet force of 0.001 N, and a compression test rate of 1 mm/min. The device automatically recorded the force-displacement curve. After the test, the "force-displacement" curve data was exported, underwent standardization processing in Excel, and was converted into a "stress-strain" curve to eliminate the influence of different dimensions on the test results.
+
+**Surface hydrophilia of the soggy-pcGelMAs.** The hydrophilia of the soggy-pcGelMAs (n = 3) was determined by a contact angle surface tension measuring instrument (Dropmeter 100P). The soggy-pcGelMAs were placed on the sample stage of the surface contact angle tester. Then, a 4 µL droplet ($${V}_{\text{d}\text{r}\text{o}\text{p}\text{l}\text{e}\text{t}}$$) was dropped on the surface of the samples ($${t}_{0}$$). A high-speed digital camera was utilized to take pictures; the shooting time interval of each photo was set to 0.3 s. The obtained photos were analyzed frame by frame and used to measure the size and change in the contact angle of the droplets in each photo. The time point at which the droplet had completely infiltrated was recorded as $${t}_{i}$$. The average infiltration rate of the droplet can be expressed by the following formula:
+
+$${\stackrel{-}{v}}_{\text{a}\text{b}\text{s}\text{o}\text{r}\text{b}}=\frac{{V}_{\text{d}\text{r}\text{o}\text{p}\text{l}\text{e}\text{t}}}{{t}_{i}-{t}_{0}}$$
+
+**PBP printing of the BioBullets.** BioBullets featuring grids were designed with Solidworks software. The pore size was set to 600 µm. To compensate for the effect of the crosslinking depth on the side pore size, the side pore height was set to 1000 µm. The model was imported into slicing software. The layer height of the slice was set to 100 µm. The printed pcGelMAs were then carefully scraped off the deposition platform with a razor blade and subsequently washed in PBS to remove the residual bioink, which was followed by immersing the pcGelMAs in PBS for 24 h to reach swelling equilibrium. The resulting pcGelMAs then underwent L&D treatment. The dampening process lasted for 2 hours.
+
+**SEM morphology of the BioBullets.** The morphology of the pore network in the BioBullets was observed with a scanning electronic microscope (SEM). The samples were frozen with liquid nitrogen and transferred to a vacuum freeze dryer at -80°C for 24 h. The resulting freeze-dried samples were then torn along the cross-section and coated with a sputtering coater.
+
+**Degradation of the BioBullets.** The BioBullets were re-swelled in deionized water for 24 hours and then placed in a 1.5 mL EP tubes in a 37°C, 5% CO₂ incubator. Then, 1 mL of PBS containing 2 U/mL type II collagenase was added to each tube. After 0, 1, 2, and 4 h of degradation, the solution in the EP tube was removed and the tubes were transferred to a -80°C refrigerator for storage. After removing all the degraded samples, those remaining were lyophilized by a vacuum freeze dryer for 24 hours. The final weights of the freeze-dried samples were then respectively weighed.
+
+**ADSC seeding on the BioBullets.** The BioBullets were placed into a sterile 96-well plate and irradiated with UV light for 30 minutes on a sterile ultra-clean bench. ADSCs were then digested in culture dishes with 0.25% EDTA-trypsin and resuspended with a small amount of medium to reach a final cell concentration of approximately 5 × 10⁶/mL. Then, 100 µL of the cell suspension was pipetted gently onto the center of the surface of the BioBullets. The cell suspension then underwent rapid absorption by the BioBullets and the volume expanded. The inoculated scaffolds were placed on a clean table for 30 minutes before being transferred to a 37°C, 5% CO₂ incubator for 4 hours of incubation. After 4 hours, 100 µL of cell culture medium was slowly added to each well to ensure that the liquid level did not cover the scaffold. This was then transferred to a CO₂ incubator to continue the incubation.
+
+**Modeling of adipose tissue defects and minimally invasive injection using BioGun.** 8-week-old female mice from the ICR (Institute of Cancer Research) were anesthetized with 1% pentobarbital before their side hair was shaved using a body hair machine. Then, the surgical area was depilated with depilatory cream and thoroughly cleaned with 75% alcohol. The subcutaneous fat in the inguinal region of the mice was selected from the surgical area. The inguinal fat of each mouse was located, a piece was cut off with sterile scissors, and the incision was sutured. For the minimally invasive injection process, each part of the BioGun was thoroughly wiped with 75% alcohol and UV-irradiated overnight on a sterile bench. Then, the BioBullet was compressed, loaded into a glass cartridge case, and pressed into the magazine. A certain amount of phosphate-buffered saline (PBS) was drawn with a 2 mL syringe and then assembled according to the operating procedure. The minimally invasive injection of the BioBullet was completed under the guidance of small animal B-ultrasound. The metal tip of the BioGun was placed into the skin of the mouse to locate the adipose tissue. Under real-time display using B-ultrasound, the metal gun barrel has a strong white echo with a sound shadow. A pipette tip was inserted deeply into the defect site of the adipose tissue and the piston was pressed to push out the compressed GelMA scaffold. The water injection trigger was squeezed to inject a certain amount of PBS *in situ*.
+
+**Histological analysis.** BioBullets within newborn tissues were removed on the 14th, 28th, and 40th days after their injection, respectively. They were then dehydrated, embedded, sectioned, and stained with H&E to show the ingrowth of cells around the scaffold. In order to more clearly show the distribution of the newly grown cells inside of the BioBullets, during the sectioning process, sections were sectioned at a thickness of 6 µm, and those near the BioBullet-containing tissue were selected after staining. In order to show the ingrowth trend of adipocytes within 40 days of scaffold repair, ImageJ software was used to quantify the newly formed vacuolar adipose tissue in the slices.
+
+**CD31 immunohistochemistry.** Dewaxing and hydration were conducted with the following at room temperature, in chronological order: xylene No. 1 for 10 minutes, xylene No. 2 for 8 minutes, xylene No. 3 for 8 minutes, absolute ethanol for 5 minutes, 90% ethanol for 3 minutes, 80% ethanol for 3 minutes, 70% ethanol for 3 minutes, ethanol for 3 min, distilled water for 3 minutes × 3 times, and PBS for 3 minutes × 3 times. For hot antigen retrieval, a new slide rack was used to replace the old one, and the hydrated slides were placed. Antigen retrieval solution was prepared using a packet of sodium citrate powder to prepare 2 L of PBS solution. This was then heated and boiled in an iron lunch box or pot. The temperature was controlled at 96–98°C. The slices were then placed in a hot pan for 15 minutes before being naturally cooled to room temperature. The slices were then rinsed with PBS for 5 minutes × 2 times, followed by distilled water for 3 minutes × 2 times. The tissue was removed, shaken to dry, and absorbed of water droplets using absorbent paper. A histochemical stroke circle was used to surround the tissue. Inactivation of endogenous enzyme activity was conducted as follows: the slices were placed in a wet box, added with a small amount of distilled water, and then added with 3% hydrogen peroxide dropwise (one drop or 50 µL of A in the secondary antibody kit; see the instructions for the secondary antibody for details). This was incubated for 10 minutes at room temperature. The slices were then washed with PBS for 3 minutes × 3 times, followed by distilled water for 3 minutes. To block non-specific sites: the water was spun dry, the water beads were absorbed, goat serum blocking solution was added (a drop of secondary antibody kit B or 50 µL), and the slides were placed in a humid box and left at room temperature for 10 minutes. Then, the blocking solution was shaken off, the primary antibody was added dropwise to cover the tissue, and the tissue was placed in a humid chamber at 4°C overnight. For the concentration of primary antibody, refer to the instruction manual of the antibody; antibodies were diluted in advance, divided into packaging, and then stored in a refrigerator at -20°C. The dose of the primary antibody differed for each tablet depending on the size of the tissue. Each sample of cancer tissue occupied approximately 20 µL in volume. The next day, the wet box was removed from the 4°C refrigerator, rewarmed at room temperature for 30 minutes, washed with PBS for 3 minutes × 3 times, and then washed with distilled water for 2 minutes. The water was then drained, the water droplets were absorbed, and either one drop or 50 µL of the secondary antibody was added (Secondary Antibody Kit C) dropwise before incubation was carried out at room temperature for 10 minutes. This was followed by washing with PBS for 3 minutes × 3 times, followed by distilled water for 3 minutes. Then, one drop or 50 µL of streptavidin-peroxidase solution (secondary antibody kit D) was added to each slide before they were incubated at room temperature for 10 minutes. They were then washed with PBS for 3 minutes × 3 times, followed by rinsing with distilled water for 3 minutes. Either two drops or 100 µL of DAB solution was then added to each slide before they were observed under a microscope for 3–10 minutes. The dyeing was terminated when appropriate before being rinsed off with tap water. Hematoxylin counterstaining was conducted for 1–2 minutes. When the nuclei turned blue, the dyeing process was stopped and the slide was rinsed with tap water or PBS (reverse blue with 0.5% ammonia). The slides were then placed in 1% hydrochloric acid alcohol for 3 seconds, followed by rinsing with tap water for 3 min. Slide dehydration was conducted as follows, in chronological order: 70% alcohol for 1 minute, 80% alcohol for 1 minute, 95% alcohol for 2 minutes, absolute ethanol for 4 minutes, xylene 1 for 3 minutes, and xylene 2 for 3 minutes. After drying the slides, add an appropriate amount of neutral gum was added to cover the slides. The slides were then covered with a cover glass to remove any air bubbles. They were finally left to dry before photos were captured under a microscope.
+
+# References
+
+1. Zhao J-t. Free iliac skin flap transplantation by anastomosing the fourth lumbar blood vessel. *Plastic and reconstructive surgery* **77**, 836–842 (1986).
+
+2. Simonacci F, Bertozzi N, Grieco MP, Grignaffini E, Raposio E. Procedure, applications, and outcomes of autologous fat grafting. *Annals of Medicine and Surgery* **20**, 49–60 (2017).
+
+3. Bellini E, Grieco MP, Raposio E. The science behind autologous fat grafting. *Annals of medicine and surgery* **24**, 65–73 (2017).
+
+4. Hashimoto I, et al. Development of skin flaps for reconstructive surgery: random pattern flap to perforator flap. *The Journal of Medical Investigation* **63**, 159–162 (2016).
+
+5. Rifkin WJ, et al. Achievements and challenges in facial transplantation. *Annals of Surgery* **268**, 260–270 (2018).
+
+6. Henkel J, et al. Bone regeneration based on tissue engineering conceptions—a 21st century perspective. *Bone research* **1**, 216–248 (2013).
+
+7. Choi JH, et al. Adipose tissue engineering for soft tissue regeneration. *Tissue Engineering Part B: Reviews* **16**, 413–426 (2010).
+
+8. Weitman E, Cuzzone D, Mehrara BJ. Tissue engineering and regeneration of lymphatic structures. *Future Oncology* **9**, 1365–1374 (2013).
+
+9. Seol YJ, Kang HW, Lee SJ, Atala A, Yoo JJ. Bioprinting technology and its applications. *European Journal of Cardio-Thoracic Surgery* **46**, 342–348 (2014).
+
+10. He Y, Gu Z, Xie M, Fu J, Lin H. Why choose 3D bioprinting? Part II: methods and bioprinters. *Bio-Design and Manufacturing* **3**, 1–4 (2020).
+
+11. Thakor J, et al. Engineered hydrogels for brain tumor culture and therapy. *Bio-Design and Manufacturing* **3**, 203–226 (2020).
+
+12. Zhao X, et al. Photocrosslinkable Gelatin Hydrogel for Epidermal Tissue Engineering. *Advanced Healthcare Materials* **5**, 108–118 (2016).
+
+13. Sun X, et al. Electrospun Photocrosslinkable Hydrogel Fibrous Scaffolds for Rapid In Vivo Vascularized Skin Flap Regeneration. *Advanced Functional Materials* **27**, 1604617 (2017).
+
+14. Chen H, et al. Quickly promoting angiogenesis by using a DFO-loaded photo-crosslinked gelatin hydrogel for diabetic skin regeneration. *Journal of Materials Chemistry B* **4**, 3770–3781 (2016).
+
+15. Mohammadpour Z, Kharaziha M, Zarrabi A. 3D-Printing of Silk Nanofibrils Reinforced Alginate for Soft Tissue Engineering. *Pharmaceutics* **15**, 763 (2023).
+
+16. Peng LH, et al. Self‐Adaptive All‐In‐One Delivery Chip for Rapid Skin Nerves Regeneration by Endogenous Mesenchymal Stem Cells. *Advanced Functional Materials* (2020).
+
+17. Hamad GG, Curet M. Minimally invasive surgery. Elsevier (2010).
+
+18. Mackenzie L, Banerjee A. Minimally invasive direct restorations: a practical guide. *British dental journal* **223**, 163–171 (2017).
+
+19. Xie M, et al. Electro-Assisted Bioprinting of Low-Concentration GelMA Microdroplets. *Small* **15**, 1804216 (2019).
+
+20. Xie M, et al. Thermo-sensitive Sacrificial Microsphere-based Bioink for Centimeter-scale Tissue with Angiogenesis. *International Journal of Bioprinting* **1** (2022).
+
+21. Xie M, et al. In situ 3D bioprinting with bioconcrete bioink. *Nature Communications* **13**, 3597 (2022).
+
+22. Sun M, Sun X, Wang Z, Guo S, Yu G, Yang H. Synthesis and Properties of Gelatin Methacryloyl (GelMA) Hydrogels and Their Recent Applications in Load-Bearing Tissue. *Polymers* **10** (2018).
+
+23. Gao Q, et al. 3D printing of complex GelMA-based scaffolds with nanoclay. *Biofabrication* **11**, 035006 (2019).
+
+24. Van Den Bulcke AI, Bogdanov B, De Rooze N, Schacht EH, Cornelissen M, Berghmans H. Structural and Rheological Properties of Methacrylamide Modified Gelatin Hydrogels. *Biomacromolecules* **1**, 31–38 (2000).
+
+25. Sun Y, Yu K, Gao Q, He Y. Projection-based 3D bioprinting for hydrogel scaffold manufacturing. *Bio-Design and Manufacturing* **5**, 633–639 (2022).
+
+26. Yu K, et al. Printability during projection-based 3D bioprinting. *Bioactive Materials* **11**, 254–267 (2022).
+
+27. Yuan S, et al. Modeling the printability of photocuring and strength adjustable hydrogel bioink during projection based 3D bioprinting. *Biofabrication* (2020).
+
+28. Smirnova I, Gurikov P. Aerogel production: Current status, research directions, and future opportunities. *The Journal of Supercritical Fluids* **134**, 228–233 (2018).
+
+29. Luo Z, et al. Biodegradable gelatin methacryloyl microneedles for transdermal drug delivery. *Advanced healthcare materials* **8**, 1801054 (2019).
+
+30. Williams RC. A method of freeze-drying for electron microscopy. *Experimental Cell Research* **4**, 188–201 (1953).
+
+31. Khan MR, Sadiq MB. Importance of gelatin, nanoparticles and their interactions in the formulation of biodegradable composite films: A review. *Polymer Bulletin* (2020).
+
+32. Ranasinghe R, Wijesekara W, Perera P, Senanayake S, Pathmalal M, Marapana R. Functional and bioactive properties of gelatin extracted from aquatic bioresources–A review. *Food Reviews International* (2020).
+
+33. Rigueto CVT, Nazari MT, Massuda LÁ, Ostwald BEP, Piccin JS, Dettmer A. Production and environmental applications of gelatin-based composite adsorbents for contaminants removal: A review. *Environmental Chemistry Letters* (2021).
+
+34. Yoon HJ, et al. Cold Water Fish Gelatin Methacryloyl Hydrogel for Tissue Engineering Application. *PLOS ONE* **11**, e0163902-e0163902 (2016).
+
+35. Zhao X, et al. Injectable Stem Cell-Laden Photocrosslinkable Microspheres Fabricated Using Microfluidics for Rapid Generation of Osteogenic Tissue Constructs. *Advanced Functional Materials* **26**, 2809–2819 (2016).
+
+36. Sterodimas A, de Faria J, Nicaretta B, Pitanguy I. Tissue engineering with adipose-derived stem cells (ADSCs): current and future applications. *Journal of Plastic, Reconstructive & Aesthetic Surgery* **63**, 1886–1892 (2010).
+
+37. Ding X, et al. Asperosaponin VI stimulates osteogenic differentiation of rat adipose-derived stem cells. *Regenerative therapy* **11**, 17–24 (2019).
+
+38. Caligiuri G. CD31 as a therapeutic target in atherosclerosis. *Circulation Research* **126**, 1178–1189 (2020).
+
+39. Rouse RV. CD31: a new marker for vascular neoplasia. *Advances in Anatomic Pathology* **1**, 16–20 (1994).
+
+40. Zhang Z, Gan Q, Han J, Tao Q, Qiu WQ, Madri JA. CD31 as a probable responding and gate-keeping protein of the blood–brain barrier and the risk of Alzheimer’s disease. *Journal of Cerebral Blood Flow & Metabolism* (2023).
+
+# Supplementary Files
+
+- [11S1XXXXXXXXXXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/78313acf89070674e9362c5d.mp4)  
+  Video S1
+
+- [11S2XXXXXXXXXXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/56a95cbb0eaaa59add1e8f14.mp4)  
+  Video S2
+
+- [11S3XXXXXXXXXXXXXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/83c62f7b161d3344677255ca.mp4)  
+  Video S3
+
+- [11S4XXXXXXXXXXXXXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/9294dc644384320902655026.mp4)  
+  Video S4
+
+- [11S5XXXXXXXXXXXXXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/87bd56d3dbbe6e0ab0a40287.mp4)  
+  Video S5
+
+- [11S6XXXXXXXXXBXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/149080c684b31ac7f22437f8.mp4)  
+  Video S6
+
+- [11S7XXXXXXXXXXXXXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/83d1ee29786350c9741cc3cb.mp4)  
+  Video S7
+
+- [11S8XXXXXXXXXXXXXXX.mp4](https://assets-eu.researchsquare.com/files/rs-3208364/v1/944958e95fac16862b25debe.mp4)  
+  Video S8
+
+- [Supplementaryinformationv2.docx](https://assets-eu.researchsquare.com/files/rs-3208364/v1/7bad709ea6733869ce39e32d.docx)  
+  Supplementary information
+
+- [SupplementaryPresentation.pptx](https://assets-eu.researchsquare.com/files/rs-3208364/v1/42c9266cec03e75416b3e3a1.pptx)  
+  Supplementary Presentation

1a6984e310935268a34427031076405a230eba7a6126a8bf2ab3fcbc999e864f/preprint/images_list.json

@@ -0,0 +1,50 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Migration of small intestinal IL-7Rhigh CD8low \u03b3\u03b4 T17 cells to meninges exacerbates SAE.\na, b Flow cytometry (FCM) plots and graphs showing CD45+ \u03b3\u03b4 T cells in the small intestine and meninges of sham and cecum ligation and puncture (CLP)-treated mice. (n = 6\u20137). c FCM plots and graphs showing IL-17A+ \u03b3\u03b4 T17 cells (CD45+ CD3+ \u03b3\u03b4 TCR+ IL-17A+) in the meninges of sham and CLP-treated mice. (n = 6\u20137). d, e FCM plots and graphs showing CD45+ IL-17A+ \u03b3\u03b4 T cells (CD45+ IL-17A+ CD3+ \u03b3\u03b4 TCR+) in the small intestine and meninges of sham and CLP-treated mice. (n = 6\u20137). f, g FCM plots and graphs showing IL-7Rhigh CD8low \u03b3\u03b4 T17 cells and IL-7Rlow CD8high \u03b3\u03b4 T17 cells in the small intestine and meninges of sham and CLP-treated mice. (n = 6\u20137). h Treatment schedule. Kaede transgenic mice were treated with a 405 nm laser and analyzed after sham or CLP surgery. i Graph of the Kaede red+ IL-7Rlow CD8high and IL-7Rhigh CD8low \u03b3\u03b4 T17 cell in the meninges of septic mice (n = 6). j, k Golgi staining image of CA1 region and dendritic spine density in septic mice (n = 6). 6X, Scale bar = 200 \u03bcm. 12X, Scale bar = 100 \u03bcm, Primary and secondary, Scale bar = 10\u03bcm. l, m Transmission electron microscopy (TEM) of synapses and mitochondria in the CA1 region of septic mice (n = 4), 15X, Scale bar = 500nm. 20X, Scale bar = 2.0\u03bcm. n, o Y-maze tests (n = 12\u201313). p\u2013r MWM tests (n = 12\u201313). Data are shown as mean \u00b1 SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant. Statistical analyses: unpaired t-tests (b, c, g, i), Welch\u2019s t-test (e), ANOVA with \u0160\u00edd\u00e1k\u2019s test (k, m, o, q, r).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "IL-17A from \u03b3\u03b4 T cells damages microglial mitochondria and activates cGAS-STING-C1q pathway.\na, b JC-1 staining shows mitochondrial membrane potential in IL-17A and LPS-treated BV2 cells (n = 6). c Transmission electron microscopy (TEM) images of mitochondria of BV2 cells. 6X, Scale bar = 5.0\u03bcm. 20X, Scale bar = 2.0\u03bcm, 40X, Scale bar = 0.5\u03bcm. Western blot (d) and quantification (e) of cGAS and STING in BV2 cells exposed to IL-17A and LPS (n = 6). f Schematic of the transwell co-culture of primary \u03b3\u03b4 T cells and microglia. g TEM images of primary microglia. 6X, Scale bar = 5.0\u03bcm. 20X, Scale bar = 2.0\u03bcm, 40X, Scale bar = 0.5\u03bcm. Western blot (h, j) and quantification (i, k) of cGAS, STING and C1q in primary microglia (n = 6). l Treatment schedule. Stereotactic injection of IL-17A+-Adv into the hippocampus, followed by behavioral assessments. Western blot (m) and quantification (n) of cGAS and STING in hippocampus (n = 6). Western blot (o) and quantification (p) of C1q, PSD95 and SYN in hippocampal synaptic proteins (n = 6). q, r TEM images of Adv-treated mice (n = 4), 20X, Scale bar = 500nm. 15X, Scale bar = 2.0\u03bcm. Western blot (s) and quantification (t) of C1q, PSD95 and SYN in hippocampal synaptic proteins (n = 6). Western blot (u) and quantification (v) of cGAS and STING in the hippocampus (n = 6). 3D reconstructed imaging (w) and analysis (x) of microglia engulfing synapses (n = 6). Scale bar, 5\u03bcm. Data are shown as mean \u00b1 SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant. Statistical tests: ANOVA with \u0160\u00edd\u00e1k\u2019s test (b, e, t, v, x), unpaired t-tests (i, k, n, p, r).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Increased STING promotes C1q \u2013 dependent synaptic pruning.\na Treatment schedule. H151 was administered intraperitoneally before cecum ligation and puncture (CLP), with brain collection 1 day post-CLP and behavioral testing starting 7 days post-CLP. Western blot (b) and quantification (c) of STING in the hippocampus (n = 6). Western blot (d) and quantification (e) of C1q, PSD95 and SYN in hippocampal synaptic proteins (n = 6). 3D imaging (f) and analysis (g) of microglia engulfing synapses (n = 6). Scale bar, 5\u03bcm. h, i MWM tests. Western blot (j) and quantification (k) of STING and C1q of primary microglia (n = 6). l Treatment schedule. Hippocampal injection of cGAS+ adeno-associated virus (AAV); H151 administered 21 days later, followed by brain collection and behavioral testing. m Western blot analysis showing cGAS overexpression in the hippocampus. 3D imaging (n) and analysis (o) of microglia engulfing synapses (n = 6). Scale bar, 5\u03bcm. Western blot (p) and quantification (q) of cGAS and STING in hippocampus (n = 6). Western blot (r) and quantification (s) of C1q, PSD95 and SYN in hippocampal synaptic proteins (n = 6). t Treatment schedule. C1q neutralizing antibody or IgG injected into hippocampus 1 day before CLP. 3D imaging (u) and analysis (v) of microglia engulfing synapses (n = 6). Scale bar, 5\u03bcm. w, x Morris-water maze tests (n = 12). Western blot (y) and quantification (z) of C1q, PSD95 and SYN in hippocampal synaptic proteins (n = 6). Data are shown as mean \u00b1 SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant. Statistical tests: unpaired t-tests (c, e, g, i, k, o, v, x, z), ANOVA with \u0160\u00edd\u00e1k\u2019s test (q, s).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "The K150 site is crucial for STING ubiquitination in microglia.\na Western blot showing STING ubiquitination levels in primary microglia. b Western blot showing STING ubiquitination levels in mouse hippocampus. c Mass spectrometry identification of STING ubiquitination site K150. d Western blot analysis of STING ubiquitination levels in primary microglia transfected with mutant plasmids for each of the seven identified ubiquitination sites. Western blot (e) and quantification (f) of K150 in hippocampal of mice (n = 6). g Western blot showing STING ubiquitination levels in hippocampus 21 days after K150R+ adeno-associated virus (AAV) injection. Representative 3D reconstructed imaging (h) and quantitative analysis (i) of microglia engulfing synapses (n = 6). Scale bar, 5\u03bcm. Western blot (j) and quantification (k) of STING in hippocampus of mice received K150R+-AAV (n = 6). Western blot (l) and quantification (m) of C1q, PSD95 and SYN in hippocampal synaptic proteins of mice treated with K150R+-AAV (n = 6). n, o Representative transmission electron microscopy (TEM) images of synapse and mitochondria in the CA1 region of mice treated with K150R+-AAV (n = 4), 15X, Scale bar = 500nm. 20X, Scale bar = 2.0\u03bcm. Representative images (p) and statistical analysis (q) of the Morris water maze (n = 12). Representative images (r) and statistical analysis (s) of the Y-maze tests (n = 12). Representative images (t) and statistical analysis (u) of the open field tests (n = 12). v Statistical analysis of the novel object recognition tests (n = 12). Data are shown as mean \u00b1 SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant. Statistical tests: two-sided Student\u2019s unpaired t-tests (f, i, k, m, o, q, s, u) and multiple unpaired t-tests (v).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "4-Octyl itaconate alleviates SAE.\na Heatmap of top 20 upregulated differentially expressed genes (DEGs) in small intestine (n = 3). b, c Western blot of ACOD1 in small intestines (n = 6). d Treatment schedule. e Kaplan\u2013Meier survival analysis. (n = 7\u201317). f, g Western blot of ZO-1, MUC2 and Occludin in small intestine (n = 6). h Immunofluorescence images of ZO-1, MUC2, and Occludin of small intestines. i Flow cytometry (FCM) plots of IL-17A+ live \u03b3\u03b4 T cells in the meninges. j FCM of Kaede red+ IL-7Rhigh CD8low \u03b3\u03b4 T17 cells in the meninges (n = 6). k Transmission electron microscopy (TEM) images of mice (n = 4), 15X, Scale bar = 500nm. 20X, Scale bar = 2.0\u03bcm. l 3D imaging and analysis of microglia engulfing synapses (n = 6). Scale bar, 5\u03bcm. m Golgi staining images of mice (n = 6). 6X, Scale bar = 200\u03bcm. 12X, Scale bar = 100\u03bcm, Primary and secondary, Scale bar = 10\u03bcm. n Morris-water maze tests (n = 12). o, p Western blot of cGAS and STING in hippocampus (n = 6). q Western blot of C1q, PSD95 and SYN in hippocampal synaptic proteins (n = 6). r Enzyme-linked immunosorbent assay (ELISA) analysis of IL-17A (n = 6). s JC-1 staining in BV2 cells (n = 6). t, u Western blot of cGAS and STING of BV2 cells (n = 6). v, w Western blot of C1q of BV2 cells (n = 6). x Volcano plot of hippocampal transcriptome from cecum ligation and puncture (CLP) and control mice. y Western blot of K150 in hippocampus (n = 6). z Western blot of STING ubiquitination of hippocampus. Data are shown as mean \u00b1 SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant. Statistical tests: unpaired t-tests (c, j\u2013n, p\u2013s, u, w, y), Log-rank test (e), ANOVA with \u0160\u00edd\u00e1k\u2019s test (g).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "Acod1 expression in microglia\na Treatment schedule. Hippocampal stereotactic injection of Acod1\u2013-adeno-associated virus (AAV); brain collection 22 days post-infection. Behavioral assessments were conducted 28 days post-infection. Western blot (b) and quantification (c) of cGAS and STING in hippocampus of mice treated with control or Acod1\u2013-AAV (n = 6). d, e Representative transmission electron microscopy (TEM) of synapse and mitochondria in CA1 region in mice treated with control or Acod1\u2013-AAV (n = 4). Western blot (f) and quantification (g) of cGAS and STING in hippocampus of mice received control or Acod1\u2013-AAV (n = 6). Representative 3D reconstructed imaging (h) and quantitative analysis (i) of microglia engulfing synapses (n = 6). Scale bar, 5\u03bcm. Western blot (j) and quantification (k) of K150 in the hippocampus of mice treated with control or Acod1\u2013-AAV (n = 6). l Western blot image of STING ubiquitination levels in mouse hippocampus. Data are shown as mean \u00b1 SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant. P values were calculated using two-sided Student\u2019s unpaired t-tests (c, e, g, i, k).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

1a6984e310935268a34427031076405a230eba7a6126a8bf2ab3fcbc999e864f/preprint/preprint.md

@@ -0,0 +1,292 @@
+# Abstract
+
+Sepsis is a severe global health issue with high mortality rates, and sepsis-associated encephalopathy (SAE) further exacerbates this risk. While recent studies have shown the migration of gut immune cells to the lungs after sepsis, their impact on the central nervous system remains unclear. Our research demonstrates that sepsis could induce the migration of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells from the small intestine to the meninges, where they secrete IL-17A, impairing mitochondrial function in microglia and activating the cGAS-STING-C1q pathway. This process is accompanied by inhibited ubiquitination of STING at the K150 site, resulting in STING accumulation and increased release of C1q-tagged hippocampal synapses, which are subsequently pruned by activated microglia. Importantly, 4-Octyl itaconate mitigates the excessive synaptic pruning by inhibiting γδ T17 cell migration and promoting STING ubiquitination, thereby alleviating SAE. Our findings reveal a novel mechanism of synaptic pruning by microglia via the cGAS-STING-C1q pathway, emphasize the critical role of gut-derived γδ T17 cell migration to the meninges in SAE, and highlight the importance of STING ubiquitination in modulating C1q-mediated excessive synaptic pruning.
+
+Health sciences/Diseases/Infectious diseases/Bacterial infection  
+Biological sciences/Immunology/Infectious diseases/Bacterial infection  
+Sepsis-associated encephalopathy  
+Gut-brain axis  
+Synaptic pruning  
+Microglia  
+cGAS-STING
+
+# Introduction
+
+Sepsis-associated encephalopathy (SAE) is a severe complication of sepsis, characterized by diffuse or multifocal neurological dysfunction, and is a leading cause of increased mortality among ICU patients. Although SAE is often considered reversible, approximately 40% of patients experience long-term neurological symptoms, including memory impairment, depression, anxiety, and cognitive dysfunction, especially in cases of severe sepsis<sup>1</sup>. The lack of in-depth understanding of the mechanisms and effective therapy for SAE remains a big challenge in clinical practice.
+
+The gut is considered the “engine” driving sepsis and multiple organ dysfunction, a concept known as “gut-origin sepsis”<sup>2</sup>. As a critical immune organ, the gut contains diverse lymphocyte subsets, including γδ T cells and Th17 cells, primarily located in the small intestinal lamina propria (SI-LP). These cells are crucial for maintaining mucosal immune balance and intestinal barrier integrity<sup>3</sup>. Although γδ T cells comprise a small fraction of total T cells, they are essential for epithelial integrity, tissue repair, host homeostasis, and pathogen defense<sup>3</sup>. When the intestinal barrier is compromised, leading to bacterial invasion and inflammation, γδ T cells could differentiate into interleukin-17A (IL-17A)-producing γδ T17 cells, which exacerbates the inflammatory response<sup>4</sup>.
+
+T cells in the meninges can secrete IL-17A, which could affect border-associated macrophages, reduces cerebral blood flow, and impairs cognition<sup>5</sup>. Similarly, γδ T17 cells can migrate from the SI-LP to the meninges, worsening the acute ischemic brain injury<sup>6</sup>. IL-17A secreted from γδ T17 cells could activate microglia and contribute to lipopolysaccharide (LPS)-induced neuroinflammation and cognitive deficits<sup>7</sup>. We recently showed that IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells could migrate from the gut to the lungs and exacerbate cecal ligation and puncture (CLP)-induced acute lung injury and mortality via secreting IL-17A<sup>8</sup>. However, it remains unclear whether gut-derived IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 could migrate to the central nervous system (CNS) and exacerbate SAE via IL-17A.
+
+Synaptic pruning, a natural process of synapse elimination, is crucial for proper neural circuit formation, but aberrant pruning can lead to CNS disorders such as schizophrenia, anxiety, and autism<sup>9</sup>. Aberrant synaptic pruning affects neural signal transmission and network functionality, which is primarily mediated by microglial phagocytosis<sup>10</sup>. Complement-mediated synaptic pruning by microglia is closely linked to the progressive synaptic loss in LPS-induced depression<sup>11</sup>. Furthermore, sepsis-activated microglia could release C1q, which tags synapses for abnormal pruning and thereby exacerbates peritoneal contamination and infection-induced SAE<sup>12</sup>. However, the connection between C1q-mediated synaptic pruning and IL-17A and its potential mechanisms remain to be clarified.
+
+This study demonstrated that the migration of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells from the small intestine to the meninges during sepsis was a critical factor contributing to SAE. IL-17A mediated mitochondrial damage in microglia and activated the cGAS-STING pathway, which was accompanied by inhibited ubiquitination of STING at the K150 site, resulting in STING accumulation. Increased levels of STING promoted the release of C1q and thus enhanced C1q-tagged hippocampal synapses, which exacerbated microglia-mediated synaptic pruning and SAE. 4-Octyl itaconate (4-OI), an itaconate derivative with potent anti-inflammatory properties<sup>13,14</sup>, inhibited IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cell migration, increased the ubiquitination of STING at the K150 site in microglia, reduced STING expression, and subsequently decreases C1q-tagged synapses and excessive synaptic pruning, thereby ameliorating SAE.
+
+# Results
+
+Gut-derived IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells migrate to the meninges and exacerbate SAE
+
+First, we investigated whether small intestinal γδ T cells could migrate to CNS after sepsis, we performed flow cytometry (FCM) analysis following CLP. We observed significant reduction of γδ T cells in the small intestine, with a concurrent increase in the meninges (Fig. 1a, b), while γδ T cells in the brain parenchyma remained low (Extended Data Fig. 1a). These findings suggest that the meninges serve as the primary CNS destination for migrating small intestinal γδ T cells after sepsis. Notably, CLP induced elevated production of IL-17A by γδ T cells in the meninges, with over 50% of these cells identified as γδ T17 cells (Fig. 1c–e). Building on previous findings that IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells in small intestine could migrate to the lungs after sepsis<sup><span citationid="CR8" class="CitationRef">8</span></sup>, we observed a similar elevation of this subset in the meninges after CLP (Fig. 1f, g). Therefore, we hypothesized that small intestinal IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells could directly migrate to the meninges following sepsis. To investigate this, the small intestine of Kaede-tg mice underwent localized irradiation to track the migration of small intestinal lymphocytes (Fig. 1h). We found that the most IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells in the meninges originated from the small intestine (Fig. 1i).
+
+We further investigate the impact of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells on SAE, γδ T cell-specific knockout mouse models targeting <em>Cd8a</em> and <em>Il7r</em> were used. Specifically, mice were developed by crossing <em>Trdc</em><sup>CreERT<span citationid="CR2" class="CitationRef">2</span></sup> mice with <em>Il7r</em><sup>flox/flox</sup> and <em>Cd8a</em><sup>flox/flox</sup> mice, leading to a conditional knockout of <em>Il7r</em> and <em>Cd8a</em> in γδ T cells, respectively. In particular, <em>Trdc</em><sup>CreERT<span citationid="CR2" class="CitationRef">2</span></sup><em>Il7r</em><sup>flox/flox</sup> mice exhibited a significant reduction in γδ T17 cells in the meninges compared to wild-type (WT) and <em>Trdc</em><sup>CreERT<span citationid="CR2" class="CitationRef">2</span></sup><em>Cd8a</em><sup>flox/flox</sup> mice (Extended Data Fig. 1c). Additionally, Golgi staining showed that CLP-induced dendritic spine loss and disorganization were reversed in <em>Trdc</em><sup>CreERT<span citationid="CR2" class="CitationRef">2</span></sup><em>Il7r</em><sup>flox/flox</sup> mice (Fig. 1j, k). Furthermore, transmission electron microscopy (TEM) revealed attenuation of synaptic vesicle loss, postsynaptic density thinning, and mitochondrial damage in <em>Trdc</em><sup>CreERT<span citationid="CR2" class="CitationRef">2</span></sup><em>Il7r</em><sup>flox/flox</sup> mice (Fig. 1l, m).
+
+Behavioral assessments highlighted significant cognitive and memory impairments in CLP-induced SAE, including reduced exploration in Y-maze (Fig. 1n, o) and Morris-water maze (MWM) tests (Fig. 1p–r), increased anxiety-like behavior in open field tests (OFT) (Extended Data Fig. 1f, g), and decreased time spent exploring the novel object in the novel object recognition test (NORT) (Extended Data Fig. 1h). However, these behavioral deficits were markedly alleviated in <em>Trdc</em><sup>CreERT<span citationid="CR2" class="CitationRef">2</span></sup><em>Il7r</em><sup>flox/flox</sup> mice, suggesting that IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells could exacerbate cognitive dysfunction in SAE.
+
+γδT cells enhance microglial cGAS-STING expression and exacerbate synaptic pruning via C1q release
+
+Having established that the migration of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells from the intestine to the meninges could exacerbate SAE, we next explored the molecular mechanisms involved. We found that more than 50% of IL-17A in the meninges was produced by γδ T17 cells (Fig. 1d, e). IL-17A from γδ T17 cells is known to activate microglia, contributing to neuroinflammation and cognitive dysfunction<sup><span citationid="CR7" class="CitationRef">7</span></sup>. Although IL-17A has been implicated in mitochondrial degradation via enhanced mitophagy in bronchial fibroblasts in severe asthma<sup><span citationid="CR15" class="CitationRef">15</span></sup>, its effect on microglial mitochondrial integrity remains unclear. To evaluate the impact of IL-17A, produced by γδ T cells, on microglial function, we treated BV2 cells with recombinant IL-17A (100 ng ml<sup>–<span citationid="CR1" class="CitationRef">1</span></sup>) and LPS (1000 ng ml<sup>–<span citationid="CR1" class="CitationRef">1</span></sup>). This treatment exhibited decreased mitochondrial membrane potential and mitochondrial swelling, as indicated by JC-1 staining and TEM (Fig. 2a–c), suggesting mitochondrial damage. Mitochondrial dysfunction can release mitochondrial DNA (mtDNA), triggering the cGAS-STING pathway, a key intracellular DNA-sensing mechanism involved in host defense and inflammation<sup><span citationid="CR16" class="CitationRef">16</span></sup>. Activation of the cGAS-STING pathway not only triggers inflammatory responses but also exacerbates various CNS injuries, including traumatic brain injury, spinal cord injury, subarachnoid hemorrhage, and hypoxic-ischemic encephalopathy<sup><span citationid="CR17" class="CitationRef">17</span></sup>.We confirmed elevated cGAS and STING expression in BV2 cells following IL-17A and LPS treatment (Fig. 2d). Importantly, we found that mitochondrial damage was necessary for cGAS-STING upregulation, as mitochondria-depleted BV2<sup>ρ0</sup> cells did not exhibit increased cGAS and STING expression (Extended Data Fig. 1i–k). Consistent with these findings, silencing IL-17A in primary γδ T cells using siRNA reduced mitochondrial damage and decreased cGAS, STING, and C1q levels in primary microglia (Fig. 2f–k).
+
+<em>In vivo</em>, hippocampal overexpression of IL-17A increased cGAS and STING expression (Fig. 2l–n) and impaired cognitive function (Extended Data Fig. 2a–g). Conversely, neutralization of IL-17A improved cognition in septic mice (Extended Data Fig. 2h–n). As lowering IL-17A levels reduced C1q expression in microglia (Fig. 2j, k). Given that C1q can bind to synapses and mediate synaptic pruning in a murine polymicrobial sepsis model<sup><span citationid="CR12" class="CitationRef">12</span></sup>, we examined synaptic proteins in the hippocampus. IL-17A overexpression elevated C1q and reduced the levels of postsynaptic density protein 95 (PSD95) and synaptophysin (SYN), indicating that IL-17A could promote increased synaptic pruning by increasing C1q levels (Fig. 2o–r). However, this synaptic damage was reversed in <em>Trdc</em><sup>CreERT<span citationid="CR2" class="CitationRef">2</span></sup><em>Il7r</em><sup>flox/flox</sup> mice (Fig. 2s, t), which also exhibited lower expression of cGAS and STING in the hippocampus (Fig. 2u, v). Confocal imaging and 3D reconstruction using Imaris software showed that microglia in the hippocampus of septic mice displayed an activated phenotype, with increased phagocytosis of PSD95, linking microglial activation to synaptic protein loss (Fig. 2w, x).
+
+Considering that STING promotes neuroinflammation<sup><span citationid="CR18" class="CitationRef">18</span></sup> and contributes to synaptic impairments, we investigated whether pharmacological inhibition of STING could reverse the associated neuroinflammatory responses and synaptic deficits (Fig. 3a–c). Treatment with the STING inhibitor H151 (750 nM) reduced C1q, increased the expression of PSD95 and SYN (Fig. 3d, e), and mitigated synaptic pruning (Fig. 3f, g), which led to improved cognitive outcomes (Fig. 3h, i and Extended Data Fig. 3a–e). H151 also reversed synaptic impairments caused by cGAS overexpression in microglia (Fig. 3j–s, Extended Data Fig. 3f–l). Additionally, direct C1q neutralization reduced microglial phagocytosis of synaptic proteins, resulting in improved cognition and restoring synaptic integrity (Fig. 3t–z). In line with this, microglial depletion with PLX3397 similarly reduced synaptic pruning and improved behavior deficts in septic mice (Extended Data Fig. 4e–k). These findings demonstrate that synaptic pruning after sepsis is closely related to the phagocytosis of C1q-labeled synapses by microglias. IL-17A produced by γδ T cells could drive C1q release via the cGAS-STING pathway, representing a key therapeutic target to reduce cognitive impairment in sepsis.
+
+K150-mediated STING ubiquitination prevents cognitive dysfunction
+
+Given that sepsis could increase STING expression and contributes to C1q-mediated synaptic pruning in our findings, we investigated whether the ubiquitin-proteasome system (UPS), a key pathway for protein degradation and maintenance of protein homeostasis<sup><span citationid="CR19" class="CitationRef">19</span></sup>, could regulate STING accumulation. We hypothesized that the ubiquitination of STING could play an important role in the pathogenesis of SAE. As expected, we found that exposure of primary microglia to LPS resulted in a significant reduction in STING ubiquitination, which was also observed in the hippocampus of CLP mice (Fig. 4a, b). Mass spectrometry identified several potential ubiquitination sites on STING, with lysine 150 (K150) emerging as critical, mutation of K150 markedly reduced STING ubiquitination in primary microglia (Fig. 4c, d). A specific antibody targeting ubiquitinated K150 confirmed a reduction in the hippocampus of septic mice (Fig. 4e, f).
+
+To further explore the role of K150, we generated an adeno-associated virus (AAV) vector carrying the K150R mutation and injected it into the hippocampus of <em>Cx3cr1</em><sup>Cre</sup> mice to mutate K150 site specifically in microglia. This mutation led to reduced STING ubiquitination (Fig. 4g), increased microglial phagocytosis of PSD95 (Fig. 4h, i), and elevated STING protein levels (Fig. 4j, k). Consequently, mice with the K150R mutation in microglia showed enhanced synaptic pruning, evidenced by increased C1q and decreased levels of PSD95 and SYN (Fig. 4l, m). Furthermore, microglial K150R-mutant mice exhibited severe synaptic and mitochondrial damage (Fig. 4n, o), along with pronounced learning and memory deficits in behavioral tests (Fig. 4p–v). These findings demonstrate that K150-mediated ubiquitination in microglia is essential for STING degradation and prevention of cognitive dysfunction.
+
+We further explored the ubiquitination process of STING. We used UbiBrowser to predict potential E3 ubiquitin ligases for STING and identified RNF5 as a key candidate (Extended Data Fig. 5a). Co-immunoprecipitation confirmed a direct interaction between RNF5 and STING, which decreased following LPS stimulation or CLP-induced sepsis (Extended Data Fig. 5b–f). Furthermore, RNF5 knockdown in microglia significantly reduced STING ubiquitination (Extended Data Fig. 5g), while RNF5 overexpression increased STING ubiquitination, reduced C1q levels, increased synaptic proteins, and improved cognitive function in septic mice (Extended Data Fig. 5h–t). Importantly, the K150 mutation disrupted the interaction between STING and RNF5, highlighting the dependence of RNF5-mediated STING ubiquitination on K150 (Extended Data Fig. 5u). Together, these findings underscore the importance of K150-mediated STING ubiquitination in mcroglia in preventing excessive synaptic pruning and cognitive dysfunction, with RNF5 serving as a key regulator of this process.
+
+4-OI Reduces γδ T17 Cell Migration and Promotes K150-Mediated STING Ubiquitination
+
+To explore potential therapeutic strategies for SAE, we considered the role of small intestinal γδ T17 cell migration and the strong correlation between sepsis, gut barrier dysfunction, and systemic inflammatory response<sup><span citationid="CR20" class="CitationRef">20</span></sup>. Proteomic analysis of the small intestine from septic mice revealed significant upregulation of aconitate decarboxylase 1 (ACOD1), which produces itaconate, known for its anti-inflammatory effects<sup><span citationid="CR13" class="CitationRef">13</span></sup> (Fig. 5a–c). Hence we further investigated the protective effects of 4-octyl-itaconate (4-OI), a blood-brain barrier-permeable itaconate derivative, on sepsis-induced systemic inflammation.
+
+Notably, administration of 4-OI led to a marked reduction in pro-inflammatory cytokines, as demonstrated by multiplex cytokine analysis (Extended Data Fig. 6a, b), with improved survival rates and enhanced intestinal barrier integrity (Fig. 5d–h). Interestingly, 4-OI significantly reduced the number of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells in the meninges, correlating with decreased migration from the small intestine (Fig. 5i, j). This reduction in migration was accompanied by mitigation of hippocampal synaptic and mitochondrial damage (Fig. 5k), reduced reactive oxygen species (ROS) levels, and enhanced antioxidant activity, as confirmed by dihydroethidium (DHE), malondialdehyde (MDA), and superoxide dismutase (SOD) assays, respectively (Extended Data Fig. 7a–d). Additionally, positron emission tomography-computed tomography (PET-CT) and laser speckle imaging demonstrated improved hippocampal glucose metabolism and restored sepsis-induced decrease in the cerebral blood flow after 4-OI treatment (Extended Data Fig. 8a, b).
+
+Moreover, 4-OI reduced microglial phagocytosis of PSD95 (Fig. 5l), increased dendritic spine density (Fig. 5m), and improved cognitive performance in the MWM, OFT, Y-maze and NORT (Fig. 5n; Extended Data Fig. 7e–i). Mechanistically, 4-OI reduced the expression of cGAS and STING, leading to decreased C1q and increased levels of PSD95 and SYN, indicating that 4-OI could attenuate C1q-mediated synaptic pruning (Fig. 5o–q). <em>In vitro</em>, 4-OI inhibited IL-17A release and mitigated mitochondrial dysfunction in BV2 cells (Fig. 5r, s). 4-OI also promoted microglial M2 polarization, enhancing the M2/M1 ratio in both <em>in vitro</em> and <em>in vivo</em> models while reducing the migration and phagocytic activity of LPS-stimulated microglia (Extended Data Fig. 9a–i). Furthermore, 4-OI inhibited STING activity in microglia (Extended Data Fig. 9g).
+
+Interestingly, transcriptomic analysis of the hippocampus from septic mice also showed upregulation of <em>Acod1</em> (Fig. 5x; Extended Data Fig. 10a). In the hippocampus, 4-OI enhanced K150-mediated STING ubiquitination and upregulated RNF5 expression (Fig. 5y, z; Extended Data Fig. 5v, w). These results confirm that 4-OI could protect against SAE not only by inhibiting γδ T17 cell migration from small intestine into the meninges, but also by promoting K150-mediated STING ubiquitination in the hippocampus.
+
+Silencing Acod1 in Microglia Aggravates Cognitive Dysfunction
+
+Given the increased expression of ACOD1 in both the intestine and hippocampus after sepsis, we further investigated its role in microglia and its impact on cognitive impairment in SAE. We silenced <em>Acod1</em> using <em>Acod1–</em>-AAV in <em>Cx3cr1</em><sup>Cre</sup> mice (Fig. 6a). This suppression significantly upregulated cGAS and STING expression in the hippocampus, leading to synaptic and mitochondrial damage (Fig. 6b–e). Silencing <em>Acod1</em> also elevated C1q levels and reduced the levels of PSD95 and SYN, indicating that <em>Acod1</em> suppression enhanced complement-mediated synaptic pruning (Fig. 6f, g). Correspondingly, microglial phagocytosis of PSD95 was notably increased following <em>Acod1</em> suppression, further confirming its involvement in excessive synaptic pruning (Fig. 6h, i).
+
+Consistent with the promotion of K150-mediated STING ubiquitination by 4-OI, <em>Acod1</em> suppression led to a significant reduction in K150-mediated STING ubiquitination (Fig. 6j–l), promoting cGAS-STING pathway activation and exacerbating synaptic injury (Fig. 6b–i). Taken together, these findings suggest that ACOD1 could regulate microglial function by promoting K150-mediated STING ubiquitination, thereby limiting cGAS-STING activation and preventing excessive synaptic pruning. Therefore, <em>Acod1</em> suppression in microglia could worsen hippocampal synaptic and mitochondrial dysfunction by impairing K150-mediated STING ubiquitination, activating the cGAS-STING-C1q pathway, and enhancing synaptic pruning, ultimately leading to cognitive dysfunction after sepsis.
+
+# Discussion
+
+Our latest study reveals that IL-7R<sup>high</sup> CD8<sup>low</sup> memory γδT17 cells, originating from the small intestine, could migrate to the lungs and contribute to acute lung injury following sepsis<sup><span citationid="CR8" class="CitationRef">8</span></sup>. Interestingly, this present research further demonstrates that the migration of small intestine-derived IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells to the meninges plays a pivotal role in exacerbating SAE, establishing a crucial link between intestinal γδT17 cells and central nervous system pathology. Our findings underscore the importance of gut-centric inter-organ interactions.
+
+Prior research has predominantly focused on gut dysbiosis and intestinal barrier dysfunction in SAE<sup><span citationid="CR21" class="CitationRef">21</span>, <span citationid="CR22" class="CitationRef">22</span></sup>. However, our study provides the first direct evidence of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cell migration playing a pathogenic role in SAE. Notably, similar mechanisms have been observed in hypertension, where IL-17-secreting T cells in meninges could contribute to CNS injury<sup><span citationid="CR5" class="CitationRef">5</span></sup>. Consistent with these findings, we observed a significant increase in γδ T17 cells in the meninges after CLP, along with an upregulation of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells. However, the number of γδ T cells in the brain parenchyma remained low in both CLP and sham-operated mice, suggesting a specific migration pattern to the meninges rather than the brain tissue.
+
+IL-17A, a pro-inflammatory cytokine known to exacerbates immune responses under inflammatory conditions<sup><span citationid="CR23" class="CitationRef">23</span></sup>, appears to be a key mediator in this “small intestine-derived γδ T17 cells induced SAE”. We found that over 50% of IL-17A-secreting immune cells in the meninges following sepsis were γδ T cells, significantly higher than that of 30% in the sham group. We further confirmed that IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells are the primary IL-17A source in the meninges after sepsis, as shown using <em>Trdc</em><sup>CreERT</sup><span citationid="CR2" class="CitationRef">2</span> <em>Il7r</em><sup>flox/flox</sup> and <em>Trdc</em><sup>CreERT</sup><span citationid="CR2" class="CitationRef">2</span> <em>Cd8a</em><sup>flox/flox</sup> transgenic mice. The IL-17A secreted by IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells activates microglia, triggering CNS inflammation and exacerbating SAE.
+
+Microglia, as CNS resident immune cells, can be neuroprotective or neurotoxic<sup><span citationid="CR24" class="CitationRef">24</span></sup>. In the CLP model of sepsis, activated microglia could exacerbate SAE by upregulation neuronal NAT10 expression<sup><span citationid="CR25" class="CitationRef">25</span></sup>, induction of hippocampal neuronal ferroptosis via the CXCL2/CXCR2 pathway<sup><span citationid="CR26" class="CitationRef">26</span></sup>, and aberrant synaptic pruning<sup><span citationid="CR27" class="CitationRef">27</span></sup>. In the peritoneal contamination and infection (PCI)-induced SAE model, microglial complement C1q-dependent synaptic pruning is shown to be able to worsen SAE<sup><span citationid="CR12" class="CitationRef">12</span></sup>, although the mechanisms underlying sepsis-induced C1q release remains unclear. Our study reveals that IL-17A-stimulated microglia upregulate C1q-dependent synaptic pruning through activation of the cGAS-STING pathway. C1q, a key initiator of the classical complement pathway, mediates synaptic pruning during both development and disease. We found that inhibiting STING with H151 or depleting microglia with PLX3397 significantly reduced C1q levels and synaptic pruning, suggesting that STING-mediated microglial C1q release contributes to the worsening SAE. Traditionally associated with antiviral immunity through the detection of cytosolic DNA and activation of type Ⅰ interferon responses, the cGAS-STING pathway has recently been implicated in neuroinflammatory diseases<sup><span citationid="CR18" class="CitationRef">18</span></sup>. In multiple sclerosis, STING activation enhances autophagy in neurons, increasing their susceptibility to glutamate-induced excitotoxicity, while STING inhibitors such as C176 or H151 have been shown to reduce neuroinflammatory damage<sup><span citationid="CR28" class="CitationRef">28</span></sup>. Furthermore, reducing cGAS-STING signaling has been found to inhibit glial cell activation in aging models<sup><span citationid="CR29" class="CitationRef">29</span></sup>.
+
+In exploring potential therapeutic interventions, we focused on itaconate, an immunomodulatory metabolite produced by the enzyme ACOD1. Elevated itaconate levels, dependent on ACOD1, have been observed in the blood during <em>Plasmodium</em> infection<sup><span citationid="CR30" class="CitationRef">30</span></sup>. Macrophages and myeloid cells produce itaconate under inflammatory stimulation, and microglia also express ACOD1 highly under pro-inflammatory conditions<sup><span citationid="CR31" class="CitationRef">31</span></sup>, which aligns with our findings. Our proteomic and transcriptomic analyses revealed significant upregulation of ACOD1 in both the small intestine and hippocampus after sepsis. Exogenous supplementation with itaconate and its derivative 4-OI has been shown to mitigate systemic inflammation and exert anti-inflammatory effects in the small intestine during CLP-induced sepsis<sup><span citationid="CR14" class="CitationRef">14</span></sup>. Given that itaconate can be released into the bloodstream<sup><span citationid="CR32" class="CitationRef">32</span></sup>, we propose that 4-OI inhibits the migration of IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells from the small intestine to the meninges, offering a theoretical strategy for SAE. Moreover, our data indicate that 4-OI enhances the expression of RNF5, an E3 ubiquitin ligase that promotes ubiquitination and degradation of STING at the K150 residue. RNF5 has been shown to target STING at K150 for ubiquitination and degradation following viral infection<sup><span citationid="CR33" class="CitationRef">33</span></sup>. The K150 site is crucial for recruiting the deubiquitinating enzyme VP1-2 to STING, influencing its ubiquitination by other E3 ligases such as TRIM32<sup>34</sup>. These findings suggest that STING ubiquitination may differ across diseases, but the role of STING ubiquitination in CNS disorders remains unclear, our study is the first to demonstrate its relevance in SAE. By enhancing STING ubiquitination, 4-OI effectively dampens neuroinflammatory processes mediated by the cGAS-STING pathway. Although other E3 ligases such as TRIM32 have been shown to promote STING ubiquitination in different contexts like in herpes simplex virus encephalitis<sup><span citationid="CR34" class="CitationRef">34</span></sup>, the role of these ligases in CNS disorders remains to be elucidated. Future research should explore the involvement of additional ubiquitin ligases and the precise molecular mechanisms governing STING regulation in neuroinflammation.
+
+In summary, our study highlights a novel mechanism wherein in the migration of small intestine-derived IL-7R<sup>high</sup> CD8<sup>low</sup> γδ T17 cells to the meninges leads to IL-17A secretion, microglia activation, and inhibition of K150-dependent STING ubiquitination. And promotes the cGAS-STING-C1q pathway, playing a key role in cognitive impairment during SAE. By enhancing STING ubiquitination and controlling γδT17 cell migration, 4-OI emerges as a promising therapeutic agent to alleviate the neurological consequences of sepsis. This study lays the foundation for “gut-brain” axis-targeted therapies in inflammatory CNS diseases, offering new hope for patients with SAE and other neuroinflammatory conditions.
+
+# Materials and Methods
+
+## Animals
+
+Male WT C57BL/6 J mice, *Trdc*<sup>CreERT</sup> mice, *Il7r*<sup>flox/flox</sup> mice, *Cd8a*<sup>flox/flox</sup> mice, and *Cx3cr1*<sup>Cre</sup> mice (all on a C57BL/6 background), weighing between 23.0 and 25.0 grams and aged between 8 and 10 weeks, were obtained from Vital River Laboratory Animal Technology Co Ltd., Beijing, China. *Trdc*<sup>CreERT</sup> mice were crossed with *Il7r*<sup>flox/flox</sup> mice and *Cd8a*<sup>flox/flox</sup> mice to generate *Trdc*<sup>CreERT</sup>*Il7r*<sup>flox/flox</sup> and *Trdc*<sup>CreERT</sup>*Cd8a*<sup>flox/flox</sup> offspring. To induce Cre activity in γδ T cells, tamoxifen (150 mg kg<sup>−1</sup>) was administered intraperitoneally to *Trdc*<sup>CreERT</sup>*Il7r*<sup>flox/flox</sup> mice and *Trdc*<sup>CreERT</sup>*Cd8a*<sup>flox/flox</sup> mice for five consecutive days. The CLP model was performed one week after tamoxifen induction. Kaede-transgenic (Kaede-Tg) mice (B6. Cg-Gt (ROSA)26Sor<lt>tm1.1(CAG-kikGR) Kgwa&gt;) were generously provided by M. Tomura of Kyoto University. The mice were housed in a controlled, specific pathogen-free environment. They were exposed to a 12:12 light/dark cycle, maintained at a regulated temperature and humidity, and provided with unrestricted access to food and water. Ethical approval for all experiments was obtained from the Experimental Animals Committee of Tongji Medical College (permission number: 4028), in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The study adhered to ARRIVE guidelines (Animals in Research: Reporting *In Vivo* Experiments).
+
+## Animal model
+
+Mice were anesthetized with sodium pentobarbital (0.3% solution) at a dose of 40 mg kg<sup>–1</sup> body weight. A midline laparotomy was performed to conduct a central lymphadenectomy, following which the abdominal region was shaved and sterilized.
+
+A 1 cm incision was made along the midline to expose the cecum, which was ligated 1 cm from the distal end using a 4−0 silk suture. A 20-gauge needle was then used to puncture the cecum, allowing a small amount of cecal content to extrude from both openings. After the procedure, the ligated cecum was returned to the abdominal cavity, and the incision was closed in multiple layers. Either 50 mg kg<sup>–1</sup> of 4-OI or 10 ml kg<sup>–1</sup> of 0.9% saline was administered intraperitoneally immediately after CLP or sham-operation. Inhibitor H151 was administered at a concentration of 750 nM via intraperitoneal injection, three times per week for three weeks prior to the CLP procedure. Survival rates were monitored and assessed 7 days post-CLP.
+
+To increase IL-17A expression in the brain, mice received adenoviruses (Adv) via lateral ventricle injection three weeks before sham surgery or CLP. Each Adv injection contained 1 × 10<sup>12</sup> plaque-forming units (PFU) of IL-17A-expressing recombinant AAV (IL-17A<sup>+</sup>-Adv). Control mice received an equivalent dose of Adv expressing GFP. In a separate experimental group, IL-17A neutralizing antibody was administered intraperitoneally at 100 µg per day for 5 days, with four doses given prior to the CLP challenge and one dose immediately after. Control mice received equivalent doses of normal hamster serum IgG.
+
+To specifically interfere with gene expression in microglia, we injected *Cx3cr1*<sup>Cre</sup> mice with DIO-AAV vectors. The AAVs were obtained from Brainvta (Wuhan, China) and included the following constructs: rAAV-SFFV-DIO-cGAS-2a-EGFP-WPREs, rAAV-SFFV-DIO-RNF5-His-2a-EGFP-WPREs, rAAV-CWV-DIO-(EGFP-U6)-shRNA1(Acod1), and rAAV-SFFV-DIO-STING1 (K150R)-2a-EGFP-W.
+
+## Kaede photoconversion
+
+To track immune cells in the small intestines *in vivo*, we performed photoconversion following the method described in a previous study<sup>8</sup>. Briefly, following CLP or sham surgery, the small intestine of Kaede-transgenic mice was exposed to a 405-nm laser for 10 min, while the surrounding tissue was shielded from light with aluminum foil.
+
+## Brain stereotaxic injection
+
+A previous study<sup>35</sup> detailed the stereotaxic injection of viruses and neutralizing antibodies into the hippocampus. Briefly, mice were anesthetized with 1–2% isoflurane. Openings were made at specific coordinates targeting the unilateral hippocampal CA1 region (x: ± 2.15 mm; y: −2.5 mm; z: −2.25 mm). A volume of 400 nanoliters (nL) was injected into each hippocampus at a rate of 26.67 nL min<sup>–1</sup>. Following the injection, the needle was left in place for 5 min to ensure proper diffusion before removal. The CLP model was performed 21 days post-viral injection and 1 day after C1q neutralizing antibody administration.
+
+## Morris water maze
+
+Experimental data were collected in a 120 cm circular pool filled with opaque water maintained at 20–22 ℃. Mice were trained four times daily for four days to locate a hidden platform submerged 1 cm below the water surface. The average latency to find the platform was calculated from four trials. Each mouse was allowed 60 seconds per trial to search for the platform. Mice remained on the platform for 15 seconds if they found it, or were placed on it for 10 seconds if they failed to locate it. A probe trial without the platform was conducted 24 h after the final hidden platform test. Mice were tested for 60 seconds to locate the platform. The number of times the mice crossed the platform target and the total time spent in the target quadrant were recorded. Mice were monitored via video cameras throughout training and probe trials. Data were analyzed using TopScan Lite (Clever Sys. Inc.).
+
+## Y-maze test
+
+Three identical arms (30 cm long, 5 cm wide, 20 cm high) were positioned at 120° angles in the Y-maze (YM). In the initial 10-minute training session, mice explored two arms while the third arm was blocked. One hour later, mice were given free access to all arms during the retention test. Mice were recorded exploring the novel arm for 5 min. Each trial was separated by cleaning the Y-maze arms with 75% ethanol. Arm entries and time spent in the novel arm were recorded, and short-term memory was calculated as the ratio of novel arm time to total exploration time. Data were examined using TopScan Lite (Clever Sys. Inc.).
+
+## Open filed test
+
+The testing area was an acrylic box measuring 40 cm long, 40 cm wide, and 30 cm high. The center of the base contained a 20 x 20 cm square core region. Each mouse was gently placed in the center of a dimly lit open-field arena and allowed to explore for 5 min. A mobile camera automatically recorded and tracked their movement. The time spent in the central region was used to assess exploratory behavior. To eliminate odors, the chamber floor was wiped with 75% ethanol after each session. Data were examined using TopScan Lite (Clever Sys. Inc.).
+
+## Novel objection recognition
+
+Two similar objects were placed at the edges of a 40 × 40 × 30 cm box. Mice were then gently placed in the box and allowed to explore the objects freely for 5 min, with the time spent on each object recorded. Exploration was defined as sniffing or touching an object within 0–2 cm with the nose. One hour later, one object was replaced with a novel one, and mice were given another 5 min to explore both objects, with the time spent on each recorded. The novel object recognition rate was calculated as (time spent on the novel object / total time spent on both objects) × 100%.
+
+## Tissue preparation
+
+Cold PBS was perfused transcardially into sedated mice. Brains and small intestines were then collected. For immunofluorescent labeling, tissues were fixed overnight at 4 ℃ in 4% paraformaldehyde and cryoprotected in 30% sucrose for at least 2 days. Brains were sectioned at 30 µm thickness using a Leica CM1950 cryostat. Small intestines were paraffin-embedded and sectioned at 4 µm thickness. Blood was collected from sedated mice via cardiac puncture into anticoagulant tubes for biochemical analysis. Plasma was obtained and stored at −80 ℃. Small intestines were stored at −80 ℃ after a gentle flush with cold PBS. After dissociation, the hippocampus was snap-frozen in liquid nitrogen and stored at −80 ℃ for protein extraction.
+
+## Immunofluorescence
+
+Brain slices were washed in PBS and blocked for 2 h at room temperature with 5% BSA (Biofroxx, Germany) and 0.3% Triton X-100 in PBS. The sections were incubated overnight at 4 ℃ with mouse anti-Iba1 (1:100, Abcam) and rabbit anti-PSD95 (1:250, CST) primary antibodies. After PBS washing, the sections were incubated with Alexa Fluor 488-conjugated goat anti-rabbit (1:1000, Abcam) and Alexa Fluor 549-conjugated goat anti-mouse (1:1000, Abcam) for 1 h at room temperature the following day. The slices were mounted with SouthernBiotech DAPI Fluoromount-G after a final wash.
+
+After deparaffinization in xylene, intestinal sections were rehydrated through a graded ethanol series to water. Sections were heated in citrate buffer (pH 6.0) for 20 min for antigen retrieval. After cooling to room temperature, the sections were blocked for 1 h with 5% BSA and 0.3% Triton X-100 in PBS. Sections were incubated overnight at 4 ℃ with rabbit anti-ZO-1, rabbit anti-MUC2, and rabbit anti-Occludin primary antibodies. The following day, sections were washed in PBS and incubated with Alexa Fluor 449-conjugated goat anti-rabbit (1:1000, Abcam) for 1 h at room temperature. The slices were mounted with SouthernBiotech DAPI Fluoromount-G after a final wash.
+
+## Microscopy and analysis
+
+Immunofluorescent-stained brain slices were imaged using a Dragonfly spinning disk confocal microscope (Andor Technology) equipped with 405, 488, and 561 nm laser lines. Z-stack images were captured using a 60x oil immersion objective. Image processing and analysis were performed using Imaris 10.2 (Bitplane). Z-stack images were generated using Imaris 10.2 for 3D reconstruction and quantification of Iba1 and PSD95 expression. Immunofluorescent signal intensity in ROIs was compared across experimental groups.
+
+Immunofluorescent-stained small intestine sections were scanned at 20x magnification using an Olympus VS200 slide scanner. High-resolution images were analyzed using Image J (1.8.0, NIH, USA). The fluorescence intensity of ZO-1, MUC2, and Occludin was quantified in randomly selected fields of view.
+
+## Primary microglia culture
+
+Based on a recent study<sup>36</sup>, we made slight modifications to the primary microglia culture procedure. Cortical brain tissues from 1-day-old C57BL/6J mice were dissected in cold HBSS (Servicebio, China, cat. no. G4203). After removing the meninges, the brain tissues were minced and rinsed three times with HBSS. The tissues were then digested in 0.25% trypsin-EDTA (ThermoFisher, cat. no. 25200056) for 20 min at 37 ℃ and triturated into a single-cell suspension. The primary cells were plated on poly-L-lysine-coated plates in Neurobasal medium supplemented with 10% FBS (Gibco, USA. cat. no. 10099-141), 1% GlutaMAX (Gibco, USA. cat. no. 35050061), and 1% penicillin-streptomycin (P.S.) (Gibco, USA. cat. no. 15140122). After 10 days of culture, the cell culture flasks were shaken at 250 rpm at 37 ℃ for 2 h. The collected culture media were centrifuged at 1000 rpm for 10 min, and the cells were resuspended in DMEM (Gibco, USA. cat. no. 11995073) with 10% FBS for inoculation.
+
+## Primary γδ T culture
+
+Primary γδ T cells were isolated from mouse spleens. Briefly, T25 cell culture flasks were coated with 5 µg mL<sup>–1</sup> TCR γδ antibody one day prior. After anesthesia, mouse spleens were extracted under sterile conditions, homogenized with 2 mL of precooled mouse lymphatic separation solution, filtered through a 70-mesh filter, gently layered over 5 mL of precooled DMEM medium, and centrifuged at 2000g for 20 min. The intermediate layer was carefully aspirated, washed with PBS, and centrifuged at 420 × g for 6 min at 4 ℃. Cells were resuspended in inoculation medium (DMEM + 10% FBS + 1% P.S + 0.1 mM β-mercaptoethanol + 5 µM Zoledronic acid monohydrate + 1000 IU IL-2 + 20 ng mL<sup>–1</sup> IL-7) at 1 × 10⁵ cells mL<sup>–1</sup> and seeded into six-well plates for growth. The culture medium was replaced every 3 days with DMEM containing 10% FBS, 1% P.S, 0.1 mM β-mercaptoethanol, 1000 IU IL-2, and 20 ng mL<sup>–1</sup> IL-7. On day 12, primary cells were transfected and stimulated.
+
+## Primary γδ T cells-microglia co-culture
+
+Primary γδ T cells were isolated from mouse spleens, and microglia were obtained from the brains of postnatal day 1 (P1) mice. For co-culture experiments, microglia were seeded in the bottom chamber of a transwell device with 0.4 µm pore-size polycarbonate membrane inserts, while γδ T cells were seeded in the top chamber. To investigate the impact of IL-17A, γδ T cells were transfected with siIL-17A or siCtrl using Lipofectamine RNAiMAX. Additionally, γδ T cells were treated with recombinant mouse IL-17A (100 ng mL<sup>–1</sup>) and LPS (1000 ng mL<sup>–1</sup>) for 6 h, followed by a medium change. The transwell inserts containing γδ T cells and microglia were co-cultured for 24 h post-treatment. After co-culture, microglia were harvested for analysis.
+
+## Cell Culture and treatment
+
+BV2 murine microglial cells were obtained from Punosai Life Science and Technology Co., Ltd. and cultured in RPMI-1640 medium (Gibco, USA. cat. no. 11875119) supplemented with 10% FBS and 1% P.S. at 37 ℃ in 5% CO<sub>2</sub>. To model SAE, BV2 cells were treated with 1000 µg mL<sup>–1</sup> LPS for 6 h. After incubation, Cells were collected and processed for further assays. To block the cGAS-STING pathway, BV2 cells were treated with H151 (0.75 µM) for 2 h prior to LPS stimulation. To generate BV2<sup>ρ0</sup> cells, which are devoid of mitochondria, BV2 cells were treated with ethidium bromide (50 ng mL<sup>–1</sup>) for 4 weeks<sup>37</sup>.
+
+## Cell Transfection
+
+For gene silencing, primary γδ T cells were transfected with IL-17A siRNA, and BV2 microglial cells transfected with RNF5 siRNA. Additionally, BV2 cells were transfected with STING protein plasmids containing site-directed mutations for ubiquitination site analysis. All siRNAs and plasmids were obtained from Obio Technology Co., Ltd., Shanghai. Following the manufacturer’s instructions. Lipofectamine RNAiMAX was used for siRNA transfections, and Lipofectamine 3000 was used for plasmid transfections. Transfection efficiency was confirmed by Western blot analysis.
+
+## Western blot
+
+Total proteins were extracted from cells and tissues, and protein concentrations were measured using the BCA assay. Equal amounts of protein were separated by SDS-PAGE and transferred to PVDF membranes (Millipore). Membranes were incubated overnight at 4 ℃ with primary antibodies against ZO-1, MUC-2, Occludin, cGAS, STING, C1q, PSD95, Synaptophysin (SYN), ACOD1, RNF5, Ubiquitin (UB), β-actin, GAPDH and a custom-made antibody for the K150 ubiquitination site on STING (Abclonal). After incubation, membranes were treated with HRP-conjugated anti-mouse or anti-rabbit IgG. Protein bands were visualized using ECL Western blot Detection Reagents (Beyotime) and captured with a UVP gel documentation system (UVP, LLC, Phoenix). Band intensity was quantified using ImageJ (1.8.0, NIH, USA).
+
+Synaptosomal proteins were extracted from mouse hippocampi using the Syn-PER Synaptic Protein Extraction Reagent (Thermofisher) following the manufacturer’s instructions. Western blot analysis was then performed on the isolated synaptosomal proteins.
+
+## FCM analysis
+
+As previously described<sup>8</sup>, cell suspensions from the meninges and small intestine were prepared. After counting 1 × 10<sup>6</sup> cells, they were resuspended in 100 µL of 1% BSA in PBS, blocked with 5 ng µL<sup>–1</sup> anti-CD16/CD32 antibody for 5 min at 4 ℃, and then stained with the desired antibodies. Antibodies used for extracellular staining included fixable viability dye, CD45, CD3, TCR γδ, IL7R, CD8, F4/80, CD206, CD86, and CD11B. For intracellular staining, cells were first labeled with surface markers, then frozen and permeabilized before being labeled with IL-17A antibody (4 ng µL<sup>–1</sup>). Samples were analyzed using a Beckman CytoFLEX or BD LSRFortessa X-20 cytometer (BD Biosciences, San Jose, CA). Data were analyzed using FlowJo 10.0 (FlowJo, Oregon, USA).
+
+## JC-1 Assay
+
+Mitochondrial membrane potential in BV2 microglial cells was measured using the JC-1 detection kit (Elabscience, China) according to the manufacturer’s instructions. BV2 cells were seeded in 6-well plates following experimental criteria. After treatment, cells were stained with JC-1 at 37 ℃ for 20 min. The cells were rinsed with buffer solution and analyzed using a fluorescence microscope or flow cytometer after incubation. The mitochondrial membrane potential was assessed by the red-to-green fluorescence ratio.
+
+## Ubiquitination level detection
+
+MG132 (100 µM) was added to cells 6 h before collection to inhibit proteasome activity and accumulate ubiquitinated proteins in cell culture studies. Hippocampus tissues were lysed in IP lysis buffer containing protease and phosphatase inhibitors to extract proteins. The lysates were incubated overnight at 4 ℃ with an anti-STING antibody and protein A/G magnetic beads for co-immunoprecipitation. After washing, the co-immunoprecipitated proteins were eluted from the beads. Eluted proteins were analyzed by Western blot using an anti-UB antibody (1:1000, Proteintech) to measure ubiquitination.
+
+## Detection of ubiquitination modification sites
+
+Immunoprecipitation (IP) was performed to pull down STING from BV2 microglial cells for the identification of ubiquitination modification sites. Ubiquitination sites were identified by mass spectrometry following STING isolation. The identified modification sites were used to create mutant plasmids for each ubiquitination site using site-directed mutagenesis. These mutant plasmids were then transfected into BV2 cells to specifically alter STING ubiquitination sites. Ubiquitination levels of WT and mutant STING proteins were compared in subsequent assays to identify key ubiquitination sites.
+
+## Microglia depletion
+
+PLX3397 (Selleck, China) was used to deplete brain microglia. As previously reported<sup>38</sup>, PLX3397 was incorporated into AIN-76A chow at a concentration of 300 mg kg<sup>–1</sup>. The PLX3397-supplemented chow was prepared by Jiangsu Xietong, Inc., Nanjing. Mice were fed AIN-76A chow or PLX3397-supplemented chow (300 mg kg<sup>–1</sup> in AIN-76A) for 21 days. Microglia depletion was evaluated by quantifying microglia numbers and assessing Iba1 expression after the 21-day feeding period.
+
+## Golgi staining
+
+After anesthesia, mouse brains were immediately placed in Golgi stain fixing solution and immersed in dye solution for 14 days at room temperature in the dark. The treatment solution was changed after 1 h and then stored at 4 ℃ in the dark for 3 days. Brain samples were sectioned into 60 µm coronal slices using a vibrating microtome. Dendritic spine morphology was examined under a microscope.
+
+## Reactive Oxygen Species (ROS) Detection by DHE Staining
+
+Oxidative stress in the hippocampal CA1 region was measured using dihydroethidium (DHE) staining. Brain slices from treated animals were processed as previously described. Sections were treated with Beyotime DHE staining reagent according to the manufacturer’s instructions. Sections were incubated with DHE (10 µM) for 30 min in a light-protected, humidified chamber at 37 ℃. Superoxide anions oxidize DHE to ethidium, which intercalates with DNA and fluoresces red, indicating tissue ROS levels.
+
+After incubation, sections were washed in PBS and mounted with anti-fade media. Fluorescent signals were captured using an Olympus fluorescence microscope (Olympus, Japan) with DHE filters. Red fluorescence intensity in the hippocampal CA1 region was quantified using ImageJ (NIH, USA) to assess oxidative stress. Oxidative stress levels between experimental groups were assessed by statistical analysis of mean fluorescence intensity.
+
+## TEM
+
+Mice were perfused transcardially with 2.5% glutaraldehyde. 1-mm coronal brain slices were collected immediately after perfusion. The CA1 region of the hippocampus was meticulously microdissected and post-fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer at 4 ℃ for 24 h. After treatment, BV2 and primary microglia were centrifuged and fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer at 4 ℃ for 24 h.
+
+Both tissue samples and cells were processed similarly after fixation. After three washes with 0.1 M sodium cacodylate buffer, the samples were post-fixed in 1% OsO₄ in the same buffer at 4 ℃ for 1.5 h. After dehydration in graded ethanol (50%, 70%, 90%, 100%), the samples were embedded in epoxy resin. Thin sections (60–70 nm) were cut using a Leica EM UC7 ultramicrotome and placed on copper grids. For contrast enhancement, sections were stained with 2% uranyl acetate for 15 min and lead citrate for 10 min. Electron micrographs were captured using an 80 kV Hitachi HT7800 transmission electron microscope (HITACHI, Japan). Ultrastructural characteristics of hippocampal CA1 synapses, BV2 cells, and primary microglia were identified and analyzed.
+
+## MDA and SOD activity determination
+
+MDA and SOD activity were measured using kits (Beyotime, China) according to the manufacturer’s instructions. The hippocampus was excised, weighed, and homogenized in MDA and SOD assay solutions. Plasma and homogenate supernatants were added to the reaction system. MDA activity was measured at 532 nm by absorbance and reported as µmol mg<sup>–1</sup> or µM. SOD activity in the samples was measured by absorbance at 450 nm and expressed as U mg<sup>–1</sup> or U mL<sup>–1</sup> of total protein.
+
+## Determination of cytokine levels
+
+Cytokine levels in mouse plasma and hippocampus were measured using the ABplex Mouse Cytokine 8-Plex Assay Kit (Abclonal). The analyzed cytokines included IL-2, IL-4, IL-12p70, IL-1β, IL-6, IFN-γ, TNF-α, and IL-17A. The samples were clarified by centrifugation and analyzed by FCM.
+
+## Scratch Assay
+
+BV2 microglial cell migration was assessed using a scratch assay. BV2 cells were seeded into 12-well plates and cultured until 90% confluence. A linear scratch was made in the cell monolayer using a sterile 200 µL pipette tip. After scratching, wells were gently rinsed with PBS to remove floating cells and debris before fresh media was added. Cell migration into the scratch region was tracked for 24 h using the Opera Phenix Plus high-content imaging system (Revvity). Images were frequently captured to monitor cell migration. The migration rate was calculated as the proportion of the scratch area covered by migrating cells relative to the control group. ImageJ (NIH, USA) was used for image analysis to compare cell migration rates between experimental conditions.
+
+## Phagocytosis Assay
+
+The phagocytic capability of primary microglial cells was assessed using the Cell Meter™ Fluorimetric Phagocytosis Assay Kit (Red Fluorescence, Cat#21225). Primary microglial cells were allowed to adhere overnight in a 96-well confocal cell culture plate. After experimental treatments, cells were incubated with Protonex™ 600-labeled beads according to the manufacturer’s instructions. After incubation, non-internalized beads were washed away, and fluorescence from ingested beads was detected using a Zeiss LSM 880 confocal laser scanning microscope. Phagocytic activity was quantified as bead engulfment (%), calculated as the ratio of red fluorescent area (engulfed beads) to total cell area. ImageJ software (NIH, USA) was used to calculate this ratio and statistically compare microglial phagocytic capability between experimental groups.
+
+## PET-CT Imaging for Brain Metabolism
+
+Brain metabolic activity was assessed via PET-CT imaging on mice 24 h after CLP. To minimize glucose fluctuations that could affect 18F-fluorodeoxyglucose (18F-FDG) uptake, mice were fasted for 4–6 h with access to water before imaging. After fasting, mice were intravenously administered 18F-FDG at a dose of 15.6±1.7 MBq in 0.5 mL saline. Following injection, mice were housed in a warm, calm environment for 30–45 min to enhance tracer uptake and minimize stress and muscular activity that could affect glucose distribution.
+
+Small-animal PET-CT scanners with high resolution and sensitivity were utilized under isoflurane anesthesia. Optimized scanning covered the entire brain, with a focus on the hippocampi. CT scans provided anatomical reference, followed by PET scans to assess glucose metabolism.
+
+The SUVmean, which represents the average 18F-FDG uptake adjusted for body weight and injected dose, was used to measure brain metabolic activity, particularly in the hippocampal region. SUVmean data were statistically compared between experimental groups to evaluate post-CLP brain metabolism.
+
+## Laser Speckle Contrast Imaging (LSCI) for Cerebral Blood Flow
+
+Cerebral blood flow dynamics in mice were assessed by LSCI 24 h post-CLP. After anesthesia, the scalp was gently removed to expose the skull. LSCI uses coherent laser light to illuminate cortical tissue. Red blood cells in cerebral vessels alter the speckle pattern captured by a high-resolution camera. These oscillations are used to map blood flow across the cortical surface in real time. Data were collected using a laser-synchronized high-sensitivity CCD camera. Relative blood flow changes were computed based on speckle contrast. To evaluate the impact of CLP on cerebral perfusion, blood flow was quantified and statistically compared between experimental groups.
+
+## TMT-based quantitative proteomics analysis
+
+### Sample preparation
+
+Proteomic analysis was conducted on CLP and sham-operated mice (n=3 per group). After anesthesia, intestinal tissues were promptly removed and snap-frozen in liquid nitrogen. Weighed protein samples were added to SDS L3-EDTA lysis buffer. The samples were homogenized (25,000 g, 4 ℃, 5 min). The homogenate was then treated with 10 mM DTT and 45 mM IAM and incubated in the dark for 45 min. Proteins were precipitated with cold acetone, followed by centrifugation (25,000g, 4 ℃, 15 min). After air-drying the pellet, SDS L3-free lysis solution was added to fully solubilize the proteins. Following another centrifugation (25,000 g, 4 ℃, 15 min), protein concentration was measured using the Bradford method. Proteins were desalted after trypsin digestion. The freeze-dried peptides were separated using a Shimadzu LC-20AB liquid chromatography system.
+
+### High-performance liquid chromatography and mass spectrometry
+
+Peptides were separated using gradient elution on a self-packed C18 column on an Easy nLC 1200 system. The isolated peptides were analyzed using Data Dependent Acquisition (DDA) tandem mass spectrometry. TMT-proteomic analysis was performed by BGI-Shenzhen, China.
+
+### Bioinformatics analysis
+
+The UniProtKB database (Release 2016_10) was used to obtain FASTA protein sequences of differentially expressed proteins for Gene Ontology (GO) mapping and annotation. KEGG Orthology (KO) identities and pathways were determined by aligning the FASTA sequences of significantly altered proteins with the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://geneontology.org/). Protein expression data from relevant KEGG pathways were visualized using a hierarchical clustering heat map.
+
+## Statistical analysis
+
+Data reported as mean ± standard deviation (Mean ± SD) were analyzed using GraphPad Prism 10.0 software. Data normality was assessed using the D’Agostino & Pearson test or Shapiro-Wilk test. One-way analysis of variance (ANOVA) and two way ANOVA with Tukey’s or Šídák’s multiple comparisons test was used for multiple-group comparisons. Unpaired Student’s t-test or Mann-Whitney test was used for group comparisons, and Log-rank (Mantel–Cox) test was performed for survival rate analysis. *P* values < 0.05 were considered statistically significant. All data analyses and statistical figures were generated using GraphPad Prism 10.0.
+
+# References
+
+1. Robba C, Crippa IA, Taccone FS (2018) Septic Encephalopathy. Curr Neurol Neurosci Rep 18:82. 10.1007/s11910-018-0895–6
+2. Assimakopoulos SF et al (2018) Gut-origin sepsis in the critically ill patient: pathophysiology and treatment. Infection 46:751–760. 10.1007/s15010-018-1178–5
+3. Nielsen MM, Witherden DA, Havran WL (2017) gammadelta T cells in homeostasis and host defence of epithelial barrier tissues. Nat Rev Immunol 17:733–745. 10.1038/nri.2017.101
+4. Wu P et al (2014) gammadeltaT17 cells promote the accumulation and expansion of myeloid-derived suppressor cells in human colorectal cancer. Immunity 40:785–800. 10.1016/j.immuni.2014.03.013
+5. Santisteban MM et al (2024) Meningeal interleukin–17-producing T cells mediate cognitive impairment in a mouse model of salt-sensitive hypertension. Nat Neurosci 27:63–77. 10.1038/s41593-023-01497-z
+6. Benakis C et al (2016) Commensal microbiota affects ischemic stroke outcome by regulating intestinal gammadelta T cells. Nat Med 22:516–523. 10.1038/nm.4068
+7. Sun J et al (2015) IL–17A is implicated in lipopolysaccharide-induced neuroinflammation and cognitive impairment in aged rats via microglial activation. J Neuroinflamm 12:165. 10.1186/s12974-015-0394–5
+8. Xie B et al (2024) Gut-derived memory gammadelta T17 cells exacerbate sepsis-induced acute lung injury in mice. Nat Commun 15:6737. 10.1038/s41467-024-51209–9
+9. Zhang YQ et al (2021) Dopamine D2 receptor regulates cortical synaptic pruning in rodents. Nat Commun 12:6444. 10.1038/s41467-021-26769–9
+10. Dejanovic B et al (2022) Complement C1q-dependent excitatory and inhibitory synapse elimination by astrocytes and microglia in Alzheimer's disease mouse models. Nat Aging 2:837–850. 10.1038/s43587-022-00281–1
+11. Hao W et al (2024) Gut dysbiosis induces the development of depression-like behavior through abnormal synapse pruning in microglia-mediated by complement C3. Microbiome 12, 34. 10.1186/s40168-024-01756–6
+12. Chung HY et al (2023) Microglia mediate neurocognitive deficits by eliminating C1q-tagged synapses in sepsis-associated encephalopathy. Sci Adv 9:eabq7806. 10.1126/sciadv.abq7806
+13. Mills EL et al (2018) Itaconate is an anti-inflammatory metabolite that activates Nrf2 via alkylation of KEAP1. Nature 556:113–117. 10.1038/nature25986
+14. Li W et al (2023) 4-octyl itaconate as a metabolite derivative inhibits inflammation via alkylation of STING. Cell Rep 42:112145. 10.1016/j.celrep.2023.112145
+15. Ramakrishnan RK et al (2020) IL–17 Induced Autophagy Regulates Mitochondrial Dysfunction and Fibrosis in Severe Asthmatic Bronchial Fibroblasts. Front Immunol 11:1002. 10.3389/fimmu.2020.01002
+16. Ghosh M, Saha S, Li J, Montrose DC, Martinez LA (2023) p53 engages the cGAS/STING cytosolic DNA sensing pathway for tumor suppression. Mol Cell 83:266–280. 10.1016/j.molcel.2022.12.023
+17. Hu X et al (2022) Emerging role of STING signalling in CNS injury: inflammation, autophagy, necroptosis, ferroptosis and pyroptosis. J Neuroinflamm 19:242. 10.1186/s12974-022-02602-y
+18. Ferecsko AS et al (2023) STING-Triggered CNS Inflammation in Human Neurodegenerative Diseases. Biomedicines 11:1375. 10.3390/biomedicines11051375
+19. Meyer-Schwesinger C (2019) The ubiquitin-proteasome system in kidney physiology and disease. Nat Rev Nephrol 15:393–411. 10.1038/s41581-019-0148–1
+20. Chancharoenthana W, Kamolratanakul S, Schultz MJ, Leelahavanichkul A (2023) The leaky gut and the gut microbiome in sepsis - targets in research and treatment. Clin Sci 137:645–662. 10.1042/CS20220777
+21. Xu K et al (2024) Phosphatidylserine improves aging sepsis survival, modulates gut microbiome, and prevents sepsis-associated encephalopathy. Biomed Pharmacother 178:117200. 10.1016/j.biopha.2024.117200
+22. Xu Y et al (2024) Palmatine ameliorated lipopolysaccharide-induced sepsis-associated encephalopathy mice by regulating the microbiota-gut-brain axis. Phytomedicine 124:155307. 10.1016/j.phymed.2023.155307
+23. Mills K (2023) IL–17 and IL–17-producing cells in protection versus pathology. Nat Rev Immunol 23:38–54. 10.1038/s41577-022-00746-9
+24. Li J et al (2021) Microglial Phenotypic Transition: Signaling Pathways and Influencing Modulators Involved in Regulation in Central Nervous System Diseases. Front Cell Neurosci 15:736310. 10.3389/fncel.2021.736310
+25. Gao S et al (2024) N-acetyltransferase 10 mediates cognitive dysfunction through the acetylation of GABA(B)R1 mRNA in sepsis-associated encephalopathy. Proc. Natl. Acad. Sci. U. S. A. 121, e1884403175. 10.1073/pnas.2410564121
+26. Yang YS et al (2024) Quercetin protects against sepsis-associated encephalopathy by inhibiting microglia-neuron crosstalk via the CXCL2/CXCR2 signaling pathway. Phytomedicine 134:155987. 10.1016/j.phymed.2024.155987
+27. Yin XY et al (2023) HMGB1 mediates synaptic loss and cognitive impairment in an animal model of sepsis-associated encephalopathy. J Neuroinflamm 20:69. 10.1186/s12974-023-02756-3
+28. Woo MS et al (2024) STING orchestrates the neuronal inflammatory stress response in multiple sclerosis. Cell 187:4043–4060. 10.1016/j.cell.2024.05.031
+29. Jimenez-Loygorri JI et al (2024) Mitophagy curtails cytosolic mtDNA-dependent activation of cGAS/STING inflammation during aging. Nat Commun 15:830. 10.1038/s41467-024-45044-1
+30. Ramalho T et al (2024) Itaconate impairs immune control of Plasmodium by enhancing mtDNA-mediated PD-L1 expression in monocyte-derived dendritic cells. Cell Metab 36:484–497. 10.1016/j.cmet.2024.01.008
+31. Chen YJ et al (2023) Targeting IRG1 reverses the immunosuppressive function of tumor-associated macrophages and enhances cancer immunotherapy. Sci Adv 9:eadg654. 10.1126/sciadv.adg0654
+32. Gu X et al (2023) Itaconate promotes hepatocellular carcinoma progression by epigenetic induction of CD8(+) T-cell exhaustion. Nat Commun 14:8154. 10.1038/s41467-023-43988-4
+33. Zhong B et al (2009) The ubiquitin ligase RNF5 regulates antiviral responses by mediating degradation of the adaptor protein MITA. Immunity 30:397–407. 10.1016/j.immuni.2009.01.008
+34. Bodda C et al (2020) HSV1 VP1–2 deubiquitinates STING to block type I interferon expression and promote brain infection. J Exp Med 217:e20191422. 10.1084/jem.20191422
+35. Wu X et al (2024) Foxq1 activates CB2R with oleamide to alleviate POCD. Brain Pathol. e13289. 10.1111/bpa.13289
+36. Liu Y et al (2024) NG2 glia protect against prion neurotoxicity by inhibiting microglia-to-neuron prostaglandin E2 signaling. Nat Neurosci 27:1534–1544. 10.1038/s41593-024-01663-x
+37. Zecchini V et al (2023) Fumarate induces vesicular release of mtDNA to drive innate immunity. Nature 615:499–506. 10.1038/s41586-023-05770-w
+38. Shi Y et al (2019) Microglia drive APOE-dependent neurodegeneration in a tauopathy mouse model. J Exp Med 216:2546–2561. 10.1084/jem.20190980
+
+# Supplementary Files
+
+- [ExtendeddataTable.docx](https://assets-eu.researchsquare.com/files/rs-5152680/v1/ed26119bf09e58aae25de77d.docx)  
+  Extended data table 1
+
+- [SupplementaryFigures.pdf](https://assets-eu.researchsquare.com/files/rs-5152680/v1/3c0e2b7ae51a9d49b3476333.pdf)  
+  Supplementary Figures

2265644aa9d24f1f1b46b422cebe9864b5b65978e8025983a87651be36d20587/preprint/images_list.json

@@ -0,0 +1,34 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Single cell RNA-seq analysis reveals vaccine-associated reduction of LPL B-cell but not plasma cell-like subpopulations in the bone marrow tumor microenvironment. (A) Schematic of individualized chemokine-idiotype DNA vaccines and vaccine treatment schedule. (B) Swimmer plot illustrating clinical responses of each patient (patients are designated as LPL-001 through -009) (C) UMAP of B-lineage cell populations extracted the from total dataset. 12 clusters were generated (left). Total cell numbers per cluster are shown for normal B-lineage cells and specifically, tumor B cells, based on known idiotype sequences (right). (D) Cell frequencies by total B-lineage cluster for each patient in paired pre- and post-vaccine bone marrow samples. Paired student t-test was used (E) Cell frequencies of major BCR clonotypes in all B-cell clusters pre- and post-vaccine for each patient. Blue = dominant tumor idiotype clone isolated for vaccine production; all other colors indicate normal B-cell clonotypes. (F) Pooled data from all patients. Volcano plots of differentially expressed genes (adjusted p-value > 0.05) pre- vs. post-vaccine for the major B-lineage clusters containing tumor cells (clusters 0-2, top). Ingenuity Pathway Analysis (Qiagen IPA) based on differentially expressed genes (z-score > 2, adjusted log p-value >1.3) of canonical pathways (bottom left) and biological processes (bottom right) contrasting tumor mature B-cell (0-2) and plasma cell-like clusters (5 and 10), respectively. (G) Heatmap of selected HLA class II gene expression by idiotype clonal tumor cells pre- and post-vaccine relative to total cells and grouped by B-lineage cluster. (G) Violin plots of expression of HLA family genes and co-inhibitory ligands by idiotype clonal tumor cells pre- and post-vaccine in cluster 2 only. Two-sided Wilcoxon test was used. *P \u2264 0.05, **P \u2264 0.01, ***P \u2264 0.001, and ****P \u2264 0.0001",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Paired single cell TCR-seq reveals T-cell clonal expansion and activation in the tumor microenvironment following vaccine treatment. (A) UMAP of T-cell subpopulations extracted the from the total dataset. A total of 12 clusters were generated with unique phenotypes (left). UMAP of T-cells highlighting cells with single cell TCR sequences detected (right). (B)\u00a0 T-cell frequencies pre- and post-vaccine for individual patients by T-cell cluster. Paired student t-test was used (C) Heatmap showing selected immune signaling pathways post vs pre vaccine (adjusted log p-value > 1.3) based on significantly differentially expressed genes for each indicated T cell subpopulation using IPA software (Qiagen). Pooled data from all patients. (D) Scatterplots of TCR clonotype frequencies pre-vaccine (y-axis) and post-vaccine (x-axis) for each patient. (E) Frequencies of the 20 most prevalent T-cell clonotypes identified within the post-vaccination T-cell repertoire, compared with their matching clonotype pre-vaccine. Paired student t-test was used. (F) Shannon entropy of TCR clonotype repertoires in paired samples pre. vs post vaccine for each patient. Paired student t-test was used (G) UMAPs of pooled T-cell subpopulations in pre- and post-vaccine samples with highlighted cells in the 20 most prevalent post-vaccine clonotypes (left panels) and the cell cycle phases inferred from gene expression signatures using Seurat package (right panels). (H) Violin plots of expression levels of selected gene markers by the top 20 T-cell clonotypes pre- vs. post-vaccine. Two-sided Wilcoxon test was used. (I)\u00a0 Functional tumor idiotype-specific T-cell responses post-vaccination. Bone marrow samples from each patient were enriched for T cells by negative selection and then 2.5x105 cells per well were stimulated in triplicate with autologous immortalized B cells as APCs (2.5x105 cells / well) which had been transfected with either patient-specific tumor idiotype (Ig VH and VL genes) or HIV Nef as a negative control.\u00a0Supernatants were harvested 3, 7, or 10 days later and analyzed by multiplex cytokine assay.\u00a0Representative cytokines are shown for each patient. MR \u2013 minor response, SD \u2013 stable disease, PD \u2013 progressive disease. Two-sided student t-test was used. *P \u2264 0.05, **P \u2264 0.01, and ***P \u2264 0.001. n.s., not significant.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "DNA vaccine significantly alters cell-cell communication networks in the tumor microenvironment. (A) Total numbers of ligand-receptor pair interactions in bone marrow samples pooled from all patients. Paired t-test was used to compare pre- and post-vaccine samples. ***p<0.001 (B) Pooled outgoing and incoming interaction strengths between the following cell types in 2D space pre- and post-treatment for all patients: LPL (mature B-lymphoid), LPL (plasma-like), myeloid, T- and NK, and normal B progenitor. Dot size indicates the number of expressed ligand-receptor pairs. Interaction strengths were calculated with Cellchat software. (C) Pooled relative information flows between pairwise pre- and post-vaccine datasets for all signaling pathways, sorted by increasing information flow post-treatment. (D) Heatmap of relative strengths of all signaling pathways pre- and post-vaccine by cell type. Outgoing and incoming signaling patterns from data pooled from all patients are shown. Paired t-test was used to compute p-values comparing signaling patterns within each cell type. (E) Circle plots of selected signaling pathways and relative contributions of ligand-receptor pairs. Cell types are color coded, dot size is proportional to the number of expressed ligand-receptor pairs, edge color indicates the source of outgoing signal, and edge weight is proportional to interaction strength.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Reduced signaling by myeloid cell subpopulations post-vaccination. (A) UMAP of myeloid cells extracted from the total dataset. 9 clusters were generated, and cluster identities were assigned based on differentially expressed gene markers. (B) Heatmap of expression of selected gene markers for each cluster. (C) Cell frequencies pre- and post-vaccine for individual patients by myeloid cell cluster. Paired student t-test was used. (D) Dot plots of differentially expressed genes pre- and post- vaccine in selected pathways identified in Fig. 3 by cluster. Pooled data from all patients. Significantly downregulated or upregulated genes in post-vaccine samples are highlighted by red and green rectangles respectively. Color intensity and dot size correspond to expression level and relative proportion of positive cells, respectively. (E) \u00a0Violin plots of SIRPa and CD47 gene expression pre- and post-vaccination by cell cluster in myeloid and B-cell tumor populations, respectively. Pooled data from all patients. Two-sided Wilcoxon test was used. *P \u2264 0.05, ***P \u2264 0.001. n.s., not significant.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

2265644aa9d24f1f1b46b422cebe9864b5b65978e8025983a87651be36d20587/preprint/preprint.md

@@ -0,0 +1,350 @@
+# Abstract
+
+Lymphoplasmacytic lymphoma (LPL) is an incurable low-grade B-cell lymphoma of the bone marrow. Despite a cumulative risk of progression, there is no approved therapy for patients in the asymptomatic phase. We conducted a first-in-human clinical trial of a novel therapeutic DNA idiotype neoantigen vaccine in nine patients with asymptomatic LPL. Treatment was well tolerated with no dose limiting toxicities. One patient achieved a minor response, and all remaining patients experienced stable disease, with median time to disease progression of 61+ months. Direct interrogation of the tumor microenvironment by single-cell transcriptome analysis revealed an unexpected dichotomous antitumor response, with significantly reduced numbers of clonal tumor mature B-cells, tracked by their unique BCR, and downregulation of genes involved in signaling pathways critical for B-cell survival post-vaccine, but no change in clonal plasma cell subpopulations. Downregulation of HLA class II molecule expression suggested intrinsic resistance by tumor plasma cell subpopulations and cell-cell interaction analyses predicted paradoxical upregulation of IGF signaling post vaccine by plasma cell, but not mature B-cell subpopulations, suggesting a potential mechanism of acquired resistance. Vaccine therapy induced dynamic changes in bone marrow T-cells, including upregulation of signaling pathways involved in T-cell activation, expansion of T-cell clonotypes, increased T-cell clonal diversity, and functional tumor antigen-specific cytokine production, with little change in co-inhibitory pathways or Treg. Vaccine therapy also globally altered cell-cell communication networks across various bone marrow cell types and was associated with reduction of protumoral signaling by myeloid cells, principally non-classical monocytes. These results suggest that this prototype neoantigen vaccine favorably perturbed the tumor immune microenvironment, resulting in reduction of clonal tumor mature B-cell, but not plasma cell subpopulations. Future strategies to improve clinical efficacy may require combinations of neoantigen vaccines with agents which specifically target LPL plasma cell subpopulations, or enable blockade of IGF-1 signaling or myeloid cell checkpoints.
+
+**Biological sciences/Immunology/Immunotherapy/Immunization**  
+**Biological sciences/Cancer/Haematological cancer/Lymphoma/Non-hodgkin lymphoma/B-cell lymphoma**
+
+# Introduction
+
+LPL is an incurable low-grade B-cell lymphoma, characterized by the presence of clonal lymphoplasmacytic cells infiltrating the bone marrow as the primary organ, and a serum monoclonal protein. IgM-secreting LPL, known as Waldenström macroglobulinemia (WM), is the most common subtype<sup>1</sup>. In the absence of end organ damage, patients are considered to have smoldering phase disease. There is no approved standard therapy for smoldering LPL, and patients are generally managed by active surveillance<sup>1</sup>. Accordingly, the availability of a well-tolerated therapeutic agent that would enable early intervention to delay progression to symptomatic phase disease and other complications, without inducing cross-resistance to subsequent chemotherapies, is desirable.
+
+Tumor neoantigens recognized by T cells are emerging as targets for the design of cancer vaccines, and initial clinical trials have demonstrated both safety and efficacy<sup>2-4</sup>. One of the first neoantigens tested was the unique B-cell receptor generated by clonal rearrangement of Ig variable region gene sequences by lymphoma cells, rather than somatic mutation, referred to as idiotype<sup>5-8</sup>. Idiotype peptides were the dominant neoantigens eluted from HLA molecules in human lymphomas<sup>9,10</sup>. Therapeutic idiotype vaccines have been shown to elicit robust CD8+ T-cell immunity in humans, and one randomized, controlled clinical trial demonstrated improved disease-free survival in a minimal residual disease setting following induction chemotherapy in follicular lymphoma<sup>11-15</sup>.
+
+Here, in a first-in-human clinical trial we tested a novel DNA vaccine platform, encoding the autologous LPL-derived Ig single chain variable fragment (scFv) fused to human chemokine CCL20 (macrophage inflammatory protein-3, MIP-3a), which was designed to trigger T-cell immunity by targeting delivery of the expressed fusion protein to antigen presenting cells<sup>16,17</sup> in smoldering LPL patients.
+
+# Results
+
+## Patient characteristics
+
+A total of nine patients were enrolled and treated on trial, three in the 500µg cohort and six in the 2500µg cohort. The baseline characteristics of patients are described in Table 1. The median age of all patients was 67, and the majority were male (78%). Of eight patients with baseline gene mutation data available, six had MYD88 mutations; of these, one had a CXCR4 WHIM mutation. The median time from diagnosis of asymptomatic LPL to first vaccination was 2.2 years.
+
+| Characteristic (range) | Cohort 1 500µg (n = 3) | Cohort 2 2500µg (n = 6) |
+|------------------------|------------------------|------------------------|
+| Median Age | 65 (56–71) | 69 (61–78) |
+| Male Sex | 100% | 67% |
+| ECOG performance status 0–1 | 100% | 100% |
+| Time from diagnosis of SWM to 1st vaccination (yrs.) | 8.1 (1.4–8.8) | 2.0 (0.7–10.8) |
+| Genotype* (no. of patients) | | |
+| MYD88WT/CXCR4WT | 1 | 1 |
+| MYD88 L265P/CXCR4WT | 2 | 3 |
+| MYD88 L265P/CXCR4WHIM | 0 | 1 |
+| Bone Marrow infiltration (%) | 30 (25–40) | 30 (10–50) |
+| Serum IgM (mg/dL) | 2900 (814–3150) | 3255 (473–7210) |
+| Monoclonal Protein (g/dL) | 2.3 (0.9–2.8) | 2.5 (0.4–6.3) |
+| Hemoglobin (g/dL) | 13.6 (12.2–14.9) | 12.3 (137–353) |
+| Platelet count (K/µL) | 204 (189–372) | 278 (137–353) |
+| Beta 2 microglobulin (mg/L) | 2.5 (2.0-3.7) | 2.5 (1.7–3.7) |
+| Albumin (g/dL) | 77.7 (76.9–80) | 4.1 (3.7–4.4) |
+| LDH (U/L) (normal range: 313–618)** | 378 (376–405) | 337 (201–613) |
+
+*Genotype not available for 1 patient in Cohort 2  
+* LDH not available for 1 patient in Cohort 2
+
+## Safety, tolerability, and response assessment
+
+All patients successfully completed planned therapy. No patients in either cohort experienced dose limiting toxicities (DLTs) or Grade 4 adverse events (AEs). Ten months after the last vaccination, LPL-005 developed a grade 3 non-malignant pleural effusion, grade 1 pericardial effusion, and leukocytopenia, accompanied by an increase in rheumatoid factor (23.1 IU/mL [normal range 0.0-15.9]) and an ANA titer of 1:80; all findings resolved within 2 months. Grade 1–2 AEs occurring in 3 or more patients were leukopenia, nausea, myalgias, fatigue, diarrhea, anemia, hyperglycemia, and increased creatinine. Details are provided in Table 2.
+
+| Most common adverse events | Total No. of patients (n = 9) | |
+|----------------------------|-------------------------------|---|
+| | Any Grade | ≥ Grade 3 |
+| Hematologic | | |
+| Leukocytopenia | 6 | |
+| Anemia | 4 | |
+| Gastrointestinal | | |
+| Nausea | 5 | |
+| Diarrhea | 3 | |
+| General | | |
+| Fatigue | 4 | |
+| Myalgia | 3 | |
+| Respiratory | | |
+| Dyspnea | 3 | |
+| Pleural Effusion | 1 | 1 |
+| Cardiac | | |
+| Pericardial Effusion | 1 | |
+| Dermatologic | | |
+| Injection Site Reaction | 3 | |
+| Lab Abnormalities | | |
+| Creatinine increase | 4 | |
+| Hyperglycemia | 6 | |
+
+Using response criteria from the 6th International WM Workshop Consensus Panel, LPL-003 achieved a minor response (MR). The best response for the remaining eight patients was stable disease (SD) (Fig. 1 B). After a median follow-up period of 77 months for all patients, four patients have experienced progression to symptomatic WM, requiring initiation of systemic therapy (LPL-005, -006, -007, and −009) at 29, 8, 32, and 25 months, respectively. LPL-006 experienced early disease progression and was lost to follow up 8.8 months after last vaccination, before a post-vaccine bone marrow sample could be obtained. All remaining patients are known to be alive.
+
+## Reduction in clonal tumor subpopulations and their gene expression pathways after vaccination in the mature B-cell, but not in the LPL plasma cell-like compartment
+
+To interrogate vaccine-induced changes directly in the tumor microenvironment, bone marrow samples were obtained a median of 3 months (range 1–13 months) after vaccine treatment from all nine patients, except patient LPL-006. We performed single cell RNA-seq analysis on matched pre- and post-vaccine bone marrow samples, paired with matched single cell BCR and TCR sequencing (Fig. S1B-F). To analyze specific changes in LPL cells following vaccination we separated and re-clustered heterogeneous B-lineage populations and obtained a total of 12 clusters based on differential gene expression (Fig. 1 C, Fig. S2A). To specifically identify clonal tumor cells across various clusters we matched single cell BCRs with the previously identified unique tumor idiotype (Ig VH and VL CDR3) sequences used for manufacturing individualized therapeutic vaccines for each patient. LPL is known to consist of distinct clonal B-cell- and plasma cell-like subpopulations. Clusters 0, 1 and 2, representing mature B cells, were comprised almost entirely of the tumor clonotype, with cluster 1 being the most abundant (Fig. 1 C, right). Plasmablast-like and mature plasma cells (clusters 5 and 10, respectively) were less abundant but also contained relatively high proportions of tumor clonotypes (Fig. 1 C, right). Analysis of paired total B-lineage cells pre- vs. post-vaccine showed significantly reduced frequencies post-vaccine for B-cell cluster 1 but not for the plasma cell-like clusters (Fig. 1 D). This reduction in B-cell frequencies in cluster 1 was entirely attributable to a specific reduction in the tumor clonotype. This reduction in tumor cell clonotype frequencies was observed in all except for three evaluable patients (Fig. 1 E).
+
+Concomitant changes in global gene expression patterns in tumor cells were associated with the reduction of the tumor mature B-cell compartment post-vaccine. Differential gene expression analysis of each of the relevant B-lineage cell clusters revealed a pattern of significant gene downregulation following vaccine in clusters 0, 1 and 2 (Fig. 1 F, top). Among the top downregulated genes were FOS, JUN, ATF3, ATF4, NFKBIA and MAP3K8 which are essential for the growth of B lymphocytes as well as proteins of the EIF (eukaryotic initiation factor) family (EIF4A1, EIF4A2, EIF4A3), GADD34, ribosomal protein L family (RPL4, RPL9, RPL13, RPL21, RP23, PRL27, RPL37, RPL38, RPL10A) and ribosomal proteins family (RPS2, RPS6, RPS9, RPS11, RPS16, RPS20, RPS26, RPS27) which are essential for lymphoma cell proliferation and protein synthesis (Fig. 1 F, top). Furthermore, pathway analysis based on differentially expressed genes identified signaling pathways significantly reduced post vaccine known to be critical for B-cell survival including IL-1, IL-6, IGF-1 and APRIL (Fig. 1 F, bottom left). BCR, PI3K/AKT and ERK/MAPK, which are involved in survival-promoting signaling by mutant MYD88 in WM cells, were also significantly downregulated. Conversely, PPAR signaling, which is known to promote tumor cell apoptosis, and ferroptosis cell death pathways were both upregulated by these clusters. Finally, the analysis predicted overall downregulation of biological processes (z-score > 2, adjusted log p-value > 1.3) including cell survival, viability, proliferation, protein synthesis and RNA transcription and upregulation of necrosis (Fig. 1 F, bottom right). Notably, no global changes were inferred for corresponding plasma cell-like clusters. These observations suggest that tumor subpopulations of LPL within a single patient may be dichotomous in their response to therapeutic vaccine treatment, with mature B-cell subpopulations more susceptible than plasma cell-like cells.
+
+A well-described mechanism of tumor cell resistance to T-cell-mediated killing is the downregulation of expression of HLA family genes, particularly HLA class II genes. To investigate this possibility, we compared expression of HLA family genes in tumor cells in relevant B- and plasma-cell clusters. Consistent with previous reports we observed downregulation of HLA class II family (HLA-DMA, HLA-DMB, HLA-DPA1, HLA-DPB1, HLA-DQA2) gene expression in clusters 5 and 10 containing plasmablast-like and plasma cells in both pre- and post-vaccine samples, but not in B-cell clusters 0 and 1 (Fig. 1 G). Interestingly, there was also a trend towards downregulation of expression of HLA class II (but not HLA class I) genes post-vaccine, compared with pre-vaccine in B-cell cluster 2 tumor cells (Fig. 1 GH). In contrast, no significant changes were observed in tumor expression of T-cell checkpoint ligands, including PDL1 (CD274), and PDL2 (PDCD1LG2) (Fig. 1 H and not shown). We also observed no significant differences in expression of genes of the death receptor family among clonal tumor B-cell or plasma cell-like clusters post-vaccine (Fig S1G). Taken together these observations suggest that plasma-cell subpopulations of clonotypic tumor cells in LPL may exhibit immune evasion to our vaccine therapy by downregulating expression of HLA genes, rather than by activation of T-cell immune checkpoints.
+
+## Paired single cell transcriptomics reveals dynamic changes in T cells in the tumor microenvironment following vaccine treatment.
+
+To investigate vaccine-induced changes in normal immune cells in the bone marrow microenvironment we re-clustered T-cell populations separately and obtained a total of 12 clusters that we identified based on differential gene expression, SingleR software analysis, and the expression of defined gene markers (Fig. 2 A and Fig. S2B-C). These T-cell subpopulations were consistent across all patient samples (not shown). We analyzed changes in T-cell frequencies within each cluster, comparing paired pre- vs. post-vaccine bone marrow samples and observed statistically significant decrease in cell frequencies of naïve CD4 T cells (cluster 0) and trends toward increases in effector memory and terminal effector T cells (clusters 1 and 3, respectively) (Fig. 2 B). We also performed differential gene expression analysis on each T-cell cluster, followed by pathway enrichment analysis pre- vs. post-vaccine using IPA software (Qiagen). We observed significant upregulation (adjusted log p-value > 1.3) of pathways involved in T cell activation, including TCR signaling, PI3K/AKT signaling, integrin signaling and leukocyte extravasation and down-regulation of PD-1/PD-L1 pathway following vaccination in effector T cells (Fig. 2 C). Notably, there were no obvious changes in frequencies or signaling pathways in Treg (Fig. 2 B cluster 9, and 2C).
+
+To analyze the clonal composition of T cells in the microenvironment we used matched single cell TCR-seq data (Fig. 2 A, right panel). A mean of 750 unique T-cell clonotypes (range 141–1596) were identified pre- and post-vaccine each for each patient. Comparing post- vs. pre-vaccine samples, we observed expansion of existing clonotypes in all patients except for two of clinical progressors LPL-005 and −009 (Fig. 2 D). Furthermore, among the 20 most prevalent clonotypes detected post-vaccination the majority increased from low frequency clonotypes that were present before vaccination, except for clinical progressors LPL-005 and −009 (Fig. 2 E). New clonotypes were also detected post-vaccination (overall 4.4%), consistent with increased T-cell clonal diversity. Increasing clonal diversity post-vaccination was also suggested in most patients, as analyzed by individual Shannon entropy scores (Fig. 2 F). Phenotypically, unique or shared T-cell clonotypes expanded in post-vaccine samples localized primarily to clusters enriched for effector memory or effector T cells (not shown). This same pattern was observed for the 20 most abundant post-vaccine clonotypes (Fig. 2 G, left) with cells localized to effector memory T cells and terminal effector T cell clusters. Post-vaccine clusters also showed increased frequencies of cells in the G2-M phase of the cell cycle, consistent with increased proliferation (Fig. 2 G, right). Phenotypically these top 20 post-vaccine clonotypes were primarily CD8 T cells expressing markers affecting activation, differentiation, or proliferation, including CD27, CXCR4, HLA-DR, PIK3RI, REL, and FKBP1A (Fig. 2 H). In contrast, clonotypes detected only in a single T-cell or uniquely in pre-vaccine samples were distributed broadly across all T-cell subpopulations, including naïve CD4 and CD8 T cells, regulatory T cells, Th1/Th2 cells, Th17, and to lesser extent central memory T cells and effector memory phenotypes (not shown). Notably, the top 20 post-vaccine clonotypes showed a mixed pattern of up- and down-regulation of co-inhibitory molecules DUSP2 and TIGIT, respectively, with most, including PD-1, LAG3, and TIM3 (HAVCR2) showing no significant change post-vaccination (Fig. 2 H). Taken together, these results suggest that vaccine therapy induced significant expansion and activation of terminal effector and effector memory T cells within the top 20 TCR clonotypes post-vaccine, with little activation of immune checkpoints.
+
+## Tumor idiotype-specific T-cell immune responses.
+
+To detect idiotype-specific T cell responses elicited by the vaccine treatment, we analyzed T cells isolated directly from the bone marrow tumor microenvironment. T cells were enriched from each patient’s post vaccine sample by negative selection and then stimulated with autologous immortalized normal B cells (as antigen presenting cells, APCs) transfected with either Ig VH and VL sequences (expressed as sFv’s) derived from the respective patient-specific tumor idiotype (used previously for therapeutic vaccine production), or HIV Nef as a negative control, described previously. Multiplex cytokine analysis was performed on culture supernatants. Representative post-vaccination samples are shown from patients achieving minor response, stable disease, or progressive disease clinically (Fig. 2 I). All patients T cells secreted cytokines in an antigen-specific manner, with the exception of the two patients who experienced progressive disease (LPL-005 and −009). Taken together these functional data are consistent with T-cell clonal expansion post-vaccination detected by transcriptomic analysis above.
+
+## Vaccine-induced reduction in cross-talk between immune cell types and tumor cells in the microenvironment.
+
+To infer and analyze global changes in cell-cell communications in the tumor microenvironment after vaccination, we employed comparative CellChat software to analyze signaling interactions among all major cell types in pre- and post-vaccination bone marrow samples. A cell-cell interaction map was constructed using aggregate sc-RNAseq data from all evaluable patients with five major interaction populations: clonal LPL mature B cells, clonal LPL plasma cell-like cells, T/NK cells, myeloid cells, and normal B progenitor cells as controls. From this cell-cell interaction map, the total number of ligand-receptor pairs contributing to communication between any two interacting cell types was analyzed. We observed that the total number of inferred interactions between the five major cell types in the tumor microenvironment significantly decreased post- compared with pre-vaccine (Fig. 3 A), with this same pattern consistently observed between individual pairs of cell types (Fig S3A-C).
+
+To investigate which cell populations contributed to the reduction in inferred interactions, we used network centrality analysis to compare incoming and outgoing interaction strengths (Fig. 3 B). Interestingly, predicted interaction strengths for myeloid and LPL mature B-cell, but not LPL plasma cell-like populations, were most dramatically reduced post-vaccine.
+
+We then analyzed the overall information flow for multiple specific signaling pathways across the pre and post-vaccine datasets. Multiple signaling pathways were implicated as active predominantly in pre- but not post-vaccine samples, including pathways such as APRIL, which is known to promote B- or plasma cell survival, and others with known roles supporting tumor cell proliferation in solid cancers, such as RESISTIN, VEGF, and IL-10, TGFβ and BMP (Fig. 3 C). Moreover, the IL-6 signaling pathway, which promotes IgM secretion and LPL and plasma cell growth via the JAK/STAT pathway, was substantially reduced in post-vaccine samples.
+
+The analysis of individual cell types revealed that myeloid cells mainly contributed to the downregulation of information flow of these signaling pathways (Fig. 3 D and Fig S3D-O). For example, we observed dramatic reductions in predicted outgoing signals provided by myeloid cells for both RESISTIN and APRIL pathways, as well as IL-6, associated with their respective ligand-receptor pairs (Fig. 3 E).
+
+Paradoxically, dichotomous upregulation of the insulin-like growth factor (IGF) signaling axis post-vaccine was inferred by plasma cell, but not mature B cell LPL subpopulations, including both autocrine and paracrine pathways, consistent with a potential mechanism of escape by the former (Fig. 3, D and E). Our scRNAseq data confirmed increased expression of IGF-1 among clonal tumor cells post-vaccine in both plasma cell-like clusters (clusters 5 and 10) but not in any B-cell clusters (0, 1, and 2). We also observed an increased proportion of clonal tumor cells expressing IGF-1 in cluster 10 (Fig S3P).
+
+## Vaccine-induced changes in myeloid cell subpopulations in the tumor microenvironment
+
+Given that vaccination was associated with significantly reduced cell-cell communication patterns in the tumor microenvironment, most pronounced in outgoing signals provided by myeloid cells to clonal LPL cells, we further analyzed subpopulations of myeloid cells by re-clustering them based on differential gene expression analysis from the combined datasets of pre- and post-vaccine bone marrow cells from all patients. We obtained a total of 9 clusters, based on the differential expression of established marker genes (Fig. 4. A, B and fig. S4A). Myeloid cell populations were consistent across all individual patient samples (not shown).
+
+We analyzed changes in cell frequencies per cluster in paired pre- vs. post-vaccine patient samples and observed significant increases in frequencies of CD14− CD16+ non-classical monocytes (cluster 3) and concomitant significant decreases in the frequencies of CD14+ CD16+ intermediate monocytes (cluster 4, Fig. 4 C). Given that monocyte differentiation is believed to proceed from classical CD14+ CD16− to non-classical CD14− CD16+ monocytes via intermediate CD14+ CD16+ monocytes, these results may suggest that monocytes in the tumor microenvironment post vaccination undergo increased differentiation from intermediate to non-classical subpopulations, thereby causing a general skewing away from classical monocytes. This hypothesis was also supported by the trend towards decreasing CD14+ CD16− classical monocyte frequencies (cluster 0) observed post- vs. pre-vaccination, although these differences did not reach statistical significance.
+
+Next, we sought to identify the specific myeloid cell clusters which contributed to the dramatic reductions in outgoing predicted signals provided by myeloid cells to LPL cells by analyzing each of the individual signaling pathways in Fig. 3 D. Comparing post- vs. pre-vaccine samples, most of the signaling pathways affected were associated with reductions in monocytic subpopulations, particularly cluster 3 non-classical monocytes, and to a lesser extent cluster 0 classical monocytes (Fig. 4 D). Changes were also observed across many of the pathways for cluster 1 of mature neutrophils, but these were generally of lesser magnitudes.
+
+Taken together, these data suggest that vaccination was associated with clear reductions in pro-tumoral outgoing signals provided by non-classical monocytes to LPL cells, but with a paradoxical expansion of this myeloid subpopulation.
+
+Finally, because of the availability of potential therapeutic intervention, we also performed cell-cell communication analysis of the CD47-SIRPα pathway which predicted overall decreased signaling after vaccination (Fig. 3 E), despite increased CD47 expression on at least one lymphoid (B-cell cluster 2) and one plasmacytoid (B-cell cluster 10) LPL tumor subpopulation. SIRPα was observed on cluster 0 classical monocytes pre-vaccination, with no significant change in expression post-vaccination (Fig. 4 E).
+
+# Discussion
+
+In the absence of any standard treatment for patients with smoldering phase LPL/WM, patients are typically managed by active surveillance alone. The median time to progression to the symptomatic phase is 3.9 years<sup>62</sup>. Early intervention with a well-tolerated therapeutic agent, such as a vaccine, that could delay progression to symptomatic phase disease, would therefore be highly desirable. The individualized therapeutic DNA vaccines used in this trial appear to be safe, with no patients experiencing dose limiting toxicities (DLTs) or Grade 4 adverse events (AEs). Most of the patients experienced potential clinical benefit, with documentation of stable disease or better for a median of 61 + months, including one patient who achieved a minor response (MR).
+
+The reasons for lack of more robust objective clinical responses were revealed by direct interrogation of the tumor microenvironment by single-cell transcriptome analysis. Comparing paired pre- and post- vaccine bone marrow samples available from eight of the nine patients, we observed a striking dichotomous pattern of significantly reduced numbers of clonal tumor idiotype-expressing B-cells post-vaccine in the majority of patients, but no change in clonally related plasma cell-like clusters of any patient (Fig.<span class="InternalRef" refid="Fig1">1</span>D). Heterogeneity within LPL, consisting of separate mature B-cell and plasma cell-like subpopulations, has been described<sup>63,64</sup>. In our dataset, plasma cell-like cells were detected at lower frequencies, compared with mature B cell-like subpopulations, but partial loss of this fragile cell subpopulation during frozen sample preparation cannot be ruled out<sup>40</sup>. Furthermore, there was an associated pattern of downregulation of expression of genes known to be essential for lymphoma cell proliferation and protein synthesis in tumor clusters of mature B cells, but not of plasma cell-like LPL subpopulations (Fig.<span class="InternalRef" refid="Fig1">1</span>F). Pathway analysis predicted global downregulation of signaling pathways known to be critical for B-cell survival and conversely, upregulation of pathways known to promote tumor cell apoptosis and other forms of cell death, in mature B-cell, but not plasma cell-like LPL clusters, consistent with a vaccine-induced antitumor response against the mature B-cell LPL compartment.
+
+Downregulation of expression of HLA molecules, particularly class II, has been recognized as one immune-evasion mechanism in cancer<sup>39,40</sup>. Indeed, we observed that plasma and plasmablast-like LPL cells expressed low levels of HLA class II genes. This represents a potentially intrinsic mechanism of resistance by plasma cell-like subpopulations, as low levels were observed in both pre- and post-vaccine samples (Fig.<span class="InternalRef" refid="Fig1">1</span>G). In contrast, the trend towards downregulation HLA class II gene expression observed in LPL cluster 2 cells post-vaccine, compared with pre-vaccine, suggests a potential mechanism of immune evasion among mature B-cell subpopulations (Fig.<span class="InternalRef" refid="Fig1">1</span>G and H). In contrast, dichotomous upregulation of the IGF signaling axis post-vaccine was inferred by plasma cell, but not mature B cell LPL subpopulations (Fig.<span class="InternalRef" refid="Fig4">3</span>D and E), suggesting a possible mechanism of acquired resistance to vaccine therapy. The IGF axis has been implicated in acquired drug resistance in various hematologic cancers, and selective IGF-1 receptor inhibitors could block tumor cell proliferation and migration and overcome resistance to treatment of multiple myeloma, and lymphomas with bortezomib, EZH2 inhibitors and crizotinib<sup>66–68</sup>. Furthermore, our finding that the proportion of tumor cells expressing IGF-1 was also increased in one of two plasma cell-like clusters post-vaccine suggests that clonal selection of IGF signaling-dependent tumor clones cannot be excluded (Fig. S3P).
+
+The chemokine-antigen fusion vaccine platform was designed to elicit robust T-cell immunity, by targeting idiotype antigen delivery to chemokine receptors on antigen presenting cells<sup>10,11</sup>. Indeed we observed that vaccine therapy induced dynamic changes in T cells in the tumor microenvironment, consistent with generation of antigen-specific immune responses. Trends toward increases in effector memory and terminal effector phenotypes post-vaccine (Fig.<span class="InternalRef" refid="Fig3">2</span>B and H) were associated with upregulation of pathways involved in T-cell activation (Fig.<span class="InternalRef" refid="Fig3">2</span>C), expansion of individual T-cell clonotypes (Fig.<span class="InternalRef" refid="Fig3">2</span>D, E), increased T-cell clonal diversity (Fig.<span class="InternalRef" refid="Fig3">2</span>F), and functional LPL idiotype-specific cytokine production (Fig.<span class="InternalRef" refid="Fig3">2</span>I).
+
+Tracking the top 20 TCR clonotypes pre- and post-vaccine suggested that they expanded from pre-existing idiotype-specific effector/effector memory cells (Fig.<span class="InternalRef" refid="Fig3">2</span>G), rather than naïve cells primed by vaccine. Recent studies have suggested that Treg cells create an immunosuppressive milieu in WM, the most common subtype of LPL<sup>69</sup>. However, we observed no obvious changes in frequencies or signaling pathways in Treg that would suggest an effect of vaccination on this subpopulation (Fig.<span class="InternalRef" refid="Fig3">2</span>B cluster 9, and 2C). T cells in the microenvironment also showed a mixed pattern of up- and down-regulation of co-inhibitory pathways (Fig.<span class="InternalRef" refid="Fig3">2</span>C), with most showing no significant change post-vaccination (Fig.<span class="InternalRef" refid="Fig3">2</span>H). Taken together, these results suggest little activation of co-inhibitory molecules by vaccine therapy.
+
+Our vaccine also globally altered the levels of the cell-cell communication networks and signaling strength across various other cell populations in the tumor microenvironment. We detected significant downregulation of signaling pathways post-vaccine that likely directly promote growth of LPL cells, such as APRIL and IL-6 which are known to promote B- or plasma cell survival (Fig.<span class="InternalRef" refid="Fig4">3</span>C to E). Other pathways were reduced post-vaccine, such as RESISTIN, which has a known role in supporting proliferation of solid cancers by binding CAP1 receptors. A role for RESISTIN in supporting LPL has not been previously inferred, but it induced multidrug resistance in human multiple myeloma<sup>56</sup>.
+
+Bioinformatic analysis also identified a predominant role for myeloid cells in the tumor microenvironment as a source of vaccine-induced, downregulated pro-tumoral signaling to LPL cells. The global signaling pathways affected were primarily associated with monocytic, rather than granulocytic or dendritic cell subpopulations, particularly non-classical CD14<sup>−</sup>CD16<sup>+</sup> monocytes, and to a lesser extent classical CD14<sup>+</sup>CD16<sup>−</sup> monocytes (Fig.<span class="InternalRef" refid="Fig5">4</span>D). Myeloid cells have been recognized as key component of the immune suppressive microenvironment in solid tumors and this observation has been extended more recently to B-cell tumor microenvironments<sup>70</sup>. The trend we observed towards reduction of classical monocytes may be of particular relevance, as this subpopulation has been recently shown to pro-tumoral in multiple myeloma<sup>71</sup>. Additionally, our gene signature analysis revealed that this cluster may contain myeloid derived suppressor cells (MDSC) which have also been extensively characterized as immune-suppressive (Fig. S4B,<em>57</em>)). Finally, reports indicate that an increased pro-inflammatory myeloid signature is an early step in the development of WM and in monoclonal gammopathy of undetermined significance (MGUS)<sup>72</sup>.
+
+One potential strategy to further overcome resistance to myeloid signaling may be therapeutic blockade of SIRPα-CD47, an emerging checkpoint utilized by cancer cells to evade immune responses. Despite increased expression of CD47 (“do not eat me” signal) on at least one mature B-cell- (cluster 2) and one plasma cell-like (cluster 10) LPL tumor subpopulation after vaccination, CellChat analysis predicted overall decreased signaling to SIRPα on myeloid cells which expression was confirmed on classical monocytes (Fig.<span class="InternalRef" refid="Fig5">4</span>E). Taken together, these results suggest that vaccine therapy was associated with significant reduction of pro-tumoral signaling by myeloid cells in the LPL microenvironment.
+
+Recent improvements in bioinformatics, design, and manufacturing are facilitating the clinical development of individualized neoantigen cancer vaccines. As a prototype, our idiotype neoantigen vaccine demonstrated safety, the ability to significantly reduce clonal mature B-cell, but not plasma cell-like, LPL subpopulations and to favorably perturb the tumor microenvironment. Future functional studies of the pathways affected are needed to confirm mechanisms of resistance elucidated and to design combination strategies to circumvent them. Such strategies could include adding IFNγ or epigenetic drugs, designed to increase HLA molecule expression on plasma cell-like LPL subpopulations<sup>73</sup> and combining neoantigen vaccines with agents that specifically target plasma cells, such as anti CD38 antibodies<sup>74</sup>, or pathways known to promote their growth, such as IGF-1 receptor inhibitors<sup>75</sup>. Furthermore, our data suggest that combinations of these vaccines with myeloid cell checkpoint blockade may be worthwhile. Finally, although little activation of co-inhibitory molecules was observed by vaccine therapy, elevated PD-1 ligands on human WM cells and exhausted CD8 T cells in the WM microenvironment have been reported by others<sup>76</sup>, suggesting that there may still be a role for therapeutic T-cell checkpoint blockade combined with this vaccine.
+
+# Materials and Methods
+
+## Experimental Design
+
+This was an open-label phase I trial conducted at the University of Texas M.D. Anderson Cancer Center. Patients were enrolled between March 2015 and August 2017. Patients received a series of three DNA vaccinations with autologous MIP-3α fused lymphoma idiotype at 4-week intervals intradermally into both thighs by needle-free injection device (PharmaJet, Golden, CO). Consecutive patients were enrolled to dose cohorts 1 (500 µg) and 2 (2500 µg) according to a standard 3 + 3 statistical design.
+
+The primary objective was to evaluate the safety profile of the vaccine and to determine its maximum tolerated dose (MTD). Clinical laboratory testing, patient reporting, and physical examination findings were used to evaluate adverse events (AEs), including serious adverse events (SAEs). Toxicities were graded according to the NCI Common Toxicity Criteria v4.0. Dose limiting toxicity (DLT) was defined as a ≥ grade 2 allergic reaction, ≥ grade 2 autoimmune reaction, and any grade 3 or 4 toxicity except for fever, grade 4 fever which subsequently required 50% dose reduction. MTD was defined as the highest dose level in which 6 patients have been treated with less than 2 instances of DLT.
+
+Initial disease response was assessed one month after the final vaccination according to International WM consensus panel response criteria from the 6th International Workshop<sup>77</sup>.
+
+## Patients
+
+Eligible patients had a diagnosis of smoldering lymphoplasmacytic lymphoma (LPL) confirmed by tissue diagnosis, with a monoclonal heavy and light chain as determined by flow cytometry. All participants were required to be able to provide informed consent. Patients were excluded if they had a history of autoimmune diseases except for Hashimoto’s thyroiditis, or either a positive antinuclear antibody titer or anti-double strand DNA titer. Conduct of this trial was approved by the University of Texas M.D. Anderson Cancer Center institutional review board (protocol 2009−0465) and was carried out in accordance with the Declaration of Helsinki and the International Conference on Harmonization Guidelines for Good Clinical Practice.
+
+## Bone marrow aspirates and cryopreservation
+
+All patients had up to 10 ml of bone marrow aspirated before treatment for morphological sorting, immunophenotyping, and characterization. A second 15 ml bone marrow aspirate sample was obtained from the contralateral side for additional tumor cell banking for vaccine production and/or correlative analysis. Bone marrow mononuclear cells (BMMNCs) were isolated by density gradient centrifugation using Ficoll. Mononuclear cells were washed three times with 45 mL PBS (800g), counted, and viably cryopreserved in 10% DMSO.
+
+## Generation of individualized DNA vaccines
+
+LPL B cells generally comprise > 30% of the total B-lymphocyte population in bone marrow<sup>78</sup>. The unique lymphoma idiotype for each patient’s tumor was identified based on the clonal amplification of a predominant Ig heavy and light chain V(D)J sequence<sup>12</sup>. MIP-3α fused lymphoma idiotype plasmid DNA vaccines were prepared from each patient’s bone marrow as described previously<sup>79</sup>. The plasmid DNA was then amplified and purified from *E. coli* according to Good Manufacturing Practices (GMP) standards by FUJIFILM Diosynth Biotechnologies U.S.A., Inc., (College Station, TX).
+
+### Bone marrow sample processing for single cell RNA sequencing.
+
+Cryopreserved bone marrow samples pre and post vaccine were processed together for each patient. Cells were thawed at 37<sup>o</sup>C and resuspended in culture media. Dead cell removal was performed using StemCell EasySep Annexin V kit (Cat#17899). Cell were resuspended in PBS with 0.04% BSA and count and viability was determined using automated cell counted (BioRad) prior to loading onto 10x Genomics Chip (Chromium Single Cell 5’ Kit). LPL-008 pre and post samples after thawing, as described above, were used for staining with hashtag antibodies (Biolegend TotalSeq-C0251, Cat #394661 and TotalSeq-C0252, Cat #393663) according to manufacturer’s protocol. Pre and post vaccine samples were mixed in a 1:1 ratio and loaded onto the 10x Genomics chip. Libraries were prepared using Chromium Single Cell 5’ Kit (10x Genomics) for gene expression, TCR and BCR and QC was performed using Agilent High Sensitivity DNA kit on the Agilent 2100 bioanalyzer. Libraries were sequenced on Illumina HiSeq 2500 (Read1: 26 cycles, i7 index: 8 cycles, Read2: 101 cycles, Sequencing depth: 20,000/ read pairs per cycle for gene expression, Read1: 151 cycles, i7 Index: 8 cycles, Read2: 151 cycles, sequencing depth: 5,000/ read pairs per cell for TCR/BCR). Raw sequencing reads were processed using CellRanger pipeline using default settings (10x Genomics, software version 3.1.0).
+
+### Single cell RNA-seq data analysis.
+
+Filtered gene expression matrices generated by CellRanger pipeline were used for downstream analysis using Seurat package (version 3)<sup>81,82</sup>. Loupe Browser was used to resolve Hashtag identities of cellular barcodes (10x Genomics version 3). Cells were filtered based on total mRNA counts, total genes detected, and a mitochondrial content. We obtained an average of 2163 cells per sample (range 830–4680), totaling 36,777 cells in the dataset. Dataset were normalized and 2000 most variable genes were selected. Datasets were merged sequentially using IntegrateData function in Seurat. To improve clustering resolution, a dataset “A single cell immune cell atlas of human hematopoietic system” from Human Cell Atlas website (<span class="ExternalRef"><span class="RefSource">https://data.humancellatlas.org/</span></span>) was obtained. Dataset file was subset to obtain samples of normal healthy bone marrow of 15 oldest individuals available based on metadata files, subject to identical filtering and processing, followed by sequential data integration as above. Feature scaling, PCA, clustering and UMAP analysis were performed on merged datasets using integrated assay. Identification of markers of cell populations were done with FindAllMarkers function. A total of 25 clusters were identified (Fig. S1B-F) which were consistent across all patient samples (Fig. S1D). For single cell BCR and single cell TCR sequencing, filtered contig annotation CellRanger output files were processed with custom R script and were added to the Seurat object as metadata based on cell barcodes (Fig. S1E). For further sub-analysis, B cell, T cell and myeloid cell populations were separated out using subset function and re-analyzed with PCA, clustering and UMAP as described above. Differential gene expression pre vs post vaccine was performed using FindMarkers function with default parameters. Volcano plots were created using EnhancedVolcano package<sup>84</sup> Pathway analysis and biological processes analysis was performed using IPA software (Qiagen) (adjusted log p-value > 1.3) and visualized using R. Heatmaps, dot plots and violin plots were generated with Seurat. TCR clonotype analysis was performed using immunarch R package according to instructions<sup>85</sup>.
+
+### Bone marrow T cell stimulation and cytokine 30-Plex human panel.
+
+Cells were thawed at 37<sup>o</sup>C and washed with ImmunoCult-XF T cell media (StemCell Cat#10981) supplemented with 1U/mL of DNAse (Thermo Sci Cat#EN0521). Cells were centrifuged and resuspended in EasySep buffer (StemCell Cat#20144), followed by exposure to CD20 magnetic microbeads (Miltenyi Biotec Cat#130-0910104). T cells were enriched by negative selection (EasySep Cat#10981) and resuspended in T cell media supplemented with IL-2 at 50U/mL (clinical grade), IL-7 at 25ng/mL (Miltenyi Biotec, Cat#130-095-361) and IL-15 at 25ng/mL (Miltenyi Biotec, Cat#130-095-762) to a concentration of 2 x10<sup>6</sup> cells/mL. 125ul of cell suspension were plated in V-bottom 96-well plates. Autologous patient derived immortalized B cells were generated as described previously<sup>52,83</sup>. Cells were suspended in PBS, irradiated with 1500 rads for 5min, spun down and resuspended in Neon Buffer R (Invitrogen). mRNA encoding the patient’s unique patient Idiotype, scFv-MITD or irrelevant antigen (Nef-MITD) were prepared as described previously<sup>52</sup> and electroporation was performed using Neon MPK5000 system (Invitrogen) with settings: pulse voltage: 1150 V, pulse width: 30 ms, pulse number: 2. Electroporated APC were resuspended in T cell media supplemented with IL-2, IL-7 and IL-15 to a final concentration of 2x10<sup>6</sup> cells/mL and 125ul of cells were added to V-bottom 96-well plate containing T-cell suspensions. Cells were co-cultured in 37<sup>o</sup>C incubator for a total of 10 days. At day 3 and 7, 125ul of culture media was replaced with fresh T cell media supplemented with IL-2, IL-7 and IL-15 at 50U/mL, 25ng/mL and 25ng/mL for day 3 and 100U/mL, 25ng/mL and 25ng/mL on day 7 respectively. On day 10 cells were washed in 500 µL of PBS, spun down and supernatants were snap-frozen at -80<sup>o</sup>C. Supernatants were used for 30-Plex human panel analysis according to manufacturer’s protocol.
+
+## Cell–cell communication analysis
+
+The CellChat package was used to infer cell–cell communications between the following cell types via interaction-network analysis: LPL (mature B-lymphoid), LPL (plasma cell-like), myeloid, T- and NK, and normal B progenitor cells. A Seurat object was used as input for CellChat following standard protocols (<span class="ExternalRef"><span class="RefSource">https://github.com/sqjin/CellChat</span></span>. NetAnalysis_signalingRole_heatmap) and was used to compute the comparison of overall signaling pathways in pre- vs. post-vaccine samples. Circle plots were generated via netVisual_aggregate, vertex.size = groupSize, respectively. In circle plots, edge color indicates the source of outgoing signal, and edge weight is proportional to interaction strength. Incoming and outgoing strength were calculated via CellChat function with default parameter. The information flow of each signaling pathway was defined by the sum of communication probabilities among all pairs of cell groups in the inferred network and was generated via CellChat rankNet. Statistical comparisons of relative interaction strengths were performed by Student’s t-test
+
+# References
+
+1. Kyle, R.A., Ansell, S.M. & Kapoor, P. Prognostic factors and indications for treatment of Waldenstrom's Macroglobulinemia. *Best Pract Res Clin Haematol* **29**, 179-186 (2016).
+
+2. Rojas, L.A. *et al.* Personalized RNA neoantigen vaccines stimulate T cells in pancreatic cancer. *Nature* **618**, 144-150 (2023).
+
+3. Desrichard, A., Snyder, A. & Chan, T.A. Cancer Neoantigens and Applications for Immunotherapy. *Clin Cancer Res* **22**, 807-812 (2016).
+
+4. Schumacher, T.N. & Schreiber, R.D. Neoantigens in cancer immunotherapy. *Science* **348**, 69-74 (2015).
+
+5. Stevenson, G.T., Elliott, E.V. & Stevenson, F.K. Idiotypic determinants on the surface immunoglobulin of neoplastic lymphocytes: a therapeutic target. *Fed Proc* **36**, 2268-2271 (1977).
+
+6. Stevenson, G.T. & Stevenson, F.K. Antibody to a molecularly-defined antigen confined to a tumour cell surface. *Nature* **254**, 714-716 (1975).
+
+7. Sirisinha, S. & Eisen, H.N. Autoimmune-like antibodies to the ligand-binding sites of myeloma proteins. *Proc Natl Acad Sci U S A* **68**, 3130-3135 (1971).
+
+8. Lynch, R.G., Graff, R.J., Sirisinha, S., Simms, E.S. & Eisen, H.N. Myeloma proteins as tumor-specific transplantation antigens. *Proc Natl Acad Sci U S A* **69**, 1540-1544 (1972).
+
+9. Khodadoust, M.S. *et al.* Antigen presentation profiling reveals recognition of lymphoma immunoglobulin neoantigens. *Nature* **543**, 723-727 (2017).
+
+10. Khodadoust, M.S. *et al.* B-cell lymphomas present immunoglobulin neoantigens. *Blood* **133**, 878-881 (2019).
+
+11. Kwak, L.W. *et al.* Induction of immune responses in patients with B-cell lymphoma against the surface-immunoglobulin idiotype expressed by their tumors. *N Engl J Med* **327**, 1209-1215 (1992).
+
+12. Bendandi, M. *et al.* Complete molecular remissions induced by patient-specific vaccination plus granulocyte-monocyte colony-stimulating factor against lymphoma. *Nat Med* **5**, 1171-1177 (1999).
+
+13. Schuster, S.J. *et al.* Vaccination with patient-specific tumor-derived antigen in first remission improves disease-free survival in follicular lymphoma. *J Clin Oncol* **29**, 2787-2794 (2011).
+
+14. Freedman, A. *et al.* Placebo-controlled phase III trial of patient-specific immunotherapy with mitumprotimut-T and granulocyte-macrophage colony-stimulating factor after rituximab in patients with follicular lymphoma. *J Clin Oncol* **27**, 3036-3043 (2009).
+
+15. Levy, R. *et al.* Active idiotypic vaccination versus control immunotherapy for follicular lymphoma. *J Clin Oncol* **32**, 1797-1803 (2014).
+
+16. Biragyn, A., Tani, K., Grimm, M.C., Weeks, S. & Kwak, L.W. Genetic fusion of chemokines to a self tumor antigen induces protective, T-cell dependent antitumor immunity. *Nat Biotechnol* **17**, 253-258 (1999).
+
+17. Biragyn, A. *et al.* Toll-like receptor 4-dependent activation of dendritic cells by beta-defensin 2. *Science* **298**, 1025-1029 (2002).
+
+18. Remstein, E.D., Hanson, C.A., Kyle, R.A., Hodnefield, J.M. & Kurtin, P.J. Despite apparent morphologic and immunophenotypic heterogeneity, Waldenstrom's macroglobulinemia is consistently composed of cells along a morphologic continuum of small lymphocytes, plasmacytoid lymphocytes, and plasma cells. *Semin Oncol* **30**, 182-186 (2003).
+
+19. Janz, S. Waldenstrom macroglobulinemia: clinical and immunological aspects, natural history, cell of origin, and emerging mouse models. *ISRN Hematol* **2013**, 815325 (2013).
+
+20. van Dam, H. & Castellazzi, M. Distinct roles of Jun : Fos and Jun : ATF dimers in oncogenesis. *Oncogene* **20**, 2453-2464 (2001).
+
+21. Hebron, E. *et al.* MAP3K8 kinase regulates myeloma growth by cell-autonomous and non-autonomous mechanisms involving myeloma-associated monocytes/macrophages. *Br J Haematol* **160**, 779-784 (2013).
+
+22. Kayastha, F. *et al.* Novel eIF4A1 inhibitors with anti-tumor activity in lymphoma. *Mol Med* **28**, 101 (2022).
+
+23. Lee, Y.Y., Cevallos, R.C. & Jan, E. An upstream open reading frame regulates translation of GADD34 during cellular stresses that induce eIF2alpha phosphorylation. *J Biol Chem* **284**, 6661-6673 (2009).
+
+24. Kovalski, J.R., Kuzuoglu-Ozturk, D. & Ruggero, D. Protein synthesis control in cancer: selectivity and therapeutic targeting. *EMBO J* **41**, e109823 (2022).
+
+25. Garlanda, C. & Mantovani, A. Interleukin-1 in tumor progression, therapy, and prevention. *Cancer Cell* **39**, 1023-1027 (2021).
+
+26. Elsawa, S.F. *et al.* Comprehensive analysis of tumor microenvironment cytokines in Waldenstrom macroglobulinemia identifies CCL5 as a novel modulator of IL-6 activity. *Blood* **118**, 5540-5549 (2011).
+
+27. Xie, J. *et al.* IGF-IR determines the fates of BCR/ABL leukemia. *J Hematol Oncol* **8**, 3 (2015).
+
+28. Elsawa, S.F. & Ansell, S.M. Cytokines in the microenvironment of Waldenstrom's macroglobulinemia. *Clin Lymphoma Myeloma* **9**, 43-45 (2009).
+
+29. Treon, S.P. *et al.* Ibrutinib in previously treated Waldenstrom's macroglobulinemia. *N Engl J Med* **372**, 1430-1440 (2015).
+
+30. Argyropoulos, K.V. *et al.* Clonal B cells in Waldenstrom's macroglobulinemia exhibit functional features of chronic active B-cell receptor signaling. *Leukemia* **30**, 1116-1125 (2016).
+
+31. Palomba, M.L. Learning from Waldenstrom Macroglobulinemia. *Best Pract Res Clin Haematol* **29**, 135 (2016).
+
+32. Treon, S.P. *et al.* Genomic Landscape of Waldenstrom Macroglobulinemia and Its Impact on Treatment Strategies. *J Clin Oncol* **38**, 1198-1208 (2020).
+
+33. Leleu, X. *et al.* The Akt pathway regulates survival and homing in Waldenstrom macroglobulinemia. *Blood* **110**, 4417-4426 (2007).
+
+34. Ray, D.M., Akbiyik, F. & Phipps, R.P. The peroxisome proliferator-activated receptor gamma (PPARgamma) ligands 15-deoxy-Delta12,14-prostaglandin J2 and ciglitazone induce human B lymphocyte and B cell lymphoma apoptosis by PPARgamma-independent mechanisms. *J Immunol* **177**, 5068-5076 (2006).
+
+35. Wang, W. *et al.* CD8(+) T cells regulate tumour ferroptosis during cancer immunotherapy. *Nature* **569**, 270-274 (2019).
+
+36. Johnson, D.B. *et al.* Melanoma-specific MHC-II expression represents a tumour-autonomous phenotype and predicts response to anti-PD-1/PD-L1 therapy. *Nat Commun* **7**, 10582 (2016).
+
+37. Roemer, M.G.M. *et al.* Major Histocompatibility Complex Class II and Programmed Death Ligand 1 Expression Predict Outcome After Programmed Death 1 Blockade in Classic Hodgkin Lymphoma. *J Clin Oncol* **36**, 942-950 (2018).
+
+38. Kreiter, S. *et al.* Mutant MHC class II epitopes drive therapeutic immune responses to cancer. *Nature* **520**, 692-696 (2015).
+
+39. Green, M.R. *et al.* Mutations in early follicular lymphoma progenitors are associated with suppressed antigen presentation. *Proc Natl Acad Sci U S A* **112**, E1116-1125 (2015).
+
+40. Han, G. *et al.* Follicular Lymphoma Microenvironment Characteristics Associated with Tumor Cell Mutations and MHC Class II Expression. *Blood Cancer Discov* **3**, 428-443 (2022).
+
+41. Silacci, P., Mottet, A., Steimle, V., Reith, W. & Mach, B. Developmental extinction of major histocompatibility complex class II gene expression in plasmocytes is mediated by silencing of the transactivator gene CIITA. *J Exp Med* **180**, 1329-1336 (1994).
+
+42. Ghosh, N., Gyory, I., Wright, G., Wood, J. & Wright, K.L. Positive regulatory domain I binding factor 1 silences class II transactivator expression in multiple myeloma cells. *J Biol Chem* **276**, 15264-15268 (2001).
+
+43. Singh, N. *et al.* Impaired Death Receptor Signaling in Leukemia Causes Antigen-Independent Resistance by Inducing CAR T-cell Dysfunction. *Cancer Discov* **10**, 552-567 (2020).
+
+44. Aran, D. *et al.* Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage. *Nat Immunol* **20**, 163-172 (2019).
+
+45. Chiffelle, J. *et al.* T-cell repertoire analysis and metrics of diversity and clonality. *Curr Opin Biotechnol* **65**, 284-295 (2020).
+
+46. Khan, A.B. *et al.* Redirection to the bone marrow improves T cell persistence and antitumor functions. *J Clin Invest* **128**, 2010-2024 (2018).
+
+47. Saraiva, D.P., Jacinto, A., Borralho, P., Braga, S. & Cabral, M.G. HLA-DR in Cytotoxic T Lymphocytes Predicts Breast Cancer Patients' Response to Neoadjuvant Chemotherapy. *Front Immunol* **9**, 2605 (2018).
+
+48. Kim, E.H. & Suresh, M. Role of PI3K/Akt signaling in memory CD8 T cell differentiation. *Front Immunol* **4**, 20 (2013).
+
+49. Visekruna, A., Volkov, A. & Steinhoff, U. A key role for NF-kappaB transcription factor c-Rel in T-lymphocyte-differentiation and effector functions. *Clin Dev Immunol* **2012**, 239368 (2012).
+
+50. Dubois, S. *et al.* Distinct pathways involving the FK506-binding proteins 12 and 12.6 underlie IL-2-versus IL-15-mediated proliferation of T cells. *Proc Natl Acad Sci U S A* **100**, 14169-14174 (2003).
+
+51. Good, C.R. *et al.* An NK-like CAR T cell transition in CAR T cell dysfunction. *Cell* **184**, 6081-6100 e6026 (2021).
+
+52. Wang, Z. *et al.* Immortalized B Cells Transfected with mRNA of Antigen Fused to MITD (IBMAM): An Effective Tool for Antigen-Specific T-Cell Expansion and TCR Validation. *Biomedicines* **11** (2023).
+
+53. Jin, S. *et al.* Inference and analysis of cell-cell communication using CellChat. *Nat Commun* **12**, 1088 (2021).
+
+54. Vu, R. *et al.* Wound healing in aged skin exhibits systems-level alterations in cellular composition and cell-cell communication. *Cell Rep* **40**, 111155 (2022).
+
+55. Moreaux, J. *et al.* BAFF and APRIL protect myeloma cells from apoptosis induced by interleukin 6 deprivation and dexamethasone. *Blood* **103**, 3148-3157 (2004).
+
+56. Pang, J. *et al.* Resistin induces multidrug resistance in myeloma by inhibiting cell death and upregulating ABC transporter expression. *Haematologica* **102**, 1273-1280 (2017).
+
+57. Deb, A., Deshmukh, B., Ramteke, P., Bhati, F.K. & Bhat, M.K. Resistin: A journey from metabolism to cancer. *Transl Oncol* **14**, 101178 (2021).
+
+58. Schmidt, C.S. *et al.* Enhanced B cell expansion, survival, and humoral responses by targeting death receptor 6. *J Exp Med* **197**, 51-62 (2003).
+
+59. Gupta, M. *et al.* Elevated serum IL-10 levels in diffuse large B-cell lymphoma: a mechanism of aberrant JAK2 activation. *Blood* **119**, 2844-2853 (2012).
+
+60. Timmins, M.A. & Ringshausen, I. Transforming Growth Factor-Beta Orchestrates Tumour and Bystander Cells in B-Cell Non-Hodgkin Lymphoma. *Cancers (Basel)* **14** (2022).
+
+61. Guilliams, M., Mildner, A. & Yona, S. Developmental and Functional Heterogeneity of Monocytes. *Immunity* **49**, 595-613 (2018).
+
+62. Bustoros, M. *et al.* Progression Risk Stratification of Asymptomatic Waldenstrom Macroglobulinemia. *J Clin Oncol* **37**, 1403-1411 (2019).
+
+63. Growkova, K. *et al.* Waldenstrom's macroglobulinemia: Two malignant clones in a monoclonal disease? Molecular background and clinical reflection. *Eur J Haematol* **99**, 469-478 (2017).
+
+64. Jakubikova, J. *et al.* High-Dimensional Heterogeneity of Waldenström Macroglobulinemia within Its Immune Tumor Microenvironment. *Blood* **134**, 3975 (2019).
+
+65. Samur, M.K., Szalat, R. & Munshi, N.C. Single-Cell Profiling in Multiple Myeloma: Insights, Problems, and Promises. *Blood* (2023).
+
+66. Mitsiades, C.S. *et al.* Inhibition of the insulin-like growth factor receptor-1 tyrosine kinase activity as a therapeutic strategy for multiple myeloma, other hematologic malignancies, and solid tumors. *Cancer Cell* **5**, 221-230 (2004).
+
+67. Vishwamitra, D., George, S.K., Shi, P., Kaseb, A.O. & Amin, H.M. Type I insulin-like growth factor receptor signaling in hematological malignancies. *Oncotarget* **8**, 1814-1844 (2017).
+
+68. Kuhn, D.J. *et al.* Targeting the insulin-like growth factor-1 receptor to overcome bortezomib resistance in preclinical models of multiple myeloma. *Blood* **120**, 3260-3270 (2012).
+
+69. Sacco, A. *et al.* Targeting the immune microenvironment in Waldenstrom macroglobulinemia via halting the CD40/CD40-ligand axis. *Blood* **141**, 2615-2628 (2023).
+
+70. Pham, L.V., Pogue, E. & Ford, R.J. The Role of Macrophage/B-Cell Interactions in the Pathophysiology of B-Cell Lymphomas. *Front Oncol* **8**, 147 (2018).
+
+71. Zavidij, O. *et al.* Single-cell RNA sequencing reveals compromised immune microenvironment in precursor stages of multiple myeloma. *Nat Cancer* **1**, 493-506 (2020).
+
+72. Kaushal, A. *et al.* Aberrant Extrafollicular B Cells, Immune Dysfunction, Myeloid Inflammation, and MyD88-Mutant Progenitors Precede Waldenstrom Macroglobulinemia. *Blood Cancer Discov* **2**, 600-615 (2021).
+
+73. Christopher, M.J. *et al.* Immune Escape of Relapsed AML Cells after Allogeneic Transplantation. *N Engl J Med* **379**, 2330-2341 (2018).
+
+74. Raedler, L.A. Darzalex (Daratumumab): First Anti-CD38 Monoclonal Antibody Approved for Patients with Relapsed Multiple Myeloma. *Am Health Drug Benefits* **9**, 70-73 (2016).
+
+75. Khan, S. *et al.* A phase-1 trial of linsitinib (OSI-906) in combination with bortezomib and dexamethasone for the treatment of relapsed/refractory multiple myeloma. *Leuk Lymphoma* **62**, 1721-1729 (2021).
+
+76. Sun, H. *et al.* Single-cell profiles reveal tumor cell heterogeneity and immunosuppressive microenvironment in Waldenstrom macroglobulinemia. *J Transl Med* **20**, 576 (2022).
+
+77. Owen, R.G. *et al.* Response assessment in Waldenstrom macroglobulinaemia: update from the VIth International Workshop. *Br J Haematol* **160**, 171-176 (2013).
+
+78. Kobrin, C.B. & Kwak, L.W. Development of vaccine strategies for the treatment of B-cell malignancies. *Cancer Invest* **15**, 577-587 (1997).
+
+79. Cha, S.C., Qin, H., Sakamaki, I. & Kwak, L. Cloning variable region genes of clonal lymphoma immunoglobulin for generating patient-specific idiotype DNA vaccine. *Methods Mol Biol* **1139**, 289-303 (2014).
+
+80. Tiller, T. *et al.* Efficient generation of monoclonal antibodies from single human B cells by single cell RT-PCR and expression vector cloning. *J Immunol Methods* **329**, 112-124 (2008).
+
+81. Stuart, T. *et al.* Comprehensive Integration of Single-Cell Data. *Cell* **177**, 1888-1902 e1821 (2019).
+
+82. Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. *Nat Biotechnol* **36**, 411-420 (2018).
+
+83. Qazilbash, M.H. *et al.* A randomized phase 2 trial of idiotype vaccination and adoptive autologous T-cell transfer in patients with multiple myeloma. *Blood* **139**, 1289-1301 (2022).
+
+84. Blighe K, Rana S, Lewis M (2023). EnhancedVolcano: Publication-ready volcano plots with enhanced colouring and labeling.
+
+85. Nazarov V, Tsvetkov V, Fiadziushchanka S, Rumynskiy E, Popov A, Balashov I, Samokhina M (2023). *immunarch: Bioinformatics Analysis of T-Cell and B-Cell Immune Repertoires*. https://immunarch.com/, https://github.com/immunomind/immunarch.
+
+# Supplementary Files
+
+- [20090465protocol.pdf](https://assets-eu.researchsquare.com/files/rs-3315017/v1/474e2823cfcab25996157d51.pdf)  
+  Protocol
+
+- [SupplementaryFigures.pdf](https://assets-eu.researchsquare.com/files/rs-3315017/v1/e4ecb9ea9f672060f4c9db94.pdf)

253d8b2ab690f7725938b1be16cba2c0f3d79f60e4fc32a5156fc9ff5c8fee9e/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Biochemical characterizations of the interaction of ATG16L1 with FIP200. (a) A schematic diagram showing the domain organizations of FIP200, ATG16L1 and mammalian ATG8 family proteins. In this drawing, the interactions of ATG16L1 with FIP200 and mammalian ATG8 family proteins are highlighted and indicated by two-way arrows. (b)Size exclusion chromatography-based analysis of the interaction of FIP200 Claw with Trx-tagged ATG16L1(78-247). (c) Multi-angle light-scattering analysis of the purified ATG16L1(78-247)/FIP200 Claw complex showing the relative light scattering signals as a function of elution volume. The molecular mass error is the \ufb01tted error obtained from the data analysis software. (d) Sequence alignment analysis of the FIR of ATG16L1 with the currently known FIP200 Claw-binding regions of NAP1, SINTBAD, CCPG1, NDP52, p62, NBR1 and Optineurin from human species. In this alignment, the highly conserved acidic residues (Asp, Glu or potentially phosphorylated Ser residue) and the following two conserved hydrophobic residues are boxed and highlighted with black triangles and stars, respectively. (c) The fluorescence polarization (FP)-based assay measuring the binding affinity of FIP200 Claw with FITC-labelled ATG16L1 FIR. The Kd value is the fitted dissociation constant with standard errors obtained by using the one-site binding model to fit the FP data. (f) Size exclusion chromatography-based analysis of the interaction of FIP200 Claw with Trx-tagged ATG16L1(78-235).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "The molecular mechanism of FIP200 and ATG16L1 interaction. (a) Ribbon diagram showing the overall structure of the dimeric FIP200 Claw/ATG16L1 FIR complex. In this drawing, two FIP200 Claw molecules are colored in orange and marine, while two bound ATG16L1 FIR motifs are colored in magenta and hot-pink, respectively. (b) Ribbon representation showing the structural comparison of apo-form FIP200 Claw dimer (green, PDB ID: 6DCE) with the FIP200 Claw/ATG16L1 FIR complex (blue/magenta). In this drawing, the two dimeric structures are overlaid by aligning selected one FIP200 Claw monomer in these two structures. (c) Ribbon-stick model showing the detailed interactions between FIP200 Claw and ATG16L1 FIR. The hydrogen bonds and salt bridges involved in the interaction are shown as dotted lines. (d) The combined surface charge representation and the ribbon-stick model showing the charge-charge interactions between FIP200 Claw and ATG16L1 FIR in the solved complex structure. (e) The measured binding af\ufb01nities between ATG16L1 FIR and FIP200 Claw, or their mutants by FP-based binding assays.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Biochemical and structural characterizations of the interactions of ATG16L1 FIR with mammalian ATG8 family proteins. (a) ITC-based measurement of the binding af\ufb01nity of GABARAPL1 with Trx-tagged ATG16L1 FIR. (b) Ribbon diagram showing the overall structure of the GABARAPL1/ATG16L1 FIR complex. In this drawing, the GABARAPL1 molecule is colored in green, while ATG16L1 FIR in magenta. (c) The ribbon-stick model showing the detailed interactions between GABARAPL1 and ATG16L1 FIR. The hydrogen bonds and salt bridges involved in the interaction are shown as dotted lines. (d) The combined surface charge representation and the ribbon-stick model showing the charge-charge interactions between GABARAPL1 and ATG16L1 FIR. (e) The combined surface representation and the ribbon-stick model showing the hydrophobic interactions between GABARAPL1 and ATG16L1 FIR. In this drawing, ATG16L1 FIR is displayed in the ribbon-stick model, and GABARAPL1 is showed in surface representation colored by different amino acid types. Specifically, the hydrophobic amino acid residues in the surface model of GABARAPL1 are drawn in yellow; the positively charged residues are drawn in blue; the negatively charged residues are drawn in red, and the uncharged polar residues are drawn in gray. (f) The measured binding af\ufb01nities between ATG16L1 FIR and six mammalian ATG8 family proteins or their mutants by ITC-based binding assays.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Biochemical characterizations of the interactions of ATG16L1 variants with FIP200 Claw and GABARAPL1. (a) Size exclusion chromatography coupled with SDS-PAGE analysis of the Trx-ATG16L1(78-247)/GST-GABARAPL1 complex incubated with increasing molar ratio of FIP200 Claw proteins. (b) The SDS-PAGE combined with Coomassie-blue staining analyses of the protein components of the indicated \"fraction 1\" and \"fraction 2\" fractions collected from the analytical gel filtration chromatography experiment at different molar ratios of FIP200 Claw in panel a. (c-d) Size exclusion chromatography coupled with SDS-PAGE analysis of Trx-tagged ATG16L1 FIR wild-type with GABARAPL1 or FIP200 Claw. In this panel, \u201cSum\u201d stands for the theoretical sum of Trx-tagged ATG16L1 FIR wild-type and GABARAPL1 or FIP200 Claw profiles, while \u201cMixture\u201d stands for the Trx-tagged ATG16L1 FIR wild-type and GABARAPL1 or FIP200 Claw mixture sample. (e-f) Size exclusion chromatography coupled with SDS-PAGE analysis of Trx-tagged ATG16L1 FIR D239R/I240F mutant with GABARAPL1 or FIP200 Claw. In this panel, \u201cSum\u201d stands for the theoretical sum of Trx-tagged ATG16L1 FIR D239R/I240F mutant and GABARAPL1 or FIP200 Claw profiles, while \u201cMixture\u201d stands for the Trx-tagged ATG16L1 FIR D239R/I240F mutant and GABARAPL1 or FIP200 Claw mixture sample. (g-h) Size exclusion chromatography coupled with SDS-PAGE analysis of Trx-tagged ATG16L1 FIR I240Q/I243Q mutant with GABARAPL1 or FIP200 Claw. In this panel, \u201cSum\u201d stands for the theoretical sum of Trx-tagged ATG16L1 FIR I240Q/I243Q mutant and GABARAPL1 or FIP200 Claw profiles, while \u201cMixture\u201d stands for the Trx-tagged ATG16L1 FIR I240Q/I243Q mutant and GABARAPL1 or FIP200 Claw mixture sample.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpg",
+        "caption": "Cell-based assays of the interactions of ATG16L1 with GABARAPL1 and FIP200 in autophagy. (a)Co-immunoprecipitation assays showing that point mutations of key interface residues of GABARAPL1 observed in the GABARAPL1/ATG16L1 FIR complex structure essentially disrupt their specific interaction in cells. The developed ATG16L1 D239R/I240F mutant rather than the ATG16L1 I240Q/I243Q mutant can well interact with GABARAPL1. \u201cIB\u201d means immunoblotting. (b) Co-immunoprecipitation assays showing that point mutations of both devised ATG16L1 mutant and key interface residues of FIP200 observed in the FIP200 Claw/ATG16L1 FIR complex structure decrease or essentially disrupt the specific interaction between FIP200 and ATG16L1 in cells. (c) Western blot-based measurements of the LC3B lipidation and p62 degradation levels in ATG16L1-knockout HeLa cells rescued with AcGFP1-tagged WT ATG16L1, AcGFP1-tagged ATG16L1 D239R/I240F mutant (DRIF), or AcGFP1-tagged ATG16L1 I240Q/I243Q mutant (IQIQ) treated for 4 hours using normal medium, amino acid starvation medium (AASM), or AASM with bafilomycin A1 at 400 nM. Cell lysates were immunoblotted for the indicated proteins. (d) The levels of p62 and \u03b2-actin in panel cwere also quantified in ImageJ and normalized to the 16KO+DRIF cells treated using \u00a0AASM with bafilomycin A1 at 400 nM. The data is presented as means\u00b1SEM from three independent experiments. Statistical analyses were both performed in GraphPad Prism 9 by two-way analysis of variance (ANOVA) followed by Bonferroni multiple comparisons test, and P value style is P = 0.1234 [not significant (ns)], *P = 0.0332, **P = 0.0021, ***P = 0.0002, and ****P < 0.0001. (e) A proposed model depicting the function of ATG16L1 in canonical autophagy.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

253d8b2ab690f7725938b1be16cba2c0f3d79f60e4fc32a5156fc9ff5c8fee9e/preprint/preprint.md

@@ -0,0 +1,158 @@
+# Abstract
+
+Macroautophagy maintains cellular and organismal homeostasis, and entails *de novo* synthesis of double-membrane autophagosome. The effective formation of autophagosome requires the recruitment of the ATG12~ATG5-ATG16L1 complex to the pre-autophagosomal structure by relevant ATG16L1-binding autophagic factors including FIP200. However, the molecular mechanism governing the specific interaction of ATG16L1 with FIP200 remains elusive. Here, we uncover that ATG16L1 contains a FIP200-interacting region (FIR), which not only can directly bind to the Claw domain of FIP200, but also can serve as an atypical ATG8-interacting motif (AIM) to selectively recognize mammalian ATG8 family proteins. We determine the high-resolution crystal structures of ATG16L1 FIR in complex with FIP200 Claw and GABARAPL1, respectively, and elucidate the molecular mechanism underlying the interactions of ATG16L1 with FIP200 and ATG8 family proteins. To distinguish the precise contribution of FIP200 from ATG8 family proteins for binding to ATG16L1 FIR in autophagy, we develop a unique ATG16L1 mutant that can exclusively interact with ATG8 family proteins but not FIP200. Finally, using relevant cell-based functional assays, we demonstrate that the interaction of ATG16L1 with FIP200 is indispensable for the effective autophagic flux. In conclusion, our findings provide mechanistic insights into the interactions of ATG16L1 with FIP200 and ATG8 family proteins, and are valuable for further understanding the function of ATG16L1 in autophagy.
+
+**Biological sciences/Structural biology/X-ray crystallography**  
+**Biological sciences/Cell biology/Autophagy/Macroautophagy**  
+Autophagy  
+ATG16L1  
+FIP200  
+GABARAPL1  
+ATG8 family proteins  
+WIPI2
+
+# Introduction
+
+Macroautophagy (hereafter referred to as autophagy) is a well-conserved and lysosome-dependent catabolic process that recycles undesired or harmful cytosolic components for providing essential building blocks to maintain cellular homeostasis in mammals<sup>1–4</sup>. Autophagy necessitates *de novo* synthesis of double-membrane autophagosome through the sophisticated cooperation of a series of autophagy-related (ATG) proteins<sup>3,5,6</sup>. During the amino acid starvation-induced canonical autophagy, calcium transients trigger liquid-liquid phase separation of FAK family kinase-interacting protein of 200 kDa (FIP200) and the FIP200-containing Unc-51-like kinase (ULK) complex in the pre-autophagosomal structure (PAS) to initiate autophagosome formation<sup>7</sup>. Concurrently, the class III phosphatidylinositol 3-kinase complex I (PI3KC3-C1) is translocated to the PAS, and is subsequently activated to phosphorylate phosphatidylinositol (PI) to generate phosphatidylinositol-3-phosphate (PI3P)<sup>8</sup>. Meanwhile, both the FIP200 subunit of the ULK complex and WD repeat domain phosphoinositide-interacting protein 2 (WIPI2) that senses the PI3P signal from PI3KC3-C1 can regulate the targeting of the E3-like ATG12~ATG5-ATG16L1 complex (the ATG16L1 complex) to the PAS<sup>9–12</sup>. Subsequently, the ATG16L1 complex catalyzes the phosphatidylethanolamine (PE) lipidation of ATG8 family proteins to facilitate the elongation of PAS to form autophagosome<sup>13–15</sup>. Although the molecular mechanism underpinning the recruitment of the ATG16L1 complex by WIPI2 is well elucidated in previous studies<sup>12,16</sup>, how FIP200 specifically interacts with ATG16L1 to recruit the ATG16L1 complex remains elusive. Moreover, why autophagy deploys two distinct approaches to recruit the ATG16L1 complex is still enigmatic.
+
+In addition to non-selective canonical autophagy, accumulating studies have uncovered considerable selective autophagy processes mediated by different types of autophagy receptors<sup>17–23</sup>. Currently identified autophagy receptors in mammals, such as p62, NBR1, Optineurin, CCPG1, NDP52 and TAX1BP1, all encompass a cargo-associating domain that can specially recognize certain types of autophagic cargoes as well as a unique ATG8-interacting motif (AIM) that recognizes the ATG8 family proteins known as LC3A, LC3B, LC3C, GABARAP, GABARAPL1 and GABARAPL2 in mammals<sup>19,20,24–26</sup>. Moreover, recent studies revealed that in order to induce *in situ* autophagosome formation to envelope and sequester the targeting cargoes, many autophagy receptors can directly interact with FIP200 to recruit the ULK complex<sup>27–31</sup>. Particularly, some autophagy receptors can directly bind to the Claw domain of FIP200 through their respective FIP200-interacting region (FIR) for recruiting the ULK complex<sup>28,29,31,32</sup>. Since the sequence pattern of AIM bears a striking resemblance to that of FIR, the AIM motifs of many autophagy receptors can also function as FIR to interact with FIP200 Claw<sup>28,29</sup>. However, due to the high similarity, it is challenging to dissect the individual contribution of FIP200 and ATG8 family proteins for binding to these autophagy receptors. Interestingly, similar to aforementioned autophagy receptors, mammalian ATG16L1 can function as an adaptor to specifically recognize invading pathogens or pathogen-containing vacuoles through its C-terminal WD40 repeats domain to mediate the antibacterial selective autophagy (also named as xenophagy)<sup>33–36</sup>. Moreover, a previous elegant study of the network organization of the human autophagy system well demonstrated that ATG16L1 can directly bind to ATG8 family proteins<sup>37</sup>. However, how ATG16L1 interacts with ATG8 family proteins and the detailed relationship between FIP200 and ATG8 family proteins in binding to ATG16L1 are largely unknown.
+
+In this study, we discover that ATG16L1 contains a typical FIR motif for directly interacting with the Claw domain of FIP200. Importantly, in addition to binding to FIP200 Claw, ATG16L1 FIR can also serve as a noncanonical AIM motif to selectively recognize mammalian ATG8 family proteins. Moreover, our determined crystal structures of the ATG16L1 FIR/FIP200 Claw complex and the ATG16L1 FIR/GABARAPL1 complex not only uncover the detailed molecular mechanism governing the specific interactions of ATG16L1 with FIP200 and ATG8 family members, but also reveal that FIP200 and ATG8 family proteins are mutually exclusive in binding to ATG16L1. On this basis, we devise a specific ATG16L1 mutant that can well interact with ATG8 family proteins but not FIP200. Finally, using this ATG16L1 mutant together with relevant cell-based functional assays, we demonstrate that the interaction of ATG16L1 FIR with FIP200 is indispensable for the effective autophagic flux in canonic autophagy.
+
+# Results
+
+Previous studies uncovered that the ATG16L1(229–242) region adjacent to the WIPI2-binding site 1 (WBS1) of ATG16L1 is essential for the interaction between ATG16L1 and FIP200 (Fig. <span class="InternalRef" refid="Fig1">1</span> a) <sup><span citationid="CR9" class="CitationRef">9</span>, <span citationid="CR10" class="CitationRef">10</span></sup>. Consistently, using size exclusion chromatography (SEC)-based assays, we demonstrated that the highly conserved central region of ATG16L1, the ATG16L1(78–247) fragment (<b>Supplementary Fig. 1</b>), can readily interact with FIP200(1490–1594) (hereafter referred to as FIP200 Claw) (Fig. <span class="InternalRef" refid="Fig1">1</span> b). Further multi-angle light scattering analysis revealed that the purified ATG16L1(78–247)/FIP200 Claw complex forms a stable 2:2 stoichiometric complex in solution (Fig. <span class="InternalRef" refid="Fig1">1</span> c). Strikingly, detailed sequence alignment analysis of the FIP200-binding region of ATG16L1 with that of NAP1, SINTBAD, CCPG1 and NDP52, all of which were demonstrated to directly interact with FIP200 Claw <sup><span citationid="CR28" class="CitationRef">28</span>, <span citationid="CR32" class="CitationRef">32</span>, <span citationid="CR38" class="CitationRef">38</span>, <span citationid="CR39" class="CitationRef">39</span></sup>, unraveled that the ATG16L1(235–247) region conforms to the criteria for a FIP200 Claw-binding FIR motif <sup><span citationid="CR28" class="CitationRef">28</span></sup>, likely representing a putative FIR motif (hereafter referred to as ATG16L1 FIR) (Fig. <span class="InternalRef" refid="Fig1">1</span> d). Indeed, further quantitative fluorescent polarization (FP)-based assay revealed that ATG16L1 FIR can specifically interact with FIP200 Claw with a <em>K</em>d value of about ~ 1.33 µM (Fig. <span class="InternalRef" refid="Fig1">1</span> e). In contrast, the removal of ATG16L1 FIR totally disrupted the association of ATG16L1(78–247) with FIP200 Claw (Fig. <span class="InternalRef" refid="Fig1">1</span> f). Taken together, all these data clearly demonstrated that ATG16L1 contains a conserved FIR motif to directly interact with the Claw domain of FIP200.
+
+Then, we intended to determine the ATG16L1 FIR/FIP200 Claw complex structure to uncover how ATG16L1 FIR binds to the Claw domain of FIP200. After numerous attempts, we finally managed to solve the crystal structure of ATG16L1 FIR in complex with FIP200 Claw at 1.61 Å resolution (<b>Supplementary Table 1</b>). The determined ATG16L1 FIR/FIP200 Claw complex structure is composed of two ATG16L1 FIR molecules and a FIP200 Claw dimer, forming a unique 2:2 stoichiometric hetero-tetramer (Fig. <span class="InternalRef" refid="Fig2">2</span> a), in line with our aforementioned multi-angle light scattering result (Fig. <span class="InternalRef" refid="Fig1">1</span> c). In the complex structure, except for a distinct N-terminal α-helix that was only found in one of the two FIP200 Claw domains owing to crystal packing (<b>Supplementary Fig. 2a</b>), the two monomeric FIP200 Claw domains adopt a highly similar core architecture assembled by six β-strands and one α-helix (<b>Supplementary Fig. 2b</b>). In parallel, the two ATG16L1 FIR molecules in the complex structure mainly form two short β-strands that symmetrically augment the β4-strand of two FIP200 Claw domains in an anti-parallel manner (Fig. <span class="InternalRef" refid="Fig2">2</span> a). Further structural comparison analyses showed that the overall structure of the monomeric FIP200 Claw domain in the ATG16L1 FIR/FIP200 Claw complex is highly akin to that of the <em>apo</em>-form FIP200 Claw domain (PDB ID: 6DCE) (<b>Supplementary Fig. 2c</b>), whereas the binding of ATG16L1 FIR to FIP200 Claw induces an obvious conformational rearrangement of the FIP200 Claw dimer (Fig. <span class="InternalRef" refid="Fig2">2</span> b). Notably, similar phenomena were also observed in our previous studies <sup><span citationid="CR28" class="CitationRef">28</span>, <span citationid="CR32" class="CitationRef">32</span></sup>.
+
+In the ATG16L1 FIR/FIP200 Claw complex structure, each ATG16L1 FIR molecule packs extensively with a highly electropositive and hydrophobic concave groove that is situated adjacent to the β4/β5 connecting region of the monomeric FIP200 Claw, burying a total interface area of ~ 505 Å<sup><span citationid="CR2" class="CitationRef">2</span></sup> (Fig. <span class="InternalRef" refid="Fig2">2</span> a). Further careful analyses of the molecular interface in the ATG16L1 FIR/FIP200 Claw complex structure revealed that the hydrophobic side chain of ATG16L1 I240 deeply inserts into a hydrophobic pocket formed by the side chains of C1565, A1567, F1574 and F1582 residues of FIP200 Claw (Fig. <span class="InternalRef" refid="Fig2">2</span> c and <b>Supplementary Fig. 2d</b>), and concurrently, the hydrophobic side chains of ATG16L1 I243 and V244 residues form hydrophobic contacts with the aromatic side chain of Y1564 and the aliphatic side chain group of K1581 from FIP200 (Fig. <span class="InternalRef" refid="Fig2">2</span> c). Furthermore, the negatively charged D238, D239 and E241 residues of ATG16L1 FIR form specific hydrogen bonding and charge-charge interactions with the positively charged K1569, R1573 and K1568 residues of FIP200 Claw (Fig. <span class="InternalRef" refid="Fig2">2</span> c and <b>d</b>). In addition, the backbone groups of E241, I243 and V244 residues of ATG16L1 FIR couple with the backbone groups of Q1566 and Y1564 residues of FIP200 to form five specific backbone hydrogen bonds (Fig. <span class="InternalRef" refid="Fig2">2</span> c). In keeping with their critical structural roles, all these key binding interface residues of ATG16L1 and FIP200 are highly conserved in different eukaryotic species (<b>Supplementary Figs. 1 and 3</b>). Consistent with our structural results, further FP-based assays showed that point mutations of key interface residues of FIP200 Claw, such as the Y1568A, K1569A, R1573E and F1574Q mutations, all significantly decrease the specific interaction of FIP200 Claw with ATG16L1 FIR (Fig. <span class="InternalRef" refid="Fig2">2</span> e and <b>Supplementary Fig. 4</b>). Reciprocally, point mutations of key interface residues from ATG16L1 FIR, including the D238R, D239R, I240Q, E241R and I243Q mutations of ATG16L1 FIR, all largely attenuate or essentially disrupt the interaction between FIP200 Claw and ATG16L1 FIR (<b>Supplementary Fig. 5</b>). Therefore, all those mutagenesis-based biochemical assays confirmed the specific interaction between ATG16L1 FIR and FIP200 Claw observed in the solved ATG16L1 FIR/FIP200 Claw complex structure.
+
+Since the reported FIR motifs have been demonstrated to directly engage with mammalian ATG8 family proteins <sup><span citationid="CR28" class="CitationRef">28</span>, <span citationid="CR32" class="CitationRef">32</span></sup>, we suspected that ATG16L1 FIR may also recognize mammalian ATG8 orthologs. Indeed, our ITC analyses revealed that ATG16L1 FIR can directly interact with six mammalian ATG8 members, and preferentially binds to GABARAPL1 and LC3C with the <em>K</em>d value of ~ 4.59 µM and ~ 6.27 µM, respectively (Fig. <span class="InternalRef" refid="Fig3">3</span> a, <span class="InternalRef" refid="Fig3">3</span> f and <b>Supplementary Fig. 6</b>). Then, we chose GABARAPL1 as a representative to further characterize the binding mechanism of ATG16L1 FIR with ATG8 family proteins. Our SEC results showed that the FIR-containing ATG16L1(78–247) and ATG16L1(235–247) fragments can readily interact with GABARAPL1 (<b>Supplementary Fig. 7a</b> and <b>b</b>). In contrast, the ATG16L1(78–235) fragment, which lacks the FIR motif of ATG16L1, is completely unable to interact with GABARAPL1 (<b>Supplementary Fig. 7c</b>). Concurrently, we also utilized NMR to validate the interaction between GABARAPL1 and ATG16L1. Titrations of <sup><span citationid="CR15" class="CitationRef">15</span></sup>N-labelled GABARAPL1 with un-labelled ATG16L1(235–247) or ATG16L1(78–247) proteins showed that many peaks in the <sup><span citationid="CR1" class="CitationRef">1</span></sup>H-<sup><span citationid="CR15" class="CitationRef">15</span></sup>N HSQC spectra of GABARAPL1 undergo significant chemical shift changes or peak-broadenings, confirming that the FIR motif-containing ATG16L1 fragments can directly bind to GABARAPL1 (<b>Supplementary Fig. 8</b>). Taken together, all these data clearly demonstrated that ATG16L1 FIR can function as an unconventional AIM motif to directly recognize mammalian ATG8 orthologs.
+
+To further uncover the detailed molecular mechanism underlying the selective recognitions of ATG8 family proteins by ATG16L1 FIR, we determined the high-resolution crystal structure of GABARAPLI in complex with ATG16L1 FIR (<b>Supplementary Table 1</b>). In the determined GABARAPL1/ATG16L1 FIR complex structure, GABARAPL1 adopts a typical ATG8 family protein architecture assembled by a ubiquitin-like structural core and two preceding N-terminal α-helices (Fig. <span class="InternalRef" refid="Fig3">3</span> b). In the complex structure, the clearly defined ATG16L1 FIR motif contains 10 highly conserved residues spanning from Q236 to D245 of ATG16L1 (<b>Supplementary Fig. 9a</b>), and adopts an extended configuration to occupy the canonical AIM-binding groove of GABARAPL1, which is mainly formed by the β1-strand, β2-strand, α2-helix and α3-helix of GABARAPL1 (Fig. <span class="InternalRef" refid="Fig3">3</span> c-e). Further structural analyses of the binding interface of the GABARAPL1/ATG16L1 FIR complex revealed that the interaction between GABARAPL1 and ATG16L1 FIR is mainly mediated by both polar and hydrophobic interactions (Fig. <span class="InternalRef" refid="Fig3">3</span> c). In particular, the negatively charged side chains of ATG16L1 D239, E241 interact with the positively charged side chains of K48, K46 and R67 of GABARAPL1 to form three charge-charge interactions, and the backbone amide and oxygen groups of ATG16L1 E241 together with the backbone amide group of ATG16L1 I243 form three backbone hydrogen bonds with the GABARAPL1 K48 and L50 residues (Fig. <span class="InternalRef" refid="Fig3">3</span> c and <b>d</b>). In addition, the GABARAPL1/ATG16L1 FIR interaction is further strengthened by four specific hydrogen bonds, two of which are formed between the backbone oxygen groups of D238 and I240 residues of ATG16L1 and the positively charged side chain of GABARAPL1 K46, while the other two are mediated by the backbone oxygen group of ATG16L1 I243 and the positively charged side chain of GABARAPL1 R28 (Fig. <span class="InternalRef" refid="Fig3">3</span> c). In parallel, the hydrophobic side chains of ATG16L1 I240 and V242 residues pack against a hydrophobic groove of GABARAPL1 formed by the hydrophobic side chains of I21, P30, L50, F104 and the aliphatic portion of the side chain of GABARAPL1 K48 residue (Fig. <span class="InternalRef" refid="Fig3">3</span> c and <b>e</b>). Moreover, the hydrophobic side chain of ATG16L1 I243 occupies a hydrophobic pocket assembled by the side chains of Y49, V51, P52, L55, F60 and L63 residues of GABARAPL1 (Fig. <span class="InternalRef" refid="Fig3">3</span> c and <b>e</b>). Consistently, all these key binding interface residues of ATG16L1 FIR and GABARAPL1 are highly conserved during evolution (<b>Supplementary Figs. 1 and 10</b>). Using ITC-based assays, we further verified the specific interactions between interface residues of GABARAPL1 and ATG16L1 FIR observed in the complex structure. Individual point mutations of key residues involved in the binding interface of GABARAPL1/ATG16L1 FIR complex either from GABARAPL1 or ATG16L1 FIR, such as the I21Q, K48E, L50Q and R67E mutations of GABARAPL1 (Fig. <span class="InternalRef" refid="Fig3">3</span> f and <b>Supplementary Fig. 11</b>), or the D239R, I240Q, E241R, V242E and I243Q mutations of ATG16L1 FIR (Fig. <span class="InternalRef" refid="Fig3">3</span> f and <b>Supplementary Fig. 12</b>), all dramatically decrease or essentially disrupt the association of GABARAPL1 with ATG16L1 FIR. Notably, previous studies showed that the core unconventional AIM motif of NAP1, NDP52 or TAX1BP1 consists of an acidic Asp followed by four consecutive hydrophobic residues, all of which participate in hydrophobic interactions with the relevant ATG8 family proteins <sup><span citationid="CR26" class="CitationRef">26</span>, <span citationid="CR32" class="CitationRef">32</span>, <span citationid="CR40" class="CitationRef">40</span></sup>. Interestingly, unlike that of the unconventional AIM motifs of NAP1, NDP52 and TAX1BP1, the third residue of the core unconventional AIM motif of ATG16L1 is an acidic Glu residue (<b>Supplementary Fig. 9b</b>), which is directly involved in the interaction with GABARAPL1 (Fig. <span class="InternalRef" refid="Fig3">3</span> c). Therefore, ATG16L1 FIR represents a unique type of unconventional AIM motif.
+
+Based on our aforementioned structural analyses, ATG16L1 FIR adopts essentially the same key residues to recognize FIP200 Claw and GABARAPL1, such as the D239, I240, E241 and I243 residues (Figs. <span class="InternalRef" refid="Fig2">2</span> c, <span class="InternalRef" refid="Fig3">3</span> c and <b>Supplementary Fig. 1</b>). Thus, FIP200 and ATG8 family proteins should be mutually exclusive in binding to ATG16L1 FIR. As expected, further SEC coupled with SDS-PAGE assays confirmed that FIP200 Claw and GABARAPL1 are competitive in binding to ATG16L1 (Fig. <span class="InternalRef" refid="Fig4">4</span> a and <b>b</b>). Given that ATG16L1 FIR is C-terminally adjacent to the WIPI2-binding site 1 (WBS1) of ATG16L1 (Fig. <span class="InternalRef" refid="Fig1">1</span> a), we also tested the relationship between WIPI2 and FIP200 or GABARAPL1 in binding to ATG16L1. Intriguingly, we revealed that FIP200 Claw but not GABARAPL1, can form a stable ternary complex with WIPI2 and the ATG16L1(207–247) fragment that contains both WBS1 and FIR (<b>Supplementary Fig. 13a</b>), suggesting that WIPI2 and FIP200 can simultaneously bind to ATG16L1, consistent with a previous study <sup><span citationid="CR11" class="CitationRef">11</span></sup>. In contrast, WIPI2 and ATG8 family protein, such as GABARAPL1, are competitive in binding to ATG16L1(207–247) (<b>Supplementary Fig. 13b</b>), likely due to the potential steric hindrance.
+
+Based on our previous study <sup><span citationid="CR28" class="CitationRef">28</span></sup>, the consensus FIR motif (Ψ-Θ-Χ1-Χ2-Φ, where Ψ represents a phosphorylated Ser/Thr residue or an acidic Asp, Glu, Θ represents a bulk hydrophobic Ile, Leu, Met or aromatic Phe, Tyr, Trp residue, Φ represents a hydrophobic Leu, Ile or Val residue, and Χ1/Χ2 represents any residue), bears a striking resemblance to the sequence pattern of the core AIM sequence. Actually, the similarity between FIR and AIM severely interferes with our assessment of the individual contribution made by FIP200 and ATG8 family proteins when binding to ATG16L1 FIR. Therefore, it is necessary to develop selective ATG16L1 mutants that can exclusively interact with FIP200 or ATG8 family proteins. Based on our previous biochemical and structural characterizations of relevant FIR and AIM motifs <sup><span citationid="CR28" class="CitationRef">28</span>, <span citationid="CR41" class="CitationRef">41</span></sup>, we realized that the first residue preceding the consensus core sequence of AIM, which is corresponding to the Ψ residue of FIR, can tolerate basic residues to some extent, such as Arg residue, while FIR cannot. In addition, the Θ residue of AIM prefers an aromatic Phe, Tyr, or Trp residue rather than a hydrophobic Ile, Leu, or Val residue. Eventually, we managed to devise a selective ATG16L1 D239R/I240F (DRIF) double mutant that can solely recognize GABARAPL1 but not FIP200 (Fig. <span class="InternalRef" refid="Fig4">4</span> c-f). In addition, we also obtained a unique ATG16L1 I240Q/I243Q (IQIQ) double mutant, which binds neither GABARAPL1 nor FIP200 (Fig. <span class="InternalRef" refid="Fig4">4</span> c, d, g, h). Importantly, in agreement with our biochemical data (Fig. <span class="InternalRef" refid="Fig4">4</span> c-h), further co-immunoprecipitation assays showed that the DRIF double mutant can selectively bind to GABARAPL1 but not FIP200 in cells (Fig. <span class="InternalRef" refid="Fig5">5</span> a and <b>b</b>). Concomitantly, the IQIQ double mutation of ATG16L1 completely abolishes the specific interactions of ATG16L1 with FIP200 and GABARAPL1 in cells (Fig. <span class="InternalRef" refid="Fig5">5</span> a and <b>b</b>). Notably, in order to avoid potential ATG8ylation of ATG16L1 in cells <sup><span citationid="CR42" class="CitationRef">42</span></sup>, the C-terminal Gly-Lys of GABARAPL1 was removed in these assays. Unfortunately, albeit with numerous attempts, we failed to obtain an ATG16L1 mutant, which can only bind to FIP200 with a comparable binding ability as the wild-type ATG16L1 but not GABARAPL1.
+
+To further unravel the functional relevance of ATG16L1 FIR in autophagy, we back-transfected the <em>ATG16L1</em> knockout HeLa cell line, which was generated in our previous study <sup><span citationid="CR12" class="CitationRef">12</span></sup>, with relevant plasmids to stably express the wild-type ATG16L1 or relevant ATG16L1 mutants, such as the DRIF mutant of ATG16L1 that only loses the FIP200-binding ability, or the ATG16L1 IQIQ mutant that simultaneously loses the abilities for interacting with FIP200 and ATG8 family proteins. In line with our previous study <sup><span citationid="CR12" class="CitationRef">12</span></sup>, the autophagic flux in the <em>ATG16L1</em> knockout cells was effectively rescued by the wild-type ATG16L1. However, both ATG16L1 DRIF and ATG16L1 IQIQ mutants only partially restored amino acid starvation-induced LC3B lipidation, and were essentially unable to recover the autophagic degradation of p62 (Fig. <span class="InternalRef" refid="Fig5">5</span> c and <b>d</b>), underscoring an indispensable role of ATG16L1 FIR in starvation-induced autophagy. Taken together, these cell-based functional data clearly demonstrated that the interaction between ATG16L1 FIR and FIP200 is essential for the effective autophagic flux in canonical autophagy.
+
+# Discussion
+
+In this work, we uncovered that in addition to interacting with FIP200 Claw, ATG16L1 FIR can also serve as an unconventional AIM to recognize ATG8 family proteins. Furthermore, our biochemical and structural analyses revealed that ATG16L1 FIR can selectively bind to six mammalian ATG8 family proteins, but preferentially bind to GABARAPL1 and LC3C (Fig. <span class="InternalRef" refid="Fig3">3</span> f and <b>Supplementary Fig. 6</b>). Of note, detailed sequence alignment analysis elucidated that several key interface residues for interacting with ATG16L1 are quite different among six mammalian ATG8 family proteins (<b>Supplementary Fig. 14</b>). For instance, the residue corresponding to the R28 residue of GABARAPL1 that forms two hydrogen bonds with the backbone oxygen group of ATG16L1 I243, is a Lys residue in the LC3 subfamily; the residues corresponding to the bulk hydrophobic L55 and F60 residues in GABARAPL1 are two much smaller Val and Leu residues in LC3A and LC3B (<b>Supplementary Fig. 14</b>). The identification of these non-conserved interface residues among different ATG8 family proteins is likely to rationalize the selective recognition of different mammalian ATG8 orthologs by ATG16L1 FIR. Notably, based on this study together with previous reports <sup><span citationid="CR28" class="CitationRef">28</span>, <span citationid="CR32" class="CitationRef">32</span>, <span citationid="CR40" class="CitationRef">40</span></sup>, the X1 position of FIR or AIM can accommodate acidic Glu residue as well as hydrophobic residues, including Ile, Val and Cys (Fig. <span class="InternalRef" refid="Fig1">1</span> d), for engaging with FIP200 or ATG8 family proteins. It is noteworthy that the X1 residues in the FIR and/or AIM of SINTBAD and p62 are Ser/Thr residues that might undergo phosphorylation (Fig. <span class="InternalRef" refid="Fig1">1</span> d). Accordingly, whether there might be a potential regulatory role of X1 phosphorylation in tuning the interactions of SINTBAD, p62 or other related autophagic factors with FIP200 and ATG8 family proteins remains an open question that is worthwhile to be addressed in the future.
+
+Based on our solid biochemical and structural results (Figs. <span class="InternalRef" refid="Fig2">2</span> and <span class="InternalRef" refid="Fig3">3</span>), ATG16L1 FIR adopts many identical interface residues to interact with FIP200 Claw and ATG8 family proteins. Therefore, the simple deletion or mutagenesis of ATG16L1 used to disrupt the ATG16L1/FIP200 interaction in previous functional studies actually leads to the loss of the interactions of ATG16L1 with both FIP200 and ATG8 family proteins <sup><span additionalcitationids="CR10" citationid="CR9" class="CitationRef">9</span>–<span citationid="CR11" class="CitationRef">11</span>, <span citationid="CR36" class="CitationRef">36</span></sup>. Consequently, the exact function of FIP200 for binding to ATG16L1 in autophagy remained unclear in previous studies. In this study, by developing and utilizing the ATG16L1 DRIF mutant that solely loses its ability to interact with FIP200 but not ATG8 family proteins, we confidently elucidated that the ATG16L1/FIP200 interaction is essential for the amino acid starvation-induced canonical autophagy. Given that the FIR/AIM motif exhibits universally in many autophagy receptors <sup><span citationid="CR28" class="CitationRef">28</span>, <span citationid="CR29" class="CitationRef">29</span></sup>, the DRIF mutation strategy utilized for ATG16L1 in this study might also be employed to evaluate the respective roles of FIP200 and ATG8 family proteins in binding to relevant autophagy receptors during selective autophagy.
+
+Previous functional studies well established that mammalian ATG16L1 can function as an adaptor to specifically recognize invading pathogens or pathogen-containing vacuoles through its C-terminal WD40 repeats domain to mediate xenophagy processes <sup><span citationid="CR33" class="CitationRef">33</span>, <span citationid="CR34" class="CitationRef">34</span>, <span citationid="CR36" class="CitationRef">36</span>, <span citationid="CR43" class="CitationRef">43</span></sup>. Meanwhile, when ectopically targeted to the mitochondria, a fragment containing the ATG16L1 FIP200-binding region is sufficient to initiate <em>de novo</em> autophagosome biogenesis through the recruitment of FIP200-containing ULK complex <sup><span citationid="CR30" class="CitationRef">30</span></sup>. Intriguingly, the deficiency of FIP200 blocks mitophagy rather than ATG16L1-mediated xenophagy, implying the dispensable role of the FIP200/ATG16L1 interaction in ATG16L1-mediated xenophagy <sup><span citationid="CR34" class="CitationRef">34</span></sup>. Based on our discovery that FIP200 and ATG8 family proteins are mutually exclusive in binding to ATG16L1 FIR, we suspected that during ATG16L1-mediated xenophagy, once the ATG16L1 complex is recruited onto the bacteria-containing vacuole, ATG16L1 FIR is likely to be rapidly saturated by surrounding excessive lipidated ATG8 family proteins, thereby depriving of the opportunity of ATG16L1 to interact with FIP200. In contrast, the mitochondria-localized ATG16L1 FIP200-binding fragment lacks E3-like enzyme activity to generate PE-conjugated ATG8 family proteins in the vicinity, thereby enabling its recruitment of the ULK complex via binding to FIP200. Unfortunately, the lack of specific ATG16L1 mutant, which selectively binds to FIP200 but not ATG8 family proteins, precludes further evaluations of our hypothesis.
+
+Notably, p62-ubiquitin condensates recruit the ULK complex upon the interaction between p62 FIR and FIP200 Claw, thereby promoting the autophagy-dependent degradation of ubiquitinated cargoes <sup><span citationid="CR27" class="CitationRef">27</span></sup>. In this study, our biochemical and structural data demonstrated that ATG16L1 FIR is directly involved in the interaction with FIP200 Claw (Figs. <span class="InternalRef" refid="Fig1">1</span> e, <span class="InternalRef" refid="Fig2">2</span> and <span class="InternalRef" refid="Fig4">4</span> d). Thus, the deletion of FIP200 Claw not only abolishes the association between p62 and FIP200, but also eliminates the interaction of FIP200 with ATG16L1 as well as many other autophagy receptors, such as TAX1BP1 that can indirectly associate with p62 through NBR1 <sup>44</sup>. Therefore, the related functional data derived from the deprivation of FIP200 Claw in the previous study should be interpreted with caution <sup><span citationid="CR27" class="CitationRef">27</span></sup>.
+
+In mammals, ATG16L1 is equipped with two distinct WIPI2-binding sites and one FIP200-interacting region to facilitate the normal progression of autophagy <sup><span citationid="CR9" class="CitationRef">9</span>, <span citationid="CR10" class="CitationRef">10</span>, <span citationid="CR12" class="CitationRef">12</span></sup>. In contrast, ATG16 in yeast encompasses only one conserved site to bind ATG21, the yeast homologue of WIPI2 <sup>45,46</sup>. In addition, the direct interaction between the ATG1 complex and the ATG16 complex is mediated by the N-terminal segment of yeast ATG12 rather than mammalian ATG16L1 FIR <sup><span citationid="CR47" class="CitationRef">47</span></sup>. Thus, although the detailed binding mechanisms are different, the associations of the ATG16/ATG16L1 complex with the ATG1/ULK complex and ATG21/WIPI2 in canonical autophagy are well conserved from yeast to mammals. Based on our functional data in this study, the ATG16L1 FIR/FIP200 interaction is essential for starvation-induced canonical autophagy (Fig. <span class="InternalRef" refid="Fig5">5</span> c and <b>d</b>). Therefore, the expeditious autophagosome biogenesis in canonical autophagy is unlikely to be adequately explained by merely two distinct linear hierarchical pathways started from the ULK complex. Accordingly, we proposed a positive feedback loop model to illustrate the function of ATG16L1 in canonical autophagy. In this model, the FIP200-containing ULK complex at PAS initiates the translocation of the PI3KC3-C1 complex (Fig. <span class="InternalRef" refid="Fig5">5</span> e). Subsequently, the activated PI3KC3-C1 complex results in the redistribution of WIPI2 through the generation of substantial PI3P molecules (Fig. <span class="InternalRef" refid="Fig5">5</span> e). Whereafter, WIPI2 recruits and activates the ATG16L1 complex, which inversely promotes the recruitment of additional ULK complexes to PAS through a binding mechanism between ATG16L1 FIR and FIP200 Claw as uncovered in this study, thereby facilitating the rapid expansion and closure of phagophore membrane to form autophagosome (Fig. <span class="InternalRef" refid="Fig5">5</span> e).
+
+# Materials and Methods
+
+## Materials
+
+HEK293T and HeLa cell lines were kindly provided by Prof. Junying Yuan from Interdisciplinary Research Center on Biology and Chemistry, CAS, Shanghai, China. The full length human *WIPI2b*, *ATG8* s, *ATG16L1* and *FIP200* genes were obtained from Prof. Jiahuai Han from School of Life Sciences, Xiamen University, Xiamen, China. The synthetic peptide "EQDDDIEVIVDET" (ATG16L1 FIR) was purchased from the China Peptides company, and the purity of the commercially synthesized peptides was > 98%.
+
+## Protein expression and purification
+
+The DNA fragments encoding human FIP200 (residues 1490–1594) and WIPI2b (residues 13–362 without 265–297) were cloned into pET-SUMO-3C vector or pET-32M-3C vector (modified versions of pET-32a vector containing an N-terminal SUMO or Trx tag). The DNA fragments encoding ATG16L1 (residues 207–247, 78–235, 235–247 and 78–247) were all cloned into pACYC-Trx1-3C vector (a modified version of pACYC vector containing an N-terminal Trx tag). Meanwhile, the DNA fragments encoding six human ATG8s were all cloned into pET-32M-3C vector or pET-GST-3C vector (modified versions of pET-32a vector containing an N-terminal GST tag). Of note, 6xHis was placed on either the N-terminal or the C-terminal of target proteins. For Co-immunoprecipitation assays, the DNA fragments encoding human GABARAPL1(1-115) and full-length FIP200 were separately cloned into pEGFP-C1 vector and pmEGFP-C1 vector (a modified version of pEGFP-C1 with an A206K mutation that monomerizes EGFP) as well as full-length ATG16L1 into pFlag-CMV-2 vector. All point mutations of WIPI2b, FIP200, GABARAPL1 and ATG16L1 used in this study were generated through standard PCR-based mutagenesis method and further confirmed by DNA sequencing.
+
+Recombinant proteins were all expressed in BL21 (DE3) *E. coli* cells induced by 200 µM IPTG overnight at 16°C. The bacterial cell pellets were re-suspended in the binding buffer (50 mM Tris, 500 mM NaCl, and 5 mM imidazole at pH 7.9), and then lysed by the ultrahigh pressure homogenizer FB-110XNANO homogenizer machine (Shanghai Litu Machinery Equipment Engineering Co., Ltd.). Then, the lysate was spun down by centrifuge at 17000 rpm (35000 g) for 35 minutes to remove the pellets fractions. All proteins were purified by Ni²⁺-NTA agarose (GE Healthcare) affinity chromatography and further purified by size-exclusion chromatography (Superdex 75 or 200 26/60 column; GE Healthcare) equilibrated with the column buffer containing 20 mM Tris, 100 mM NaCl, 1 mM DTT and 1 mM EDTA at pH 7.5. To obtain the FIP200 Claw/ATG16L1 FIR complex used for crystallization, the FIP200/ATG16L1 complex was obtained through the co-expression of Trx-FIP200(1490–1594) and Trx-ATG16L1(235–247). The N-terminal Trx tags of relevant FIP200 and ATG16L1 proteins were cleaved by 3C protease and removed by HisTrap excel column (GE Healthcare). Finally, the FIP200 Claw/ATG16L1 FIR complex and the GABARAPL1/ATG16L1 FIR complex were further purified through Superdex 75 column equilibrated with the aforementioned column buffer. Special for six human ATG8 family proteins, their Trx tags were removed by MonoQ 10/10 ion-exchange column (GE Healthcare). Meanwhile, uniformly ¹⁵N-labelled GABARAPL1 proteins were prepared by growing bacteria in M9 minimal medium using ¹⁵NH₄Cl (Cambridge Isotope Laboratories Inc.) as the sole nitrogen source.
+
+## ITC assay
+
+ITC measurements were all carried out on a MicroCal PEAQ-ITC (Malvern) calorimeter at 25°C. All protein samples were prepared in the same buffer containing 20 mM Tris, 100 mM NaCl, 1 mM DTT at pH 7.5. For each ITC experiment in this study, the concentrated (~ 50 µM) proteins were loaded into the cell, and the other titrated proteins (~ 500 µM) were loaded into the syringe. The titration processes were performed by injecting proteins from syringe into the cell at time intervals of 2 minutes to ensure that the titration peak returned to the baseline. The titration data were analyzed using the Malvern MicroCal PEAQ-ITC analysis program and fitted using the one-site binding model.
+
+## Size exclusion chromatography
+
+Size exclusion chromatography was carried out on an AKTA FPLC system (GE Healthcare). Purified proteins were loaded on to a Superdex 200 or 75 increase 10/300 GL column (GE Healthcare) equilibrated with the same column buffer. The fitting results were further output to the Origin 9 software and aligned with each other.
+
+## NMR spectroscopy
+
+The ¹⁵N-labelled protein samples for NMR titration experiments were concentrated to ~ 0.1 mM. All the protein samples for NMR studies were prepared in the 50 mM potassium phosphate buffer containing 100 mM NaCl and 1 mM DTT at pH 6.5, and NMR spectra were acquired at 25°C on an Agilent 800 MHz spectrometer equipped with an actively z gradient shielded triple resonance cryogenic probe at the Shanghai Institute of Organic Chemistry.
+
+## Multi-angle light scattering
+
+For multi-angle light-scattering measurement, the FIP200(1490–1594)/ATG16L1(78–247) complex samples were injected into an AKTA FPLC system (GE Healthcare) with a Superdex 200 increase 10/300 GL column (GE Healthcare) with the same column buffer mentioned before. The chromatography system was coupled to a static light scattering detector (miniDawn, Wyatt Technology) and a differential refractive index detector (Optilab, Wyatt Technology). Data were collected every 0.5 s with a flow rate of 0.5 mL/min. Data were analyzed using the ASTRA 6 software (Wyatt Technology) and drawn using the Origin 9 software.
+
+## Fluorescence polarization assay
+
+Fluorescence anisotropy binding assays were performed on the SpectraMax i3x Multi-Mode Detection Platform from Molecular Devices, using a 485 nm excitation filter and a 535 nm emission filter. Peptides were labeled with fluorescein isothiocyanate isomer I (Sigma-Aldrich) at their N-terminal NH₂. In this assay, the FITC-labeled peptide (~ 0.3 µM) was titrated with increasing amount of testing proteins in the column buffer at 25 °C. The *K*d values were obtained by fitting the titration curves with the classical one-site binding model using GraphPad Prism 9 software.
+
+## Protein crystallization and structural elucidation
+
+Crystals of the FIP200(1490–1594)/ATG16L1(235–247) complex and the GABARAPL1/ATG16L1(235–247) complex were all obtained using the sitting-drop vapor-diffusion method at 16°C. The crystal-growing condition of the FIP200(1490–1594)/ATG16L1(235–247) complex (38 mg/mL) contains 10% v/v 2-Propanol, 0.1 M BICINE (pH 8.5), and 30% w/v Polyethylene glycol 1500. As for the GABARAPL1/ATG16L1(235–247) complex, the purified GABARAPL1 protein (27 mg/mL) was saturated with ATG16L1 FIR peptide with a molar ratio up to 1:10. Crystals were observed in the condition containing 0.1 M Sodium cacodylate (pH 6.5), 40% v/v MPD and 5% w/v PEG 8000. Before diffraction experiments, relevant amount of glycerol was added as the cryo-protectant. A 1.76 Å resolution X-ray data set for the FIP200(1490–1594)/ATG16L1(235–247) complex and a 1.54 Å resolution X-ray data set for the GABARAPL1/ATG16L1(235–247) complex were collected at the beamline BL19U1, BL02U1 and BL10U2 of the Shanghai Synchrotron Radiation Facility⁴⁸. The diffraction data were processed upon autoPROC⁴⁹. The phase problems of the FIP200/ATG16L1 complex and the GABARAPL1/ATG16L1 complex were all solved by molecular replacement method by using the FIP200 Claw structure (PDB ID: 6DCE) and the GABARAPL1 structure (PDB ID: 5LXI) respectively as the search model with PHASER⁵⁰. The initial structural models were rebuilt manually using COOT⁵¹, and then refined through PHENIX⁵². Further manual model building and adjustments were completed via COOT⁵¹. The qualities of the final models were validated by MolProbity⁵³. The final refinement statistics of solved structures in this study were listed in Supplementary Table 1. All the structural diagrams were prepared using the program PyMOL (http://www.pymol.org/).
+
+## Co-immunoprecipitation assay
+
+Flag-tagged ATG16L1 plasmids (wild-type or mutants) were co-transfected into HEK293T cells using Lipofectamine 2000 transfection reagent (Thermo Fisher Scientific) or Lipofectamine 6000 transfection reagent (Beyotime) with mEGFP-tagged FIP200 or EGFP-tagged GABARAPL1(1-115) plasmids. Cells were collected 24 hours after transfection and lysed in ice-cold cell lysis buffer (50 mM Tris, 150 mM NaCl, 0.5% NP-40, 1 mM PMSF, 1% protease inhibitor cocktail at pH 7.5) for 20 to 40 minutes at 4°C. Lysates were centrifuged at 14500 g for 15 minutes at 4°C to separate soluble fractions and cell debris. Supernatants were applied to anti-GFP mAb-Agarose (Medical & Biological Laboratories) and incubated for 40 to 60 minutes at 4°C. The beads and non-bound proteins were separated by centrifugation at 800 g for 3 minutes at 4°C. After washing several times with the cold wash buffer (50 mM Tris, 150 mM NaCl and 0.1% or 0.5% NP-40 at pH 7.5), the beads were re-suspended with the 1X SDS-PAGE sample buffer and boiled at 65°C for 10 minutes. The prepared samples were analyzed by SDS-PAGE. The EGFP-tagged GABARAPL1(1-115), mEGFP-tagged FIP200 and Flag-tagged ATG16L1 were detected by western blot using the anti-GFP (Proteintech, 50430-2-AP, 1:1000 dilution), anti-GFP (Proteintech, 66002-1-Ig, 1:2000 dilution), anti-Flag (Proteintech, 20543-1-AP, 1:1000 dilution) and anti-Flag (Proteintech, 66008-4-Ig, 1:2000 dilution) primary antibodies.
+
+## Generation of relevant ATG16L1 stable cell lines
+
+The AcGFP1-tagged mutant ATG16L1 was cloned into the pMSCV-blasticidin vector and was co-transfected into HEK293T cells with VSV-G and gag/pol using Lipofectamine 6000 transfection reagent (Beyotime). Notably, the sgRNA-targeting region of ATG16L1 at pMSCV-blasticidin vector was synonymously mutated to avoid being targeted again by Cas9 enzyme. *ATG16L1*-knockout cells were incubated with polybrene (Sigma-Aldrich) and concentrated virus-containing medium filtered through a 0.45-µm-pore syringe filter. Transfected cells were treated with blasticidin (5 µg/mL; InvivoGen) to generate stable polyclonal cell lines.
+
+### Autophagy induction
+
+The *ATG16L1*-knockout HeLa cells, which were generated in our previous study¹², were rescued by lentiviral transduction with AcGFP1-tagged WT ATG16L1, ATG16L1 D239R/I240F mutant (DRIF), or ATG16L1 I240Q/I243Q mutant (IQIQ). Rescued HeLa cells were separately seeded on a six-well plate. The following day, cells were incubated for 4 hours with DMEM (Thermo Fisher Scientific) supplemented with 10% FBS (Thermo Fisher Scientific) and 1% penicillin-streptomycin (Thermo Fisher Scientific), amino acid starvation medium (BOSTER), and amino acid starvation medium with bafilomycin A1 (Selleck) at 400 nM. After starvation treatment, cells were resuspended with the 1X SDS-PAGE sample buffer and boiled for 7 minutes at 100°C. The samples were detected by Western blot using specific ATG16L1 antibody (1:1000; Abcam, catalog no. ab187671), LC3B antibody (1:1000; Abcam, catalog no. ab192890), β-actin antibody (1:5000; Proteintech, catalog no. 66009-1-lg), and p62 antibody (1:1000; Cell Signaling Technology, catalog no. #39749). The data are presented as means ± SEM from three independent experiments. Statistical analyses were performed in GraphPad Prism 9 by two-way analysis of variance (ANOVA) followed by Bonferroni multiple comparisons test, and P value style is P = 0.1234 (not significant (ns)), *P = 0.0332, **P = 0.0021, ***P = 0.0002, and ****P < 0.0001.
+
+# References
+
+1. Lamb CA, Yoshimori T, Tooze SA (2013) The autophagosome: origins unknown, biogenesis complex. Nat Rev Mol Cell Biol 14:759–774
+2. Levine B, Kroemer G (2019) Biological Functions of Autophagy Genes: A Disease Perspective. Cell 176:11–42
+3. Yamamoto H, Zhang S, Mizushima N (2023) Autophagy genes in biology and disease. Nat Rev Genet 24:382–400
+4. Dikic I, Elazar Z (2018) Mechanism and medical implications of mammalian autophagy. Nat Rev Mol Cell Biol 19:349–364
+5. Feng YC, He D, Yao ZY, Klionsky DJ (2014) The machinery of macroautophagy. Cell Res 24:24–41
+6. Hurley JH, Young LN (2017) Mechanisms of Autophagy Initiation. Annu Rev Biochem 86:225–244
+7. Zheng Q et al (2022) Calcium transients on the ER surface trigger liquid-liquid phase separation of FIP200 to specify autophagosome initiation sites. Cell 185:4082–4098e22
+8. Matsunaga K et al (2010) Autophagy requires endoplasmic reticulum targeting of the PI3-kinase complex via Atg14L. J Cell Biol 190:511–521
+9. Gammoh N, Florey O, Overholtzer M, Jiang XJ (2013) Interaction between FIP200 and ATG16L1 distinguishes ULK1 complex-dependent and -independent autophagy. Nat Struct Mol Biol 20:144–149
+10. Nishimura T et al (2013) FIP200 regulates targeting of Atg16L1 to the isolation membrane. EMBO Rep 14, 284–91
+11. Dooley HC et al (2014) WIPI2 links LC3 conjugation with PI3P, autophagosome formation, and pathogen clearance by recruiting Atg12-5-16L1. Mol Cell 55:238–252
+12. Gong X et al (2023) ATG16L1 adopts a dual-binding site mode to interact with WIPI2b in autophagy. Sci Adv 9:eadf0824
+13. Otomo C, Metlagel Z, Takaesu G, Otomo T (2013) Structure of the human ATG12~ATG5 conjugate required for LC3 lipidation in autophagy. Nat Struct Mol Biol 20:59–66
+14. Fujita N et al (2008) The Atg16L complex specifies the site of LC3 lipidation for membrane biogenesis in autophagy. Mol Biol Cell 19:2092–2100
+15. Bozic M et al (2020) A conserved ATG2-GABARAP family interaction is critical for phagophore formation. EMBO Rep 21:e48412
+16. Strong LM et al (2021) Structural basis for membrane recruitment of ATG16L1 by WIPI2 in autophagy. Elife 10
+17. Gatica D, Lahiri V, Klionsky DJ (2018) Cargo recognition and degradation by selective autophagy. Nat Cell Biol 20:233–242
+18. Stolz A, Ernst A, Dikic I (2014) Cargo recognition and trafficking in selective autophagy. Nat Cell Biol 16:495–501
+19. Johansen T, Lamark T (2011) Selective autophagy mediated by autophagic adapter proteins. Autophagy 7, 279–96
+20. Rogov V, Dotsch V, Johansen T, Kirkin V (2014) Interactions between autophagy receptors and ubiquitin-like proteins form the molecular basis for selective autophagy. Mol Cell 53:167–178
+21. Green DR, Levine B (2014) To Be or Not to Be? How Selective Autophagy and Cell Death Govern Cell Fate. Cell 157:65–75
+22. Kirkin V (2020) History of the Selective Autophagy Research: How Did It Begin and Where Does It Stand Today? J Mol Biol 432:3–27
+23. Farre JC, Subramani S (2016) Mechanistic insights into selective autophagy pathways: lessons from yeast. Nat Rev Mol Cell Biol 17:537–552
+24. Kirkin V, Rogov VV (2019) A Diversity of Selective Autophagy Receptors Determines the Specificity of the Autophagy Pathway. Mol Cell 76:268–285
+25. Johansen T, Lamark T, Selective Autophagy (2020) ATG8 Family Proteins, LIR Motifs and Cargo Receptors. J Mol Biol 432:80–103
+26. Zhang MF et al (2024) Mechanistic insights into the interactions of TAX1BP1 with RB1CC1 and mammalian ATG8 family proteins. Proc Natl Acad Sci USA 121
+27. Turco E et al (2019) FIP200 Claw Domain Binding to p62 Promotes Autophagosome Formation at Ubiquitin Condensates. Mol Cell 74:330–346e11
+28. Zhou Z et al (2021) Phosphorylation regulates the binding of autophagy receptors to FIP200 Claw domain for selective autophagy initiation. Nat Commun 12:1570
+29. Turco E et al (2019) FIP200 Claw Domain Binding to p62 Promotes Autophagosome Formation at Ubiquitin Condensates. Mol Cell 74:330–
+30. Vargas JNS et al (2019) Spatiotemporal Control of ULK1 Activation by NDP52 and TBK1 during Selective Autophagy. Mol Cell 74:347–
+31. Smith MD et al (2018) CCPG1 Is a Non-canonical Autophagy Cargo Receptor Essential for ER-Phagy and Pancreatic ER Proteostasis. Dev Cell 44:217–
+32. Fu T et al (2021) Structural and biochemical advances on the recruitment of the autophagy-initiating ULK and TBK1 complexes by autophagy receptor NDP52. Sci Adv 7
+33. Sorbara MT et al (2018) Complement C3 Drives Autophagy-Dependent Restriction of Cyto-invasive Bacteria. Cell Host Microbe 23:644–652e5
+34. Xu Y et al (2019) A Bacterial Effector Reveals the V-ATPase-ATG16L1 Axis that Initiates Xenophagy. Cell 178:552–566e20
+35. Gammoh N (2020) The multifaceted functions of ATG16L1 in autophagy and related processes. J Cell Sci 133
+36. Fujita N et al (2013) Recruitment of the autophagic machinery to endosomes during infection is mediated by ubiquitin. J Cell Biol 203:115–128
+37. Behrends C, Sowa ME, Gygi SP, Harper JW (2010) Network organization of the human autophagy system. Nature 466:68–76
+38. Ravenhill BJ et al (2019) The Cargo Receptor NDP52 Initiates Selective Autophagy by Recruiting the ULK Complex to Cytosol-Invading Bacteria. Mol Cell 74, 320–329 e6
+39. Smith MD et al (2018) CCPG1 Is a Non-canonical Autophagy Cargo Receptor Essential for ER-Phagy and Pancreatic ER Proteostasis. Dev Cell 44:217–232e11
+40. von Muhlinen N et al (2012) LC3C, bound selectively by a noncanonical LIR motif in NDP52, is required for antibacterial autophagy. Mol Cell 48:329–342
+41. Cheng XF et al (2016) Structural basis of FYCO1 and MAP1LC3A interaction reveals a novel binding mode for Atg8-family proteins. Autophagy 12:1330–1339
+42. Nguyen TN et al ATG4 family proteins drive phagophore growth independently of the LC3/GABARAP lipidation system. Mol Cell 81, 2013–2030 e9 (2021).
+43. King BC et al (2019) Complement Component C3 Is Highly Expressed in Human Pancreatic Islets and Prevents beta Cell Death via ATG16L1 Interaction and Autophagy Regulation. Cell Metab 29, 202–210 e6
+44. Turco E et al (2021) Reconstitution defines the roles of p62, NBR1 and TAX1BP1 in ubiquitin condensate formation and autophagy initiation. Nat Commun 12:5212
+45. Juris L et al (2015) PI3P binding by Atg21 organises Atg8 lipidation. EMBO J 34:955–973
+46. Munzel L et al (2021) Atg21 organizes Atg8 lipidation at the contact of the vacuole with the phagophore. Autophagy 17:1458–1478
+47. Harada K et al (2019) Two distinct mechanisms target the autophagy-related E3 complex to the pre-autophagosomal structure. Elife 8
+48. Wang Z et al (2016) Automatic crystal centring procedure at the SSRF macromolecular crystallography beamline. J Synchrotron Radiat 23:1323–1332
+49. Vonrhein C et al (2011) Data processing and analysis with the autoPROC toolbox. Acta Crystallogr D Biol Crystallogr 67:293–302
+50. Storoni LC, McCoy AJ, Read RJ (2004) Likelihood-enhanced fast rotation functions. Acta Crystallogr Sect D-Biological Crystallogr 60:432–438
+51. Emsley P, Cowtan K (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr Sect D-Biological Crystallogr 60:2126–2132
+52. Adams PD et al (2002) PHENIX: building new software for automated crystallographic structure determination. Acta Crystallogr Sect D-Biological Crystallogr 58:1948–1954
+53. Davis IW et al (2007) MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res 35:W375–W383
+
+# Supplementary Files
+
+- [SupplementalMaterialsL1FIP200ATG16L1V4.pdf](https://assets-eu.researchsquare.com/files/rs-5058991/v1/36eda79d4875934709391b93.pdf)

2c30acfc4937547ac84a9f760356bb8c998144198f558d679632fb7d33d25240/preprint/images_list.json

@@ -0,0 +1,34 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Motivation and design of electrocatalyst. a Tuning strategy of the electronic structure of the MoS2 surface. b Design ideas of hydrogen evolution catalyst. c-f the energetics of hydroxyl species on 2Hphase-MoS2 (002), 1Tphase-MoS2 (002), NiS2 (210), and Ni2P (111) HER electrocatalyst surfaces. g Schematics of the 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P synthesis steps.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Electronic structure characterizations of 1T0.72-MoS2@NiS2 and 1T0.81-MoS2@Ni2P catalysts. a-c HRTEM image of 1T0.72-MoS2@NiS2. (b) shows NiS2 lattice. (c) shows 2H and 1T lattices. Scale bars are 5 nm (a), 1 nm (b), and 2 nm (c). d-f Typical HR-TEM image of 1T0.81-MoS2@Ni2P. (g) shows Ni2P lattice fringes. (h) shows 2H and 1T lattices. Scale bars are 5 nm (f), 1 nm (g) and 2 nm (h). g HR Mo 3d core-level XPS spectra of 1T0.72-MoS2@NiS2, 1T0.81-MoS2@Ni2P, 1T0.41-MoS2 and 2Hphase-MoS2. h S 2p core-level XPS spectra of 1T0.72-MoS2@NiS2, 1T0.81-MoS2@Ni2P, 1T0.41-MoS2 and 2Hphase-MoS2, respectively. i Ni 2p XPS spectrum for 1T0.72-MoS2@NiS2, 1T0.81-MoS2@Ni2P, 1T0.41-MoS2@Ni(OH)2, and Ni(OH)2.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "HER performed in alkaline and acidic electrolytes. a LSV curves in 1M KOH. b \u03b710 and Tafel slopes for various Mo-based HER electrocatalysts in 1.0 M KOH. c. LSV curves in 0.5 M H2SO4. d. \u03b710 and Tafel slopes for various Mo-based HER catalysts in 0.5 M H2SO4.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Theoretical calculation and mechanisms analysis of the surface structure and HER activation energy of the as-prepared electrocatalysts. a, b the deformation of the electronic density of 2Hphase-MoS2@NiS2 and 1Tphase-MoS2@NiS2 interface, in which yellow/green isosurfaces correspond to positive/negative spin densities (0.00295308 e/\u00c53). Band structure and density of states (DOS) for NiS2 (c), 1Tphase-MoS2 (d) and 1Tphase-MoS2@NiS2 (e). f Free-energy diagrams of H2 adsorption by the 2Hphase-MoS2, 1Tphase-MoS2, 2Hphase-MoS2@Ni2P, 2Hphase-MoS2@NiS2, 1Tphase-MoS2@NiS2 and 1Tphase-MoS2@Ni2P. g Schematics showing water activation, *H intermediate formation and hydrogen generation on multi-heterojunction interface electrocatalysts.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

2c30acfc4937547ac84a9f760356bb8c998144198f558d679632fb7d33d25240/preprint/preprint.md

@@ -0,0 +1,134 @@
+# Abstract
+
+Molybdenum disulfide, as an electronic highly-adjustable catalysts material, tuning its electronic structure is crucial to enhance its intrinsic hydrogen evolution reaction (HER) activity. Nevertheless, there are yet huge challenges to the understanding and regulation of the surface electronic structure of molybdenum disulfide-based catalysts. Here we address these challenges by tuning its electronic structure of phase modulation synergistic with interfacial chemistry and defects from phosphorus or sulfur implantation, and we then successfully design and synthesize electrocatalysts with the multi-heterojunction interfaces (e.g., 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P), demonstrating superior HER activities and good stabilities with a small overpotentials of 38.9 and 98.5 mV at 10 mA/cm<sup>2</sup>, a low Tafel slopes of 41 and 42 mV/dec in acidic as well as alkaline surroundings, outperforming commercial Pt/C catalyst and other reported Mo-based catalysts. Theoretical calculation verified that the incorporation of metallic-phase and intrinsic HER-active Ni-based materials into molybdenum disulfide could effectively regulate its electronic structure for making the bandgap narrower. Additionally, reduced nickel possesses empty orbitals, which is helpful for additional H binding ability. All these factors can decrease Mo-H bond strength, greatly improving the HER catalytic activity of these materials.
+
+[Catalysis](/browse?subjectArea=Catalysis) [Nanoscience](/browse?subjectArea=Nanoscience) [Electrochemistry](/browse?subjectArea=Electrochemistry) [Molybdenum disulfide](/browse?subjectArea=Molybdenum%20disulfide) [Hydrogen evolution reaction](/browse?subjectArea=Hydrogen%20evolution%20reaction) [Surface electronic structure](/browse?subjectArea=Surface%20electronic%20structure) [Hetero geneous-phase-interface](/browse?subjectArea=Hetero%20geneous-phase-interface)
+
+# Introduction
+
+Extensive use and depletion of fossil fuels resulting in serious pollution. Therefore, green and renewable fuel resources are required for continuing sustainable economic development.¹⁻³ Electrocatalysis acts as a vital role in the conversion of clean energy to achieve a sustainable approach to various commercial processes, including HER.⁴,⁵ However, electrochemical water splitting is hindered by the large kinetic barrier and slow kinetics.⁶⁻⁹ Pt-based electrocatalysts are recognized as highly efficient electrocatalysts due to good electrical conductivity,¹⁰ fast kinetics, and the preference to overcome the large kinetic energy barrier involved in the above-mentioned process.¹¹ Unfortunately, high price and not desirable stability hinder the extended Pt-based catalysts’ application.¹² Thus, it is very urgent to develop cost-effective Pt-free electrocatalysts with comparable activity and better stability.
+
+Researchers recently have designed a wide range of low-cost catalysts, including transition-metal chalcogenides (TMDCs),¹³,¹⁴ metal nitrides,¹⁵,¹⁶ metal carbides,¹⁷,¹⁸ and metal phosphides.¹⁹,²⁰ Among these candidates, MoS₂, a typical layered 2D TMDCs formed by Van der Waals interaction and stacking of S-Mo-S layers, attracts extensive interests with its adjustable bandgap, unique band structure, high energy-conversion efficiency, and earth abundance.²¹⁻²³ However, the electrocatalytic activity of MoS₂ is closely associated with its surface electric structure, such as phase structure or phase composition, interface active site, and defect. Interestingly, two main phases of MoS₂ were widely justified: 2H and 1T phases.²⁴ 2H phase has the most thermodynamical stability among the molybdenum sulfide family, whose HER activities are restrained by the amount and active site types as well as conductivity. Unlike the 2H phase, 1T phase one demonstrates higher catalytic activity since it has numerous active sites on the edges and a fast transfer rate. However, it is remaining a giant challenge of directly synthesizing the high percentage 1T phase molybdenum sulfide due to the thermodynamic instability of 1T phase-MoS₂.²⁵ To solve this problem, a feasible strategy is to efficiently realize the 2H → 1T phase transformation to improve HER capability. Wang et al. found that a partial 2H → 1T MoS₂ phase transition by facile one-pot annealing of a large amount of 2H phase-MoS₂ under phosphorus vapor is able to enhance HER catalytic activities.²⁶ A synergistic strategy of doping nitrogen and intercalating PO₄³⁻ is reported, which can convert 2H- to 1T-phase with a conversion rate of up to 41%, and has excellent HER performance.²⁷ However, the electronic transport capacity and phase stability of the phase boundary of a single component (pure 1T-phase) are generally poor. In order to overcome the puzzles, the HER activity of the pure phase can be improved by constructing a heterogeneous boundary. Therefore, it is expected to further enhance the HER performance and its stability of traditional single 1T-phase or 2H-phase interface by constructing a composite heterojunction between 1T-phase and the other phases.²⁸
+
+Interface modification could be an effective approach to construct a composite heterojunction.²⁹,³⁰ Ni-based materials (such as Ni₂P, NiS₂, Ni₂S₃, etc.) with high activity and conductivity have been considered as highly efficient electrocatalysis materials for HER,¹⁹,²⁰,³¹ as another heterogeneous interface, which is also very important to control the electronic structure of the MoS₂ interface. Kim et al. reported that Ni₂P nanoparticles were used to activate the MoS₂ base surface, which exhibits Pt-like HER performance in 0.5 M HCl solution.³¹ Because the electronic structure of Ni₂P is a P62m space group, which could facilitate recombination at the atomic scale. Moreover, Ni has a unique α and β orbital integral asymmetric d orbital, which makes it easy for the lone pair of electrons to recombine with the d orbital of the exposed Mo atom on MoS₂ to generate new interface electrons, thereby improving HER performance. Lin et al. reported that a defect-rich heterogeneous interfacial catalyst (MoS₂/NiS₂) could provide abundant active sites to promote electron transfer, thereby further rapidly promoting electrocatalytic hydrogen evolution.³² More importantly, the introduction of NiS₂ hybridization on the surface of MoS₂ generates new form of interface electrons, and Niδ⁺ is reduced to low-valence Ni to improve the binding energy with hydrogen elements, thereby weakening the Mo-H strength. To sum up, although the heterojunction-phase catalyst synthesized by the above-mentioned approach further improves the HER activity and good stability, the understanding and regulation of the surface electronic structure on the MoS₂ interface are still huge challenges, and thus it is very necessary to develop an efficient synthesis approach to obtain stable multi-heterogeneous interface catalyst.
+
+Here, we address these challenges by tuning its electronic structure through phase modulation synergistic with interfacial chemistry and defects of phosphorus or sulfur implantation, and we then successfully design and prepare a series of heterojunction-phase-interface electrocatalysts (denoted 1T₀.₈₁-MoS₂@Ni₂P and 1T₀.₇₂-MoS₂@NiS₂) with an outstanding HER activity and are stable in dual-pH surroundings. The strategies to control the electronic characteristics of the MoS₂ surface include surface phase modulation, surface defects, and the construction of hetero-structure (Fig. 1a). Furthermore, we control the hydrogen and hydroxyl adsorption energy through the synergistic effect of heterojunction-phase-interface catalysts (Fig. 1b and Fig. 1c–f) because the energy of the hydroxyl species is very important for the hydrolysis accelerator. Starting from hydrothermally synthesized MoS₂ nanosheets, we develop a simple surface electronic structure modulation strategy of constructing multi-heterogeneous-phase-interface 1T₀.₈₁-MoS₂@Ni₂P and 1T₀.₇₂-MoS₂@NiS₂ electrocatalysts (Fig. 1a) by citric acid-induced hydrothermal synthesis, electrodeposition and then phosphorus (or sulfur) vapor thermal treatment approach for the first time. Our approach can not only realize the construction of abundant catalytic reactive sites but also improve the conversion rate of 2H to 1T (81%), and it is also convenient to introduce Ni₂P or NiS₂ heterogeneous interfaces. As to the surface electronic structure of catalysts, high-resolution transmission electron microscopy (HRTEM) images show that such phase-structures, heterojunction-phase-interface edges, and defects are derived by the featured electronic states and Ni atomic coordination. Additionally, X-ray photoelectron spectra (XPS) showed that citric acid induces hydrothermal synthesis of stable 1T₀.₄₁-MoS₂ (41% of 1T phase), and the 1T₀.₈₁-MoS₂ or 1T₀.₇₂-MoS₂ (81% or 72% of 1T phase) conversion rate is further improved after phosphorus or sulfur vapor thermal treatment. Electrodes containing 1T₀.₈₁-MoS₂@Ni₂P (or 1T₀.₇₂-MoS₂@NiS₂) only require 38.9 (or 186) and 98.5 mV (or 128) to achieve HER current density equal to 10 mA/cm². They also need Tafel slopes equal to 41 (or 79) and 42 (or 68) mV/dec in 1M H⁺ or OH⁻ media, good stability during testing for 16 h in both media, respectively. The 1T₀.₈₁-MoS₂@Ni₂P (or 1T₀.₇₂-MoS₂@NiS₂) catalysts exhibited superior activities with Tafel slope values and the over-potentials lower than the values reported for Mo-base HER catalysts in both alkaline and acidic media.²⁵,²⁶,²⁸,³¹⁻³⁴ The mechanism of hydrogen formation at the heterogeneous interface was elucidated by DFT calculation, which is attributed to the shortened bandgap, the decreased Mo-H bond strength, and the reduced electron cloud density around Ni providing enough empty d-orbitals to bind with H atoms. This work provides useful insights for exploring the enhancement mechanisms of HER with an optimized surface electronic structure on the MoS₂ interface, which provides an effective insight of constructing invaluable metal electrocatalysts for HER and other fields.
+
+# Results
+
+Preparation and characterizations of multi-heterojunction interface electrocatalysts. The formation process of multi-heterojunction interface electrocatalysts is schematically illustrated in Fig 1g. 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> catalysts were synthesized by a three-step procedure. First, 1T<sub>0.41</sub>-MoS<sub>2</sub> catalyst was obtained on carbon cloth (CC) by acid-induced hydrothermal approach at 200<sup>o</sup>C for 12 h (see details in “Methods” section). The as-obtained 1T<sub>0.41</sub>-MoS<sub>2</sub> catalyst shows a large number of microspheres (Supplementary Fig. 1b-d) with a narrow diameter distribution of 2.0 ~ 4.0 µm distributed uniformly on the surface of CC substrate. Flower-shaped MoS<sub>2</sub> microspheres are consisted of many aligned 1T<sub>0.41</sub>-MoS<sub>2</sub> nanosheets, on which the Ni(OH)<sub>2</sub> nanoparticles were then electro-deposited (see details in “Methods” section). 1T<sub>0.41</sub>-MoS<sub>2</sub>@Ni(OH)<sub>2</sub> material inherited its morphology from spherical MoS<sub>2</sub>. Subsequently, 1T<sub>0.41</sub>-MoS<sub>2</sub>@Ni(OH)<sub>2</sub> material was loaded into a quartz tube mixed with red phosphorus or sulfur powder and sealed by oxyacetylene flame. Finally, these were heated to 600<sup>o</sup>C for the reaction with red phosphorus or sulfur to synthesize 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> catalysts, respectively (Supplementary Fig. 2, 3). As to 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalyst, the MoS<sub>2</sub> microspheres are very rough, on which there distributes many random Ni<sub>2</sub>P nanoparticles (Supplementary Fig. 3). It is because that the 1T/2H mixed-phase and heterojunction-interface structure reduces the adhesion of the gas-solid interface and facilitates releasing hydrogen from the catalyst surface, which is essential for enhancing HER.<sup>29</sup>
+
+Next, the phase composition and crystal properties of 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> were obtained by X-ray diffraction (XRD) and Raman spectroscopy. There are some obvious characteristic diffraction peaks of 14.3°, 33.4°, and 59.2° (Supplementary Fig. 4a), which can be ascribed to 2H<sub>phase</sub>-MoS<sub>2</sub> according to the JCPDS card number #37-1429. However, the XRD peak of 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> located at 2θ ≈ 28.8° can be indexed as the (004) peak of 1T<sub>phase</sub>-MoS<sub>2</sub>, which indicates that 1T- and 2H-mixed phases were successfully hydrothermally synthesized.<sup>35</sup> The other characteristic peaks (2θ ≈ 31.3°, 35.2°, 38.8°, 44.9°, and 53.3°) demonstrate that the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> is a hybrid of NiS<sub>2</sub> (JCPDS#11-0099), which verifies the presence of NiS<sub>2</sub> nanoparticles. Similarly, as to 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalyst, its XRD results also showed the presence of Ni<sub>2</sub>P nanoparticles (JCPDS#21-0590) on the 1T<sub>0.41</sub>-MoS<sub>2</sub> surface. Raman spectroscopy showed E<sub>2g</sub><sup>1</sup> and A<sub>1g</sub> vibrational bands at 376.2 and 402.9 cm<sup>-1</sup> peaks typical for 2H<sub>phase</sub>-MoS<sub>2</sub>.<sup>36</sup> J<sub>1</sub>, J<sub>2</sub> and J<sub>3</sub> vibrations at 147.3, 235.4 and 335.2 cm<sup>-1</sup> are characteristic for 1T<sub>phase</sub>-MoS<sub>2</sub><sup>37</sup> (Supplementary Fig. 4b). These results prove that the 1T phase of MoS<sub>2</sub> is formed by the hydrothermal reaction induced by organic acids (e.g., citric acid).<sup>35</sup> 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> or 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P demonstrated three characteristic peaks of 1T<sub>phase</sub>-MoS<sub>2</sub> and the two characteristic peaks (E<sub>2g</sub><sup>1</sup> and A<sub>1g</sub>) of 2H<sub>phase</sub>-MoS<sub>2</sub>. Additionally, they showed a vibrational peak (437.3 cm<sup>-1</sup>) of Ni-S<sup>32</sup> or three vibrational peaks (216.2 cm<sup>-1</sup>, 249.7 cm<sup>-1</sup>, and 269.5 cm<sup>-1</sup>) of Ni-P.<sup>31</sup> More importantly, the E<sub>2g</sub><sup>1</sup> and A<sub>1g</sub> vibrations of 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> at 382.2 and 408.1 cm<sup>-1</sup> were red-shifted by 6.0 and 5.2 cm<sup>-1</sup>, respectively (Supplementary Fig. 4b). Thus, NiS<sub>2</sub> nanoparticles are between the 1T<sub>0.41</sub>-MoS<sub>2</sub> layers. Similarly, the E<sub>2g</sub><sup>1</sup> and A<sub>1g</sub> peaks for the 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalyst slightly red-shifted by 7.3 and 3.0 cm<sup>-1</sup>, respectively. These results confirm that rich multi-heterojunction interface edges active sites catalysts were successfully synthesized.
+
+Electronic structure characterizations of 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalysts. To further identify the surface electronic structure of multi-heterogeneous interface catalysts, we applied the high-resolution transmission electron microscopy (HRTEM) to assess the morphology and crystal structures of 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> catalysts. Supplementary Fig. 5a, b shows the typical low-magnification image of the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> on the Cu grid, which confirms the flower-like nanosphere morphologies of 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>. TEM and corresponding elemental distribution map obtained for the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> sample demonstrated uniformly distributed Mo, Ni, and S (Supplementary Fig. 5c-c<sub>4</sub>). As revealed by the HRTEM image (Fig. 2a-c and Supplementary Fig. 5e, f), NiS<sub>2</sub> nanoparticles are decorated on MoS<sub>2</sub> nano-sheets edge. The HRTEM of the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> catalyst clearly shows the crystal lattice of 0.25 nm, referring to the NiS<sub>2</sub> (210). Interestingly, Fig. 2a shows the HRTEM image of 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> flower-like nano-sheets, in which there demonstrates the lattice fringes perpendicularly to the electron beam direction circled by blood-color, justifying the S defect (Fig. 2c). The trigonal lattice in the yellow circle (Fig. 2c) implies the presence of 1T phase MoS<sub>2</sub>, while the hexagonal lattice in the blue circle (Fig. 2c) suggests the presence of 2H phase MoS<sub>2</sub>. The above-described results further confirm the successful preparation of the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> multi-heterojunction interface catalyst. The anion is changed to be P to produce 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P multi-heterojunction interface catalyst by phosphorus vapor thermal treatment. Supplementary Fig. 6a, b displays the morphologies of 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalyst, overlapping nanosheets with many embedded particles can be clearly identified. There is an obvious alternation of 1T and 2H phases, and a large number of defects (Fig. 2f). As shown in Supplementary Fig. 6c, there are the distributions of Mo, Ni, S, and P over the whole 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalyst, verifying that Ni<sub>2</sub>P nanoparticles are encapsulated by MoS<sub>2</sub> edges. The interplanar spacings, equal to 0.62 and 0.22 nm, match (002) and (111) interplanar distances of MoS<sub>2</sub> and Ni<sub>2</sub>P, respectively (Fig. 2d, and e). Similarly, Fig. 2e, f displays two amplified HRTEM images truncated from Fig. 2d. Fig. 2f demonstrates some hexagonal and trigonal lattice areas of semi-conductor 2H<sub>phase</sub>- and metallic 1T<sub>phase</sub>-MoS<sub>2</sub>, respectively. The HRTEM results further confirm the successful preparation of the 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P multi-heterojunction interface catalyst.
+
+Next, we performed XPS measurement to assess the elemental valence states of all the as-synthesized samples (Fig. 2g-i and Supplementary Fig. 7). Full XPS spectrum for 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> electrode (Supplementary Fig. 7a) showed that atomic ratios of Mo, S and Ni were equal to 13.96%, 36.96% and 4.39 %, respectively, and close to that measured by HRTEM elemental mapping (∼14.30 %, 35.87 %, and 4.76 %). Mo 3d spectra obtained for the multi-heterojunction interfaces of the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>, 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P, 1T<sub>0.41</sub>-MoS<sub>2</sub> and 2H<sub>phase</sub>-MoS<sub>2</sub> electrodes showed Mo 3d<sub>3/2</sub> and 3d<sub>5/2</sub> peaks of the 2H phase at 233.2 and 229.9 eV, and of the 1T phase at 232.4 and 229.1 eV in 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P, respectively. These results suggest Mo<sup>4+</sup> presence (Fig. 2g). Interestingly, the 1T phase contents in the 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> samples (equal to 81% and 72%, respectively) were higher than the 41% value observed for the 1T<sub>0.41</sub>-MoS<sub>2</sub> phase. Thus, phosphorus or sulfur implantation further facilitates the phase transformation of 1T<sub>phase</sub>-MoS<sub>2</sub>.<sup>22,27</sup> The S 2p spectra also displayed similar results (Fig. 2h). All as-prepared electrodes also have two new doublet peaks of 161.9 eV and 163.2 eV, indicating that these belong to the characteristic peaks of 1T<sub>phase</sub>-MoS<sub>2</sub>, further confirming 2H → 1T phase transformation.<sup>22</sup> Ni 2p spectrum showed peaks at 855.4 and 872.9 eV corresponding to Ni 2p<sub>1/2</sub> and 2p<sub>3/2</sub>, respectively (Fig. 2i) and two satellite peaks, suggesting the existence of Ni<sup>2+</sup> in the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P samples. Both peaks were slightly shifted (by 0.5 eV) towards higher binding energies, suggesting that there is an interaction between NiS<sub>2</sub> (or Ni<sub>2</sub>P) and MoS<sub>2</sub> via as-formed hetero-structures. For the 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalyst, the XPS survey spectrum displays that the atomic ratio of Mo, S, P and Ni are 19.29%, 32.7%, 8.23%, 8.23% (Supplementary Fig. 7a). Also, as shown in Supplementary Fig. 7b, there are two doublets of P 2p peaks (129.4 eV, 130.2 eV), which further confirms the formation of Ni<sub>2</sub>P. These results indicate the successful synthesis of multi-heterogeneous-interface catalysts.
+
+Electrocatalytic HER performances in alkaline and acidic media. 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrodes exhibited attractive multi-heterogeneous interface edges, plentiful active sites and abundant mass transfer and gas release channels and are expected to be used as very effective and stable catalysts for H<sub>2</sub> production. First, we analyzed HER activities (in 1.0 M KOH) of the electrodes containing these electrodes. The overpotentials of the electrodes containing 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P were 128 and 168 mV at 10 mA/cm<sup>2</sup>, respectively (see linear sweep voltammetry (LSV) results in Fig. 3a). These values are close to the Pt/C electrode potential equal to 84 mV. For the reaction kinetics analysis, we adjusted the Tafel slopes of these electrodes using the Tafel equation<sup>38,39</sup> and obtained the smallest slopes equal to 68 and 79 mV/dec for the electrodes containing 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P, respectively (Supplementary Fig. 8a). These values are even closer to the corresponding slope of the Pt/C electrode (equal to 56 mV/dec). Thus, electrodes containing 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P as active materials exhibit the fastest HER processes and better reactivity, which is attributed to the multi-heterogeneous interface effect, a large number of defects, and a higher proportion of 1T<sub>phase</sub>-MoS<sub>2</sub>. Next, we evaluated the long-term cycling stability of the as-prepared electrodes using the chronopotentiometry technique at 10 and 30 mA/cm<sup>2</sup>, respectively. The electrodes containing 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P were very robust and exhibited negligible damping after 16 h measurement (Supplementary Fig. 8b), and the LSV curves measured before and after the long-term tests are almost the same (Supplementary Fig. 8c), demonstrating excellent long-term stability. Supplementary Fig. 8d lists the overpotential values for the 20.0 wt % Pt/C, 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrodes in 1.0 M KOH at various current densities. 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> electrodes exhibited lower overpotential. Generally, low overpotential and Tafel slope values demonstrated the superior HER catalytic activities, which was the case for our 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrodes. 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> containing electrode has such excellent HER activity comparable to those of as-reported Mo-based materials (Fig. 3b) and composites and various representative catalysts<sup>25,26,28,31-34</sup> (Supplementary Table 1). Thus, 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> electrode is a catalyst with the best HER activity in alkaline solutions.
+
+To obtain the electrochemically active area (ECSA) of the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrodes, the double-layer capacitance (C<sub>dl</sub>) was calculated because the two values are proportional to each other. Therefore, we tested their cyclic voltammetry (CV) by continuously increasing scanning speed (Supplementary Fig. 9a-c) in order to obtain the CV curve of the electrode materials in the non-Faraday region (-0.2-0.4 V). Then, as shown in Supplementary Fig. 9d, the C<sub>dl</sub> was calculated from the plot slope (slope = 2C<sub>dl</sub>) between current-density difference (∆j) (0.15 V vs. RHE) and scan rate. The 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> electrodes possessed the highest C<sub>dl</sub> value (C<sub>dl</sub> = 359.7 mF/cm<sup>2</sup>), suggesting a multi-heterogeneous interface could be effectively enhanced conductivity and exposed more active sites of as-prepared electrodes. We recorded the electrochemical impedance spectra (EIS). The corresponding Nyquist (Supplementary Fig. 10) of the 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> electrode showed the lowest value for the charge transfer resistance (R<sub>ct</sub>). Thus, it possessed very favorable charge transfer kinetics.
+
+Next, we also studied the HER performance of all the as-prepared electrodes in 0.5 M H<sub>2</sub>SO<sub>4</sub> (Fig. 3c). The HER catalytic performance of the electrodes containing 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> were significantly improved their HER activities according to the LSV data: their overpotential values at 10 mA/cm<sup>2</sup> were as low as 38.9 and 98.5 mV, respectively, which is lower than the values for the electrodes containing 1T<sub>0.41</sub>-MoS<sub>2</sub>@Ni(OH)<sub>2</sub> (236 mV), 1T<sub>0.41</sub>-MoS<sub>2</sub> (389 mV), 1T<sub>phase</sub>-MoS<sub>2</sub> (392 mV), and 2H<sub>phase</sub>-MoS<sub>2</sub> (354 mV). The Tafel slopes for the electrodes containing 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> were 41 and 42 mV/dec (Supplementary Fig. 11a). These values were lower than the values obtained for electrodes containing 1T<sub>0.41</sub>-MoS<sub>2</sub> (65 mV/dec), 1T<sub>phase</sub>-MoS<sub>2</sub> (76 mV/dec), and 2H<sub>phase</sub>-MoS<sub>2</sub> (71 mV/dec) and were close to the electrode based on 20 wt% Pt/C (38 mV/dec). It is probably because, in the acidic environment, the H<sub>2</sub> desorption is the limiting step because H<sup>+</sup> are abundant. The 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrode had a weaker adsorption capacity toward H<sub>ads</sub> so it exhibits a better catalytic effect than 2H<sub>phase</sub>-MoS<sub>2</sub>.<sup>40</sup> Meanwhile, compared to the other electrodes, 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P also has a higher ECSA because it has a larger C<sub>dl</sub> (C<sub>dl</sub> = 106.15 mF/cm<sup>2</sup>, Supplementary Fig. 12) and, as a result, more catalytical sites, which significantly contributed to the overall activity. Furthermore, 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P also possesses a much smaller R<sub>ct</sub>, in contrast to other electrodes at 50 mV overpotential vs. RHE (Supplementary Fig. 13), revealing satisfied electron transport and good catalytic kinetics, which leads to high activity and low Tafel slope. Supplementary Fig. 11b shows that at 10 and 45 mA/cm<sup>2</sup>, 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P and 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> electrodes were very durable and possesses negligible damping after 16 h measurement, which displays excellent long-term stability. In addition, even after 16 h of a chronoamperometric stability test of the electrodes, the current density remains above 95% (Supplementary Fig. 11c), and there is only a slight deviation for the LSV recorded after the stability test, indicating that as-prepared electrodes have very good stability in an acidic environment. As to 20.0 wt % Pt/C, 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>, and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrodes in 0.5 M H<sub>2</sub>SO<sub>4</sub>, Supplementary Fig. 11d shows overpotentials vs. various current densities. 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P exhibits lower overpotential. We also compared the overpotentials (at 10 mA/cm<sup>2</sup> in acidic medium) and Tafel slopes with previously excellent Mo-based electrocatalysts<sup>27,31,41-44</sup> (Fig. 3d and Supplementary Table 2). Catalytic HER performance of 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P is also superior. Based on the above results, 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P multi-heterogeneous interface catalyst shows the remarkable intrinsic HER activities in acidic medium mainly attributed to multi-heterointerface interface edges active sites.
+
+Theoretical calculation and mechanisms analysis of the surface electronic structure and HER activation energy for the as-prepared electrocatalysts. To explain the distinguished synergistic effect of 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> (or 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P) multi-heterogeneous interface catalysts, Density functional theory (DFT) calculations were also performed. Model building and computational parameters can be seen in the “Methods” section. Firstly, the interfacial electron interaction was investigated. The charge difference images (Fig. 4a, b and Supplementary Fig. 14) reveal the charge transfer from 1T<sub>0.41</sub>-MoS<sub>2</sub> to the Ni<sub>2</sub>S or/and Ni<sub>2</sub>P interface, and the introduction of 1T phase is more conducive to charge transfer from MoS<sub>2</sub> to NiS<sub>2</sub> or Ni<sub>2</sub>P interface, which significantly increase the interface electron concentration and thus improve its activity. To better understand the surface electronic structure reconfiguration of MoS<sub>2</sub> through coordinated phase transition and interface regulation in theory, the band structure and density of states (DOS) of bare NiS<sub>2</sub>, Ni<sub>2</sub>P, 2H<sub>phase</sub>-MoS<sub>2</sub>, 1T<sub>phase</sub>-MoS<sub>2</sub>, 2H<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>, 2H<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P, 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>, and 1T<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P (Fig. 4c-e and Supplementary Fig. 15) obtained using the hybrid DFT-HSE06 exchange–correlation functional, which is presented in the Supplementary Information. The calculation results show that the bare NiS<sub>2</sub> exhibits typical semiconductor characteristics (Fig. 4c), with a narrow bandgap equal to 0.68 eV (Supplementary Fig. 15a). The band structure of 1T<sub>phase</sub>-MoS<sub>2</sub> (Fig. 4d) and 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> (Fig. 4e) exhibited a certain zero bandgap, indicating a complete transition from the semiconductor phase (0.91 eV, Supplementary Fig. 18b) to the metallic phase (0 eV) with improved conductivities.<sup>27</sup> Notably, the intensity of PDOS of 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> was higher than that of 1T<sub>phase</sub>-MoS<sub>2</sub> and NiS<sub>2</sub> at the Fermi level (Supplementary Fig. 15). Thus, the electron mobility of the 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> catalysts was more favorable for the efficient charge transfer, which agrees consistent with the EIS test results.<sup>45</sup> Moreover, the PDOS results imply that the NiS<sub>2</sub> interface hybrid generates some new interface electronic states in 1T<sub>phase</sub>-MoS<sub>2</sub> (Supplementary Fig. 15c), which was very likely because of hybridization of the d-orbital of Mo and an empty d-orbital of Ni. Thus, higher HER activity of 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> in comparison to 1T<sub>phase</sub>-MoS<sub>2</sub> agrees with the Fermi level DOS (Fig. 4d. e). Thus, the actual electrochemical performance would show even faster conductivity and charge transfer kinetics.
+
+The HER effect was mainly studied by a three-state diagram containing original H<sup>+</sup> and intermediate H* states and 1/2 H<sub>2</sub> formation<sup>30,46</sup>. However, the energy of the intermediate state H*(ΔG<sub>H*</sub>) is a critical indicator of the ability of hydrogen evolution.<sup>47</sup> To reveal further the relationship of HER activity of catalysts with phase structure and heterojunction-interface, we used DFT to calculate the ΔG<sub>H*</sub> for HER on 2H<sub>phase</sub>-MoS<sub>2</sub>, 1T<sub>phase</sub>-MoS<sub>2</sub>, 2H<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>, 2H<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P, 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>, and 1T<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalysts with partially multi-heterojunction interface modification. Fig. 4f displays the calculated free energy diagram on the most stable energy of the 2H phase, 1T phase, 2H<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P, 2H<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>, 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P catalysts (Supplementary Fig. 15). For 2H<sub>phase</sub>-MoS<sub>2</sub>, the ΔG<sub>H*</sub> is very positive (1.18 eV), indicating that there is a strong interaction between H* and 2H<sub>phase</sub>-MoS<sub>2</sub>, showing poor HER reaction kinetics. The introduction of the 1T-phase into MoS<sub>2</sub> can obviously increase the value of ΔG<sub>H*</sub> to – 0.36 eV, implying promoted HER activity compared to 2H<sub>phase</sub>-MoS<sub>2</sub>. However, constructing multi-heterointerface interface edges active sites with NiS<sub>2</sub> (1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>) would lead to the ΔG<sub>H*</sub> value equal to almost 0 eV (- 0.17 eV). For comparison, ΔG<sub>H*</sub> for the 2H<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> was equal to 0.74 eV. The reason is that H* adsorbed on the surface of 2H<sub>phase</sub>-MoS<sub>2</sub> bounds to Mo atoms, and strong Mo-H strength and poor conductivity. However, H* can be absorbed not only by the 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> surface. Ni atoms possess empty d orbitals capable of binding H atoms, thereby weakening the Mo-H strength. More importantly, the introduction of the 1T-phase not only increases its electrical conductivity but also creates abundant active sites at the multi-heterojunction interface edges, which synergistically promote HER activity (Fig. 4g). Thus, our work demonstrates a novel and efficient design to create multi-heterogeneous interfacial electrocatalysts without noble metal materials and with excellent HER activity.
+
+# Discussion
+
+In summary, we have constructed multi-heterogeneous-interface catalysts (1T$_{0.81}$-MoS$_2$@Ni$_2$P and 1T$_{0.72}$-MoS$_2$@NiS$_2$) by tuning its electronic structure of phase modulation synergistic with interfacial chemistry and defects to phosphorus or sulfur implantation strategies, which is an efficient approach to obtain abundant reactive sites of long-cycling and stable electrocatalysts for HER in alkaline and acidic surroundings. The as-achieved 1T$_{0.81}$-MoS$_2$@Ni$_2$P and 1T$_{0.72}$-MoS$_2$@NiS$_2$ electrodes only require small overpotentials of 38.9 (or 186) and 98.5 (or 128) mV to drive HER at 10 mA/cm$^2$ and have low Tafel slopes: 41 (or 79) and 42 (or 68) mV/dec in 0.5 M H$_2$SO$_4$ (or 1.0 M KOH). Accordingly, these results show varieties of multi-heterogeneous interfaces in 1T$_{0.81}$-MoS$_2$@Ni$_2$P and 1T$_{0.72}$-MoS$_2$@NiS$_2$ electrodes, which can be considered versatile electroactive sites and facilitate electron transfer because of their unique heterogeneous-effects. DFT calculation results also display that the introduction of metallic-phase MoS$_2$ and intrinsic HER-active Ni-based materials can regulate MoS$_2$ electronic structure effectively (a narrower bandgap and reduced Ni supply empty d-orbitals to facilitate H atom capture) and decrease Mo-H strength of 1T$_{0.81}$-MoS$_2$@Ni$_2$P (or 1T$_{0.72}$-MoS$_2$@NiS$_2$) catalysts that account for the outstanding HER properties with lower Tafel slopes and overpotentials compared with 2H$_{\text{phase}}$-MoS$_2$, 1T$_{\text{phase}}$-MoS$_2$ counterparts and other Mo-based catalysts. Thus, our work provides a new horizon for rationally designing multi-heterogeneous interfaces of non-precious electrocatalysts to realize excellent HER activities.
+
+# Methods
+
+**Synthesis of 1T<sub>0.41</sub>-MoS<sub>2</sub>**
+
+MoS<sub>2</sub> microspheres were grown on carbon cloth (CC) hydrothermally. First, a CC (2 × 4 cm) was cleaned (for 15 min) using acetone and then sonicated in deionized water and ethanol for 10 minutes. Then, sodium molybdate (Na<sub>2</sub>MoO<sub>4</sub>·2H<sub>2</sub>O, 411.9 mg) and thiourea (CS(NH<sub>2</sub>)<sub>2</sub>, 608.96 mg) were added to deionized water (40 mL) and citric acid (20 mL). The mixture was magnetically stirred to form a cleaning solution, then placed into a 100 mL Teflon-lined autoclave and held in it at 180 °C for 12 h. Finally, the CC substrates with 1T<sub>0.41</sub>-MoS<sub>2</sub> microspheres (denoted through the paper as 1T<sub>0.41</sub>-MoS<sub>2</sub>) were rinsed using deionized water and ethanol and vacuum-dried for 6 h at 60<sup>o</sup>C. For comparison, deionized water was used as the solvent, and 2H<sub>phase</sub>-MoS<sub>2</sub> microspheres were synthesized hydrothermally at 220<sup>o</sup>C for 24 h from the same precursors.
+
+**Synthesis of 1T<sub>phase</sub>-MoS<sub>2</sub>**
+
+We used Li-intercalated bulk MoS<sub>2</sub> to prepare 1T<sub>phase</sub>-MoS<sub>2</sub>. <sup>48</sup> In an Ar-filled glove box, bulk MoS<sub>2</sub> (1.0 g) prepared by stripping were dispersed in 15 mL of 2M n-BuLi/hexane solution and stirred at ambient conditions for 48 h. The resulting black materials were repeatedly rinsed with anhydrous n-hexane and then centrifuged to eliminate n-butyl lithium excess and other solution impurities. The 1T<sub>phase</sub>-MoS<sub>2</sub> powder was prepared and was then coated on the CC substrate.
+
+**Synthesis of 1T<sub>0.41</sub>-MoS2@Ni(OH)<sub>2</sub>**
+
+We use a standard three-electrode system to prepare 1T<sub>0.41</sub>-MoS<sub>2</sub>@Ni(OH)<sub>2</sub>. 1T<sub>0.41</sub>-MoS<sub>2</sub> acted as a working electrode, while Pt sheet and Ag/AgCl/3.5 M KCl acted as counter and reference electrodes. Ni(OH)<sub>2</sub> was electrodeposited on the 1T<sub>0.41</sub>-MoS<sub>2</sub> using 0.1 M Ni(NO<sub>3</sub>)<sub>2</sub> at 5.0 mA/cm<sup>2</sup> cathode current density applied for 300 s. 1T<sub>0.41</sub>-MoS<sub>2</sub>@Ni(OH)<sub>2</sub> samples were rinsed with deionized water and ethanol several times and vacuum-dried at 60<sup>o</sup>C.
+
+**Synthesis of 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub>**
+
+The 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> multi-heterogeneous interfaces were prepared by the solid-vapor reaction method. First, a piece of 1T<sub>0.41</sub>-MoS<sub>2</sub>@Ni(OH)<sub>2</sub> grew on CC was put into the quartz tube with 32.0 mg S powder and was then sealed. Secondly, the quartz tube was positioned inside a tube furnace and was calcinated at 500<sup>o</sup>C for 60 min to obtain a 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> electrode.
+
+**Synthesis of 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P**
+
+Similarly, the 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P multi-heterogeneous interfaces were also obtained by the solid-vapor reaction method. First, a piece of 1T<sub>0.41</sub>-MoS<sub>2</sub>@Ni(OH)<sub>2</sub> grew on CC was put into the quartz tube with 31.0 mg red phosphorus and was then sealed. Secondly, the quartz tube was also calcinated at 580<sup>o</sup>C for 1.0 h to prepare the 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrode. Additionally, 20 wt% Pt/C was also coated on CC substrate (2.0 mg/cm<sup>2</sup>) and was labeled as 20% Pt/C for comparison.
+
+**Materials characterization**
+
+All as-synthetized electrodes were characterized by XRD (performed by Bruker D8 Advance instrument) and Raman spectroscopy (performed using Horiba LabRAB HR800 instrument). The sample morphologies were studied using SEM performed by Hitachi SU8010 instrument and TEM (performed by FEI Tecnai F30 instrument). XPS spectra were collected by the r ESCALAB 250Xi instrument manufactured by Thermo Fisher) using Al Kα radiation.
+
+**Electrochemical measurements**
+
+The loadings of 1T<sub>0.72</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> and 1T<sub>0.81</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P electrodes were 0.685 and 0.762 mg/cm<sup>2</sup>, respectively. The tests were conducted by a three-electrode system containing 1.0 M KOH or 0.5 M H<sub>2</sub>SO<sub>4</sub> electrolytes using a CHI 660E workstation. All materials were used as the working electrodes (1 × 1 cm in size) as-synthesized. A graphite rod and Hg/HgO or Ag/AgCl acted as the counter and reference electrodes, respectively. The LSV scan rate was 5.0 mV/s. CV tests were performed 50.0 mV/s scan rate. N<sub>2</sub> was passed through the electrolyte for 16 min) before each test to minimize the dissolved O<sub>2</sub> content.
+
+**DFT theoretical calculation**
+
+Model Building: According to the HRTEM micrographs, 1T<sub>phase</sub>-MoS<sub>2</sub>, 2H<sub>phase</sub>-MoS<sub>2</sub>, Ni<sub>2</sub>P, and NiS<sub>2</sub> formed a multiphase heterojunction. 1T<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P interface, 1T<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> interface, 2H<sub>phase</sub>-MoS<sub>2</sub>@Ni<sub>2</sub>P interface, 2H<sub>phase</sub>-MoS<sub>2</sub>@NiS<sub>2</sub> interface, bulk 1T<sub>phase</sub>-MoS<sub>2</sub>, 2H<sub>phase</sub>-MoS<sub>2</sub> were also constructed as comparisons. Considering the Van der Waals forces between the two phases, the unrelaxed heterojunction interface distance was set to 3.0 Å. These original structures were obtained from <em>Materials Project Database</em>. <sup>49</sup>
+
+Computational parameters: DFT calculation was applied to calculate electronic structures of two crystal structures by the partial augmented plane wave method (PAW) implemented in the VASP <sup>50</sup> using VASPKIT code for post-processing. Considering the heterojunction structure, the long-range force correction was considered by using the DFT-D3 correction method of Grimme. <sup>51</sup> The Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation <sup>52</sup> was implemented for exchange-correlation energy calculations using 550 eV kinetic energy cut off for the plane-wave basis. Then structural optimizations using a conjugate gradient (CG) method based on the pre-optimized structure were repeated until the maximum force component on each atom remained below 0.01 eV/Å. Monkhorst-Pack k-point meshes in the first Brillouin zone of the primitive cell were used the VASPKIT code recommended accuracy levels of 0.04 for the optimization calculation and 0.02 for the static calculation, respectively. After fully relaxing the structures, one final (electronic scf) step with the tetrahedron method using Blöchl corrections and denser k-meshes was employed for DOS calculation. In addition to the H adsorbed energy calculations, the frequency calculation of free H and free energy correction at 298.15 K (including the entropy and zero-point energy contributions) were also calculated. To avoid abnormal entropy contribution, frequencies less than 50 cm<sup>-1</sup> are set to be 50 cm<sup>-1</sup>.
+
+# References
+
+1. Choi, W. et al. High-detectivity multilayer MoS₂ phototransistors with spectral response from ultraviolet to infrared. Adv. Mater. **24**, 5832–5836 (2012).
+2. Shi, J. et al. Two-dimensional metallic tantalum disulfide as a hydrogen evolution catalyst. Nat. Commun. **8**, 958 (2017).
+3. Yang, L. et al. Combining photocatalytic hydrogen generation and capsule storage in graphene based sandwich structures. Nat. Commun. **8**, 16049 (2017).
+4. Yi, J. et al. Large-scale production of ultrathin carbon nitride-based photocatalysts for high-yield hydrogen evolution. Appl. Catal. B: Environ. **281**, 119475 (2021).
+5. Wei, Z. et al. Simultaneous realization of sulfur-rich surface and amorphous nanocluster of NiS₁₊ₓ cocatalyst for efficient photocatalytic H₂ evolution. Appl. Catal. B: Environ. **280**, 119455 (2021).
+6. Peng, L. et al. Accelerated alkaline hydrogen evolution on M(OH)ₓ/M-MoPOₓ (M = Ni, Co, Fe, Mn) electrocatalysts by coupling water dissociation and hydrogen ad-desorption steps. Chem. Sci. **11**, 2487–2493 (2020).
+7. Wang, P. et al. Precise tuning in platinum-nickel/nickel sulfide interface nanowires for synergistic hydrogen evolution catalysis. Nat. Commun. **8**, 14580 (2017).
+8. Chen, Y. et al. Electrocatalytically inactive SnS₂ promotes water adsorption/dissociation on molybdenum dichalcogenides for accelerated alkaline hydrogen evolution. Nano Energy **64**, 103918 (2019).
+9. Mahmood, N. et al. Electrocatalysts for hydrogen evolution in alkaline electrolytes: mechanisms, challenges, and prospective solutions. Adv. Sci. **5**, 1700464 (2018).
+10. Sun, Y. et al. Strongly coupled dual zerovalent nonmetal doped nickel phosphide nanoparticles/nitrogen, boron-graphene hybrid for pH-universal hydrogen evolution catalysis. Appl. Catal. B: Environ. **278**, 119284 (2020).
+11. Walter, M. et al. Solar Water Splitting Cells. Chem. Rev. **110**, 6446 (2010).
+12. Yang, B. et al. Amorphous phosphatized ruthenium-iron bimetallic nanoclusters with Pt-like activity for hydrogen evolution reaction. Appl. Catal. B: Environ. **283**, 119583 (2021).
+13. Gao, B. et al. 3D flower-like defected MoS₂ magnetron-sputtered on candle soot for enhanced hydrogen evolution reaction. Appl. Catal. B: Environ. **263**, 117750 (2020).
+14. Ren, J. et al. Molybdenum-based nanoparticles (Mo₂C, MoP and MoS₂) coupled heteroatoms-doped carbon nanosheets for efficient hydrogen evolution reaction. Appl. Catal. B: Environ. **263**, 118352 (2020).
+15. Yao, N. et al. Oxygen-Vacancy-Induced CeO₂/Co₄N heterostructures toward enhanced pH-Universal hydrogen evolution reactions. Appl. Catal. B: Environ. **277**, 119282 (2020).
+16. Fan, H. et al. Plasma-heteroatom-doped Ni-V-Fe trimetallic phospho-nitride as high-performance bifunctional electrocatalyst. Appl. Catal. B: Environ. **268**, 118440 (2020).
+17. Wang, H. et al. Confined growth of pyridinic N–Mo₂C sites on MXenes for hydrogen evolution. J. Mater. Chem. A. **8**, 7109–7116 (2020).
+18. Zhang, X. et al. Structure and phase regulation in MoₓC (α-MoC₁₋ₓ/β-Mo₂C) to enhance hydrogen evolution. Appl. Catal. B: Environ. **247**, 78–85 (2019).
+19. Ge, Y. et al. Transforming nickel hydroxide into 3D prussian blue analogue array to obtain Ni₂P/Fe₂P for efficient hydrogen evolution reaction. Adv. Energy Mater. **8**, 1800484 (2018).
+20. Xu, K. et al. Yin-Yang Harmony: Metal and nonmetal dual-doping boosts electrocatalytic activity for alkaline hydrogen evolution. ACS Energy Lett. **3**, 2750–2756 (2018).
+21. Anjum, M.A.R. et al. Efficient Hydrogen Evolution Reaction Catalysis in Alkaline Media by All‐in‐One MoS₂ with Multifunctional Active Sites. Adv. Mater. **30**, 1707105 (2018).
+22. Chang, K. et al. Targeted Synthesis of 2H‐ and 1T‐Phase MoS₂ Monolayers for Catalytic Hydrogen Evolution. Adv. Mater. **28**, 10033–10041 (2016).
+23. Kim, M. et al. Covalent 0D–2D heterostructuring of Co₉S₈–MoS₂ for enhanced hydrogen evolution in all pH electrolytes. Adv. Funct. Mater. **30**, 2002536 (2020).
+24. Liu, Z. et al. Vertical nanosheet array of 1T phase MoS₂ for efficient and stable hydrogen evolution. Appl. Catal. B: Environ. **246**, 296–302 (2019).
+25. Lei, C. et al. Efficient alkaline hydrogen evolution on atomically dispersed Ni–Nₓ Species anchored porous carbon with embedded Ni nanoparticles by accelerating water dissociation kinetics. Energy Environ. Sci. **12**, 149–156 (2019).
+26. Wang, S. et al. Ultrastable In‐Plane 1T–2H MoS₂ heterostructures for enhanced hydrogen evolution reaction. Adv. Energy Mater. **8**, 1801345 (2018).
+27. Deng, S. et al. Synergistic doping and intercalation: realizing deep phase modulation on MoS₂ arrays for high‐efficiency hydrogen evolution reaction. Angew. Chem. **58**, 16289–16296 (2019).
+28. Chen, W. et al. Achieving rich and active alkaline hydrogen evolution heterostructures via interface engineering on 2D 1T‐MoS₂ quantum sheets. Adv. Funct. Mater. **30**, 2000551 (2020).
+29. Luo, Y. et al. Morphology and surface chemistry engineering toward pH-universal catalysts for hydrogen evolution at high current density. Nat. Commun. **10**, 269 (2019).
+30. Zhang, B. et al. Interface engineering: the Ni(OH)₂/MoS₂ heterostructure for highly efficient alkaline hydrogen evolution. Nano Energy **37**, 74–80 (2017).
+31. Kim, M. et al. Activating MoS₂ basal plane with Ni₂P nanoparticles for Pt‐Like hydrogen evolution reaction in acidic media. Adv. Funct. Mater. **29**, 1809151 (2019).
+32. Lin, J. et al. Defect‐rich heterogeneous MoS₂/NiS₂ nanosheets electrocatalysts for efficient overall water splitting. Adv. Sci. **6**, 1900246 (2019).
+33. Jia, Y. et al. A heterostructure coupling of exfoliated Ni–Fe hydroxide nanosheet and defective graphene as a bifunctional electrocatalyst for overall water splitting. Adv. Mater. **29**, 1700017 (2017).
+34. Liu, Y. et al. Interface engineering of (Ni, Fe)S₂@ MoS₂ heterostructures for synergetic electrochemical water splitting. Appl. Catal. B: Environ. **247**, 107–114 (2019).
+35. Liu, Z. et al. Heterogeneous nanostructure based on 1T-phase MoS₂ for enhanced electrocatalytic hydrogen evolution. ACS Appl. Mater. Inter. **9**, 25291–25297 (2017).
+36. Ding, W. et al. Highly ambient-stable 1T-MoS₂ and 1T-WS₂ by hydrothermal synthesis under high magnetic fields. ACS Nano **13**, 1694–1702 (2019).
+37. Wang, X. et al. 2D/2D 1T‐MoS₂/Ti₃C₂ MXene heterostructure with excellent supercapacitor performance. Adv. Funct. Mater. **30**, 0190302 (2020).
+38. Chang, B. et al. Bimetallic NiMoN nanowires with a preferential reactive facet: an ultraefficient bifunctional electrocatalyst for overall water splitting. Chem. Sus. Chem. **11**, 3198–3207 (2018).
+39. Zhang, J. et al. Synergistic interlayer and defect engineering in VS₂ nanosheets toward efficient electrocatalytic hydrogen evolution reaction. Small **14**, 1703098 (2018).
+40. Deng, J. et al. Enhanced electron penetration through an ultrathin graphene layer for highly efficient catalysis of the hydrogen evolution reaction. Angew. Chem. Int. Ed. **54**, 2100–2104 (2015).
+41. Jiao, Y. et al. Porous plate-like MoP assembly as an efficient pH-universal hydrogen evolution electrocatalyst. ACS Appl. Mater. Inter. **12**, 49596–49606 (2020).
+42. Mishra, I.K. et al. Hierarchical CoP/Ni₅P₄/CoP microsheet arrays as a robust pH-universal electrocatalyst for efficient hydrogen generation. Energy Environ. Sci. **11**, 2246–2252 (2018).
+43. Li, H. et al. Activating and optimizing MoS₂ basal planes for hydrogen evolution through the formation of strained sulphur vacancies. Nat. Mater. **15**, 48–53 (2016).
+44. Chen, Y. et al. Electrocatalytically inactive SnS₂ promotes water adsorption/dissociation on molybdenum dichalcogenides for accelerated alkaline hydrogen evolution. Nano Energy **64**, 103918 (2019).
+45. Zhang, B. et al. Simultaneous interfacial chemistry and inner Helmholtz plane regulation for superior alkaline hydrogen evolution. Energy Environ. Sci. **13**, 3007–3013 (2020).
+46. Zhang, J. et al. Engineering water dissociation sites in MoS₂ nanosheets for accelerated electrocatalytic hydrogen production. Energy Environ. Sci. **9**, 2789–2793 (2016).
+47. Wu, Y. et al. Electron density modulation of NiCo₂S₄ nanowires by nitrogen incorporation for highly efficient hydrogen evolution catalysis. Nat. Commun. **9**, 1425 (2018).
+48. Luo, Y. et al. Two-Dimensional MoS₂ Confined Co(OH)₂ Electrocatalysts for Hydrogen Evolution in Alkaline Electrolytes. ACS Nano **12**, 4565–4573 (2018).
+49. Jong, M. et al. Charting the complete elastic properties of inorganic crystalline compounds. Sci. Data. **2**, 150009 (2015).
+50. Intan, N. et al. Ab initio modeling of transition metal dissolution from the LiNi₀.₅Mn₁.₅O₄ cathode. ACS Appl. Mater. Inter. **11**, 20110–20116 (2019).
+51. Grimme, S. et al. Effect of the damping function in dispersion corrected density functional theory. J. Comput. Chem. **32**, 1456–1465 (2011).
+52. Perdew, J. et al. Generalized gradient approximation made simple. Phys. Rev. Lett. **78**, 1396–1396 (1997).
+
+# Supplementary Files
+
+- [Supplementaryinformation.doc](https://assets-eu.researchsquare.com/files/rs-319079/v1/1ee4d29dcc3c0ea9f0e57d0c.doc)

3146ffcca05f88471f84512ce00b7eb2be385dff7546029ba1b0fa8e71247ee2/preprint/preprint.md

@@ -0,0 +1,210 @@
+# Abstract
+
+The application of organoids has been limited by the lack of methods for producing uniformly mature organoids at scale. This study introduces an organoid culture platform, called UniMat, which addresses the challenges of uniformity and maturity simultaneously. UniMat is designed to not only ensure consistent organoid growth but also facilitate an unrestricted supply of soluble factors by a 3D geometrically-engineered, permeable membrane-based platform. Using UniMat, we demonstrate the scalable generation of kidney organoids with enhanced uniformity in both structure and function compared to conventional methods. Notably, kidney organoids within UniMat matured significantly better, showing increased expression of nephron transcripts, more *in vivo*-like cell-type balance, and better vascularization. Moreover, UniMat's design offers a more standardized organoid model for drug testing, as demonstrated by its consistent response to a polycystic-kidney-disease drug. In essence, UniMat presents a transformative platform for organoid technology, promising applications in organ development, disease modeling, and drug screening.
+
+Biological sciences/Biotechnology/Nanobiotechnology/Nanofabrication and nanopatterning  
+Biological sciences/Biotechnology/Biomaterials  
+Biological sciences/Biotechnology/Stem-cell biotechnology
+
+# Main
+
+Organoids, which self-organize into complex multicellular structures from human stem cells, have the potential to recapitulate the structural and functional units of their corresponding organs in the human body<sup>1, 2</sup>. As a result, organoids have emerged as promising *in vitro* tools for investigating human organogenesis and disease, as well as for screening therapeutic molecules<sup>3, 4</sup>. Despite significant progress in generating organoids that possess a complexity akin to *in vivo* physiological systems, their widespread adoption in clinical trial validation and pharmaceutical drug development remains limited. This limitation is primarily attributed to the high variability observed in organoid morphology, function, and formation efficiency<sup>5</sup>. Such variability is especially prevalent in organoids formed without physical constraints<sup>6</sup>. For example, organoids generated through spontaneous self-organization within extracellular matrix (ECM) hydrogels, such as Matrigel or Geltrex<sup>7, 8</sup>, frequently exhibit considerable variations in size, shape, and differentiation<sup>6</sup>. This phenomenon stems from the inherently non-linear and deterministic nature of organogenesis and organoid formation, where even slight deviations in initial condition can lead to significant differences in the final morphogenesis<sup>9</sup>. This inconsistency complicates accurate experimental comparisons and compromise the reliability of outcomes, thereby posing challenges to the standardization of organoid-based assays<sup>10</sup>.
+
+To address these issues, researchers have employed microwell platforms. These are engineered polymeric substrates designed with microcavities that serve as physical constraints in organoid culture. Microwells offer a solution to reduce variability in organoid development by controlling cellular density and the geometry of either embryonic bodies<sup>11–13</sup> or aggregates of differentiated progenitor cells<sup>8, 14, 15</sup>, thereby allowing for precise guidance over the initial geometry of organoids. Furthermore, microwell platforms designed with a microwell array facilitate the formation of an array of organoids, thus improving the scalability essential for high-throughput organoid assays. Recently, various organoids, including those resembling the pancreas<sup>11</sup>, intestinal<sup>14</sup>, lung<sup>12</sup>, and kidney<sup>15</sup>, have been successfully generated using microwell platforms, exhibiting enhanced uniformity. Nonetheless, conventional microwell platforms have significant limitations. Their small dimensions and confined spaces of the impermeable microwells can obstruct the efficient diffusion of soluble factors from the surrounding medium to the organoids. This can lead to an accumulation of metabolic wastes around the organoids and a restricted supply of nutrients, growth factors, and oxygen vital for organoid growth and maturation<sup>16–18</sup>. As organoids grow, their need for nutrients, growth factors, and oxygen escalates, potentially exceeding the diffusion capacities of impermeable microwells. While the previous studies have proposed integrating bioreactors, such as microfluidic chips, with microwells to enhance the diffusion of soluble factors via a continuous medium flow<sup>19–21</sup>, implementing such a solution is not straightforward. This complexity limits the widespread use of these methods.
+
+Here, we propose an innovative and versatile organoid culture platform designed to address current challenges, facilitating the scalable generation of organoids that are not only uniform but also mature. Our unique approach encompasses the development of 3D geometrically-engineered, permeable membranes that provide geometrical constraints, yet allow for the unhindered exchange of soluble factors. By incorporating this 3D permeable membrane into a cell culture insert, our platform seamlessly adapts to standard cell culture plates, obviating the need for alterations to established protocols or workflows. Employing this platform for the scalable production of human induced pluripotent stem cell (hiPSC)-derived kidney organoids, we have demonstrated the superiority of our method. It not only improves the uniformity and maturity of kidney organoids but also exhibits enhanced performance in organoid-based disease modeling and drug testing. This underscores the potential of our 3D geometrically-engineered, permeable membrane as a benchmark tool for organoid-based assays.
+
+# Results
+
+To generate organoids that are not only uniform but also mature at a scalable level, we first aimed to establish an organoid culture platform. This platform offers both geometrical constraints for precise organoid regulation and an unrestricted supply of soluble factors, ensuring the efficient diffusion and exchange of nutrients, growth factors, and oxygen (Fig. 1a). We named this platform “UniMat” (Uniform and Mature organoid culture platform), and developed it based on three primary design criteria:
+
+(1) UniMat should efficiently partition individual organoids, providing geometrical constraints that ensure organoid uniformity;
+
+(2) The individual microwells in UniMat should promote cell aggregation, which is essential for the initial formation of cohesive 3D aggregates crucial to organoid development;
+
+(3) UniMat should possess a porous structure, creating a highly permeable environment that facilitates unrestricting supply of soluble factors necessary for organoid maturation.
+
+To realize this concept, we engineered a flat nanofiber (NF) membrane into a 3D permeable microwell array, maintaining a highly porous membrane structure to ensure high permeability to both gases and solutes.
+
+To fabricate the UniMat, we employed the electrospinning process, followed by a subsequent matched-mold thermoforming process, based on our previous work (Fig. 1b). Notably, this matched-mold thermoforming process can precisely microstructure a thin and flat NF membrane, ranging in thickness from 50 µm to 150 µm, into a 3D shape. This results in superior shape flexibility and replication fidelity without compromising the porous structure of the NF membrane. Leveraging this capability, we successfully engineered an electrospun free-standing NF membrane with a thickness of 50 µm into a 3D microwell array-structured membrane, resulting in a UniMat. Subsequently, the UniMat was incorporated into the bottom opening of a custom-made 24-well insert wall in a free-standing configuration, allowing for compatibility with standard cell culture plates (Fig. 1c). The successful fabrication of UniMat, with its 3D microwell array-structured NF membrane, was confirmed using scanning electron microscopy (SEM) analysis (Fig. 1d).
+
+In this study, we fabricated the UniMat by microstructuring an electrospun NF membrane to form a V-shaped microwell array (Fig. 1e), promoting the collection and aggregation of seeded cells at the bottom regions of microwells. The V-shaped design enhances cell confinement owing to its inclined walls, ensuring a more defined region for cell placement within every microwell. Importantly, by merely modifying the mold geometry used in the matched-mold thermoforming process, the size of the microwells could be tuned to accommodate the growth of organoids of various sizes. We fabricated three different types of UniMats, each with distinct width and depth dimensions: UniMat400 (width: 400 µm, depth: 343 µm), UniMat600 (width: 600 µm, depth: 517 µm), and UniMat800 (width: 800 µm, depth: 691 µm) (Supplementary Fig. 1). The microwell in the UniMat preserved a highly porous architecture similar to the original NF membrane (Fig. 1f), and possessed a thin wall thickness of approximately 30 µm (Fig. 1g). These characteristics enabled the UniMat to display a high permeability for molecules of varying molecular weights, which was more than twice that of a conventional PET porous membrane (Fig. 1h).
+
+To validate the capability of the UniMat to generate uniform and mature organoids, we selected kidney organoids as a representative lineage of organs. Kidney organoids are notably complex and hold significant biomedical interest because of their potential applications in disease modeling and drug testing. For differentiation of kidney organoids, we meticulously optimized the Morizane protocol (Supplementary Notes 1 and 2 and Supplementary Figs. 2 and 3) and seamlessly incorporated this refined protocol into the UniMat. In accordance with this protocol, nephron progenitor cells (NPCs) derived from hiPSCs were seeded onto the UniMat400 on day 9 and subjected to the subsequent differentiation processes within the UniMat400, resulting in the formation of kidney organoids by day 21 (Fig. 1i). Upon differentiation, a vast majority of kidney organoids exhibited nephron-like structures, including podocytes (PODXL+), proximal tubules (LTL+), and distal tubules (CDH1+), within the UniMat (Fig. 1j). Remarkably, approximately 87 ± 5% of all pretubular aggregates were successfully developed into nephron-like kidney organoids, achieving around 5 organoids per mm² within the UniMat400.
+
+## UniMat improves uniformity of organoids
+
+Having demonstrated the successful production of kidney organoids in the UniMat400, we utilized it as a model to explore its potential to reduce the variability of organoids. We assessed the morphological consistency in size and structure of kidney organoids in the UniMat platforms (Fig. 2a), which provide geometric guidance from the NPC stage. This was further compared to the uniformity observed in organoids that were exclusively generated on a plate coated with Geltrex hydrogel (hereinafter referred to as “hydrogel layer”), without any physical constraints (Fig. 2a). Notably, kidney organoids cultured in each UniMat demonstrated a significantly more consistent size distribution compared to those on the hydrogel layer (Fig. 2b and Supplementary Table 1). Moreover, the size of organoids appeared to be regulated through the UniMat-based culture, correlating with the size of the V-shaped microwells in the UniMat (Fig. 2b).
+
+To delve deeper into the mechanism by which UniMat controls the organoid size, we examined the initial aggregate size, which is known to influence organogenesis and organoid formation. Given the distinct sizes and arrays of V-shaped microwells in each of the UniMat400, UniMat600, and UniMat800 (Supplementary Fig. 4a), we noted that within 6 h of seeding, aggregates were formed in sizes corresponding to each specific UniMat (Supplementary Fig. 4b). These measured sizes closely matched with the estimated sizes based on the geometrical data of the microwells and the average number of cell sedimentations per microwell in each UniMat (Supplementary Note 3 and Supplementary Fig. 4c–f). Our findings suggest that by adjusting both the microwell size and the cell seeding density in the UniMat, the initial aggregate size can be effectively modulated. Notably, the size variances among the organoids grown in different UniMats were consistent with the trends observed in the initial aggregate sizes, underscoring the pivotal role of UniMat in determining organoid size.
+
+Having established that the UniMat can effectively regulate organoid size, we further investigated the structural variability of kidney organoids. Using confocal imaging rendered with the Imaris 3D surface rendering software, we quantified the presence of glomerular podocytes (PODXL+), proximal tubules (LTL+), and distal tubules (CDH1+) in individual kidney organoids (Fig. 2c). Notably, the kidney organoid cultured in the UniMat exhibited significantly less structural variability compared to those cultured on the hydrogel layer. These findings indicate that the geometrical constraints provided by the UniMat promote the production of organoids with more uniform sizes and structures (Fig. 2d and Supplementary Table 2). Especially in the UniMat800, we observed a remarkable reduction in structural variability, with an 8-fold decrease in podocytes, a 9-fold decrease in proximal tubules, and a 5-fold decrease in distal tubules, as compared to the organoids cultured on the hydrogel layer (Supplementary Table 2). Based on our data, the UniMat800-cultured kidney organoids exhibited the highest degree of uniformity and they had an average size comparable to those cultured on the hydrogel layer. In this regard, UniMat800 was selected as the representative UniMat for further experiments.
+
+To further assess the potential of UniMat in improving the functional uniformity of organoids, we randomly selected 10 organoids from both the hydrogel layer and UniMat. We then performed a quantitative real-time polymerase chain reaction (qRT-PCR) analysis to quantify the expression levels of key genes associated with podocytes (PODXL), proximal tubules (SLC34A1), and distal tubules (CDH1). Intriguingly, in accordance with the reduced structural variability observed in UniMat, the organoids cultured in UniMat exhibited less gene expression heterogeneity compared to those grown on the hydrogel layer (Fig. 2e and Supplementary Table 3). These results underscore the UniMat’s ability to enhance functional uniformity in organoids, demonstrating its potential to yield more precise and consistent experimental results. Moreover, intriguingly, UniMat showed a capacity to improve the uniformity of organoid formation efficiency, which is quantified as the number of organoids formed per mm². In determining this efficiency, the variability in the UniMat-based culture was significantly lower than in the hydrogel layer-based culture (Fig. 2f and Supplementary Table 4). Impressively, the formation efficiency was relatively consistent across three independent batches for the UniMat-based cultures (Fig. 2f and Supplementary Table 4). These findings highlight the reproducibility of organoid formation through the UniMat, obviously indicating that the UniMat provides a consistent and reliable platform for generating uniform organoids in a scalable manner.
+
+## UniMat enhances maturity of organoids
+
+After examining organoid uniformity, we investigated if the utilization of the permeable UniMat could promote the maturation of organoids, thereby addressing the limitations associated with conventional impermeable microwell platforms. For this experiment, we employed the AggreWell™800 Microwell Plate (hereinafter referred to as “AggreWell”), as a representative conventional microwell platform, with a width of 800 µm, similar to the dimensions of the UniMat used in this study. Confocal microscopic analysis revealed the formation of kidney organoids with nephron-like structures, including glomerular podocytes (PODXL+), proximal tubules (LTL+), and distal tubules (CDH1+) (Fig. 3a and Supplementary Fig. 5a). Notably, the differentiated regions of glomerular podocytes, proximal tubules, and distal tubules were more prevalent in the UniMat (Fig. 3b), indicating a higher presence of nephron cells in kidney organoids cultured in the UniMat compared to the AggreWell. Furthermore, the podocytes in the UniMat displayed mature characteristics with the basally localized expression of ZO-1 at the edge of cell-cell joints (Fig. 3c), which is a marker associated with tight junction observed within podocyte slit diaphragms in the native kidney tissue and mature kidney organoids. qRT-PCR analysis also highlighted the upregulated expressions of NPHS1 and PODXL (Supplementary Fig. 5b), both essential for the formation of the glomerular filtration barrier. This suggests that our UniMat can generate mature kidney organoids with mature glomerular cells, a contrast to those produced by the AggreWell-based culture. Furthermore, the kidney organoids cultured in UniMat exhibited morphogenesis of tubular structures, as evidenced by polarized proximal tubules with apical enrichment of the brush border marker LTL (Fig. 3d). Consistent with this polarization, the mRNA expression levels of AQP1, SLC34A1, and ABCB1 (proximal tubule) as well as UMOD (Loop of Henle) and CDH1 (distal tubule) were also upregulated in the UniMat-cultured kidney organoids (Supplementary Fig. 5c). This indicates enhanced functional potential compared to the kidney organoids cultured in the AggreWell. To further examine the physiological relevance of the tubular functions of these kidney organoids, we performed an in vitro dextran uptake assay. After a 24-h exposure to 10 kDa dextran, selective uptake of dextran within LTL+ proximal tubules confirmed the absorptive functionality of the UniMat-cultured kidney organoids (Fig. 3e). Specifically, these kidney organoids demonstrated the ability to uptake and retain dextran within LTL+ proximal tubule epithelial cells (Fig. 3e), indicating their capacity for reabsorption.
+
+Having identified that kidney organoids formed in the permeable UniMat exhibited higher maturity than those in the impermeable microwells, we next addressed the question of how the unconstrained supply of soluble factors through the 3D geometrically-engineered, permeable membrane in the UniMat influences the maturity of organoids. This question also arose in light of the limitation of conventional impermeable microwell platforms. To explore the impact of UniMat’s permeability on organoid maturity, we fabricated two different UniMat variants with distinct permeability characteristics. We achieved these characteristics by strategically blocking the pores of the NF membrane through precise adjustments of the matched-mold thermoforming process conditions (Fig. 3f). One variant, named UniMat-LP, was designed to have a lower permeability than the original UniMat, resembling the permeability of a conventional PET membrane (Fig. 3g). The second variant, UniMat-ZP, exhibited zero permeability (Fig. 3g). These modifications allowed for controlled studies on the influence of different levels of UniMat permeability on organoid maturation. Our spatiotemporal numerical simulations predicted active glucose diffusion around a single kidney organoid situated in a microwell of the original UniMat, supplied from the culture medium through its highly permeable nanofibrous wall within 72 h (Fig. 3h). However, with the reduced permeability, the glucose diffusion level around the kidney organoid decreased, as observed in both UniMat-LP and UniMat-ZP (Fig. 3h). This glucose diffusion simulation highlighted the effectiveness of UniMat’s permeability in facilitating nutrient supply to organoids. Notably, in concordance with the simulation results, mRNA expression levels of NPHS1, PODXL (podocyte), ABCB1, AQP1, SLC34A1 (proximal tubule), UMOD (loop of Henle), and CDH1 (distal tubule) significantly increased in response to the enhanced permeability (Fig. 3i). Therefore, our findings obviously indicated that the unconstrained supply of soluble factors within the UniMat, facilitated by high permeability through the 3D geometrically-engineered, permeable membrane, enables the generation of mature kidney organoids.
+
+Engineering vascularization of organoids stands as one of the primary objectives in ongoing research efforts to advance organoid technology. In a previous study, researchers transplanted kidney organoids generated in the impermeable microwell platform into mice to promote vascularization. Our study aimed to determine if our UniMat could promote in vitro vascularization in kidney organoids without relying on an animal host. We observed a significant increase in vascularization in the kidney organoids when cultured in the UniMat, as evidenced by the increased presence of renal vasculature (PECAM1+) (Fig. 3j). To quantify this vascularization, we evaluated confocal z-stack images of kidney organoids using the AngioTool (Supplementary Fig. 6a). The vessel percent area of PECAM1+ vasculature in the UniMat-cultured organoids exhibited an increase of over two-fold compared to the AggreWell-cultured organoids (Supplementary Fig. 6b). Additionally, we found that the PECAM1+ vasculature of UniMat-cultured organoids had more than a two-fold increase in both junctional density (i.e., branch points per unit area) and average vessel length (i.e., interjunctional distance) compared to the AggreWell-cultured organoids (Supplementary Fig. 6b). qRT-PCR analysis further confirmed the upregulated expression of PECAM1 in kidney organoids cultured in the UniMat (Fig. 3k). To ascertain whether the vascularization induced by a permeable environment extended to the glomerular compartments of organoids, we employed confocal imaging to quantify PODXL+ podocyte clusters invaded by PECAM1+ vascular structures in the kidney organoids cultured in both AggreWell and UniMat (Supplementary Fig. 6c and Supplementary Videos 1 and 2). While PECAM1+ vascular invasion into the glomerular structures was rarely observed in the AggreWell-cultured organoids (Supplementary Fig. 6d and Supplementary Video 1), there was a significant increase in PECAM1+ vascular invasion in the UniMat culture (Supplementary Fig. 6d and Supplementary Video 2). Intriguingly, our analysis using confocal 3D rendering revealed that the vascular structures in the kidney organoids cultured in the AggreWell displayed the characteristics of endothelial cell precursors, expressing both PECAM1+ and PODXL+ (Fig. 3l), compared to the kidney organoid in UniMat (Fig. 3l), suggesting an immature vascularization of organoids. These findings suggest that the vascular structures within the kidney organoids cultured in the UniMat are more advanced in their development compared to those cultured in the AggreWell.
+
+## Single-cell transcriptomic profiling of kidney organoids in UniMat
+
+To investigate the comprehensive capability of UniMat, we performed single-cell RNA sequencing (scRNA-seq) using the 10X Genomics platform and analyzed the data using the Seurat R package. We compared the kidney organoids cultured in UniMat to those cultured in the AggreWell to assess the functionality of UniMat’s permeability in the maturation of kidney organoids. After implementing quality control measures, we isolated 17,412 cells from day-25 kidney organoids in AggreWell and 13,098 cells from those in UniMat. These 30,510 cells were subsequently integrated and visualized using the Uniform Manifold Approximation and Projection (UMAP). The cells were categorized into 16 distinct clusters, each of which was annotated by comparing differentially expressed genes to known markers specific for cell types (Fig. 4a). A dot plot across different clusters exhibited unique transcript expression patterns for each cluster (Fig. 4b). Notably, 11 of these clusters were identified as differentiated kidney cells: podocytes (POD; PODXL, MAFB, and NPHS2), early proximal tubules (EPT; SPP1 and FXYD2 with low LRP2 and SLC3A1 expression), proximal tubules (PT; UGT3A1, LRP2, and SLC3A1), loop of Henle/distal tubules (LOH/DT; MAL, WFDC2, and EPCAM), endothelial cells (EC; PECAM1, CDH5, and CAV1), juxtaglomerular cells (JG; REN and PDGFRB), and five mesenchymal clusters (Mesen; COL1A1 and COL3A1). The proportion of differentiated kidney cells in both platforms was similar, accounting for 78.70% of the cells in UniMat and 74% in AggreWell among whole cells (Fig. 4c). The remaining cells included nephron progenitor cells (NPC), defined by expression of SIX2 and EYA1, along with proliferative premature tubular cells (Tub.pre; MKI67, SPP1, and FXYD2), proliferative mesenchymal cells (Prolif.Mesen; CENPF, COL1A1, and COL3A1), and muscle cells (Muscle; MYPLF and MYOG) (Fig. 4c).
+
+We further examined the cell-type proportions in each kidney organoid from both AggreWell and UniMat (Supplementary Table 5 and Supplementary Fig. 7). These findings were then juxtaposed with a recently published scRNA-seq dataset for human and mouse nephrons with a focus on core nephron segments such as podocytes, broad proximal tubules (including early proximal tubules), and loop of Henle/distal tubules (Fig. 4d). In the AggreWell samples, podocytes constituted 35.74%, broad proximal tubules 48.58%, and loop of Henle/distal tubules 15.68% (Fig. 4d). This distribution aligns with prior findings using the Morizane protocol, which showed high percentages of podocytes and proximal tubules. Remarkably, kidney organoids from the UniMat exhibited a 4.5-fold decrease in podocytes (7.97%), but higher proportions of broad proximal tubules (60.93%) and loop of Henle/distal tubules (31.10%). These proportions more closely resemble the cellular composition of natural nephrons in both mice and humans (Fig. 4d). Significantly, within the broad proximal tubule category, the proportion of early proximal tubules was 18.7% in UniMat, as compared to 62% in AggreWell, highlighting a higher presence of mature proximal tubules in the former. Beyond nephron cells, there was a noticeable difference in the NPC cluster: AggreWell exhibited a fraction 1.7 times higher than that of UniMat (Fig. 4e). This further demonstrated the enhanced maturity of kidney organoids cultivated on the UniMat platform (Fig. 4e and Supplementary Table 5). Given that tissue structure and function hinge on cellular compositions, these results emphasize the superior in vivo relevance of kidney organoids generated using UniMat.
+
+We next examined gene expression profiles of podocytes, proximal tubules, loop of Henle/distal tubules as well as endothelial cells clusters (Fig. 4f and Supplementary Fig. 8). The analysis revealed elevated expression of genes related to differentiation in the podocytes, loop of Henle/distal tubules, and endothelial cells clusters. This suggests a greater degree of maturity in kidney organoids cultivated using the UniMat. Although the gene expression profiles in the proximal tubule cluster appeared similar between UniMat and AggreWell, a low proportion of early proximal tubules in UniMat (Supplementary Fig. 9) indicates a more mature state of these tubules.
+
+To explore the potential pathways and functions that differentiate kidney organoids in UniMat compared to those in AggreWell, we carried out Biological Process Gene Ontology (GO) enrichment analysis on genes that were significantly upregulated in podocytes, proximal tubules, loop of Henle/distal tubules, and endothelial cells clusters within the UniMat (Fig. 4g). In the podocyte cluster, UniMat showed elevated expression of genes associated with glomerulus development (GO:0032835) and glomerular basement membrane development (GO:0032836) (Supplementary Table 6). For both the proximal tubule and the loop of Henle/distal tubule clusters, the most enriched GO terms were associated with the negative regulation of mesenchymal cell apoptosis in metanephros development (GO:1900212), metanephric nephron tubule formation (GO:0072289), and ureter morphogenesis (GO:0072197) (Supplementary Table 7). These observations corroborate the advanced tubular structures observed in the UniMat-cultured kidney organoids. In the loop of Henle/distal tubule cluster, enriched GO terms revealed elevated expression of genes like CLDN3, CLDN19, and CRB3, which are linked to the positive regulation of cell junction assembly (GO:1901890) (Supplementary Table 7). This suggests enhanced epithelial junctions in kidney organoids cultured in UniMat. Additionally, the GO term related to metanephric distal convoluted tubule development (GO:0072221) was enriched in UniMat (Supplementary Table 7), indicating a higher propensity for kidney organoids to differentiate into distal tubules in UniMat as compared to AggreWell, as shown in Fig. 4d. In the endothelial cell cluster, upregulated genes such as SOX17, SOX18, NRP1, and CLIC4, played a role in endothelial cell differentiation (GO:0060956) (Supplementary Table 8). These findings are in line with the advanced development of kidney organoids in UniMat.
+
+## Potential of UniMat in standardizing organoid-based assays
+
+While earlier researches have utilized kidney organoids to simulate polycystic kidney disease (PKD) and mimic its pathological features, the inconsistent morphology of these organoids often resulted in considerable variability in both size and cystic regions. Recognizing that UniMat facilitates the generation of kidney organoids with shape and functional consistency and higher maturity in tubular structures across different nephron segments, we employed these organoids for PKD modeling to evaluate UniMat’s potential as a drug testing platform (Fig. 5a). We exposed the UniMat-cultured kidney organoids to a 30 µM concentration of forskolin for 48 h, replicating the cAMP-induced cystic growth commonly seen in PKD. This enabled us to evaluate the platform’s suitability for generating PKD organoids. In response to the forskolin treatment, we observed both an enlargement and the formation of cysts within organoids. This confirmed the UniMat’s ability to reliably produce PKD organoids on a scalable level (Fig. 5b).
+
+To evaluate if the uniformly mature kidney organoids generated in UniMat enhance the reliability of PKD modeling, we compared the cyst-forming features of organoids cultured in the hydrogel layer, AggreWell, and UniMat. To facilitate closer observation, the kidney organoids cultured in UniMat and AggreWell were transferred to individual ultra-low attachment plates prior to 48-h exposure to forskolin at concentrations of 10 and 30 µM (Fig. 5a). Notably, cysts became evident within 24 h of forskolin exposure, predominantly in the LTL+ and CDH1+ tubular regions (Supplementary Fig. 10). We quantified the dimensions of kidney organoids collectively, both before and after the induction of cyst formation, without monitoring each organoid individually. As expected, a considerable variability in the size and morphology of kidney organoids cultured on the hydrogel layer resulted in inconsistent cyst formation, in terms of both location and dimensions (Fig. 5c). A similar pattern of variability and incomplete cyst formation was observed in organoids cultured in the AggreWell, implying the existence of undifferentiated, non-renal structures unresponsive to forskolin treatment (Fig. 5c). Additionally, cyst formation in kidney organoids from both hydrogel layer and AggreWell was erratic, regardless of forskolin dosage and time of exposure (Fig. 5e). In contrast, organoids in the UniMat exhibited uniform cysts throughout their entire structure, resulting in consistent increases in overall sizes. This can be attributed to the high and consistent differentiation of nephrons (Fig. 5c). When comparing the percentage area of individual forskolin-treated PKD organoids to their average size before treatment, we found that significant dose-dependent responses were observed uniquely in the UniMat (Fig. 5e).
+
+To explore the impact of uniform disease modeling on drug testing results, we exposed PKD organoids, induced by forskolin (30 µM), to CFTR inhibitor-172 at concentrations of 50 or 100 µM (Fig. 5a). This PKD drug, which targets the cytoplasmic side of CFTR, had previously demonstrated its potential to inhibit cyst growth in PKD mice and in vitro models. The PKD organoid induced from kidney organoids cultured in UniMat displayed consistent reductions in size (Fig. 5d), with the percentage area decreasing significantly in a dose-dependent manner (Fig. 5f). In contrast, the PKD organoids from both hydrogel layer and AggreWell experienced inconsistent changes in cyst dimensions. Our results indicate that, within the UniMat platform, we can validate not only the formation but also the reduction of cysts by simply comparing average size changes across the organoid population, without the need to track individual organoids. In summary, our study clearly demonstrated that UniMat offers significant advantages in disease modeling and drug testing by facilitating the uniform and mature differentiation of organoids.
+
+# Discussion
+
+Here, we present a simple, ready-to-use 3D geometrically-engineered, permeable membrane-based organoid culture platform, called UniMat. This platform uniquely serves a dual purpose: it provides specific geometrical constraints on organoid growth while also facilitating an unconstrained supply of soluble factors. While previous approaches in organoid engineering have largely focused on either regulating biophysical cues<sup>10–12, 14, 15</sup> to improve organoid uniformity or manipulating the microenvironment to foster mature organoids<sup>41–44</sup>, UniMat offers a significant advancement by concurrently addressing both uniformity and maturity.
+
+Our in-depth examinations, which included confocal microscopic analysis and qRT-PCR analysis, revealed that UniMat enhances uniformity not just in terms of physical appearance but also at the fundamental levels of structure and function of organoids when compared to those cultured on the hydrogel layer. The enhanced uniformity of kidney organoids is hypothesized to arise from the consistent initial aggregates formed within the V-shaped microwell array of the UniMat. Provided that initial cell seeding remains uniform across individual V-shaped microwells, almost consistent organogenesis and organoid formation can be anticipated. Furthermore, intriguingly, kidney organoids grown in the Aggrewell, which is a representative of impermeable microwell platforms, and those in the UniMat showed significant differences; the latter displayed greater structural and functional consistency (<b>Supplementary Tables 9 and 10</b>), even though both platforms imposed similar geometrical constraints on organoid growth. This emphasizes the crucial role of an unrestricted and spatially uniform supply of soluble factors in enhancing organoid uniformity<sup>42, 45</sup>. These findings show that UniMat stands out in its remarkable ability to boost organoid uniformity, a feat achieved through both its geometric design and the efficient diffusion of soluble factors.
+
+This study also highlights that kidney organoids matured significantly better in the UniMat, offering a permeable environment compared to conventional microwell platforms. The UniMat facilitated a remarkable increase in the expression of nephron transcripts, as observed in both scRNA-seq and qRT-PCR analysis. Beyond the cellular level, the increased maturity was also evident in structural aspects, such as the presence of basally located ZO1<sup>+</sup> tight junctions, a higher polarity of LTL<sup>+</sup> proximal tubules, and enhanced vascularization, featuring a greater vascular invasion into podocytes, akin to the configuration observed in <em>in vivo</em> glomeruli. Importantly, our scRNA-seq confirmed that kidney organoids differentiated in the UniMat exhibit a more <em>in vivo</em>-like cell-type balance of nephron elements, such as podocytes, proximal, and distal tubules, addressing the limitation of Morizane's protocol generating overpopulated podocytes and a smaller portion of tubular cells<sup>30, 31</sup>. Consistent with morphological findings, GO term analysis also pointed to an enrichment of genes associated with tubular morphogenesis in the UniMat-cultured organoids. Given that differences in cell-type proportions often carry biological significance, organoids developed in the UniMat could offer a more reliable nephron physiology model. Additionally, the reduction in both NPC and immature proximal tubule proportions suggests that our proposed approach accelerated kidney organoid differentiation. While recent researches have advanced kidney organoid maturity using methods like animal transplants<sup>26, 46</sup>, fluidic systems<sup>43, 47</sup>, 3D printing<sup>48</sup>, and using decellularized ECM<sup>49</sup>, our UniMat achieves this advancement without the need for complicated devices or drastic protocol changes. For further refinement of UniMat, a deeper exploration into its impact on the temporal molecular differentiation process of organoids is crucial.
+
+Organoid-based tests have inherently faced challenges due to variability in organoids, particularly when used in high-throughput experiments. Specifically, for kidney organoid tests, previous research has investigated cytogenesis and PKD pathophysiology using non-uniform organoids. This led to inconsistent outcomes in disease modeling or pharmaceutical testing<sup>36, 37, 50, 51</sup>. As a result, the individual monitoring of cyst induction or inhibition for each organoid has become a standard protocol, adding complexity to experimental processes. By leveraging UniMat's capability to produce uniform and mature organoids, we demonstrated the potential to establish a highly standardized PKD model with significantly reduced pathological heterogeneity across both organoids and batches. Our method also consistently showed responses to a PKD drug, CFTR inhibitor-172, based on dosage, a feat not achieved using conventional methods. These findings indicate that UniMat could be as a robust and reliable tool for nonclinical drug testing with organoids. Moreover, we expect that the UniMat's capacity to enhance vascularization will allow for more precise modeling of various diseases associated with vasculopathies, such as diabetic kidney disease and nephrosclerosis<sup>52</sup>.
+
+UniMat has demonstrated notable potential as a pioneering platform for the practical applications of organoids, yet to emulate native organs more closely, strategies beyond the unconstrained supply of soluble factors are imperative. This includes simulating the complexities inherent in native organs through tissue-specific microenvironments, mechanical forces, and cellular interactions. The versatility of UniMat could be instrumental in these sophisticated applications. For example, it can be readily integrated with microfluidic systems in a free-standing configuration, allowing for precise mechanical force stimulations for enhanced vascularization and the replication of complex cell interactions found in native tissues. Although this study demonstrated the potential of UniMat with kidney organoids, leveraging its innovative features can advance the maturity of various organoid models by incorporating tailored microenvironmental factors.
+
+In conclusion, UniMat represents a major advancement in organoid culture technology, efficiently generating organoids that are not only uniform but also mature at scale. Its considerable potential impact on organoid technology is noteworthy, providing a practical and effective platform with broad applications across diverse biomedical research fields. Given its ability to ensure both the uniformity and maturity of organoids, UniMat could serve as a valuable tool for deepening our understanding of organ development, disease modeling, and drug screening.
+
+# Methods
+
+## Fabrication of NF membrane
+For the fabrication of the electrospun NF membrane, a solution was prepared by mixing polycaprolactone (PCL; average *M*<sub>n</sub> = 80,000 g/mol; Sigma-Aldrich) and Pluronic F108 (PF108; Sigma-Aldrich). This mixture was dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (Sigma-Aldrich) to achieve a PCL/PF108 ratio of 1:1 (w/w) and a final concentration of 8% (w/w). This solution was then loaded into a 5-ml gastight syringe (Hamilton) equipped with a 23-gauge metal needle (NanoNC) and ejected at a constant flow rate of 0.7 ml/h via a syringe pump of an electrospinning machine (NanoNC, ESR200R2). An electric voltage of 15 kV was applied between the needle and a cylindrical aluminum collector, which was positioned 14 cm below the needle. The electrospinning was conducted for 2 min, at a relative humidity of 30–40% and a temperature range of 20–25°C, which resulted in a flat NF membrane with a thickness of 50 µm.
+
+## Fabrication of UniMat and integration with well insert wall
+A UniMat with a V-shaped microwell array was fabricated by a matched-mold thermoforming process<sup>22</sup>. The size of each V-shaped microwell ranged from 400 to 800 µm. A female mold, featuring the V-shaped microwell array, was manufactured via micromachining of a PMMA substrate using a tapered end milling cutter. A male mold was subsequently prepared through PDMS replica molding against this PMMA female mold. The male mold was fixed beneath the top plate of a customized thermoforming machine, which was connected to a motorized stage. For the thermoforming process, the electrospun NF membrane, after preheating at 60°C, was placed on the female mold. It was then microstructured via a linear movement of the male mold exerting a force of 4.8 N. Following this process, the fabricated 3D geometrically-engineered NF membrane was finally integrated into the bottom opening of a custom-made 24-well insert wall, which was produced using an injection molding machine (Sumitomo, SE50D), to finally fabricate a UniMat-integrated 24-well insert.
+
+## Scanning electron microscopy (SEM) analysis
+The UniMat was sputter-coated with Pt for 180 s at an electric current of 10 mA. The geometry and structure of the UniMat were examined by SEM imaging using a field-emission scanning electron microscope (Hitachi, SU660) at an accelerating voltage of 15 kV.
+
+## Permeability measurement
+Permeability was evaluated by measuring the diffusion of Lucifer yellow (Invitrogen™), and FITC-dextran tracers (3 kDa, 10 kDa, and 40 kDa; Sigma Aldrich) across the membrane. A commercially available 24-well Transwell insert with a 0.4 µm pore PET membrane (surface area of membrane = 0.33 cm<sup>2</sup>; Corning) and the UniMat400-integrated 24-well insert (surface area of UniMat400 = 0.38 cm<sup>2</sup>, while the projection area was 0.33 cm<sup>2</sup>) were placed in a 24-well plate. Subsequently, 0.6 ml of phosphate-buffered saline (PBS) was added to the basolateral side, and 0.1 ml of 200 µg/ml Lucifer yellow or FITC-dextran tracer was introduced to the apical side. After a 1-h incubation at 37°C, the solution from the basolateral side was collected, and the concentration of the Lucifer yellow or dextran tracer was determined using a microplate reader (Wazobia Enterprise, VICTOR3). The permeability coefficient (*P*<sub><em>app</em></sub>) was calculated using the following equation:
+
+$$
+P_{app}= \frac{\text{d}Q}{\text{d}t} \times \frac{1}{A{C}_{0}}
+$$
+
+where d*Q*/d*t* is the diffusive transport rate of Lucifer yellow or FITC-dextran, *A* is the surface area of the membrane, and *C*<sub>0</sub> is the initial concentration of the solution on the apical side<sup>53</sup>.
+
+## hiPSC culture
+The WTC-11 hiPSC cell line (Coriell Institute for Medical Research, GM25256) was maintained under feeder-free conditions on 1% GelTrex (Thermo Fisher Scientific, A1413302)-coated 6-well cell culture plates using mTeSR<sup>TM</sup> 1 (Stem Cell Technologies, 85850) in a 37°C incubator with 5% CO<sub>2</sub>. The hiPSCs were passaged every 4 days using ReLeSR™ (Stem Cell Technologies, 05872) according to the manufacturer’s protocol. For all experiments, hiPSCs from passages 45 to 70 were used. The IMR90-4 hiPSC cell line (WiCell) was also maintained in 1% GelTrex-coated 6-well cell culture plates using mTeSR<sup>TM</sup> 1. These cells were passaged every 3–4 days using Versene Solution (Thermo Fisher Scientific, 15040066) as per the manufacturer’s protocol.
+
+## Differentiation of hiPSCs into nephron progenitor cells (NPCs)
+hiPSCs were differentiated into NPCs using our optimized kidney organoid differentiation protocol, which was developed based on a previously reported protocol<sup>23</sup>. Briefly, hiPSCs, which were singularized using Accutase (Stem Cell Technologies, 07920), were plated at a density of ~235,000 cells/well in a 6-well plate in mTeSR<sup>TM</sup> 1 supplemented with a 10 µM ROCK inhibitor Y27632 (Tocris Bioscience, 1254). Once the confluency of the hiPSCs reached about 50%, the cells were treated with CHIR99021 (8–10 µM; TOCRIS, 4423) and Noggin (10 ng/ml; Peprotech, 120-10C) in a basal differentiation medium, which consisted of Advanced RPMI 1640 (Thermo Fisher Scientific, 12633-020) and GlutaMAX (Thermo Fisher Scientific, 35050-061), for 4 days. On day 4, the medium was replaced with the basal differentiation medium containing Activin A (10 ng/ml; R&D Systems, 338-AC), and on day 7, it was replaced with the medium containing FGF9 (10 ng/ml; Peprotech, 100−23). After a 48-h treatment with FGF9, the cells were differentiated into NPCs.
+
+## Generation of kidney organoids
+For generating kidney organoids in UniMat, the sterilized UniMat was coated with 0.025% agarose (Thermo Fisher Scientific, A4018) for 12 h to improve the low attachment condition at the bottom of UniMat. On day 9, the dissociated NPCs were seeded into UniMat at a density of ~960,000 cells/UniMat (~2×10<sup>6</sup> cells/cm<sup>2</sup>, considering the projection area of UniMat) with the basal differentiation medium containing 3 µM CHIR and 10 ng/ml FGF9. In each UniMat400, UniMat600, and UniMat800, the cells were seeded at an average density of 5,000, 11,000, and 20,000 cells per individual microwell, respectively. After 2 days, the medium was switched to the basal differentiation medium but supplemented with 30 ng/ml FGF9 and 1.5% FBS. On day 14, the medium was replaced with the medium containing 1.5% FBS, and from day 14 to 25, cultures were maintained in the basal differentiation medium with 1.5% FBS. The same procedure was also applied to AggreWell™ Microwell Plates (Stem Cell Technologies, AggreWell™800), which were pre-treated with an AggreWell™ Rinsing Solution (Stem Cell Technologies, 07010) according to the manufacturer's protocol, where a cell seeding density was ~4,500,000 cells/well (~2×10<sup>6</sup> cells/cm<sup>2</sup>, considering the projection area of AggreWell).
+
+## Immunofluorescence imaging
+After washing of organoids with DPBS (Welgene, LB001-01), the organoids were fixed in 4% paraformaldehyde for 1 h at room temperature. After washing the samples with PBS three times, they were blocked overnight at 4°C with 1 vol% donkey serum (Sigma-Aldrich) in PBS with 0.125 vol% Triton X-100. Then, primary antibodies were incubated with the samples for 2 days at 4°C at the dilutions listed in <b>Supplementary Table 11</b> in a solution of 0.5 wt% bovine serum albumin (BSA) and 0.1 vol% Triton X-100. After washing for 1 day with a solution of 0.5 wt% BSA and 0.1 vol% Triton X-100 in PBS, secondary antibodies were incubated for 3 h at a 1:500 dilution in a solution of 0.5 wt% BSA and 0.1 wt% Triton X-100 in PBS. After washing for 1 day with a solution of 0.5 wt% BSA and 0.1 vol% Triton X-100 in PBS, samples were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, D9542) at a 1:1000 dilution for 1 h and then washed with PBS. Before imaging, the stained organoids were cleared with a fructose (Sigma-Aldrich)–glycerol (Sigma-Aldrich) clearing agent as described previously<sup>54</sup>. Images were taken using a confocal microscope (Olympus, FV3000). A full antibody list can be found in <b>Supplementary Table 11</b>.
+
+## Image analysis and quantification
+To analyze the area of kidney organoids, bright-field images of organoids were acquired on a microscope (Carl Zeiss). Image reconstruction of the z-stacks of bright-field images was generated using Mosaic 2.0 (Tucsen). Kidney organoids were randomly selected for all experiments. The area of kidney organoid was determined by manually outlining z-projected images using the freehand selection tool in ImageJ software (NIH). To quantify the volume of PODXL<sup>+</sup>, LTL<sup>+</sup>, and CDH1<sup>+</sup> cells within the kidney organoids, confocal z-stacks of PODXL, LTL, and CDH1 were acquired, ensuring that these z-stacks were taken at the maximal limit of the confocal depth for each sample. These z-stack images were then imported into the Imaris imaging software, where confocal 3D renderings were generated to delineate glomerular and tubular volumes. The Imaris software was subsequently employed to quantitatively assess the volume occupied by PODXL<sup>+</sup>, LTL<sup>+</sup>, and CDH1<sup>+</sup> cells.
+
+## Quantitative real-time polymerase chain reaction (qRT-PCR)
+Kidney organoids were extracted from the hydrogel layer, the AggreWell, or the UniMat using a pipette. RNA was isolated from the organoid samples using the ReliaPrep™ RNA Miniprep Systems (Promega), as per the manufacturer’s protocol. cDNA was synthesized with a cDNA synthesis kit (Thermo Fisher Scientific, K1641) according to the manufacturer’s protocol. qRT-PCR was carried out with Power SYBR® Green Master Mix (Applied Biosystems™) on a StepOnePlus™ Real-Time PCR System (Applied Biosystems™). Relative mRNA expression levels were normalized to glyceraldehyde 3-phosphate dehydrogenase (*GAPDH*) expression and analyzed by the 2<sup>–ΔΔCt</sup> method. Sequences of the primers used are listed in <b>Supplementary Table 12</b>.
+
+## Numerical simulation of glucose concentration
+The spatiotemporal distribution of glucose concentration around the kidney organoid situated in a single microwell was numerically simulated using COMSOL Multiphysics software (COMSOL, Ver. 5.0). This model simulated the diffusive transport of diluted glucose species in response to glucose consumption by the organoid. The geometries and dimensions were designed to match to those of the corresponding experimental setup. The parameters used for this simulation are listed in <b>Supplementary Table 13</b>.
+
+## Dextran uptake assay
+The kidney organoids cultured in both UniMat and AggreWell were incubated with 100 µg/ml of 10,000 Mw cascade blue-labelled dextran (Thermo Fisher Scientific, D1976) in the basal differentiation medium for 24 h. After this period, the medium was replaced with the fresh basal differentiation medium without dextran and incubated for an additional 24 h. Subsequently, the organoids were fixed with PFA 4% and stained with LTL.
+
+## Single-cell RNA sequencing and data processing
+Kidney organoids were collected from both the AggreWell and UniMat using a pipette, and then dissociated into single cells using Accutase at 37°C for 20 min. The single cells were resuspended in the basal differentiation medium containing 1.5% FBS. After singularization, samples were prepared following the manufacturer’s protocol (10X Genomics). The single-cell libraries were sequenced using the Chromium Next GEM Single Cell 3’ RNA library and gel bead v3.1 kit in HiSeqXten (Illumina). Alignment to the GRCh38 human reference genome, filtering, barcode counting, and UMI counting of FASTQ files were performed using the Cell Ranger (v6.1.2) pipeline (10X Genomics). Pre-processing of raw scRNA-seq data was carried out by Macrogen, Inc.
+
+### Analysis of scRNA-seq data
+The output from Cell Ranger was imported into R (v4.3.1) and Seurat (v4.3.0.1) for downstream analyses. Cells expressing fewer than 200 genes, more than 6,000 genes, or with over 10% of reads assigned to mitochondrial genes were excluded. After this quality control step, we retained a total of 30,510 cells, comprising 17,412 cells from AggreWell and 13,098 cells from UniMat. We normalized each dataset and identified highly variable features using the <em>NormalizeData</em> and <em>FindVariableFeatures</em> functions, respectively. The two Seurat objects were combined using canonical correlation analysis (CCA) through the <em>IntegrateData</em> function, following the workflow proposed by the Satija group<sup>55</sup>. Principal component analysis was carried out using the <em>RunPCA</em> function, and the results were employed for clustering with the <em>FindNeighbors</em> and <em>FindClusters</em> functions. The UMAP projection was generated using the <em>RunUMAP</em> function. In total, cells were classified into 16 distinct clusters. Each of these clusters was annotated based on differential expression testing performed using the <em>FindAllMarkers</em> function with the Wilcoxon Rank Sum test. The enriched biological processes of the upregulated genes in each cluster were analyzed using the DAVID database<sup>56</sup>.
+
+## PKD modeling and drug testing
+<em>In situ</em> PKD modeling of kidney organoids within UniMat was carried out by introducing the basal differentiation medium containing 30 µM forskolin (LC Laboratories, F-9929) directly to the UniMat for 48 h. Cytogenesis initiation in the kidney organoid was achieved upon treatment of forskolin. For PKD organoid modeling, kidney organoids cultured in both the UniMat and AggreWell were individually transferred to ultra-low attachment 6-well plates (Corning, 3471). Subsequently, they were exposed to forskolin (10 and 30 µM) for 48 h. Inhibition of cytogenesis was achieved by treating CFTR inhibitor-172 (MedChemExpress, HY-16671). After a 48-h treatment with 30 µM forskolin, CFTR inhibitor-172 was introduced at concentrations of 50 and 100 µM in the medium with 30 µM forskolin for an additional 24 h. For the case of hydrogel layer, both cyst induction and inhibition were carried out by changing the medium containing forskolin and CFTR inhibitor-172, respectively, in an <em>in-situ</em> manner. The concentrations of forskolin<sup>36</sup> and CFTR inhibitor-172<sup>57</sup> were determined based on the previous studies. The percent area was calculated by comparing the area of individual organoids treated with either forskolin or CFTR inhibitor-172 to the average area of original or forskolin-treated organoids, respectively. Areas of organoids were quantified using the freehand selection tool in ImageJ software (NIH).
+
+### Statistical analysis and reproducibility
+The sample size for each experiment was determined based on a minimum of n = 3 independent samples for every experimental group. Statistical analyses were carried out using the Student’s t-test for comparing two data sets and a one-way analysis of variance (ANOVA) with Tukey's multiple comparisons test for evaluating differences among multiple data sets. These analyses were performed using the Origin PRO software (OriginLab). All data are presented as mean ± s.e.m. (standard error of the mean). Significance levels are denoted by one asterisk for *P*-values < 0.05, two asterisks for *P* < 0.001, and three asterisks for *P* < 0.0001.
+
+# References
+
+1. Lancaster, M.A. & Knoblich, J.A. Organogenesis in a dish: Modeling development and disease using organoid technologies. *Science* **345** (2014).
+2. Clevers, H. Modeling Development and Disease with Organoids. *Cell* **165**, 1586-1597 (2016).
+3. Dutta, D., Heo, I. & Clevers, H. Disease Modeling in Stem Cell-Derived 3D Organoid Systems. *Trends Mol Med* **23**, 393-410 (2017).
+4. Rossi, G., Manfrin, A. & Lutolf, M.P. Progress and potential in organoid research. *Nat Rev Genet* **19**, 671-687 (2018).
+5. Hofer, M. & Lutolf, M.P. Engineering organoids. *Nat Rev Mater* **6**, 402-420 (2021).
+6. Brassard, J.A. & Lutolf, M.P. Engineering stem cell self-organization to build better organoids. *Cell Stem Cell* **24**, 860-876 (2019).
+7. Kozlowski, M.T., Crook, C.J. & Ku, H.T. Towards organoid culture without Matrigel. *Commun Biol* **4** (2021).
+8. Morizane, R. et al. Nephron organoids derived from human pluripotent stem cells model kidney development and injury. *Nature biotechnology* **33**, 1193-1200 (2015).
+9. Dahl-Jensen, S. & Grapin-Botton, A. The physics of organoids: a biophysical approach to understanding organogenesis. *Development* **144**, 946-951 (2017).
+10. Jung, Y.H. et al. Drug screening by uniform patient derived colorectal cancer hydro-organoids. *Biomaterials* **276** (2021).
+11. Wiedenmann, S. et al. Single-cell-resolved differentiation of human induced pluripotent stem cells into pancreatic duct-like organoids on a microwell chip. *Nat Biomed Eng* **5**, 897-913 (2021).
+12. Loebel, C. et al. Microstructured Hydrogels to Guide Self-Assembly and Function of Lung Alveolospheres. *Adv Mater* (2022).
+13. Chen, C., Rengarajan, V., Kjar, A. & Huang, Y. A matrigel-free method to generate matured human cerebral organoids using 3D-Printed microwell arrays. *Bioact Mater* **6**, 1130-1139 (2021).
+14. Brandenberg, N. et al. High-throughput automated organoid culture via stem-cell aggregation in microcavity arrays. *Nat Biomed Eng* **4**, 863-+ (2020).
+15. Tran, T. et al. A scalable organoid model of human autosomal dominant polycystic kidney disease for disease mechanism and drug discovery. *Cell Stem Cell* **29**, 1083-+ (2022).
+16. Lee, G. et al. Enhanced oxygen permeability in membrane-bottomed concave microwells for the formation of pancreatic islet spheroids. *Acta Biomater* **65**, 185-196 (2018).
+17. He, J.Y. et al. Scalable Formation of Highly Viable and Functional Hepatocellular Carcinoma Spheroids in an Oxygen-Permeable Microwell Device for Anti-Tumor Drug Evaluation. *Adv Healthc Mater* **11** (2022).
+18. Ornoff, D.M., Wang, Y.L., Proctor, A., Shah, A.S. & Allbritton, N.L. Co-fabrication of chitosan and epoxy photoresist to form microwell arrays with permeable hydrogel bottoms. *Biomaterials* **74**, 77-88 (2016).
+19. Jun, Y. et al. In vivo-mimicking microfluidic perfusion culture of pancreatic islet spheroids. *Sci Adv* **5** (2019).
+20. Payne, M.C. et al. Microwell-based flow culture increases viability and restores drug response in prostate cancer spheroids. *Biotechnol J* **18** (2023).
+21. Ma, L.D. et al. Design and fabrication of a liver-on-a-chip platform for convenient, highly efficient, and safe in situ perfusion culture of 3D hepatic spheroids. *Lab Chip* **18** (2018).
+22. Kim, D. et al. A deep and permeable nanofibrous oval-shaped microwell array for the stable formation of viable and functional spheroids. *Biofabrication* **13** (2021).
+23. Morizane, R. & Bonventre, J.V. Generation of nephron progenitor cells and kidney organoids from human pluripotent stem cells. *Nat Protoc* **12** (2017).
+24. Schnabel, E., Anderson, J.M. & Farquhar, M.G. The Tight Junction Protein Zo-1 Is Concentrated Along Slit Diaphragms of the Glomerular Epithelium. *J Cell Biol* **111**, 1255-1263 (1990).
+25. Freedman, B.S. et al. Modelling kidney disease with CRISPR-mutant kidney organoids derived from human pluripotent epiblast spheroids. *Nat Commun* **6** (2015).
+26. van den Berg, C.W. et al. Renal Subcapsular Transplantation of PSC-Derived Kidney Organoids Induces Neo-vasculogenesis and Significant Glomerular and Tubular Maturation In Vivo. *Stem Cell Rep* **10**, 751-765 (2018).
+27. Grebenyuk, S. & Ranga, A. Engineering Organoid Vascularization. *Front Bioeng Biotech* **7** (2019).
+28. Zudaire, E., Gambardella, L., Kurcz, C. & Vermeren, S. A Computational Tool for Quantitative Analysis of Vascular Networks. *Plos One* **6** (2011).
+29. Sequeira-Lopez, M.L.S. & Torban, E. New insights into precursors of renal endothelium. *Kidney Int* **90**, 244-246 (2016).
+30. Wilson, S.B. et al. DevKidCC allows for robust classification and direct comparisons of kidney organoid datasets. *Genome Med* **14**, 19 (2022).
+31. Wu, H. et al. Comparative Analysis and Refinement of Human PSC-Derived Kidney Organoid Differentiation with Single-Cell Transcriptomics. *Cell Stem Cell* **23**, 869-881 e868 (2018).
+32. Chen, L. et al. Renal-Tubule Epithelial Cell Nomenclature for Single-Cell RNA-Sequencing Studies. *J Am Soc Nephrol* **30**, 1358-1364 (2019).
+33. Clark, J.Z. et al. Representation and relative abundance of cell-type selective markers in whole-kidney RNA-Seq data. *Kidney Int* **95**, 787-796 (2019).
+34. McEvoy, C.M. et al. Single-cell profiling of healthy human kidney reveals features of sex-based transcriptional programs and tissue-specific immunity. *Nat Commun* **13**, 7634 (2022).
+35. Lake, B.B. et al. A single-nucleus RNA-sequencing pipeline to decipher the molecular anatomy and pathophysiology of human kidneys. *Nat Commun* **10**, 2832 (2019).
+36. Czerniecki, S.M. et al. High-Throughput Screening Enhances Kidney Organoid Differentiation from Human Pluripotent Stem Cells and Enables Automated Multidimensional Phenotyping. *Cell Stem Cell* **22**, 929-940 e924 (2018).
+37. Tran, T. et al. A scalable organoid model of human autosomal dominant polycystic kidney disease for disease mechanism and drug discovery. *Cell Stem Cell* **29**, 1083-1101 e1087 (2022).
+38. Yamaguchi, T. et al. Calcium restriction allows cAMP activation of the B-Raf/ERK pathway, switching cells to a cAMP-dependent growth-stimulated phenotype. *J Biol Chem* **279**, 40419-40430 (2004).
+39. Cabrita, I. et al. Cyst growth in ADPKD is prevented by pharmacological and genetic inhibition of TMEM16A in vivo. *Nature communications* **11**, 4320 (2020).
+40. Yang, B., Sonawane, N.D., Zhao, D., Somlo, S. & Verkman, A. Small-molecule CFTR inhibitors slow cyst growth in polycystic kidney disease. *Journal of the American Society of Nephrology: JASN* **19**, 1300 (2008).
+41. Homan, K.A. et al. Flow-enhanced vascularization and maturation of kidney organoids in vitro. *Nat Methods* **16**, 255-+ (2019).
+42. Park, S.E. et al. Geometric engineering of organoid culture for enhanced organogenesis in a dish. *Nat Methods* **19**, 1449-+ (2022).
+43. Hiratsuka, K. et al. Organoid-on-a-chip model of human ARPKD reveals mechanosensing pathomechanisms for drug discovery. *Sci Adv* **8** (2022).
+44. Cho, A.N. et al. Microfluidic device with brain extracellular matrix promotes structural and functional maturation of human brain organoids. *Nat Commun* **12** (2021).
+45. Shin, W. et al. Spatiotemporal Gradient and Instability of Wnt Induce Heterogeneous Growth and Differentiation of Human Intestinal Organoids. *Iscience* **23** (2020).
+46. Garreta, E. et al. Fine tuning the extracellular environment accelerates the derivation of kidney organoids from human pluripotent stem cells. *Nat Mater* **18**, 397-405 (2019).
+47. Homan, K.A. et al. Flow-enhanced vascularization and maturation of kidney organoids in vitro. *Nat Methods* **16**, 255-262 (2019).
+48. Lawlor, K.T. et al. Cellular extrusion bioprinting improves kidney organoid reproducibility and conformation. *Nat Mater* **20**, 260-271 (2021).
+49. Kim, J.W. et al. Kidney Decellularized Extracellular Matrix Enhanced the Vascularization and Maturation of Human Kidney Organoids. *Adv Sci (Weinh)* **9**, e2103526 (2022).
+50. Cruz, N.M. et al. Organoid cystogenesis reveals a critical role of microenvironment in human polycystic kidney disease. *Nat Mater* **16**, 1112-1119 (2017).
+51. Cruz, N.M. et al. Modelling ciliopathy phenotypes in human tissues derived from pluripotent stem cells with genetically ablated cilia. *Nat Biomed Eng* **6**, 463-475 (2022).
+52. Zafrani, L. & Ince, C. Microcirculation in acute and chronic kidney diseases. *American Journal of Kidney Diseases* **66**, 1083-1094 (2015).
+53. Ma, B. et al. Permeability of rhynchophylline across human intestinal cell in vitro. *Int J Clin Exp Patho* **7**, 1957-1966 (2014).
+54. Dekkers, J.F. et al. High-resolution 3D imaging of fixed and cleared organoids. *Nat Protoc* **14**, 1756-1771 (2019).
+55. Stuart, T. et al. Comprehensive Integration of Single-Cell Data. *Cell* **177**, 1888-+ (2019).
+56. Huang, D.W. et al. DAVID Bioinformatics Resources: expanded annotation database and novel algorithms to better extract biology from large gene lists. *Nucleic Acids Res* **35**, W169-175 (2007).
+57. Low, J.H. et al. Generation of Human PSC-Derived Kidney Organoids with Patterned Nephron Segments and a De Novo Vascular Network. *Cell Stem Cell* **25**, 373-387 e379 (2019).
+
+# Supplementary Files
+
+- [vidoe1.mp4](https://assets-eu.researchsquare.com/files/rs-3425714/v1/054a5146241ae8eba7c280ba.mp4)  
+  Supplementary Video1
+
+- [video2.mp4](https://assets-eu.researchsquare.com/files/rs-3425714/v1/0d0503bd5ad8f02836366e14.mp4)  
+  Supplementary Video2
+
+- [supplementaryinformation.docx](https://assets-eu.researchsquare.com/files/rs-3425714/v1/23510ead10023a1f4be294af.docx)  
+  Supplementary Information

33a330003676e700e4ac7f2b88e7eedf7b3d4903364456ff9c176981a164e3bb/preprint/images_list.json

@@ -0,0 +1,50 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpeg",
+        "caption": "Fabrication of gear-shaped MPs by OPMF system. a, Schematic illustrations for the OPMF system and synthesizing mechanism. b, \u201cGroove & tongue\u201d assembling for gear-shaped MC fabrication inspired by the ancient Chinese \u201cmortise & tenon joint\u201d timber structure. c, UV-vis spectra and images of bulk solutions (BSs) and curing solution (CS). d, Viscosity characterization of three BSs determined by shear rate. e, Modulus transformation imitation of the precursor inside MCs during synthesizing. f, Optical images of the obtained gear-shaped MPs. g, Distribution curves for outside diameter Do, root diameter Dr and cross-section area ag of the obtained gear-shaped MPs.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpeg",
+        "caption": "\u201cGroove & tongue\u201d and sliding assembling MC fabrication strategies for preparing MP green bodies with various shapes and materials. a, Schematic illustrations for \u201cgroove & tongue\u201d assembling with MC cross-sections including regular triangle, hexagon, isosceles triangle and trapezoid. b, Optical micrographs of the corresponding-shaped MP green bodies containing nanoparticles (NPs, including SiO2, Al2O3 and Si3N4) and binding polymer polyacrylamide (PAM). c, Schematic illustrations of the triangular MC cross-section transformation by tuning the sliding distance s1. d, SEM images of i, triangular and ii-iv, star-shaped MP green bodies. e, Schematic illustrations of the square MC cross-section transformation through tuning the sliding distances s1 and s2. f, SEM images of the MP green bodies with shapes resembling i, square, ii, \u201cZ\u201d, iii, \u201cI\u201d and iv, v, quadrangular stars.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpeg",
+        "caption": "Characterization of MPs before, during and after sintering. a, Thermogravimetric analysis, b, Derivative of weight loss and c, Differential scanning calorimetry characterizations of samples during sintering. d, Scanning electron microscope (SEM) image of several star-shaped Al2O3 MPs. e, SEM image of a single star-shaped Al2O3 MP consistent with the model\u2019s shape at the upper right corner. f, High-magnification SEM image of the crystalline grains on Al2O3 MP\u2019s surface with energy-dispersive X-ray (EDX) spectra floating above. g, TEM image of the crystalline lattice fringes of Al2O3 MPs and its converted h, diffraction spots. i, SEM image of the isosceles triangular SiO2 MPs. j, High-magnification SEM image of SiO2 MP\u2019s surface with EDX spectra floating above. k, SEM image of a star-shaped Si3N4 MP. l, High-magnification SEM image of Si3N4 MP\u2019s surface with EDX spectra floating above. Atomic force microscope (AFM) images showing the surface topographies of m, Al2O3, n, SiO2 and o, Si3N4 MPs. p, Surface roughness of three sintered MPs. X-ray diffraction (XRD) patterns of q, Al2O3, r, SiO2 and s, Si3N4 MPs.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpeg",
+        "caption": "Compression tests for sintered MPs with cross-sections including isosceles triangle and star-shape. a, Schematic illustrations of compression tests for sintered MPs with two shapes and two postures, P1 for triangular standing MPs, P2 for triangular lying MPs, P3 for star-shaped standing MPs and P4 for star-shaped lying MPs. b, Force\u2212displacement curves of P1, P2, P3 and P4 with SiO2, Al2O3 and Si3N4 materials. c, The maximum force (Fmax) and d, energy absorption (Ea) of various MPs during the compression process before failure, data are represented as mean values\u2009\u00b1\u2009s.d. from three samples.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpeg",
+        "caption": "Scratch tests for isosceles triangular and star-shaped ceramic MPs with SiO2, Al2O3 and Si3N4 materials. a, Schematic illustrations of scratch tests using various MPs and multiple substrates including copper (Cu), aluminum (Al), polystyrene (PS), polymethyl methacrylate (PMMA) and wood. b, Summary of the maximum forces for different MPs\u2019 cusps to withstand (CFmax) before failure, data are represented as mean values\u2009\u00b1\u2009s.d. from three samples. c, Hardness characterization for various substrates, data are represented as mean values\u2009\u00b1\u2009s.d. from three samples. d, Optical images and e, 3D topographical maps of different scratches by Al2O3 MPs. f, 2D cloud maps of scratches by Al2O3 MPs with cross-sectional profiles floating above. g, Summary of width (w), depth (h) and volume loss of different scratches using different MPs and substrates, data are represented as mean values\u2009\u00b1\u2009s.d. from three samples.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.jpeg",
+        "caption": "Comparisons of MPs obtained in this work with microtools reported by other works. a, Schematic illustrations of critical parameters of microtools, including circumcircle diameter (Dc), edge radius (R), roughness (Ra) and length (L). Ashby plots of microtools in terms of b, edge number (N) versus Dc, c, 2R/Dc \u00a0versus Dc and d, Ra versus processing rate. The references are listed in Supplementary Table 1.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

33a330003676e700e4ac7f2b88e7eedf7b3d4903364456ff9c176981a164e3bb/preprint/preprint.md

@@ -0,0 +1,147 @@
+# Abstract
+
+In the quest for miniaturization across technical disciplines, microscale ceramic blocks emerge as pivotal components, with performance critically dependent on precise scales and intricate shapes. Sharp-edged ceramic microparticles, applied from micromachining to microelectronics, require innovative fabrication techniques for high-throughput production while maintaining structural complexity and mechanical integrity. This research unveils a revolutionary "one-pot microfluidic fabrication" technique, blending two innovative device fabrication strategies, "groove & tongue" and sliding assembling. This approach yields a remarkable variety of microparticles, each intricately shaped and precisely crafted, significantly surpassing traditional methods in both production speed and quality. Optimally designed sintering profiles enhance microparticles’ shape retention and structural strength. Extensive compression and scratch tests validate the superiority of microparticles, suggesting their practicability for diverse applications, such as precise micromachining, sophisticated microrobotics and delicate microsurgical tools. This advancement marks a paradigm shift in microscale manufacturing, offering a scalable solution to meet the demanding specifications of miniaturized technology components.
+
+- Physical sciences/Materials science/Structural materials
+- Physical sciences/Engineering/Mechanical engineering
+- Physical sciences/Materials science/Techniques and instrumentation/Design, synthesis and processing
+
+# Introduction
+
+The evolution of microscale material fabrication is a defining trend in numerous industries, from electronics<sup>1, 2</sup> to micro-robotics<sup>3, 4</sup> and surgical instrument<sup>5, 6</sup> manufacturing. The ability to process microparticles (MPs) with precision is increasingly crucial, with the choice of material being central to their function. Ceramics are in the spotlight for their exceptional properties—hardness, wear resistance, resilience to high temperatures and chemicals, coupled with low thermal conductivity. The utility of ceramic MPs is diverse, influenced by their shape and material composition<sup>7, 8, 9, 10, 11</sup>. A multitude of techniques have been explored for the creation of MPs<sup>12, 13</sup>. Methods like micro electrical discharge machining (µEDM) and micro grinding offer contactless fabrication of MPs, yet their application is confined to conductive materials<sup>12</sup>. Alternative methods employing ion or laser beams can sculpt MPs from a broader range of materials with high fidelity<sup>14, 15</sup>, but such processes come at a steep cost. Mechanical machining has been engaged to produce shanked MPs, though achieving high precision remains elusive<sup>16</sup>. Projection micro stereolithography excels in crafting ultrafine microparts; however, it’s predominantly suitable for UV-curable resins<sup>17</sup> and hampered by low production rates. Micro injection molding (µIM), while capable of processing a diverse array of materials at increased production speeds, is hindered by its inherently batch-wise production flow, which impedes continuous manufacturing and thus limits throughput.
+
+Microfluidic lithography has been utilized to fabricate diverse MPs with high monodispersity, precision, and throughput<sup>18, 19</sup>, which includes sharp-edged 3D anisotropic transparent<sup>20, 21</sup> and spherical/bowl-shaped nontransparent MPs<sup>22, 23</sup>. However, the UV light path distortion caused by reflection, scattering, and refraction of the dispersed nanoparticles (NPs) makes it almost impossible to obtain sharp-edged MPs through microfluidic lithography, let alone fabricating them with higher strength by increasing NP content. Therefore, a “one-pot microfluidic fabrication” (OPMF) system incorporating thermocuring modules is innovatively presented. The production rate of MP tools is significantly improved by more than two orders of magnitude while maintaining the equivalent roughness, sharpness, size range and shape complexity as the microtools obtained through existing methods.
+
+Owning to the stringent precision requirements for the microfluidic channel (MC) mold manufacturing and considerable difficulties in MC demolding, it remains challenging to create precise PDMS MCs with cross-sections that are not based on rectangles<sup>24, 25, 26</sup>. Drawing upon the ancient Chinese carpentry technique of the “mortise & tenon joint”, the OPMF system introduces a precise “groove & tongue” assembling (GTA) strategy for various non-rectangular MCs fabrication, which offers a creative leap in shaping MPs by allowing for intricate and precise structures previously unattainable. Furthermore, the established practice in microfluidics often restricts the use of templating molds to produce MCs with uniform cross-sections<sup>8, 20, 21, 27</sup>, limiting their versatility. Inspired by another pearl of ancient wisdom, the “Chinese Tangram”, our research transcends this boundary by pioneering a “sliding assembling” (SA) technique. This innovation enables the use of a singular mold to fabricate a variety of MCs, not only significantly broadening the scope of potential MC and MP designs but also promoting a more efficient and eco-friendly manufacturing process.
+
+To date, sharp-edged anisotropic MPs crafted through microfluidics have found their primary applications in cell manipulation<sup>28</sup>, bioassays<sup>29, 30</sup>, anticounterfeiting<sup>31, 32, 33</sup> and so forth. However, their use as micro tools/parts has been stymied by challenges such as low material density, fragility and limitations with nontransparent materials<sup>34</sup>. Addressing these issues, our work significantly enhances the density and strength of MPs by augmenting the solid content and employing optimally designed sintering profiles. Moreover, we delve into the realm of nontransparent materials, such as Al<sub>2</sub>O<sub>3</sub> and Si<sub>3</sub>N<sub>4</sub>, achieving MPs that exhibit the requisite durability and toughness to process substrates of various materials, including metals, plastics and wood. These advancements showcase the potential of these robust MPs to function as key components in micro-electromechanical systems (MEMS), micro robots, and as precise instruments in micro machining and microsurgery, marking a progressive stride in microscale material fabrication.
+
+# OPMF system for gear-shaped MP fabrication
+
+OPMF system separates the acrylamide-based polymerization precursor into two parts based on characteristics of component reagents: bulk solution (BS) for the mixture of NP dispersion and prepolymers (acrylamide and N, N’-methylenebisacrylamide), curing solution (CS) for an aqueous mixture of ammonium persulfate and N, N, N’, N’-tetramethyl ethylenediamine that initiate and accelerate the polymerization (Fig. 1a). BS and CS remain stable individually but react promptly upon meeting each other. A microfluidic device is designed to be divided into three functional zones for precursor feeding, mixing and curing, respectively. A mixer, comprising a magnetic bead confined within the workspace by two fixed pillars, is magnetically driven to enhance convection between BS and CS. Mixed precursor uniformly fills MC before the curing zone (“A−A”) and is gradually cured by the heat transferred from the peripherally wrapped copper sheet after entering the curing zone (“B−B”), where the crosslinking reaction occurs massively and forms a 3D netlike polymer structure. Oxygen diffuses through the PDMS wall and reacts with initiator species to form chain-terminating peroxide radicals<sup>35</sup>, inhibiting polymerization reaction<sup>36</sup>. Oxygen concentration C(O<sub>2</sub>) increases with proximity to the MC wall<sup>37</sup>. Therefore, a thin lubricating layer forms, facilitating the smooth extrusion.
+
+Inspired by the ancient Chinese timber structure “mortise & tenon joint”, GTA is first presented to precisely fabricate MCs with both rectangular and non-rectangular cross-sections. Precisely coordinated grooves and tongues control the relative positions of MC substrate pieces and ensure the precise alignment of MC inner walls. GTA can be utilized to fabricate MCs possessing multiple separated pieces, increasing MPs’ shape complexity. Here, a gear-shaped MC with ten teeth is created (Fig. 1b, Supplementary Figs. 1−3). Each separated MC piece features an intact contour of a gear tooth, encompassing contour lines for one top land, one bottom land, two faces, two flanks and two fillets. A tongue and groove are set on the center of two opposite oblique side walls. MC is secured after assembling. The gradually enlarged views of MC show good cooperation between grooves and tongues, as well as a clear gear-shaped cross-section.
+
+NPs with three materials, SiO<sub>2</sub>, Al<sub>2</sub>O<sub>3</sub> and Si<sub>3</sub>N<sub>4</sub>, are integrated into BS, respectively. SiO<sub>2</sub>−BS is about half transparent on average across the UVA wavelength range (315−400 nm); nevertheless, Al<sub>2</sub>O<sub>3</sub>/Si<sub>3</sub>N<sub>4</sub>−BSs show zero light transmittance in the 250−800 nm range due to light scattering, reflection and refraction caused by the NPs (Fig. 1c). Precursors made from such opaque BSs can only be solidified into desire-shaped MPs through high-intensity/long-time UV light curing within the templates such as droplets<sup>22, 23</sup>, concave molds<sup>7</sup> and so forth. Here, the OPMF system allows us to create sharp-edged MPs using both transparent and nontransparent precursors.
+
+The average viscosities of three shear-thinning BSs are 14.7, 115.7 and 571.1 mPa∙s, respectively (Fig. 1d, Supplementary Fig. 4), indicating the OPMF system is well-suited for BSs with a viscosity ranging from tens to hundreds of mPa∙s. BS and CS contact and mix at the start of the mixing zone, where storage modulus G’ dramatically increases and exceeds the loss modulus G’’ of the precursor (Fig. 1e), indicating the fluid-gel transition occurs and crosslinked polyacrylamide (PAM) is formed. As temperature increases, G’ and G’’ increase and gradually stabilize at ~10<sup>5</sup> Pa after reaching ~60 ℃, recognized as the optimal heating temperature of the OPMF system.
+
+Microfiber green bodies are extruded at a rate of ~12 mm<sup>3</sup>/min and then cut into MPs with intact gear outlines and smooth surfaces (Fig. 1f). Distributions of outer diameter *D*<sub>o</sub>, root diameter *D*<sub>r</sub> and cross-section area *a*<sub>g</sub> demonstrate high homogeneity of MPs with coefficients of variation calculated as 2.09%, 2.17% and 3.72%, respectively (Fig. 1g). The OPMF system exhibits significantly higher production rates compared to traditional methods (Supplementary Table 1), positioning it as a promising technology for the mass production of micro torque-transmitting components in MEMS and microrobots.
+
+# GTA and SA for diverse MPs fabrication
+
+GTA aforementioned is further exemplified by introducing more cross-section shapes, including regular triangle, hexagon, isosceles triangle and trapezoid (Fig. 2a). MPs incorporated with SiO₂, Al₂O₃ and Si₃N₄ NPs respectively show transparent, opaque white and brown looks under the microscope (Fig. 2b). Sharp corners and clear edges of these MPs verify the superiority of GTA. SiO₂, Al₂O₃ and Si₃N₄ MP green bodies shrink about 35%, 14% and 16% after drying. Oxygen inhibition is prevalent in numerous free-radical polymerization reactions, rendering our OPMF system suitable for producing MPs with more diverse materials. However, this one-to-one correspondence of GTA between mold and MC limits the mold’s shape-creation capacity. SA, inspired by another ancient Chinese wisdom, “Tangram”, is presented pertinently to overcome this limitation by allowing adjacent MC pieces to slide at the smooth interface in between, leading to the generation of MCs with various cross-sections and sharp cutting edges (Supplementary Fig. 5). The shape-creation capacity of MCs obtained using one mold is infinite due to the variation of sliding distance (*s*). Assuming the number of MC pieces is *N*, the degree of sliding freedom (DSF) for MC assembling is *N* − 2. Immobilizing one of the separated pieces (defined as piece 0), the number of pieces that can be moved flexibly under the driving force is *N* − 2, while the remaining one piece is a follower.
+
+There is only one driving piece for triangular MC. As *s*₁ increases from 0 to infinity, the cross-section of MC transforms from a regular triangle (*s*₁ = 0) to a three-pointed star (*s*₁ = *l*), then to a regular triangle with three negligible triangular barbs extended from each corner (*s*₁ = ∞), where *l* is defined as the side length of the cross-section (Fig. 2c,d). When piece 1 is driven along the negative direction, the obtained MC cross-section shapes are mirror-symmetrical to those driven along the positive direction. For rectangular MC, pieces 1 and 2 are both defined as driving pieces. The MC cross-section transforms as sliding distances *s*₁ and *s*₂ vary (Fig. 2e). MCs with numerous cross-sections can be obtained (Supplementary Fig. 6), indicating a further enlarged shape-creation capacity of the mold. The cross-section shapes are mirror-symmetrical about the line *y* = *x*. Additionally, the cross-section shapes on one side of the line *y* = −*x* can be transformed into the shapes on the other side by rotating 90° about their centroids. Cross-section shapes at some special points are selected to manifest the shape variation trend as *s*₁ and *s*₂ vary, among which square, “Z/I” shapes and four-pointed stars are chosen to be experimentally fabricated (Fig. 2f).
+
+# Characterization of MPs before, during and after sintering
+
+The compositions and structures of the obtained MP green bodies are verified by Fourier transform infrared (FTIR, Supplementary Fig. 7) and X-ray photoelectron spectroscopy (XPS, Supplementary Fig. 8). The compositional elements of each sample are well-mixed and distributed uniformly (Supplementary Fig. 9), indicating the homogeneous mixing of our OPMF system. MP green bodies are then debound and densified through high-temperature sintering (Fig. 3 a,b,c). They dehydrate before ~ 113 ℃ with slight mass loss due to prior 24-hour air drying. Then, polymer PAM in four samples mainly goes through two endothermic processes, glass transition and melting, subsequently at 113−345 ℃, during which the polymer transforms from a stiff “glassy” state to a soft “rubbery” state<sup>38, 39</sup>, then to a liquid state. As the temperature keeps increasing, PAM decomposes into multiple smaller molecules, H<sub>2</sub>O and gases<sup>40</sup> before 480 ℃ and is finally combusted under the synthetic effect of heat and oxygen at 480−708 ℃. Samples’ weights decrease significantly, and plenty of heat is released during decomposition and combustion. Notably, there is a prominent weight increase and a conspicuous exothermic peak for the Si<sub>3</sub>N<sub>4</sub> sample at 1173−1327 ℃ due to oxidation of the sample surface with the main chemical reaction described as follows<sup>41, 42</sup>.
+
+Si<sub>3</sub>N<sub>4</sub> (s) + 3 O<sub>2</sub> (g) = 3 SiO<sub>2</sub> (s) + 2 N<sub>2</sub> (g) (<span citationid="CR1" class="CitationRef">1</span>)
+
+Weight loss and heat flow of the samples vary with their polymer content. Due to the limited PAM content, SiO<sub>2</sub>, Al<sub>2</sub>O<sub>3</sub> and Si<sub>3</sub>N<sub>4</sub> green bodies have relatively lower weight/heat change during debinding. To preserve MPs from cracking caused by drastic mass/volume change and also to consider production efficiency, the sintering profiles for different MPs are designed pertinently (see Methods for details). After sintered, ~ 25.6%, ~ 19% and ~ 5.3% linear shrinkages are respectively detected for SiO<sub>2</sub>, Al<sub>2</sub>O<sub>3</sub> and Si<sub>3</sub>N<sub>4</sub> samples; meanwhile, their monodispersity, shape retentivity and shape fidelity remain great (Fig. 3 d,e,i,k).
+
+Densely packaged grains ranging from 1 to 5 µm (Fig. 3 f) make a roughness (<em>R</em><sub>a</sub>) ~ 0.255 µm of the Al<sub>2</sub>O<sub>3</sub> MP surface (Fig. 3 m,p). The distance (0.255 nm) between adjacent lattice fringes is in good accordance with the Al<sub>2</sub>O<sub>3</sub> (1 0 4) crystal face at 2<em>θ</em> = 35.136° (Fig. 3 g,h). Hexagonal corundum (PDF#10–0173, Fig. 3 q), a high-temperature polymorph of Al<sub>2</sub>O<sub>3</sub>, is obtained. SiO<sub>2</sub> MP comprising nanosized partially molten grains has a dense surface with a <em>R</em><sub>a</sub> ~0.391 µm (Fig. 3 j,n). Its crystal structure corresponds with cristobalite (PDF#39-1425, Fig. 3 r), while Si<sub>3</sub>N<sub>4</sub> MP has multiple crystal structures (Fig. 3 s), including cristobalite (PDF#89-3607), β-Si<sub>3</sub>N<sub>4</sub> (PDF#33-1160) and Si<sub>2</sub>N<sub>2</sub>O (PDF#83-2149). Cristobalite is formed after the oxidation of Si<sub>3</sub>N<sub>4</sub> NPs through Reaction (<span citationid="CR1" class="CitationRef">1</span>). Sintering additives Y<sub>2</sub>O<sub>3</sub> and Al<sub>2</sub>O<sub>3</sub> react with SiO<sub>2</sub> and form a silicate liquid phase Y-Si-Al-O-N, promoting the sintering process<sup>43, 44</sup>. α-Si<sub>3</sub>N<sub>4</sub> grains are transformed into β-Si<sub>3</sub>N<sub>4</sub> grains through a solution-precipitation process in the presence of the liquid phase<sup>45</sup>. The rod-like elongated β-Si<sub>3</sub>N<sub>4</sub> grains (Fig. 3 l) act as whisker-reinforcing agents in the matrix, strengthening the composite<sup>46, 47</sup>. Si<sub>2</sub>N<sub>2</sub>O is formed from the reaction of SiO<sub>2</sub> and Si<sub>3</sub>N<sub>4</sub> in the presence of a liquid phase, with the main chemical reaction described as follows<sup>48, 49</sup>.
+
+SiO<sub>2</sub> (s) + Si<sub>3</sub>N<sub>4</sub> (s) = 2 Si<sub>2</sub>N<sub>2</sub>O (s) (<span citationid="CR2" class="CitationRef">2</span>)
+
+Besides, the liquid phase finds its equilibrium at grain boundaries, making the MPs of high density and great smoothness<sup>44</sup>. As a result, the <em>R</em><sub>a</sub> of some localized surface regions is as low as ~ 0.249 µm, while it increases to 0.754 µm for an enlarged surface region due to the bumpy topography (Fig. 3 o).
+
+# Performance assessment of MPs and OPMF system
+
+Prismatic ceramic MPs possessing isosceles triangular and star-shaped cross-sections are selected as model MPs obtained from GTA and SA MCs respectively to test their mechanical strength and toughness. Besides, two postures (standing and lying) are adopted to test MPs thoroughly with different materials. Triangular MPs with standing and lying postures are designated P1 and P2, while star-shaped ones are designated P3 and P4 (Fig. 4a). Compared with P1, P2 has shorter displacement and lower maximum force (\(F_{\text{max}}\)) owing to the high intensity of the pressure exerted on its top edge (Fig. 4b). P4 withstands larger pressing displacement and higher \(F_{\text{max}}\) than P3. The destructions of four lateral sharp edges of P4 absorb and dissipate more compressive energy, and lead to several sharp force drop-offs. The force−displacement curves vary prominently with the structures of MPs (Fig. 4c). P3 has larger \(F_{\text{max}}\) and energy absorption (\(E_{\text{a}}\)) than P1, mainly due to the increase in cross-sectional area. P4 has larger \(F_{\text{max}}\) and \(E_{\text{a}}\) than P2 because of the increased pressure-bearing edges. For MPs with the same shape and posture, Si\(_3\)N\(_4\) MPs have the highest \(F_{\text{max}}\) and \(E_{\text{a}}\), while SiO\(_2\) MPs have the lowest.
+
+The practical application performance of the rigid MPs is tested through scratch tests. Six types of MPs, including triangular/star-shaped Al\(_2\)O\(_3\) (T/S-Al\(_2\)O\(_3\)), SiO\(_2\) (T/S-SiO\(_2\)), Si\(_3\)N\(_4\) (T/S-Si\(_3\)N\(_4\)) MPs, are used to scratch the substrates made of five materials including copper (Cu), aluminum (Al), polystyrene (PS), polymethyl methacrylate (PMMA) and wood (Fig. 5a). The MPs are positioned at a slant angle of 70° intersecting the substrates with their cusps. Star-shaped MPs have lower maximum force withstood by their cusps before failure (C\(F_{\text{max}}\)) than triangular ones (Fig. 5b) because their cutting cusps are sharper and relatively easier to break. SiO\(_2\) MPs have the lowest C\(F_{\text{max}}\), while Si\(_3\)N\(_4\) MPs possess the highest, consistent with the ranking order of compressive strength. To prevent MP cusps from being broken during scratching, we conservatively set the forces applied to T-SiO\(_2\)/Al\(_2\)O\(_3\)/Si\(_3\)N\(_4\) and S-SiO\(_2\)/Al\(_2\)O\(_3\)/Si\(_3\)N\(_4\) MPs as 1.5, 9, 13, 1, 6 and 9 N, respectively. Z-stack scanning is applied to reconstruct 3D shapes of various scratches and measure their width (\(w\)), height (\(h\)) and volume loss. The scratches on metal (Cu, Al) and polymeric plastic (PS, PMMA) substrates have clear traces and distinct profiles, while the scratches obtained on wood substrates are somewhat out of shape due to the interference of the intertwined component wood fibers (Fig. 5d,e,f, Supplementary Figs. 10−12). After analyzing \(w\), \(h\) and volume loss of 477-µm-long scratches in the images, substrate materials can be ranked as Al > Cu > PMMA > PS > wood based on their subtractive manufacturing difficulties, which is positively correlated with the hardness ranking of substrate materials (Fig. 5c). It’s evident that Si\(_3\)N\(_4\) MP tools are capable of removing more material than Al\(_2\)O\(_3\) ones, while SiO\(_2\) MP tools remove the least (Fig. 5g). These scratch tests demonstrate the practicability of using ceramic MP tools obtained through our work in micromachining fields.
+
+The MP tools obtained through the OPMF system are compared with the microtools fabricated by other reported methods, including micro-wire electrical discharge machining (µWEDM), µIM, micro grinding, focused ion beam and femtosecond pulsed laser machining (Supplementary Table 1). The edge number of the cross-section (\(N\)) is positively correlated with the shape complexity of the microtools. Limited by the machining precision, improving shape complexity becomes more difficult as microtools’ size decreases. The circumcircle diameter (\(D_c\)) of the cross-section is used to describe the microtool’s size universally (Fig. 6a). Inferred from the statistical results of \(N\) and \(D_c\), the MP tools obtained in this study exhibit a broad size distribution, spanning across the size ranges of other methods, while maintaining equivalent shape complexity (Fig. 6b). In terms of the same machining methods, the shape fidelity of the microtool becomes more difficult to control as the size decreases. Here, the ratio between edge radius (\(R\)) and \(D_c\)/2 is used to evaluate the shape fidelity, where \(R\) is one of the most critical parameters for tools design. \(R\) of T-Al\(_2\)O\(_3\) and S-Al\(_2\)O\(_3\) MPs are representatively measured as 4.78 and 1.32 µm (Supplementary Fig. 13), and the corresponding 2\(R\)/\(D_c\) are calculated as 0.02529 and 0.00295, respectively, which are comparable with 2\(R\)/\(D_c\) from other methods (Fig. 6c). Last and most importantly, \(R_a\) of microtools tends to be positively correlated with the processing rate in terms of the same method. This work is capable of increasing the processing rate by two orders of magnitude while keeping \(R_a\) below 1 µm, which is comparable with \(R_a\) obtained by other methods (Fig. 6d).
+
+In summary, we present an OPMF system to fabricate various sharp-edged MPs with materials, including transparent SiO\(_2\) and nontransparent Al\(_2\)O\(_3\) and Si\(_3\)N\(_4\). Besides, two types of MC designs inspired by ancient Chinese wisdom, including GTA and SA, are invented to increase the MCs’ shape complexity, precision and diversity. Compared with existing methods for microtools fabrication, the production rate of our methods is significantly improved by more than two orders of magnitude while keeping the equivalent MP shape complexity, diversity, size range, corner sharpness and surface roughness. By increasing the solid content and adopting optimally designed sintering procedures, the obtained MPs are endowed with great density, strength and toughness, being able to scratch the substrates made of various materials, including metals, plastics and wood. The obtained MPs show groundbreaking prospects for practical use as microtools in micromachining and microsurgery, as well as functional microcomponents in microrobots, MEMS, microelectronics and so forth.
+
+# Methods
+
+## Fabrication of Microfluidic Device
+The metal molds made of stainless steel were fabricated by micro-wire electrical discharge machining using a machine tool (Shenzhen Wanqi CNC Equipment CO., Ltd.; DK7735). SYLGARD® 184 Silicone Elastomer Kit (Dow Corning) was used to fabricate PDMS reverse mold and MC substrate. A weight ratio of 5: 1 between the base fluid and the crosslinking oligomer was adopted to prepare the PDMS prepolymer. The prepolymer was stirred, degassed and poured into the metal mold. The filled mold was also degassed and then heated in the oven at 65°C for 1.5 h. After solidified, the PDMS reverse mold was demolded. Trichloro(octadecyl)silane (OTS, Sigma-Aldrich) was coated on the surface of the reverse mold as the releasing agent. PDMS prepolymer was poured into the reverse mold, degassed, and then heated at 65°C for 45 min. The obtained semi-solidified MC substrate was demolded, and divided into several separated pieces, which were then assembled in a specific manner to form a closed MC. Polycarbonate clamps were applied to secure the MC with a certain pressure. The cross-section shapes of the MC were tailored by adjusting the relative positions between the MC pieces and polycarbonate clamps. After being heated for another 45 min, the assembled fully cured MCs were trimmed to possess a square outer contour. A metal needle (60 µm in inner diameter) was used to punch holes through the side wall of the MC. A flat-ended metal needle (160 µm in inner diameter) was inserted into the center of the MC through the punched hole. A copper sheet (50 µm in thickness, 5 mm in width) was wrapped peripherally around the exit of MC. A magnetic bead (300 µm in diameter) was placed inside the MC with two steel rods (230 µm in diameter) confining its workspace.
+
+## Materials and Reagents
+The solution for synthesizing MP green bodies was divided into BS and CS. BS was prepared by mixing acrylamide (AM, Sinopharm Chemical Reagent Co., Ltd.; AR), N, N’-methylenebisacrylamide (MBAM, Coolaber Science and Technology Co., Ltd.; purity > 99.9%) and NP dispersions. We chose the commercially available colloidal silica (30 nm NP average size, 45 wt% suspensions in H₂O, LUDOX CL-X, Sigma-Aldrich) as the SiO₂ NP dispersion. Al₂O₃ NP dispersion was prepared by ball-milling the mixture containing α-Al₂O₃ ceramic powder (Almatis Co., Ltd.; 1 µm, purity ≥ 99.99 wt%), deionized (DI) water and triammonium citrate (TAC, Sinopharm Chemical Reagent Co., Ltd.; AR) at 450 rpm for 24 h. The volume ratio between the ceramic powder and the DI water was 1: 1, TAC was added at 1 wt% based on the powder. 45 vol% α-Si₃N₄ ceramic powder (Shanghai Zhiliyejin Co., Ltd.; 1 µm, purity ≥ 99.9 wt%) with the addition of 7.5 wt% Y₂O₃ (Shanghai Macklin Biochemical Co., Ltd.; 1 µm, 99.99% metals basis) and 2.5 wt% α-Al₂O₃ (Almatis Co., Ltd.; 1 µm, purity ≥ 99.99 wt%) was suspended in DI water. Tetramethylammonium hydroxide solution (TMAH, Shanghai Aladdin Biochemical Technology Co., Ltd.; 10 wt% in H₂O) was added at 6 vol% based on Si₃N₄ powder. Si₃N₄ NP dispersion was obtained after ball-milling the mixture at 450 rpm for 24 h. AM was added to the above NP dispersions at 20% (w/v), and MBAM was added at 16.7 wt% based on AM. The CS of the system was the mixed aqueous solution comprised of the initiator ammonium persulfate (APS, Coolaber Science and Technology Co., Ltd.; purity ≥ 98%) and the catalyst N, N, N’, N’-tetramethyl ethylenediamine (TEMED, Shanghai Macklin Biochemical Co., Ltd.; purity ≥ 99%). The concentrations of APS and TEMED were 8.4 wt% and 4.5 vol%, respectively, based on CS. The washing solution was prepared by adding APS to DI water at 3 wt%. The aqueous aging solution was prepared by adding 3 wt% APS and 0.2 vol% TEMED.
+
+## MP Green Bodies Fabrication
+BS and CS were respectively injected into the microfluidic device at flow rates of 10 µL/min and 2 µL/min by syringes (TERUMO®SYRINGE) using Teflon tubes (Anzijie Trade Co., Ltd.). Syringe pumps (LongerPump® LSP01-3A) were used to drive the fluids. The temperature of the copper sheet was controlled at 60 ℃ by a heating system consisting of a ceramic heating plate (XH-RP4040, 12W, Jiangsu Xinghe Electron Technology Co., Ltd), temperature sensor (XH-T112) and temperature controller (XH-W2030). The copper sheet transferred heat into the MC evenly. The magnetic bead was driven by a magnetic stirrer (DLAB, MS-H280-Pro) placed under the MC. Green microfibers were produced and collected at the exit of the MC. After washed, they were aged in the oven at 65 ℃ for 1.5 h. A blade (Gillette, double edge super stainless) polished by abrasive paper (Starcke, ST 5000) was applied to cut the green microfibers vertically under the microscope (Olympus, DSX-1000) to obtain MPs. Afterward, the MPs were air-dried on Teflon film at room temperature for 12 h.
+
+## MPs Sintering
+The SiO₂ MP green bodies were heated from room temperature to 705°C at a rate of 0.5°C/min, then to 1150°C at a rate of 1°C/min. The temperature was kept at 78, 282, 394, 581°C for 0.5 h, and 705, 1150°C for 2 h. The Al₂O₃ MP green bodies were heated from room temperature to 632°C at a rate of 1°C/min, then to 1550°C at a rate of 5°C/min. The temperature was preserved at 147, 316, 365, 561 ℃ for 0.5 h, and 632, 1550 ℃ for 2 h. The Si₃N₄ MP green bodies were heated from room temperature to 522°C at a rate of 1°C/min, then to 1750°C at a rate of 5°C/min. The temperature was preserved for 0.5 h at 163, 352, 522°C, and 2 h at 570 and 1750°C, respectively. All the sintered MPs were finally obtained after the furnace was cooled.
+
+## Materials Characterization
+The morphologies of the materials were observed through the stereo microscope (Nikon, SMZ1000), upright optical microscope (Olympus, BX60F5) and scanning electron microscope (FE-SEM, 7600F). The transmittance of various BSs and CS was measured using an ultraviolet-visible spectrophotometer (UV-vis, SHIMADZU, UV-2700) over a wavelength range of 190 − 800 nm. The viscosity and modulus of the samples were measured using a rotational rheometer (Anton Paar, MCR 702e) with 12-mm-diameter steel parallel-plate geometry. An annular wall sprayed with water covers the experimental set-up peripherally to prevent rapid sample water evaporation during the measuring process. The temperature was increased at 6.49 ℃/min. All rheological characterizations were conducted with a preliminary equilibration time of 3 min. FTIR spectra of samples were examined with an attenuated total reflection FTIR (PerkinElmer, Frontier). XPS (Kratos AXIS Supra) with 225W Al Kα radiations was carried out to confirm the sample's composition. SEM-associated energy-dispersive X-ray (EDX) spectroscopy was used to analyze the elements and their distribution in samples. Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) were simultaneously conducted under air atmosphere using a thermal analysis instrument (TA, SDT Q600). Temperature of PAM, SiO₂ + PAM, Al₂O₃ + PAM and Si₃N₄ + PAM samples ramped up at 15 ℃/min from room temperature to 800, 1200, 1400 and 1400 ℃, respectively. The transmission electron microscope (TEM, FEI, Technai F20) was used to test the ceramic structure of the sintered MPs. TEM samples were prepared using the precision ion polishing system (Gatan, PIPS II 695). The sintered MPs were ground to powder and tested by X-ray diffraction (XRD, Bruker, D8 Advance). Atomic force microscope (AFM) images were acquired by an AFM (Park Systems, NX10). Double-sided carbon tape was used to attach the rigid ceramic MPs to the sample stage.
+
+## Mechanical Test
+Compression tests of sintered MPs were performed using a mechanical tester (MTS, C42) with a 250 N load cell (MTS, LSB.252). The top press head went down at a constant speed of 0.1 mm/min. The energy absorption of the MPs was calculated from the area below the force–displacement curve until fracture.
+
+## Scratch Sample Preparation
+A piece of glass slide adhered with a layer of double-sided carbon tape (2×2 mm²) was utilized as the substrate for the scratch samples preparation. MP was attached to the straight slender beam that was bedewed with DI water. Interfacial tension of the water held the MP immobilized on the beam for about 10 s before the water was fully evaporated, during which the central axis of MP was tuned to be 70° intersecting with the substrate, MP was quickly moved forward and stuck on the surface of tape. The beam was then removed. After drying for about 5 min, a small amount of Cyanoacrylate glue (ergo, 5800, Switzerland) was poured around the MP and immersed the bottom of MP. The scratch sample preparation was finally completed after drying overnight.
+
+## Scratch Test
+The substrate of the scratch sample was attached to the center of the top press head upside down using cyanoacrylate glue. The test substrate was placed beneath the MP. The top press head was moved forward until the pressure reached a certain value (1.5, 9, 13, 1, 6 and 9 N for T-SiO₂, T-Al₂O₃, T-Si₃N₄, S-SiO₂, S-Al₂O₃ and S-Si₃N₄ MPs, respectively), and then the test substrate was dragged out at a constant speed along a ruler fixed beside. Scratches were characterized and analyzed using a digital microscope (Olympus, DSX-1000). The half-embedded MPs were released using adhesive remover (ergo, 9153, Switzerland) for further characterization.
+
+## Statistical Analysis
+Excel and Origin 2022 were used to analyze and plot the data. Image J was used for image analysis and measurement. Solidworks 2020 was used to do the schematic illustration and the modeling of MC transformations. XPS profiles were analyzed by ESCApe software system. Jade 6.5 was used to analyze the XRD result.
+
+# References
+
+1. Wang S et al (2018) Skin electronics from scalable fabrication of an intrinsically stretchable transistor array. Nature 555:83–88
+2. Zhang Z et al (2022) High-brightness all-polymer stretchable LED with charge-trapping dilution. Nature 603:624–630
+3. Joh H, Fan DE (2021) Materials and schemes of multimodal reconfigurable micro/nanomachines and robots: review and perspective. Adv Mater 33:2101965
+4. Halder A, Sun Y (2019) Biocompatible propulsion for biomedical micro/nano robotics. Biosens Bioelectron 139:111334
+5. Randall CL, Gultepe E, Gracias DH (2012) Self-folding devices and materials for biomedical applications. Trends Biotechnol 30:138–146
+6. Shakoor A, Gao W, Zhao L, Jiang Z, Sun D (2022) Advanced tools and methods for single-cell surgery. Microsyst Nanoeng 8:47
+7. Chen H, Zhou C, Li Y, Liang S, Li J (2020) Dielectric tetrahedrons as terahertz resonators switched from perfect absorber to reflector. Sci Rep 10:17134
+8. Zhou C, Liang S, Li Y, Li J, Chen H (2022) Microfluidic assembly of micro-blocks into interlocked structures for enhanced strength and toughness. ACS Appl Mater Interfaces 14:7261–7269
+9. Shepherd RF et al (2008) Stop-flow lithography of colloidal, glass, and silicon microcomponents. Adv Mater 20:4734–4739
+10. Wang S-H, Wang P-H, Hsieh I-M, Hsieh C-C (2018) Microfluidic synthesis of silica microcomponents using sol-gel process and stop-flow lithography. J Taiwan Inst Chem Eng 93:103–108
+11. Oliaei SNB, Karpat Y, Paulo Davim J, Perveen A (2018) Micro tool design and fabrication: a review. J Manuf Process 36:496–519
+12. O'Hara J, Fang F (2019) Z. Advances in micro cutting tool design and fabrication. Int J Extreme Manuf 1:032003
+13. Gao S, Huang H (2017) Recent advances in micro- and nano-machining technologies. Front Mech Eng 12:18–32
+14. King PC, Zahiri SH, Jahedi M (2008) Focused ion beam micro-dissection of cold-sprayed particles. Acta Mater 56:5617–5626
+15. Palima D, Glückstad J (2013) Gearing up for optical microrobotics: micromanipulation and actuation of synthetic microstructures by optical forces. Laser Photon Rev 7:478–494
+16. Boswell B, Islam MN, Davies IJ (2018) A review of micro-mechanical cutting. Int J Adv Manuf Technol 94:789–806
+17. Ge Q et al (2020) Projection micro stereolithography based 3D printing and its applications. Int J Extreme Manuf 2,
+18. Dendukuri D, Doyle PS (2009) The synthesis and assembly of polymeric microparticles using microfluidics. Adv Mater 21:4071–4086
+19. Jo YK, Lee D (2020) Biopolymer microparticles prepared by microfluidics for biomedical applications. Small 16:1903736
+20. Zhou C, Liang S, Li Y, Chen H, Li J (2021) Fabrication of sharp-edged 3D microparticles via folded PDMS microfluidic channels. Lab Chip 22:148–155
+21. Zhou C et al (2022) Tailoring 3D shapes of polyhedral microparticles by adjusting orthogonal projection in a microfluidic channel. J Polym Sci 60:1750–1757
+22. Zhou C et al (2021) Fabrication of crescent-shaped ceramic microparticles based on single emulsion microfluidics. Ceram Int 47:10866–10872
+23. Zhou C et al (2022) Double UV lights intersection shaping for bowl-shaped ceramic microparticles based on microfluidics. Ceram Int 48:27590–27596
+24. Paulsen KS, Di Carlo D, Chung A (2015) J. Optofluidic fabrication for 3D-shaped particles. Nat Commun 6:6976
+25. Paulsen KS, Deng Y, Chung AJ (2018) DIY 3D microparticle generation from next generation optofluidic fabrication. Adv Sci 5:1800252
+26. Chung SE, Park W, Shin S, Lee SA, Kwon S (2008) Guided and fluidic self-assembly of microstructures using railed microfluidic channels. Nat Mater 7:581–587
+27. Yuan R et al (2018) Designable 3D microshapes fabricated at the intersection of structured flow and optical fields. Small 14:1803585
+28. Chen L et al (2016) Flexible octopus-shaped hydrogel particles for specific cell capture. Small 12:2001–2008
+29. Pregibon DC, Toner M, Doyle PS (2007) Multifunctional encoded particles for high-throughput biomolecule analysis. Science 315:1393–1396
+30. Lee H, Kim J, Kim H, Kim J, Kwon S (2010) Colour-barcoded magnetic microparticles for multiplexed bioassays. Nat Mater 9:745–749
+31. Han S et al (2012) Lithographically encoded polymer microtaggant using high-capacity and error-correctable QR code for anti-counterfeiting of drugs. Adv Mater 24:5924–5929
+32. Rehor I et al (2017) Biodegradable microparticles for simultaneous detection of counterfeit and deteriorated edible products. Small 13:1701804
+33. Lee J et al (2014) Universal process-inert encoding architecture for polymer microparticles. Nat Mater 13:524–529
+34. Zhou C, Cao Y, Liu C, Guo W (2023) Microparticles by microfluidic lithography. Mater Today 67:178–202
+35. Decker C, Jenkins AD (1985) Kinetic approach of oxygen inhibition in ultraviolet-and laser-induced polymerizations. Macromolecules 18:1241–1244
+36. Dendukuri D, Pregibon DC, Collins J, Hatton TA, Doyle PS (2006) Continuous-flow lithography for high-throughput microparticle synthesis. Nat Mater 5:365–369
+37. Hakimi N, Tsai SSH, Cheng C-H, Hwang DK (2013) One-step two-dimensional microfluidics-based synthesis of three-dimensional particles. Adv Mater 26:1393–1398
+38. Buchwalter SL (2001) Encyclopedia of Materials: Science and Technology. Elsevier
+39. Xie R et al (2020) Glass transition temperature from the chemical structure of conjugated polymers. Nat Commun 11:893
+40. Kitahara Y et al (2012) Thermal decomposition of acrylamide from polyacrylamide. J Therm Anal Calorim 110:423–429
+41. Fox DS, Opila EJ, Nguyen QN, Humphrey DL, Lewton SM (2003) Paralinear oxidation of silicon nitride in a water-vapor/oxygen environment. J Am Ceram Soc 86:1256–1261
+42. Qadir A, Fogarassy Z, Horváth ZE, Balazsi K, Balazsi C (2018) Effect of the oxidization of Si₃N₄ powder on the microstructural and mechanical properties of hot isostatic pressed silicon nitride. Ceram Int 44:14601–14609
+43. Loehman RE, Rowcliffe DJ (1980) Sintering of Si₃N₄-Y₂O₃-Al₂O₃. J Am Ceram Soc 63:144–148
+44. Ceja-Cárdenas L, Lemus-Ruíz J, Jaramillo-Vigueras D, de la Torre (2010) D. Spark plasma sintering of α-Si₃N₄ ceramics with Al₂O₃ and Y₂O₃ as additives and its morphology transformation. J Alloys Compd 501:345–351
+45. Wang L, Tien TY, Chen IW (2003) Formation of β-silicon nitride crystals from (Si, Al, Mg, Y)(O, N) liquid: I, phase, composition, and shape evolutions. J Am Ceram Soc 86:1578–1585
+46. Shen Z, Zhao Z, Peng H, Nygren M (2002) Formation of tough interlocking microstructures in silicon nitride ceramics by dynamic ripening. Nature 417:266–269
+47. Perera DS, Mitchell DRG, Leung S (2000) High aspect ratio β-Si₃N₄ grain growth. J Eur Ceram Soc 20:789–794
+48. Huang ZK, Greil P, Petzow G (1984) Formation of silicon oxinitride from Si₃N₄ and SiO₂ in the presence of Al₂O₃. Ceram Int 10:14–17
+49. Bergman B, Heping H (1990) The influence of different oxides on the formation of Si₂N₂O from SiO₂ and Si₃N₄. J Eur Ceram Soc 6:3–8
+
+# Supplementary Files
+
+- [GraphicalAbstract.docx](https://assets-eu.researchsquare.com/files/rs-3795530/v1/76aa9314f7fb621228d0e17f.docx)
+- [SupplementaryInformation.docx](https://assets-eu.researchsquare.com/files/rs-3795530/v1/b5ac8c2a064e04315e826407.docx)
+
+Supplementary Information

343aafacc2f44222df20fc3de21cdba58896c3b95ea6f9fe56178820d7e57bdc/preprint/images_list.json

@@ -0,0 +1,58 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Anion and cation migration and electrode reaction. The ions transfer in SPEs of all-solid-state batteries using different ion acceptors as the cathode (left: anion-hosting cathode (this work), e.g., p-type polymer, right: Li+-hosting cathode).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Synthesis and redox properties of polyvinyl ferrocene. (a) The optical image, SEM image and elemental mapping results of the polyvinyl ferrocene, and the color change significantly after polymerization. (b) The frontier orbital electronic structure of ferrocene/ferrocenium, while the electron transfer in the redox reaction is marked in red. (c) FT-IR Spectroscopy of vinyl ferrocene and polyvinyl ferrocene and the inset shows the weakening of the peak intensity at 1625 cm-1. (d) The electrode redox reaction of PVF with electrolyte anions, where the theoretical capacity is based on the molecular weight of the active unit and the charge transfers numbers. (e), (f) show the capacity retention and charge-discharge curves of PVF|Li battery at 60 \u00b0C, 300 \u03bcA cm-2, respectively. The battery was cycled at 50 \u03bcA cm-2 in the first 5 cycles.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Ion aggregation in SPEs and the impact on battery performance. (a) Lithium-ion transference number performed by steady-state current method (b) Ion motion in SPEs dominated by aggregated ion cluster (left) and single ion (right), where dashed lines between ions represent short-range intracluster interactions and arrows correspond to the ion motion considered in SPEs. Considering ion cluster as noninteracting species provide an approximation more precise than the usual Nernst-Einstein equation to the tLi+, able to explain the mechanism responsible for the negative value36,39. (c) Nyquist plots of PVF|Li batteries with different anion species before cycles, and the solid line represent the impendence fitting results. (d) Charge-discharge curves of PVF|Li batteries performed at 50 \u03bcA cm-2, the theoretical capacity was marked as a dashed line.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Electrode reaction affected by anion species. (a) Cyclic voltammetry curves of PVF|Li batteries constructed with different SPEs in five cycles, and the dashed line indicate the first cycle. The gray parts show the disparity of \u2206E in the electrode reaction. (b) The anions\u2019 binding energy with ethyl ferrocenium and size trends. (c) The negative correlation between the ionic conductivity of electrolyte and \u2206E at 60 \u00b0C. (d) The electrode potential of PVF|Li batteries, calculated by the average of oxidation and reduction peak in CV results.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpg",
+        "caption": "Carrier expansion improves rate performance and cycle stability. Schematic diagram of anode morphology changes during cycling where (a) PVF and (b) LiFePO4 served as the cathode. (c) The voltage change curves of PVF|Li (top) and LFP|Li (bottom) batteries with time at 300 \u03bcA cm-2 and (d) Coulombic efficiency of PVF|Li battery over 4000 cycles (e) Magnified view of voltage change with time during LFP|Li battery cycles (f) The rate performance from 100 to 1000 \u03bcA cm-2 and (g) the comparison between this work and the other reported advanced SPEs, classified in the graph by design strategy (detail seen in Table S5).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.jpg",
+        "caption": "PVF|Li Batteries performance with insufficient ionic conductance SPEs (a) Cyclic voltammetry curves of PVF|Li battery at 0.2 mV s-1 assembled with 5 wt% SN added SPE at 30 \u00b0C (b) The cycle performance and (c) charge-discharge curves of PVF|Li battery with 10, 30, 50 \u03bcA cm-2, respectively (d) Comparison of the effective carrier number of this work and other types of polymer-based solid electrolytes, where the ionic conductivity was specifically collected at batteries operating temperature.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/[IMAGE_MATERIALS_AND_METHODS_1].png",
+        "caption": "",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

343aafacc2f44222df20fc3de21cdba58896c3b95ea6f9fe56178820d7e57bdc/preprint/preprint.md

@@ -0,0 +1,192 @@
+# Abstract
+
+The non-reactive anion migration deteriorates the limited ionic conductivity of the solid polymer electrolytes (SPEs) and accelerates solid-state batteries failure. Here, we introduce an integrated approach in which polyvinyl ferrocene (PVF) cathode encourage anions and Li<sup>+</sup> to act as effective carriers simultaneously. The concentration polarization and poor rate performance, caused by insufficient effective carriers, were addressed by the participation of anions in electrode reaction. Specifically, the PVF|Li battery matched with unmodified SPE (PEO-LiTFSI) showed 107 mAh g<sup>−1</sup> initial capacity at 100 µA cm<sup>−2</sup> and maintained 70% retention for more than 2800 cycles at 300 µA cm<sup>−2</sup> and 60°C. Moreover, the slight capacity decrease at 1000 µA cm<sup>−2</sup> and the successful batteries operation at minimal ionic conductivity (8.13×10<sup>−6</sup> S cm<sup>−1</sup>) show that the current carrying capacity of SPEs was greatly improved without complex design. This strategy weakens the strict requirements for ion conductance and interface engineering of SPEs, and provides an efficient scenario for constructing advanced polymer-based all-solid-state batteries.
+
+**Energy Engineering**  **Catalysis**  **Materials Engineering**  **solid polymer electrolytes**  **energy storage**  **batteries**
+
+# Main
+
+The safety requirement for lithium-ion batteries and the demand for lithium metal anode has prompted researchers to look for solid-state alternatives to liquid organic electrolytes<sup>1-3</sup>. Since Wright and coworkers reported the ionic conductivity of the mixture of polyethylene oxide (PEO) and alkali metal salts<sup>4</sup>, solid polymer electrolytes (SPEs) had been considered as a potential solution for constructing all-solid-state lithium secondary batteries<sup>5,6</sup>. However, the weak polymer chain motion leads to insufficient ion transport for SPEs even at high temperature<sup>3,6</sup>. Many investigators have committed to developing high ion conductive SPEs. Typical strategies include designing the polymer segment structure<sup>7,8</sup> and combining polymer matrix with ceramic<sup>9,10</sup> or inorganic solid electrolytes<sup>11-13</sup>.
+
+One critical problem that has to be solved for constructing advanced SPEs is the movement of anions. The dissociation of lithium salt in SPEs depends on the dipolar interaction between the polymer chain (e.g., -CH<sub>2</sub>-CH<sub>2</sub>-O- in PEO) and Li<sup>+</sup><sup>14,15</sup>. Similar to the liquid electrolyte, the large solvation structure significantly slows the migration of Li<sup>+</sup><sup>16,17</sup>. In contrast, the non-coordinating anions contribute most of the ionic conductance in SPEs, meaning that the majority of ion migration is irrelevant to energy generation. Hence, the rate performance of polymer-based solid-state batteries was restricted to a great extent by the limited effective carrier concentration. Another issue caused by anion migration is concentration polarization, attributed to anions aggregation on the electrode surface during cycles<sup>18,19</sup>. This polarization results in critical performance degradation, such as increasing internal resistance and decreasing operating voltage<sup>20</sup>, most importantly, promoting lithium metal dendrite growth<sup>21</sup>. The short-circuit in polymer-based solid-state batteries generally predates the combination of liquid components and commercial separators, which runs counter to the original intention of solid-state electrolytes. Single-ion conducting solid polymer electrolytes (SISPEs) was suggested to alleviate the problems caused by anion migration<sup>22,23</sup>. By constructing polymer segments with weak interaction with Li<sup>+</sup><sup>24</sup> or grafting anions to the polymer backbone<sup>25</sup>, SISPEs can achieve high t<sub>Li+</sub> (>0.9). However, this comes at the cost of ionizing less free Li<sup>+</sup> and also the low ionic conductivity due to the lack of solvation ability.
+
+The participation of anions in electrode reaction has promoted the development of attractive energy conversion systems in the liquid or extended gel phase<sup>26-28</sup>. These systems were intended as low-cost alternatives to lithium-ion batteries, focus on some key parameters such as sustainability and material availability<sup>29,30</sup>. Apart from the requirement for high electrochemical stability, the electrolyte of the anion reaction system is similar to that of lithium-ion batteries. The benefit of the effective carrier expansion is hidden by the abundant ionic conductivity of the liquid/gel electrolyte. However, for solvent-free SPEs with limited ion movement, the strategy of enhancing the correlation between ion migration and electrode reaction is expected to play a more crucial role. In this work, we introduced anion-hosting cathode to overcome the negative impact owing to non-reactive anion migration in SPEs. The ferrocene unit, anchored to the long-chain polymer, encourage anions as the effective charge carrier similar to Li<sup>+</sup>. The expansion of carriers significantly improves current carrying capacity of unmodified SPEs (Fig. 1), and avoids the short-circuit failure lead by concentration polarization. Besides, the impact of anion species on ion mobility and interacting with the cathode is also investigated in depth.
+
+## Materials design and characterization
+
+The electronic structure of the cyclopentadiene and iron atom hybrid orbital provides ferrocene with stable and reversible redox properties (Fig. 2b)<sup>31,32</sup>. To avoid the diffusion of active units in a non-flow system, we anchored the anion-hosting unit to the polymer chain by free-radical polymerization of vinyl ferrocene, as illustrated in Fig. 2a. The EDX mapping proves the homogeneous distribution of Fe element across the PVF. The polymerization process was confirmed by the Fourier transform infrared spectroscopy (FT-IR) in Fig. 2c, which shows a significant weakening of double bond vibration peaks at 1625 cm<sup>-1</sup> after polymerization. The gel permeation chromatography (GPC) results (Table S1) prove that PVF has a high molecular weight (~4800 g mol<sup>-1</sup>) with wide distribution (M<sub>w</sub>/M<sub>n</sub> = 1.71). The redox of PVF can provide a theoretical capacity of 124 mAh g<sup>-1</sup> (Fig. 2d), locating at the top level of the anion-hosting organic cathode<sup>33,34</sup>. Moreover, the theoretical redox potential of ~3.45 V vs. Li<sup>+</sup>/Li can be tolerated by most SPEs. The thermogravimetric analysis (TGA) in Fig. S1 shows that PVF did not undergo significant thermal weight loss or phase change below 300 °C, ensuring the electrode stability in the case of high temperature operation.
+
+Classic SPEs are composed of lithium salt and polymer with solvation ability. Benefiting from the high dielectric constant and chain flexibility, PEO is one of the most widely studied polymers matrix<sup>10,11,35</sup>. Based on the well-designed anion hosting material, the PVF|Li battery matched with PEO-LiTFSI electrolyte exhibits excellent cycle stability. It maintained 70 % capacity retention after 2800 cycles (Figure 2e) at 60 °C, while avoiding battery failure for more than 4000 cycles (Figure 2e, 2f). Considering the current case, where the migration of anions and cations is related to the electrodes’ reaction, we set several types of lithium salt in SPEs to obtain a deep insight into the anion electrode reaction. The ionic conductivity of the SPEs was conducted through AC impedance (seen in Fig. S2). Fig. S3 plots the ionic conductivity as a function of temperature. Owing to the high delocalized negative charge of anions, PEO-LiTFSI displays the highest ionic conductivity (3.53×10<sup>-4</sup> S cm<sup>-1</sup> at 333 K). Except for LiClO<sub>4</sub>, other electrolytes have comparable conductivity at high temperature. The differential scanning calorimetry (DSC) results of SPEs correspond to the ionic conductivity, where PEO-LiTFSI has the lowest melting point (Fig. S4).
+
+Inefficient utilization of ions in SPEs severely restrict the batteries performance. To explore the ion mobility influence in the designed systems, we measured the lithium-ion transference number (t<sub>Li+</sub>) of SPEs through the steady-state current method (results are shown in Fig. S5, Table S2). As seen in Fig. 3a, the electrolytes with LiTFSI, LiFSI and LiClO<sub>4</sub> as salts pose low t<sub>Li+</sub> which are all-around 0.1, proving that anions contribute the most ion movement in these SPEs. Note that the calculated t<sub>Li+</sub> of PEO-LiBOB electrolyte is negative (-0.38) differ from other threes. In fact, negative cation transfer numbers are not rare, and usually attributed to the presence of ionic aggregates<sup>36,37</sup>. As shown in Fig. 3b, while the negatively charged ion clusters dominate the charge transfer in electrolyte, the t<sub>Li+</sub> of SPEs can be low to negative due to the short-range interactions<sup>38,39</sup>. The domination of ion clusters could reduce the carrier loading capacity at high current density, which affects the battery performance, as will be discussed later. The AC impendence results of PVF|Li batteries shown in Fig. 3c proves that negatively charged clusters significantly increase the batteries' charge transfer resistance with PEO-LiBOB electrolytes, despite the high ionic conductivity compared to other SPEs. Generally, for a typical Li<sup>+</sup>-hosting cathode, SPE with low t<sub>Li+</sub> operate poorly. However, the anions participation in the electrode reaction breaks through the strict requirement on ion aggregation in the SPEs<sup>20</sup>. The PVF|Li batteries assembled with SPEs all exhibit reversible charge-discharge process shown in Fig. 3d, and the overpotential of batteries is controlled by the ion conductivity of SPEs. Moreover, the anion species lead a difference on operating voltage of batteries. This finding prompted us to further study of anion impact.
+
+## Anion impact on electrode reaction
+
+The electrochemical stability of the SPEs was evaluated using Li|SPE|stainless steel cells. The linear scanning voltammetry (LSV) curves in Fig. S6 shows a low redox current within a voltage window up to 4 V vs. Li<sup>+</sup>/Li, ensuring that the electrode reaction is not disturbed by electrolyte oxidation. The electrochemical behavior of PVF with anions was evaluated through cyclic voltammetry (CV) tests. As shown in Fig. 4a, the PVF cathode shows good redox reversibility provided by the active unit. Notably, the stabilization of the unit into polymer engages a crucial role in the electrode reaction reversibility. With the free ferrocene electrode (Fig. S7), even with a high content of the conductive agent, the assembled solid-state battery exhibits oxidation peak only in the first cycle. This indicates that the ion pair formed by ferrocenium and anion cannot undergo further reduction. In contrast, PVF with its long-chain structure prevents the diffusion of active materials by anchored ion pairs into the polymer. The peak potential separation (Δ<em>E</em>) shown in CV results is informative of the electrochemical reaction kinetics. Judging from Fig. 4c, the highest ionic conductivity SPE, PEO-LiTFSI, has the lowest potential gap (∆<em>E</em> = 0.103 V), while PEO-LiClO<sub>4</sub> has the highest ∆<em>E</em> (Fig. 3d), corresponding to the ionic conductance trend.
+
+The influence of anion species on the electrode reaction in liquid electrolyte, related to anion couples, was previously investigated by Redepenning et al<sup>40</sup>. The results show that the ion pairs' formation could negative shift the electrode potential from theoretical<sup>41</sup><sup>,42</sup>. To determine the ion pair effect on the electrode in SPEs, we conducted density functional theory (DFT) simulations to calculate the binding energy (BE) of different anions to the cathode. The cathode was simplified by substituting ethyl ferrocenium for the polyvinyl ferrocenium (Table S4). The calculations indicated that ethyl ferrocenium has the highest BE to ClO<sub>4</sub><sup>-</sup>, and decrease with the order FSI<sup>-</sup>, BOB<sup>-</sup>, and TFSI<sup>-</sup> (Fig. 4b). However, the experimental results are telling a different story. Apart from the PEO-LiClO<sub>4</sub>, the batteries with PEO-LiTFSI and PEO-LiBOB electrolytes exhibit high electrode potential close to theoretical with small difference, which are 3.464 and 3.466 V, respectively. And, the FSI<sup>-</sup> lead the electrode potential to drop slightly to 3.443 V. Considering the steric hindrance of the polymer chain can further explain the contradiction. Unlike the small ethyl ferrocenium monomer considered in the DFT calculation, the folded long-chain polyvinyl ferrocenium suppress the combination of large anions, weakening the effect of ion-pairing on the electrode potential. Therefore, ClO<sub>4</sub><sup>-</sup>, with the smallest structure (shown in Fig. S8) and the strongest BE among four anions (Fig. 4b), significantly shift the electrode potential to negative (3.383 V vs. Li<sup>+</sup>/Li) while the larger anions are not significantly affected by the binding effect. Before further examining the all-solid-state batteries, we evaluated the capacity and stability of the PVF electrode with liquid electrolyte (1.0 M LiPF<sub>6</sub> in EC/DMC wt %). The capacity increase in the first five cycles can be attributed to the cathode electrochemical activation, previously observed in several organic electrodes<sup>43</sup>. The initial capacity of 113 mAh g<sup>-1</sup> and high retention of capacity in 100 cycles (Fig. S9) exhibit the stable redox property of PVF. Besides, the combination of PF<sub>6</sub><sup>-</sup> and cathode also shifts voltage plateau to a lower value (~3.27 V) than the theoretical, which can be explained by the combined effects discussed above.
+
+## Enhanced batteries rate performance and anode stability
+
+Motivated by the results that the batteries with PEO-LiTFSI and PEO-LiBOB electrolytes have small polarization and high capacity (Fig. 3d), we focused on these two SPEs for further studies of long cycles. After a few cycles of cathode activation in PEO-LiBOB electrolyte, the PVF capacities were measured to be 112, 104, and 107 mAh g<sup>-1</sup> at currents of 20, 50, and 100 μA cm<sup>-2</sup>, respectively (Fig. S10). The capacity has not been significantly attenuated after a long period cycling. However, as shown in Fig. S11, the battery with PEO-LiBOB electrolyte shows substantial polarization and significant capacity fade at higher current density (~67 mAh g<sup>-1</sup> at 300 μA cm<sup>-2</sup>), which is as weak as the SPEs containing ClO<sub>4</sub><sup>-</sup> and FSI<sup>-</sup> (Fig. S12), despite the ionic conductance disparity among threes. The poor rate performance of batteries with LiBOB salt can be ascribed to the formation of ions aggregations demonstrated in the previous section. The sluggish electrode reaction deteriorates the battery performance at high current densities.
+
+Benefit from high ionic conductivity, large anion structure and the ability to prevention aggregation, the battery with PEO-LiTFSI showed excellent cycle stability and rate performance (Fig. S13, S14). Specifically, the initial capacity reaches 97 mAh g<sup>-1</sup> at 300 μA cm<sup>-2</sup> and maintained 90 % after 1000 cycles (Fig. 2e). Benefiting from the extended carriers, the battery exhibits good rate performance, with capacities of 108, 107, 97, and 94 mAh g<sup>-1</sup> at current densities of 100, 200, 300, and 500 μA cm<sup>-2</sup>, respectively (Fig. 5f). Even when the current increases to 1 mA cm<sup>-2</sup>, which is intolerable by many reported advanced SPEs, the battery still maintains 78 mAh g<sup>-1</sup> capacity with a stable plateau. The PVF electrode exhibits pseudo-capacitance, which also contributes to the excellent rate performance to a certain extent (Fig. S15). It is worth noting that the electrode reaction rate is not the rate-determining step of most solid-state device owing to the limited ion transport capability of solid electrolytes. Despite using a large amount of the conductive agent to overcome PVF low conductivity, the result of the control experiment proves that the capacity contribution of the conductive agent at the tested voltage range is negligible (Fig. S16). The decrease of conductive agent proportion reduces the battery's rate performance significantly (Fig. S17). However, this does not affect the effectiveness of the strategy proposed in this report. The extended carrier makes the ionic conductivity of SPEs no longer a bottleneck of the battery’s performance at high current loading. The addition of single-walled carbon nanotubes in the electrode can significantly increase the proportion of active materials with the reduced polarization (Fig. S18), which provides massive space for the high conductive anion-hosting electrode.
+
+Concentration polarization could induce serious consequences in SPEs when only Li<sup>+</sup> act as effective carriers, significantly accelerate anode degradation and battery failure especially at high current density (Fig. 5b). Taking the charging process as an example, the Li<sup>+</sup>-hosting cathode (e.g., LiFePO<sub>4</sub>) undergoes an anodic reaction. In an ideal situation, the ion number in SPEs keep constant throughout the process. The concentration of Li<sup>+</sup> increase on the cathode side and decrease on the counter. This leads to a salt concentration gradient and ion diffusion barriers (Fig. S19a). The SPEs can be regarded as liquid electrolyte with extremely high viscosity. The minimal ion diffusion results in a stable massive concentration gradient, which is more severe than that in liquid. In the present work where anions were involved in energy storage, the ions and ions clusters migration under electric field is similar to that of LiFePO<sub>4</sub>. The difference between two cases is that the PVF electrode reaction consumes anions similar to Li<sup>+</sup> on the anode in SPEs (Fig. 5a). Therefore, the distribution of SPEs salt concentration is homogeneous, facilitating the ions' diffusion and avoiding the detrimental consequences of the concentration polarization (Fig. S19b).
+
+To verify that minimizing the concentration polarization could effectively enhance the lithium anode stability, we tested LiFePO<sub>4</sub>|PEO-LiTFSI|Li batteries under the same condition (Fig. S20). As shown in Fig. 5c, 5e, a micro short circuit occurred in LiFePO<sub>4</sub>|Li at the 24th hour at 300 μA cm<sup>-2</sup>.<sup></sup> The short circuit has not been repaired in subsequent cycles, leading to a continuous decayed in coulombic efficiency (Fig. S20b). Similar failures are common in other polymer-based solid-state batteries, even for SPEs with improved ionic conductivity and interface properties. In contrast, the coulombic efficiency of the PVF|PEO-LiTFSI|Li battery maintained about 99.7 % with no short circuit observed over 4000 cycles at 300 μA cm<sup>-2</sup> (Fig. 5c, d), which is the maximum level for most polymer-based solid-state-batteries. The excellent cycle stability proves the success of our concept in controlling the polarization and suppressing lithium anode deterioration.
+
+## Performance with insufficient ionic conductivity
+
+As an unmodified SPE, the combination of PEO-LiTFSI does not have advantages in terms of ionic conductivity and interfacial properties, usually performed as negative comparison. However, the expansion of the effective carrier through the anion-hosting cathode eliminates the demand of complex design for advanced SPEs. We compared the rate performance of this work with the reported advanced SPEs in Fig. 5g<sup>35,44-51</sup>. The strategy we proposed strongly enhances capacity retention at high current density with simple PEO-LiTFSI without any modifications, better than other SPEs of complex design (details in Table S5).
+
+The improvement of the current-carrying capacity of SPEs inspired us to examine the battery performance with very low ionic conductance. PEO-LiTFSI exhibits extremely limited ionic conductivity of 2.65×10<sup>-7</sup> S cm<sup>-1</sup> at 30 °C, far from the usual battery test requirement. In this condition, PVF|Li batteries exhibited capacities of 83, 60, 48 mAh g<sup>-1</sup> at 10, 30, and 50 μA cm<sup>-2</sup> with increased polarization, respectively (Fig 6c). The inferior ionic conductivity leads the unsatisfactory performance. As an ideal additive, succinonitrile (SN) could strengthen the segment movement ability of polymer thereby enhancing SPEs’ ionic conductivity<sup>52,53</sup>. After doping PEO-LiTFSI with 5% SN, the ionic conductivity increased to 8.13×10<sup>-6</sup> S cm<sup>-1</sup> at 30 °C (Fig. S21a). Therefore, the battery assembled by PEO-LiTFSI-SN electrolyte show a lower impedance compared with the electrolyte without plasticizer (Fig. S21b). The CV curves exhibit stable and reversible redox performance (Fig. 6a). As shown in Fig. 6b, 6c, the recorded capacities at 10, 30, and 50 μA cm<sup>-2</sup> rise to 94, 86, and 70 mAh g<sup>-1</sup>, respectively. The PVF|Li batteries with these two SPEs both maintain more than 100 cycles at 30 μA cm<sup>-2</sup> without significant capacity decayed. The anion-hosting cathode makes full use of each dissociated ion in electrolytes, resulting in a battery system with high tolerance to SPEs with very low ionic conductivity. In short, this strategy avoids plenty problems devoted to the low ionic conductivity and utilization faced by previous reported SPEs<sup>5,6,23</sup> (Fig. 6d), reinforcing the correlation between ion movement and electrode reaction.
+
+# Discussion
+
+In summary, we developed advanced polymer-based solid-state batteries by inducing anions as effective carriers simultaneously with Li<sup>+</sup>. The anion-hosting cathode PVF put the entire ion movement of SPEs into energy storage, which produces an updated rate performance and promotes batteries operation at very limited ionic conductance. In addition, the ultra-stable cycles of PVF|Li batteries prove that the anode deterioration, mainly contributed by concentration gradients, were avoided effectively by reactive anion migration, which is essential in building safer metal anode batteries. Besides, experiments and theoretical calculations clarified the effects of anion structure, binding energy and ion aggregation on battery performance. This work provides a pioneering strategy for the design of advanced solid-state energy storage systems. However, future investigations into high capacity and conductivity anion-hosting cathode are certainly warranted. Since the migration and aggregation of anions differ from Li<sup>+</sup> in most SPEs, the disparity of reaction status between cathode and anode could present some minor complications and addressing this issue could be the subject of the following study.
+
+# Materials And Methods
+
+## Materials
+
+PEO (Mw=6×10⁵), LiTFSI (99%), LiFSI (99%), LiClO₄ (99.9%), LiBOB (98%) were purchased from Aladdin. Vinyl ferrocene (98%) was purchased from Meryer. 2,2'-azoisobutyronitrile (AIBN) was recrystallized before use. PVDF (HSV900), conductive agent (Super P) and separator (Celgard 2325) were fully dried before use. LiFePO₄ and liquid electrolyte were purchased from DoDoChem. N-methyl pyrrolidone (NMP) and acetonitrile were of analytical grade and used directly without further purification. Dry toluene was obtained from a VSPS-5 solvent purification system.
+
+## Synthesis of polyvinyl ferrocene and solid polymer electrolytes
+
+Polyvinyl ferrocene (PVF) was synthesized by free-radical polymerization. In a typical process, vinyl ferrocene was dissolved in dry toluene, and AIBN was used as the initiator. The ratio of [monomer] / [Initiator] = 100, [monomer]₀ = 2 M. The reaction was continued at 60 °C for 48 hours. The obtained dark red solution was washed with a large amount of methanol and dried under vacuum to obtain yellow powder.
+
+Polyethylene oxide and lithium salts were dissolved in acetonitrile. The obtained solution was coated on a polytetrafluoroethylene plate and dried under reduced pressure to obtain a self-supported film. The electrolyte membrane was kept in Ar atmosphere glove box to prevent moisture contamination. The solid electrolyte containing succinonitrile adopts the same preparation method.
+
+## Materials characterization
+
+The thermal gravimetric analysis was carried out with METTLER TOLEDO TGA/DSC³ at a temperature range of 30–800 °C under nitrogen atmosphere, with a heating rate of 10 K min⁻¹. Fourier transform infrared (FT-IR, Bruker Tensor 27) were recorded between 400 and 4000 cm⁻¹. A field emission scanning electron microscope (FE-SEM, SU-6600) equipped with an energy dispersive spectrometer was used to characterize the samples’ morphology. All the electrochemical characterization was performed using the electrochemical workstation (PARSTAT 1000 and CHI 660E).
+
+## Electrochemical measurements
+
+The active material (PVF), conductive agent (Super P) and binder (PVDF) were dispersed in N-methyl pyrrolidone at a ratio of 4:5:1 or 6:3:1. The dispersed slurry was coated on aluminum foil and dried under reduced pressure at 60 °C. The prepared electrodes were stored in a glovebox until use. The same method was used to prepare LiFePO₄ electrodes with a ratio of 7:2:1 and the Super P electrode in a ratio of 85:15 (conductive agent: binder). The mass loading of the active material was controlled at 1.0 mg cm⁻². The electrodes were cut into discs (d = 12 mm) for subsequent testing. The cells assembly was carried out in a glove box filled with argon gas (H₂O, O₂ < 0.1 ppm), using metallic lithium as the counter electrode.
+
+The ionic conductivity of the electrolyte was measured by sandwiching the polymer electrolyte film between two stainless steel electrodes and then record the electrochemical impedance. The ionic conductivity was calculated by the following equation
+
+[IMAGE_MATERIALS_AND_METHODS_1]
+
+*l* is the thickness of the polymer electrolyte, *R* is the bulk resistance of the polymer electrode, and *S* is the electrolyte area.
+
+The electrochemical impedance spectroscopy (EIS) of the batteries was tested under an open circuit with a frequency range of 10⁻¹–10⁵ Hz. The cyclic voltammetry (CV) and linear sweep voltammetry (LSV) test were performed at a scan rate of 0.2 mV s⁻¹. Galvanostatic charge/discharge tests were performed within a voltage window of 2.8~4 V (v.s. Li/Li⁺) on a LAND battery tester.
+
+## DFT computational methods
+
+All of the optimization and frequency calculations were carried out on the B3LYP/6-311+G(d,p) level, and the optimized structures are free from imaginary frequency. The binding energies were calculated by E_binding = E_compound - E_anion - E_cation. The results are shown in Table S4. All of the above calculations were carried out with Gaussian 16 program package.
+
+# References
+
+1. M. Armand, J. M. Tarascon, Building better batteries. *Nature* **451**, 652–657 (2008).
+
+2. J. B. Goodenough, K. S. J. A. C. S. Park, The Li-ion rechargeable battery: A perspective. *J. Am. Chem. Soc.* **135**, 1167–1176 (2013).
+
+3. A. Manthiram, X. W. Yu, S. F. Wang, Lithium battery chemistries enabled by solid-state electrolytes. *Nat. Rev. Mater.* **2**, 16103 (2017).
+
+4. D. E. Fenton, J. M. Parker, P. V. Wright, Complexes of alkali-metal ions with poly(ethylene oxide). *Polymer* **14**, 589–589 (1973).
+
+5. J. Mindemark, M. J. Lacey, T. Bowden, D. Brandell, Beyond PEO-alternative host materials for Li⁺-conducting solid polymer electrolytes. *Prog. Polym. Sci.* **81**, 114–143 (2018).
+
+6. D. Zhou, D. Shanmukaraj, A. Tkacheva, M. Armand, G. X. Wang, Polymer electrolytes for lithium-based batteries: advances and prospects. *Chem-Us* **5**, 2326–2352 (2019).
+
+7. R. Khurana, J. L. Schaefer, L. A. Archer, G. W. Coates, Suppression of lithium dendrite growth using cross-linked polyethylene/poly(ethylene oxide) electrolytes: A new approach for practical lithium-metal polymer batteries. *J. Am. Chem. Soc.* **136**, 7395–7402 (2014).
+
+8. J. Lopez, Y. Sun, D. G. Mackanic, M. Lee, A. M. Foudeh, M. S. Song, Y. Cui, Z. Bao, A dual-crosslinking design for resilient lithium-ion conductors. *Adv. Mater.* **30**, 1804142 (2018).
+
+9. W. Liu, N. Liu, J. Sun, P. C. Hsu, Y. Li, H. W. Lee, Y. Cui, Ionic conductivity enhancement of polymer electrolytes with ceramic nanowire fillers. *Nano Lett.* **15**, 2740–2745 (2016).
+
+10. W. Liu, S. W. Lee, D. Lin, F. Shi, S. Wang, A. D. Sendek, Y. Cui, Enhancing ionic conductivity in composite polymer electrolytes with well-aligned ceramic nanowires. *Nat. Energy* **2**, 17035 (2017).
+
+11. H. Zhai, P. Xu, M. Ning, Q. Cheng, J. Mandal, Y. Yang, A flexible solid composite electrolyte with vertically aligned and connected ion-conducting nanoparticles for lithium batteries. *Nano Lett.* **17**, 3182–3187 (2017).
+
+12. Z. Xue, T. Liu, S. Zhang, H. Xin, B. Xu, Y. Lin, B. Xu, L. Li, C. W. Nan, S. Yang, Synergistic coupling between Li₆.₇₅La₃Zr₁.₇₅Ta₀.₂₅O₁₂ and poly(vinylidene fluoride) induces high ionic conductivity, mechanical strength and thermal stability of solid composite electrolytes. *J. Am. Chem. Soc.* **139**, 13779 (2017).
+
+13. Jin, Zheng, Dr., Mingxue, Tang, Prof., Yan-Yan, Hu, Lithium ion pathway within Li₇La₃Zr₂O₁₂-polyethylene oxide composite electrolytes. *Angew. Chem. Int. Ed.* **55**, 12538–12542 (2016).
+
+14. M. A. Ratner, D. F. Shriver, Ion-transport in solvent-free polymers. *Chem. Rev.* **88**, 109–124 (1988).
+
+15. M. Li, C. Wang, Z. Chen, K. Xu, New concepts in electrolytes. *Chem. Rev.* **120**, 6783–6819 (2020).
+
+16. O. Borodin, G. D. Smith, Mechanism of ion transport in amorphous poly(ethylene oxide)/LiTFSI from molecular dynamics simulations. *Macromolecules* **39**, 1620–1629 (2006).
+
+17. D. Diddens, A. Heuer, O. Borodin, Understanding the Lithium Transport within a Rouse-Based Model for a PEO/LiTFSI Polymer Electrolyte. *Macromolecules* **43**, 2028–2036 (2010).
+
+18. I. Rey, J. Lassègues, P. Baudry, H. Majastre, Study of a lithium battery by confocal Raman microspectrometry. *Electrochimica Acta* **43**, 1539–1544 (1998).
+
+19. I. Rey, J. L. Bruneel, J. Grondin, L. Servant, J. C. Lassegues, Raman spectroelectrochemistry of a lithium/polymer electrolyte symmetric cell. *J. Electrochem. Soc.* **145**, 3034–3042 (1998).
+
+20. M. Doyle, T. F. Fuller, J. Newman, The Importance of the Lithium Ion Transference Number in Lithium Polymer Cells. *Electrochimica Acta* **39**, 2073–2081 (1994).
+
+21. M. Rosso, C. Brissot, A. Teyssot, M. Dolle, L. Sannier, J. M. Tarascon, R. Bouchetc, S. Lascaud, Dendrite short-circuit and fuse effect on Li/polymer/Li cells. *Electrochimica Acta* **51**, 5334–5340 (2006).
+
+22. R. Bouchet, S. Maria, R. Meziane, A. Aboulaich, L. Lienafa, J. P. Bonnet, T. N. T. Phan, D. Bertin, D. Gigmes, D. Devaux, R. Denoyel, M. Armand, Single-ion BAB triblock copolymers as highly efficient electrolytes for lithium-metal batteries. *Nat. Mater.* **12**, 452–457 (2013).
+
+23. H. Zhang, C. Li, M. Piszcz, E. Coya, T. Rojo, L. M. Rodriguez-Martinez, M. Armand, Z. Zhou, Single lithium-ion conducting solid polymer electrolytes: advances and perspectives. *Chem. Soc. Rev.* **46**, 797–815 (2017).
+
+24. K. M. Diederichsen, E. J. Mcshane, B. D. Mccloskey, The most promising routes to a high Li⁺ transference number electrolyte for lithium ion batteries. *ACS Energy Lett.* **2**, 2563–2575 (2017).
+
+25. D. Zhou, A. Tkacheva, X. Tang, B. Sun, D. Shanmukaraj, P. Li, F. Zhang, M. Armand, G. Wang, Stable conversion chemistry-based lithium metal batteries enabled by hierarchical multifunctional polymer electrolytes with near-single ion conduction. *Angew. Chem. Int. Ed.* **58**, 6001–6006 (2019).
+
+26. C. Yang, J. Chen, X. Ji, T. P. Pollard, X. Lu, C. J. Sun, S. Hou, Q. Liu, C. Liu, T. Qing, Aqueous Li-ion battery enabled by halogen conversion–intercalation chemistry in graphite. *Nature* **569**, 245–250 (2019).
+
+27. M. C. Lin, M. Gong, B. Lu, Y. Wu, D. Y. Wang, M. Guan, M. Angell, C. Chen, J. Yang, B. J. Hwang, H. Dai, An ultrafast rechargeable aluminium-ion battery. *Nature* **520**, 325–328 (2015).
+
+28. X. F. Xu, K. Lin, D. Zhou, Q. Liu, X. Y. Qin, S. W. Wang, S. He, F. Y. Kang, B. H. Li, G. X. Wang, Quasi-solid-state dual-ion sodium metal batteries for low-cost energy storage. *Chem.* **6**, 902–918 (2020).
+
+29. W. Meng, C. Jiang, S. Zhang, X. Song, Y. Tang, H. M. Cheng, Reversible calcium alloying enables a practical room-temperature rechargeable calcium-ion battery with a high discharge voltage. *Nat. Chem.* **10**, 667–672 (2018).
+
+30. G. Chen, F. Zhang, Z. Zhou, J. Li, Y. Tang, A flexible dual‐ion battery based on PVDF‐HFP‐modified gel polymer electrolyte with excellent cycling performance and superior rate capability. *Adv. Energy Mater.* **8**, 1801219 (2018).
+
+31. Y. Zhao, Y. Ding, J. Song, G. Li, G. B. Dong, J. B. Goodenough, G. H. Yu, Sustainable electrical energy storage through the ferrocene/ferrocenium redox reaction in aprotic electrolyte. *Angew. Chem. Int. Ed.* **53**, 11036–11040 (2014).
+
+32. H. Bo, C. Debruler, Z. Rhodes, T. L. Liu, Long-cycling aqueous organic redox flow battery (AORFB) toward sustainable and safe energy storage. *J. Am. Chem. Soc.* **139**, 1207–1214 (2016).
+
+33. Z. P. Song, H. S. Zhou, Towards sustainable and versatile energy storage devices: an overview of organic electrode materials. *Energy Environ. Sci.* **6**, 2280–2301 (2013).
+
+34. Y. Lu, J. Chen, Prospects of organic electrode materials for practical lithium batteries. *Nat. Rev. Chem.* **4**, 127–142 (2020).
+
+35. J. Wan, J. Xie, X. Kong, Z. Liu, K. Liu, F. Shi, A. Pei, H. Chen, W. Chen, J. Chen, X. Zhang, L. Zong, J. Wang, L. Q. Chen, J. Qin, Y. Cui, Ultrathin, flexible, solid polymer composite electrolyte enabled with aligned nanoporous host for lithium batteries. *Nat. Nanotechnol.* **14**, 705–711 (2019).
+
+36. T. Xie, A. France-Lanord, Y. Wang, Y. Shao-Horn, J. C. Grossman, Graph dynamical networks for unsupervised learning of atomic scale dynamics in materials. *Nat. Commun.* **10**, 2667 (2019).
+
+37. H. K. Kim, N. P. Balsara, V. Srinivasan, Continuum description of the role of negative transference numbers on ion motion in polymer electrolytes. *J. Electrochem. Soc.* **167**, 110559 (110511pp) (2020).
+
+38. D. M. Pesko, K. Timachova, R. Bhattacharya, M. C. Smith, I. Villaluenga, J. Newman, N. P. Balsara, Negative transference numbers in poly(ethylene oxide)-based electrolytes. *J. Electrochem. Soc.* **164**, E3569–E3575 (2017).
+
+39. A. France-Lanord, J. C. Grossman, Correlations from ion-pairing and the Nernst-Einstein equation. *Phys. Rev. Lett.* **122**, 136001 (2018).
+
+40. J. Redepenning, E. Castro-Narro, G. Venkataraman, E. Mechalke, Influence of supporting electrolyte activity on formal potentials measured for dissolved internal standards in acetonitrile. *J. Electrochem. Soc.* **498**, 192–200 (2001).
+
+41. K. Uosaki, Y. Sato, H. Kita, Electrochemical characteristics of a gold electrode modified with a self-assembled monolayer of ferrocenylalkanethiols. *Langmuir* **7**, 1510–1514 (1991).
+
+42. G. Inzelt, L. Szabo, The effect of the nature and the concentration of counter ions on the electrochemistry of poly(vinylferrocene) polymer film electrodes. *Electrochimica Acta* **31**, 1381–1387 (1986).
+
+43. Y. Wang, Y. Deng, Q. Qu, X. Zheng, J. Zhang, G. Liu, V. S. Battaglia, H. Zheng, Ultrahigh-capacity organic anode with high-rate capability and long cycle life for lithium-ion batteries. *ACS Energy Lett.* **2**, 2140–2148 (2017).
+
+44. W. Zhou, Z. Wang, Y. Pu, Y. Li, J. B. Goodenough, Double-layer polymer electrolyte for high-voltage all-solid-state rechargeable batteries. *Adv. Mater.* **31**, 1805574 (2019).
+
+45. H. Wang, Q. Wang, X. Cao, Y. He, K. Wu, J. Yang, H. Zhou, W. Liu, X. Sun, Thiol‐branched solid polymer electrolyte featuring high strength, toughness, and lithium ionic conductivity for lithium‐metal batteries. *Adv. Mater.* **32**, 2001259 (2020).
+
+46. D. G. Mackanic, W. Michaels, M. Lee, D. W. Feng, J. Lopez, J. Qin, Y. Cui, Z. N. Bao, Crosslinked poly(tetrahydrofuran) as a loosely coordinating polymer electrolyte. *Adv. Energy Mater.* **8**, 1800703 (2018).
+
+47. Y. Liu, R. Hu, D. Zhang, J. Liu, M. Zhu, Constructing Li-rich artificial SEI layer in alloy-polymer composite electrolyte to achieve high ionic conductivity for all solid-state lithium metal batteries. *Adv. Mater.* **33**, 2004711 (2021).
+
+48. C. K. Zhang, Z. H. Niu, J. Bae, L. Y. Zhang, Y. Zhao, G. H. Yu, Polyeutectic-based stable and effective electrolytes for high-performance energy storage systems. *Energy Environ. Sci.* **14**, 931–939 (2021).
+
+49. Q. Liu, D. Zhou, D. Shanmukaraj, P. Li, F. Y. Kang, B. H. Li, M. Armand, G. X. Wang, Self-Healing Janus interfaces for high-performance LAGP-based lithiummetal batteries. *ACS Energy Lett.* **5**, 1456–1464 (2020).
+
+50. P. Jaumaux, Q. Liu, D. Zhou, X. Xu, T. Wang, Y. Wang, F. Kang, B. Li, G. Wang, Deep‐Eutectic‐solvent‐based self‐healing polymer electrolyte for safe and long‐life lithium‐metal batteries. *Angew. Chem. Int. Ed.* **132**, 9219–9227 (2020).
+
+51. M. Martinez-Ibaez, E. Sanchez-Diez, L. Qiao, Y. Zhang, H. J. A. F. M. Zhang, Unprecedented improvement of single Li-ion conductive solid polymer electrolyte through salt additive. *Adv. Funct. Mater.* **30**, 2000455 (2020).
+
+52. P. J. Alarco, Y. Abu-Lebdeh, A. Abouimrane, M. Armand, The plastic-crystalline phase of succinonitrile as a universal matrix for solid-state ionic conductors. *Nat. Mater.* **3**, 476–481 (2004).
+
+53. L. Z. Fan, Y. S. Hu, A. J. Bhattacharyya, J. Maier, Succinonitrile as a Versatile Additive for Polymer Electrolytes. *Adv. Funct. Mater.* **17**, 2800–2807 (2010).
+
+# Supplementary Files
+
+- [SupplementaryMaterials.docx](https://assets-eu.researchsquare.com/files/rs-583179/v1/b16e6589a8012a350f4875d3.docx)  
+  Supplementary Materials

3496b010fadd7d3394e5b61e1c2c42ff7e5a30126c294ca855e71942bda8ed27/preprint/images_list.json

@@ -0,0 +1,34 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Timeseries of the model boundary conditions (a-b), proxy datasets of Saharan moisture availability (with potential NAHPs labelled) and Mediterranean sapropels (c-h), and HadCM3BB-v1.0 model output (i-l) for 0-800kyr BP. a) Precession and eccentricity orbital parameters, b) CO2 concentration (ppm) and ice volume (106km3) as calculated by the ice sheet model of de Boer, et al. 1 c) marine-isotope stages, d) North African humidity/aridity index derived from the Eastern Mediterranean2, e) dust ln[Zr/Rb] ratio derived from a marine core off the Western Sahara3, f) D\u00adwax data derived from the Gulf of Aden4, g) stacked sapropel record which indicates humid phases derived from three cores (964, 966, 969) in the Eastern Mediterranean5, h) estimate of past NAHP occurrence based on Eastern Mediterranean data6, i) the numbered NAHPs identified from the model data (defined as where Saharan precipitation exceeds one standard deviation above the mean; 317mm/yr) and possible NAHPs (?) as indicated by observations, j) modelled Saharan (15:30\u00b0N, -15:35\u00b0E) annual precipitation (mm/yr) with NAHPs marked, k) a standardised West African Monsoon index7, l) latitude (\u00b0N) of maximum North African (-10:30\u00b0N, -15:35\u00b0E) summer (JJA) precipitation.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Composite climatologies during periods of precession maxima (PMax; right panels), the 20 NAHPs identified in Figure 1j (middle panels), and the anomaly between the two (NAHP-PMax; right panels). a) Annual average precipitation (mm/yr) and summer ITCZ position. b) West Saharan JJA latitude/height plots for the region identified in pink, showing lagrangian tendency of air pressure (filled contours; pa/s) where negative (positive) values represent upward (downward) motion, and zonal wind (contours; m/s) where solid/positive (dashed/negative) contours represent westward (eastward) air flow. The major components of the WAM are identified; SHL \u2013 Saharan heat low, AEJ and TEJ - African and Tropical Easterly Jets, STWJ - sub-tropical westerly jet. c) West Saharan JJA latitude/height plots showing temperature (filled contours; \u00b0C), geopotential height (black contours; m) and relative humidity (green contours; %). Solid (dashed) contours represent positive (negative) anomalies. Anomalies are 95% confident according to a student t-test.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Vegetation reconstructions during composite periods of precession maxima (left panels) and the 20 NAHPs identified in Figure 1j (right panels). a) Output from the HadCM3BB-v1.0 dynamic global vegetation model (TRIFFID), showing only the most abundant plant function type (PFT) in each grid square, b) reconstruction utilising the North African vegetation classification of Larrasoa\u00f1a, et al. 8 which demonstrates regional vegetation types.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Saharan precipitation timeseries and associated wavelet plots for the three forcing sensitivity experiments; Orb_Only, Orb_GHG, and NAHP_All. a) Annual Saharan precipitation timeseries, b-d) associated morlet wavelet power spectrum for the named experiment, the shaded areas highlight regions greater than 95% confidence for a red-noise process.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

3496b010fadd7d3394e5b61e1c2c42ff7e5a30126c294ca855e71942bda8ed27/preprint/preprint.md

@@ -0,0 +1,287 @@
+# Abstract
+
+The Sahara region has experienced periodic wet periods over the Quaternary and beyond. These North African Humid Periods (NAHPs) are astronomically paced by precession which controls the position of the African monsoon system. However, most IPCC-class climate models cannot generate enough precipitation to reconcile the magnitude of these events and so the driving mechanisms remain poorly constrained. Here, we present an 800kyr climate dataset produced using a recently developed version of the HadCM3B coupled climate model that simulates 20 NAHPs over the past 800kyr which have good agreement with the timing and amplitude of NAHPs identified in proxy data. Our results confirm that precession determines their pacing, but we identify that their amplitude is strongly linked to eccentricity via its control over ice sheet extent. During glacials, cooling due to enhanced ice-sheet albedo suppresses the amplitude of the NAHPs during periods of precession minima. Our results highlight the importance of both precession and eccentricity, and the role of high latitude processes in determining the timing and amplitude of the NAHPs. This may have implications for the out of Africa dispersal of plants and animals throughout the Quaternary.
+
+Earth and environmental sciences/Climate sciences/Palaeoclimate  
+Earth and environmental sciences/Climate sciences/Climate change/Climate and Earth system modelling  
+Earth and environmental sciences/Climate sciences/Atmospheric science/Atmospheric dynamics  
+Earth and environmental sciences/Climate sciences/Cryospheric science  
+Earth and environmental sciences/Ecology/Animal migration
+
+# Introduction
+
+There is widespread palaeoclimatological evidence indicating that the Sahara has experienced wetter periods in the past, with proliferation of vegetation, rivers and lakes into what is now desert<sup>1–16</sup>. Currently 230 of these North African humid periods (NAHPs) have been identified in proxy records over the past 8 myr<sup>8</sup>. They were ultimately driven by the precession cycle, with periods of increased boreal summer insolation intensifying the African monsoon system<sup>6,17−21</sup>. Internal biogeophysical feedbacks then amplify the external orbital forcing and further enhance precipitation<sup>15,22−25</sup>. Although precession is evident in numerous monsoon proxies<sup>12,13,20</sup>, some records show gaps at intervals of precession minima termed “skipped beats”<sup>12</sup>. These coincide with glacial periods, suggesting that the monsoon response could be modulated by eccentricity and/or obliquity. The glacial boundary conditions, namely CO<sub>2</sub> and the extent and shape of the ice sheets may also influence the African monsoon response to insolation change<sup>26,27</sup>. Problematically, most climate models fail to reproduce the precipitation change seen in the Holocene NAHP<sup>28–30</sup>. Furthermore studies investigating the role of glacial boundary conditions are not conclusive, with some indicating that the NAHPs are sensitive to ice sheet extent<sup>9,15,18</sup>, whereas other conclude that they have little impact<sup>6,19,31</sup>. Therefore, our ability to understand the mechanisms of the NAHPs remains poorly constrained.
+
+Here, we address these outstanding research questions using a recently developed version of a coupled general circulation model (HadCM3BB-v1.0, see methods) that enables us to simulate the dynamics and spatial footprint of NAHP events over the past 800kyr. This allows us to clarify the forcings and mechanisms that drive the NAHPs and the skipped beats. Our results show that the NAHPs are paced by precession due to its influence on the West African Monsoon (WAM) and regional jet streams. Glacial CO<sub>2</sub> concentrations do not significantly impact NAHP timing or amplitude. In contrast, we show that extensive ice sheets during glacials suppress NAHPs, this drives the skipped beats and elucidates a high latitude control on the NAHPs. Eccentricity therefore indirectly forces the NAHPs via its control over northern hemisphere glaciation.
+
+# Simulating The Nahps
+
+Climate models largely fail to reproduce the intensification of NAHP precipitation implied by observations <sup>28–30,32−35</sup>. This disparity between models and observations has persisted for decades across numerous climate models, and has been linked to misrepresentation of vegetation, ocean and dust feedbacks <sup><span citationid="CR24" class="CitationRef">24</span>, <span citationid="CR25" class="CitationRef">25</span>, <span citationid="CR36" class="CitationRef">36</span>, <span citationid="CR37" class="CitationRef">37</span></sup>. How these could be reconfigured in models remains unclear <sup><span citationid="CR38" class="CitationRef">38</span>, <span citationid="CR39" class="CitationRef">39</span></sup>. Instead, recent work has modified convection and land surface parameterisations <sup><span citationid="CR40" class="CitationRef">40</span>, <span citationid="CR41" class="CitationRef">41</span></sup>. This palaeo-conditioning was shown to markedly improve the simulation of the Holocene NAHP <sup><span citationid="CR40" class="CitationRef">40</span></sup>, however it has not been tested beyond this time period.
+
+Here we utilise the HadCM3B coupled climate model <sup><span citationid="CR42" class="CitationRef">42</span></sup> and incorporate the updated atmospheric parameters and vegetation parameterisations identified in Hopcroft, et al. <sup><span citationid="CR41" class="CitationRef">41</span></sup> in order to palaeo-condition the model and constrain NAHP timing and mechanisms over the past 800kyr (methods). The model is herein termed HadCM3BB-v1.0. We performed 219 snapshot simulations with results splined to a continuous timeseries on a 100 year timestep (methods). Although there are limitations to the snapshot approach (supplementary information), it has been shown to accurately simulate millennial palaeoclimate variability <sup><span citationid="CR43" class="CitationRef">43</span></sup>. Moreover, it has advantages over accelerated approaches because it is not distorting the underlying physics. Model validation (supplementary information) confirms that our methodology and HadCM3BB-v1.0 accurately reproduces present day and long-term patterns in global climate. Furthermore, HadCM3BB-v1.0 more accurately simulates mid-Holocene precipitation in the Sahara and other regions, not only relative to HadCM3B, but also against the majority of CMIP6 models <sup><span citationid="CR29" class="CitationRef">29</span></sup> (Extended Fig. <span class="InternalRef" refid="Fig1">1</span>).
+
+# Timing, Amplitude And Spatial Pattern Of The Nahps
+
+The timeseries for Saharan precipitation (15N:30°N, -15:35°E; Fig. 1j) indicates 20 NAHPs (Fig. 1i) over the past 800kyr, defined as where precipitation exceeds one standard deviation above the mean (317 mm/yr). In all cases, the timing of the 20 NAHPs correspond to precession minima (Fig. 1a), with 17 possible precession minima periods where the model simulates a ‘skipped beat’, although precipitation still increases during these periods compared to precession maxima (Fig. 1j, Extended Fig. 4). NAHP timing correlates well with those inferred from Mediterranean Sapropels<sup>7, 44</sup> and proxy reconstructions of North African humidity<sup>2</sup>, Gulf of Aden leaf-wax isotopes<sup>13</sup>, and west Saharan dust records<sup>1</sup> (Fig. 1d-h).
+
+The average NAHP Saharan precipitation is 410mm/yr (Fig. 1j). This varies spatially (Fig. 2a, Extended Fig. 2), with west Saharan (15:30°N, -15:15°E) average NAHP precipitation of 552mm/yr and maximum amplitude of 763mm/yr (NAHP 6). In contrast there is only marginal increase in North-east Africa and the Arabian Peninsula (Fig. 2a). Precipitation increase occurs contemporaneously in the Western and Eastern Sahara (Extended Fig. 2). The annual increase is dominated by anomalies in the summer months, with a small increase in winter precipitation in the Iberian Peninsula and Western Mediterranean, albeit not to the extent identified in other modelling studies<sup>6, 18</sup>. Proxy evidence for the Holocene NAHP indicates an increase ranging from ~400mm/yr (15–30°N, -10:30°E)<sup>45</sup> to 640mm/yr in the Western Sahara (15-33N, -15:0E)<sup>14</sup>. This compares to a modelled 494mm/yr and 584mm/yr for the same regions for 6-11kyr BP.
+
+During the NAHPs, HadCM3BB-v1.0 simulates the proliferation of forest to 18°N, and C4 grasses and shrubs across the central and western Sahara (Fig. 3a). When utilising a specific North African vegetation classification<sup>8</sup> (Fig. 3b), vegetation thresholds for sparse savannah grassland are estimated to be as little as 100mm/yr, which is also the current lower boundary for Sahelian vegetation<sup>46</sup>. With this classification (Fig. 3b) the modelled NAHP precipitation sustains woodland to approximately 16°N, wooded grassland and grassland across the central and western Sahara, and southward migration of Mediterranean vegetation. Note that during the 17 skipped beats, precipitation increase is still significant enough to support the proliferation of grassland across the Western Sahara (Extended Fig. 4). This may influence proxy records and may be why some West Saharan dust and leaf-wax records indicate almost equally enhanced vegetation and precipitation at each precession minima<sup>1, 10, 12</sup>.
+
+The modelled NAHP amplitude in the North-eastern Sahara (Egypt and N.E. Libya) and the Arabian Peninsula is likely too small (Fig. 2a), with widespread evidence indicating regional river networks and lakes during the NAHPs<sup>47–53</sup>. In N.E. Africa, precipitation in the Tibesti mountains (360 mm/yr) and the upper Nile sources (Ethiopia, South Sudan, and southern Sudan) may have been enough to support vegetation growth downstream. However, the Arabian Peninsula poses a more marked conundrum. Modelled precipitation is limited to the south-west (580mm/yr), which deteriorates to <100mm/yr within the interior. This may be linked to an inaccurate response of the East African or Indian Monsoon systems to insolation change, or to the response of tropical plumes which affect Northern Arabia and may have been important during the NAHPs<sup>54–57</sup>. Additionally, Mediterranean storm tracks, which are a key winter regional moisture source<sup>50, 58</sup>, may be poorly resolved in the model. High-resolution or regional modelling which better resolves these processes<sup>59–61</sup> may reconcile these biases and is a focus of future work.
+
+# Precession Driven Mechanism Of The Nahps
+
+Our results confirm that precession is the dominant pacemaker of the NAHPs, as verified by wavelet analysis (Fig. 4d). The greater precipitation anomaly in the western Sahara indicates that the WAM is important in driving the modelled NAHPs. Periods of precession-driven increased summer boreal insolation shifts the position of maximum precipitation and the Intertropical Convergence Zone (ITCZ) northward (Fig. 1l, Fig. 2a), and increases the strength of the WAM as indicated by the monsoon index (Fig. 1k).
+
+The WAM is a complex system comprised of several key components (Fig. 2b)⁶²–⁶⁴. The summer monsoon, indicated by westerly (positive) zonal wind, extends to 18°N during precession maxima (Fig. 2b). Its northward penetration is dependent on the position of the anticyclonic Saharan Heat Low⁶⁵, atmospheric uplift here is not strong enough to cause precipitation. The rain belt is comprised of upward motion involving deep convection throughout the troposphere. The Sahara is located below the descending branch of the Hadley Cell which forms the sub-tropical westerly jet, this subsidence inhibits convective rainfall. The African and Tropical Easterly Jets form in the middle and top of the troposphere, with the former sustained by dry-northerly and moist-southerly convection⁶²,⁶⁶,⁶⁷. The region between, and to the south in HadCM3BB-v1.0, of the African and Tropical Easterly Jets is characterised by strong uplift which favours precipitation⁶⁸ (Fig. 2b).
+
+Observations indicate that the relationship between the African and Tropical Easterly Jets determines Sahelian rainfall, with a more northerly weaker African Easterly Jet and/or stronger Tropical Easterly Jet associated with increased precipitation and vice versa⁶⁷,⁶⁸. Our results show that this same pattern is apparent during the NAHPs (Fig. 2b), in addition to an enhanced WAM that extends further north. Furthermore, there is northward displacement, strengthening and warming of the sub-tropical westerly jet (Fig. 2b), with a positive mid-tropospheric geopotential height anomaly and increase in relative humidity (Fig. 2c). As a result, the descending air that characterises the sub-tropical westerly jet at precession maxima changes to ascending air during the NAHPs, indicating formation of a jet ridge over the Northern Sahara (Fig. 2b). It is the combined WAM, African/Tropical Easterly Jets, and the sub-tropical westerly jet response which results in increased precipitation during the NAHPs. This initiates vegetation change and enhances biogeophysical feedbacks including evapotranspiration which is dominant in the subtropics and tropics in HadCM3B⁶ (9). This further enhances precipitation.
+
+# Nahp Forcing – Orbits, Greenhouse Gases And The Ice Sheets
+
+In addition to precession, a significant eccentricity signal and a weaker intermittent obliquity signal is evident in the wavelet analysis of modelled Saharan precipitation (Fig. 4d). Eccentricity may be present due to modulation of the precession cycle. However, the signal may also highlight the influence of the glacial boundary conditions which fluctuate on these orbital timescales <sup>70</sup>. These forcings may influence the response of the monsoon system to precession and may consequently cause the skipped beats. Past modelling studies that have investigated greenhouse gas and ice sheet forcing on the NAHPs have proved inconclusive. Some show that ice sheets and greenhouse gases had little impact on precession-controlled Saharan precipitation <sup>6, 19, 31</sup>. In contrast, others <sup>9, 15, 18</sup> showed that glacial ice sheets suppress NAHP formation and that ice sheet retreat is required for initiating an NAHP, implying an ice-sheet determined orbital signal in Saharan precipitation. However these studies used either a lower resolution EMIC <sup>9, 15, 18, 19, 31</sup>, did not simulate a NAHP with significant magnitude <sup>6</sup>, or covered a relatively short temporal timescale <sup>6, 9, 15, 18</sup>.
+
+We performed additional sensitivity experiments to evaluate the role of greenhouse gas and ice sheet forcing (Fig. 4a; methods). When incorporating only orbital variations (Orb_Only; Fig. 4a,b), Saharan precipitation is determined primarily by precession with a weaker obliquity signal (Fig. 4b). The absence of an eccentricity signal indicates firstly that precession enhanced precipitation is not directly modulated by eccentricity (i.e. the magnitude of precession minima precipitation is not directly correlated with precession forcing), and secondly that the eccentricity signal is a response to the glacial boundary conditions. The addition of greenhouse gas forcing (Orb_GHG; Fig. 4a, 4c) has only a small impact on the amplitude and variability of Saharan precipitation. It is therefore the addition of the ice sheets (NAHP_All, Figs. 1j, 4a, 4d) that produces the strong eccentricity signal in Saharan precipitation and generates the ‘skipped beats’ in the NAHPs. During glacial periods, extensive ice sheets suppress precession-induced precipitation change in the Sahara. Ice sheet volume is determined by eccentricity (Extended Fig. 5) <sup>70</sup> and therefore eccentricity indirectly forces the NAHPs via its influence on the ice sheets.
+
+O’Mara, et al. <sup>10</sup> concluded that obliquity was a dominant control on NAHPs between MIS 10 and 13. Our modelled results indicate that obliquity does have a significant influence on Saharan precipitation during MIS 10–12 (Fig. 3f). However, its longer-term influence over the past 800kyr is secondary to precession and eccentricity. We also do not see a relationship between AMOC strength and NAHP amplitude as identified in Menviel, et al. <sup>9</sup>. Furthermore, the important influence of ice sheets on NAHP amplitude, and uncertainties regarding their extent, growth and decay phases <sup>71</sup>, may explain the few inconsistences between the modelled NAHPs and the observations as labelled in Fig. 1i.
+
+Now that we have shown the key role of the ice sheets in controlling NAHP amplitude, we need to identify how they suppress the NAHPs. Studies that have investigated the influence of ice sheets on monsoon evolution <sup>72–74</sup> identify both a thermodynamic and topographic impact. The former reflects albedo change, which alters absorbed insolation and consequently atmospheric and oceanic temperatures. For example sea surface temperatures have long been implicated in the extent of Sahelian precipitation due to their control on the land-sea thermal contrast and WAM strength <sup>75–78</sup>. The topographic impact reflects ice sheet elevation, which influences synoptic scale wind patterns.
+
+During glacial precession minima the ice sheets cool the atmosphere, weaken the WAM and cause the sub-tropical westerly jet to descend which inhibits precipitation (Extended Fig. 6), similar to periods of precession maxima (Fig. 2b). Using the atmosphere-only version of our model (HadAM3BB-v1.0; methods), we identify that this response is primarily due to the thermodynamic impact of the ice sheet. This is dominated by the atmospheric response to albedo change, with a smaller contribution from sea surface temperature changes (Extended Fig. 7; supplementary information). During glacials, the increase in albedo and resultant colder atmosphere acts to weaken the WAM, and results in a more southerly, weaker, colder and descending sub-tropical westerly jet, indicative of a jet trough which inhibits convective precipitation (Extended Fig. 6). This is similar to the precession maxima mechanism. The glacial SSTs reduce precipitation by weakening the WAM and shifting the position of the African/Tropical Easterly Jets (Extended Fig. 6; supplementary information). It is therefore a combination of the sea surface temperature and atmospheric response to albedo change that is the primary mechanism by which ice sheets suppress NAHPs during precession minima.
+
+In summary, the ability of the HadCM3BB-v1.0 model to accurately simulate the timing and amplitude of the NAHPs in Western Africa confirms that palaeo-conditioning a model can improve palaeo-hydrographic modelling in this region throughout the late Quaternary. The NAHPs were predominantly paced by precession and, in the west, by its influence on the WAM and sub-tropical westerly jet. Our results show that during glacials, expansive ice sheets suppressed NAHP formation predominantly due to the impact of glacial albedo change, thus highlighting the role of high latitude processes in determining the timing and amplitude of the NAHPs. The ability to model NAHPs is important for understanding climate dynamics past and future, and enables the opportunity to study in much more detail the role of the Sahara as a factor controlling the dispersal of plants and animals, including humans, and the consequent movement of genetic and cultural information within and out of Africa.
+
+# Methods
+
+The Hadley Centre Coupled Model 3 Bristol (HadCM3B) is a coupled climate model that consists of a 3d dynamical atmospheric component with a resolution of 2.5°x3.75°, 19 vertical levels, and a 30 minute timestep <sup>79</sup>, and an ocean model with a resolution of 1.25°x1.25°, 20 vertical levels and a 1 hour timestep <sup>80</sup>. Levels have a finer resolution towards the Earth surface. The model is a variant of HadCM3 that has been developed at the University of Bristol, and is described in detailed in Valdes, et al. <sup>42</sup>. Despite its relative old age, the model has been shown to accurately simulate the climate system and remains competitive with more modern climate models. A key advantage of the model is its computationally efficient, which permits long simulations and large ensemble studies. We also utilise the atmosphere only version of the model; HadAM3B <sup>2</sup>. This incorporates the same atmospheric component as HadCM3B but with prescribed sea-surface temperatures (SSTs).
+
+The model incorporates the Met Office Surface Exchange Scheme (MOSES) version 2.1 <sup>81</sup> which simulates water and energy fluxes and physiological processes such as photosynthesis, transpiration and respiration which is determined by stomatal conductance and consequently CO<sub>2</sub> concentration. The fractional coverage of nine surface types are incorporated by MOSES 2.1 and simulated by the dynamic global vegetation model (DGVM) TRIFFID. Of the nine surface types, five are plant functional types (PFTs); deciduous and needleleaf trees, C3 and C4 grasses, shrubs, with the residual assigned to bare soil. Vegetation evolves throughout a model simulation depending on temperature, moisture, CO<sub>2</sub> and competition with other PFTs. HadCM3B does not include an interactive carbon/methane cycle or ice model, so these boundary conditions have been imposed as discussed below.
+
+## Palaeoconditioning
+
+The update to the atmospheric and vegetation parameters follows the work of Hopcroft, et al. <sup>4140</sup>. The most important change was to modify the vertical profile of convective entrainment and detrainment, decreasing the entrainment rate near the surface and increasing the entrainment rate higher up in the atmosphere globally. This resulted in greater mixing between convective plumes and the environment in the upper troposphere relative to the lower troposphere increasing the precipitation response from African Easterly Wave events. Hopcroft, et al. <sup>41</sup> showed that this permitted the model to more accurately simulate mid-Holocene greening of the Sahara. At the same time, it did not adversely influence present day climate simulations. The moisture stress function of vegetation was also altered to be linear in soil moisture potential rather than soil moisture, and then optimised to reproduce vegetation cover-climate relationships for the present day and the mid-Holocene <sup>40</sup>. Together these updates allow a dynamic simulation of the Holocene greening of the Sahara that shows many similarities with the reconstructions of this time period <sup>82</sup>, including a more gradual and later desertification in the East and South of the Sahara and a centennial-scale transition over North Western Sahara.
+
+The palaeo-condition model has not been tested for periods beyond the mid-Holocene. Here, we incorporate the changes described above into HadCM3B, the updated model configuration is termed HadCM3BB-v1.0.
+
+## Boundary Conditions
+
+The simulations have been forced with well constrained orbital parameters <sup>83</sup> and greenhouse gas concentrations (CO<sub>2</sub>, N<sub>2</sub>O and CH<sub>4</sub>) taken from the Vostok Ice core <sup>84, 85</sup>.
+
+The extent and elevation of the Antarctic, Greenland, North American and Fennoscandian ice sheets has been imposed utilising the reconstruction of de Boer, et al. <sup>86</sup>. This provides ice extent and thickness which are used within the model to calculate continental elevation (depending on ice thickness and isostatic rebound), bathymetry, ice-sheet extent, and consequently the land-sea mask. There remain uncertainties regarding Ice sheet reconstruction beyond the LGM, in part due to poor preservation following the last deglaciation. Although the ice area might be approximated based on sea-level data, uncertainties remain associated with ice sheet structure during growth and decay phases. As stated in the text, this may cause uncertainty regarding the suppression of some the NAHP events and be the reason for inconsistencies between the model and observations.
+
+## Snapshot Experiments
+
+The boundary conditions have been incorporated into 219 snapshot experiments covering 0kyr BP (corresponding to the year 1950 albeit with pre-industrial GHG concentrations) to 800kyr BP. These are set at 1kyr intervals for 0-24kyr BP and 4kyr intervals to for 24-800kyr BP. Each simulation has been run for 500 years with analysis conducted on the final 50 years of each simulation unless stated otherwise. This spin-up permits the atmosphere and surface ocean to reach a state of near equilibrium. This approach allows the simulations to be run simultaneously and is therefore very efficient.
+
+The snapshot climatologies were then splined to a 100-yr timeseries for all variables used in the study, including the land sea mask and ice fraction (subsequently rounded to 0 or 100% coverage in any grid cell). Splining has been done using the ftcurv function of the NCAR command language (NCL) <sup>87</sup> which utilises spline under tension. This is the same methodology outlined in Armstrong, et al. <sup>43</sup>. The timeseries data was then bilinearly interpolated to 1° resolution for surface variables. There has been <b>no</b> bias correction applied to the data in this study.
+
+## Boundary Condition Sensitivity Experiments
+
+In order to identify the contributing role of three forcing factors on the NAHPs; orbital variations, GHGs and ice sheets, we performed two additional sets of snapshot experiments (2x 219 simulations) using HadCM3BB-v1.0. The original set of snapshot simulations incorporate all three forcings (as discussed above) and is termed NAHP_All. In order to identify just the orbital forcing, the simulations were re-run with varying orbital parameters, but with constant pre-industrial GHGs (CO<sub>2</sub> − 280ppm, CH<sub>4</sub> – 760ppbv, and N<sub>2</sub>O – 270ppbv) and ice sheet extent and elevation, this is termed Orb_Only. The third set of snapshot simulations incorporates both orbital and GHG forcing, with constant pre-industrial ice sheets, termed Orb_GHG.
+
+The snapshot climatologies were similarly splined to a 100-yr timeseries and bilinearly interpolated to 1° resolution for surface variables.
+
+## Topographic Vs. Thermodynamic Ice-sheet Sensitivity Experiments
+
+In order to resolve the topographic vs thermodynamic contribution of the ice sheets to the suppression of the NAHPs during precession minima, we have used the atmosphere-only version of the model (HadAM3B <sup>2</sup>) and incorporated the updates described above (HadAM3BB-v1.0). We have run the model controlling for SSTs, ice sheet elevation, and ice sheet albedo.
+
+Due to computational limitations, we have not rerun the whole suite of snapshot simulations. Instead, we have performed a set of singular snapshot experiments corresponding to where the impact of the ice sheet is shown to have a significant suppressing effect on the amplitude of precession minima induced Saharan precipitation. We identify the eighth precession minima, corresponding to 176Kyr BP, as where ice sheet extent suppresses precipitation the greatest amount, from ~ 510mm/yr for Orb_Only/Orb_GHG to 205mm/yr for NAHP_All (Fig. <span class="InternalRef" refid="Fig2">4</span> a). As the 176kyr BP precipitation is equivalent for both Orb_Only and Orb_GHG, and because we only want to compare the impact of the ice sheet (rather than any possible impact of GHGs), we utilise input from the Orb_GHG experiment. We assume that our findings for this time period are consistent for all the suppressed NAHPs during glacials.
+
+The first experiment is the control HadAM3BB-v1.0 simulation (termed ‘Control’). This incorporates the boundary conditions and SST input from the 176kyr NAHP_All snapshot experiment. The climatology is therefore very similar to the 176kyr BP HadCM3BB-v1.0 NAHP_All experiment (supplementary information). In order to isolate the thermodynamic impact of SSTs, we have run HadAM3BB-v1.0 with 176kyr boundary conditions, but with SSTs prescribed from the 176kyr Orb_GHG snapshot simulation, termed ‘glacial_noSST’. This simulation therefore has glacial forcings except for SSTs. To identify the thermodynamic impact of ice sheet albedo, we have run HadAM3BB-v1.0 with 176kyr boundary conditions including SSTs, but with ice sheet albedo from the 176kyr Orb_GHG experiment (this is equivalent to PI ice sheet albedo). This is termed ‘glacial_noAlb’. This simulation therefore has glacial forcings except for non-glacial (PI) surface albedo. SSTs are fixed throughout the simulation so the albedo change will only impact the atmosphere. Note however that this is not a clean comparison because the height of the ice sheet will cause it to snow, so there will be a contribution from albedo in some locations. To identify the topographic impact of the ice sheet, we have run HadAM3BB-v1.0 with 176kyr boundary conditions, but with ice sheet topography from the 176kyr Orb_GHG experiment (this is equivalent to PI ice sheet topography). This is termed ‘glacial_noTopo’. This simulation has glacial forcings except for non-glacial (PI) topography. Each simulation was run for 200 years and analysis conducted on the final 30 years. An analysis of these simulations is provided in the Supplementary.
+
+## Data Availability
+
+A total of 661 model simulations were performed for this study. The raw model output and an overview of all the simulations is available at
+
+<span class="ExternalRef">
+<span class="RefSource">
+https://www.paleo.bristol.ac.uk/ummodel/scripts/papers/Armstrong_et_al_2022.html
+</span>
+<span address="https://www.paleo.bristol.ac.uk/ummodel/scripts/papers/Armstrong_et_al_2022.html" class="RefTarget" targettype="URL">
+</span>
+</span>
+
+## Code Availability
+
+All scripts used to analyse the data and produce the Figures have been written using the NCAR command language (NCL, Version 6.4.0) and are available on request
+
+# References
+
+1. Crocker, A. J. et al. Astronomically controlled aridity in the Sahara since at least 11 million years ago. *Nature Geoscience* **15**, 671-+, doi: [10.1038/s41561-022-00990-7](https://doi.org/10.1038/s41561-022-00990-7) (2022).
+
+2. Grant, K. M. et al. A 3 million year index for North African humidity/aridity and the implication of potential pan-African Humid periods. *Quaternary Science Reviews* **171**, 100–118, doi: [10.1016/j.quascirev.2017.07.005](https://doi.org/10.1016/j.quascirev.2017.07.005) (2017).
+
+3. Hely, C., Lezine, A. M. & Contributors, A. P. D. Holocene changes in African vegetation: tradeoff between climate and water availability. *Climate of the Past* **10**, 681–686, doi: [10.5194/cp-10-681-2014](https://doi.org/10.5194/cp-10-681-2014) (2014).
+
+4. Hoffmann, D. L. et al. Timing and causes of North African wet phases during the last glacial period and implications for modern human migration. *Scientific Reports* **6**, 36367, doi: [10.1038/srep36367](https://doi.org/10.1038/srep36367) (2016).
+
+5. Jennings, R. P. et al. The greening of Arabia: Multiple opportunities for human occupation of the Arabian Peninsula during the Late Pleistocene inferred from an ensemble of climate model simulations. *Quaternary International* **382**, 181–199, doi: [10.1016/j.quaint.2015.01.006](https://doi.org/10.1016/j.quaint.2015.01.006) (2015).
+
+6. Kutzbach, J. E. et al. African climate response to orbital and glacial forcing in 140,000-y simulation with implications for early modern human environments. *Proceedings of the National Academy of Sciences of the United States of America* **117**, 2255–2264, doi: [10.1073/pnas.1917673117](https://doi.org/10.1073/pnas.1917673117) (2020).
+
+7. Larrasoaña, J. C. A review of West African monsoon penetration during Green Sahara periods; implications for human evolution and dispersals over the last three million years. *Oxford Open Climate Change* **1**, doi: [10.1093/oxfclm/kgab011](https://doi.org/10.1093/oxfclm/kgab011) (2021).
+
+8. Larrasoaña, J. C., Roberts, A. P. & Rohling, E. J. Dynamics of Green Sahara Periods and Their Role in Hominin Evolution. *PLoS ONE* **8**, e76514, doi: [10.1371/journal.pone.0076514](https://doi.org/10.1371/journal.pone.0076514) (2013).
+
+9. Menviel, L. et al. Drivers of the evolution and amplitude of African Humid Periods. *Communications Earth & Environment* **2**, doi: [10.1038/s43247-021-00309-1](https://doi.org/10.1038/s43247-021-00309-1) (2021).
+
+10. O'Mara, N. A. et al. Pleistocene drivers of Northwest African hydroclimate and vegetation. *Nature Communications* **13**, doi: [10.1038/s41467-022-31120-x](https://doi.org/10.1038/s41467-022-31120-x) (2022).
+
+11. Pausata, F. S. R. et al. The Greening of the Sahara: Past Changes and Future Implications. *One Earth* **2**, 235–250, doi: [10.1016/j.oneear.2020.03.002](https://doi.org/10.1016/j.oneear.2020.03.002) (2020).
+
+12. Skonieczny, C. et al. Monsoon-driven Saharan dust variability over the past 240,000 years. *Sci Adv* **5**, doi: [10.1126/sciadv.aav1887](https://doi.org/10.1126/sciadv.aav1887) (2019).
+
+13. Tierney, J. E., deMenocal, P. B. & Zander, P. D. A climatic context for the out-of-Africa migration. *Geology* **45**, 1023–1026, doi: [10.1130/g39457.1](https://doi.org/10.1130/g39457.1) (2017).
+
+14. Tierney, J. E., Pausata, F. S. R. & deMenocal, P. B. Rainfall regimes of the Green Sahara. *Sci Adv* **3**, e1601503-e1601503, doi: [10.1126/sciadv.1601503](https://doi.org/10.1126/sciadv.1601503) (2017).
+
+15. Timm, O., Kohler, P., Timmermann, A. & Menviel, L. Mechanisms for the Onset of the African Humid Period and Sahara Greening 14.5–11 ka BP. *Journal of Climate* **23**, 2612–2633, doi: [10.1175/2010jcli3217.1](https://doi.org/10.1175/2010jcli3217.1) (2010).
+
+16. Claussen, M., Dallmeyer, A. & Bader, J. in *Oxford Research Encyclopedia of Climate Science* (Oxford University Press, 2017).
+
+17. Timmermann, A., Lorenz, S. J., An, S.-I., Clement, A. & Xie, S.-P. The effect of orbital forcing on the mean climate and variability of the tropical Pacific. **20**, 4147–4159 (2007).
+
+18. Blanchet, C. L. et al. Drivers of river reactivation in North Africa during the last glacial cycle. *Nature Geoscience* **14**, 97–103, doi: [10.1038/s41561-020-00671-3](https://doi.org/10.1038/s41561-020-00671-3) (2021).
+
+19. Duque-Villegas, M., Claussen, M., Brovkin, V. & Kleinen, T. Effects of orbital forcing, greenhouse gases and ice sheets on Saharan greening in past and future multi-millennia. *Climate of the Past* **18**, 1897–1914, doi: [10.5194/cp-18-1897-2022](https://doi.org/10.5194/cp-18-1897-2022) (2022).
+
+20. Ziegler, M., Tuenter, E. & Lourens, L. J. The precession phase of the boreal summer monsoon as viewed from the eastern Mediterranean (ODP Site 968). *Quaternary Science Reviews* **29**, 1481–1490, doi: [10.1016/j.quascirev.2010.03.011](https://doi.org/10.1016/j.quascirev.2010.03.011) (2010).
+
+21. DeMenocal, P. B. & Tierney, J. E. Green Sahara: African humid periods paced by Earth’s orbital changes. *Nature Education Knowledge* **3**, 12 (2012).
+
+22. Levis, S., Bonan, G. B. & Bonfils, C. Vol. 23 791–802 (2004).
+
+23. Brovkin, V., Claussen, M., Petoukhov, V. & Ganopolski, A. On the stability of the atmosphere-vegetation system in the Sahara/Sahel region. *Journal of Geophysical Research: Atmospheres* **103**, 31613–31624, doi: [10.1029/1998jd200006](https://doi.org/10.1029/1998jd200006) (1998).
+
+24. Kutzbach, J., Bonan, G., Foley, J. & Harrison, S. P. Vegetation and soil feedbacks on the response of the African monsoon to orbital forcing in the early to middle Holocene. *Nature* **384**, 623–626, doi: [10.1038/384623a0](https://doi.org/10.1038/384623a0) (1996).
+
+25. Chandan, D. & Peltier, W. R. African Humid Period Precipitation Sustained by Robust Vegetation, Soil, and Lake Feedbacks. *Geophysical Research Letters* **47**, doi: [10.1029/2020gl088728](https://doi.org/10.1029/2020gl088728) (2020).
+
+26. Claussen, M., Brovkin, V., Ganopolski, A., Kubatzki, C. & Petoukhov, V. Climate Change in Northern Africa: The Past is Not the Future. *Climatic Change* **57**, 99–118, doi: [10.1023/a:1022115604225](https://doi.org/10.1023/a:1022115604225) (2003).
+
+27. Marzin, C., Braconnot, P. & Kageyama, M. Relative impacts of insolation changes, meltwater fluxes and ice sheets on African and Asian monsoons during the Holocene. *Climate Dynamics* **41**, 2267–2286, doi: [10.1007/s00382-013-1948-9](https://doi.org/10.1007/s00382-013-1948-9) (2013).
+
+28. Harrison, S. P. et al. Evaluation of CMIP5 palaeo-simulations to improve climate projections. *Nature Climate Change* **5**, 735–743, doi: [10.1038/nclimate2649](https://doi.org/10.1038/nclimate2649) (2015).
+
+29. Brierley, C. M. et al. Large-scale features and evaluation of the PMIP4-CMIP6 midHolocene simulations. *Climate of the Past* **16**, 1847–1872, doi: [10.5194/cp-16-1847-2020](https://doi.org/10.5194/cp-16-1847-2020) (2020).
+
+30. Harrison, S. P. et al. Climate model benchmarking with glacial and mid-Holocene climates. *Climate Dynamics* **43**, 671–688, doi: [10.1007/s00382-013-1922-6](https://doi.org/10.1007/s00382-013-1922-6) (2014).
+
+31. Weber, S. L. & Tuenter, E. The impact of varying ice sheets and greenhouse gases on the intensity and timing of boreal summer monsoons. *Quaternary Science Reviews* **30**, 469–479, doi: [10.1016/j.quascirev.2010.12.009](https://doi.org/10.1016/j.quascirev.2010.12.009) (2011).
+
+32. Braconnot, P. et al. Evaluation of climate models using palaeoclimatic data. *Nature Climate Change* **2**, 417–424, doi: [10.1038/nclimate1456](https://doi.org/10.1038/nclimate1456) (2012).
+
+33. Otto-Bliesner, B. L. et al. Large-scale features of Last Interglacial climate: results from evaluating the lig127k simulations for the Coupled Model Intercomparison Project (CMIP6)-Paleoclimate Modeling Intercomparison Project (PMIP4). *Climate of the Past* **17**, 63–94, doi: [10.5194/cp-17-63-2021](https://doi.org/10.5194/cp-17-63-2021) (2021).
+
+34. Biasutti, M. et al. Global energetics and local physics as drivers of past, present and future monsoons. *Nature Geoscience* **11**, 392-+, doi: [10.1038/s41561-018-0137-1](https://doi.org/10.1038/s41561-018-0137-1) (2018).
+
+35. Perez-Sanz, A., Li, G., Gonzalez-Samperiz, P. & Harrison, S. P. Evaluation of modern and mid-Holocene seasonal precipitation of the Mediterranean and northern Africa in the CMIP5 simulations. *Climate of the Past* **10**, 551–568, doi: [10.5194/cp-10-551-2014](https://doi.org/10.5194/cp-10-551-2014) (2014).
+
+36. Braconnot, P., Joussaume, S., Marti, O. & de Noblet, N. Synergistic feedbacks from ocean and vegetation on the African monsoon response to mid-Holocene insolation. *Geophysical Research Letters* **26**, 2481–2484, doi: [10.1029/1999gl006047](https://doi.org/10.1029/1999gl006047) (1999).
+
+37. Pausata, F. S. R., Messori, G. & Zhang, Q. Impacts of dust reduction on the northward expansion of the African monsoon during the Green Sahara period. *Earth and Planetary Science Letters* **434**, 298–307, doi: [10.1016/j.epsl.2015.11.049](https://doi.org/10.1016/j.epsl.2015.11.049) (2016).
+
+38. Chen, W. Z. et al. Feedbacks of soil properties on vegetation during the Green Sahara period. *Quaternary Science Reviews* **240**, doi: [10.1016/j.quascirev.2020.106389](https://doi.org/10.1016/j.quascirev.2020.106389) (2020).
+
+39. Hopcroft, P. O., Valdes, P. J., Harper, A. B. & Beerling, D. J. Multi vegetation model evaluation of the Green Sahara climate regime. *Geophysical Research Letters* **44**, 6804–6813, doi: [10.1002/2017gl073740](https://doi.org/10.1002/2017gl073740) (2017).
+
+40. Hopcroft, P. O. & Valdes, P. J. Paleoclimate-conditioning reveals a North Africa land–atmosphere tipping point. *Proceedings of the National Academy of Sciences* **118**, e2108783118, doi: [10.1073/pnas.2108783118](https://doi.org/10.1073/pnas.2108783118) (2021).
+
+41. Hopcroft, P. O., Valdes, P. J. & Ingram, W. Using the Mid-Holocene “Greening” of the Sahara to Narrow Acceptable Ranges on Climate Model Parameters. *Geophysical Research Letters* **48**, doi: [10.1029/2020gl092043](https://doi.org/10.1029/2020gl092043) (2021).
+
+42. Valdes, P. J. et al. The BRIDGE HadCM3 family of climate models: HadCM3@Bristol v1.0. *Geoscientific Model Development* **10**, 3715–3743, doi: [10.5194/gmd-10-3715-2017](https://doi.org/10.5194/gmd-10-3715-2017) (2017).
+
+43. Armstrong, E., Hopcroft, P. O. & Valdes, P. J. A simulated Northern Hemisphere terrestrial climate dataset for the past 60,000 years. *Scientific Data* **6**, doi: [10.1038/s41597-019-0277-1](https://doi.org/10.1038/s41597-019-0277-1) (2019).
+
+44. Emeis, K. C., Sakamoto, T., Wehausen, R. & Brumsack, H. J. The sapropel record of the eastern Mediterranean Sea - results of Ocean Drilling Program Leg 160. *Palaeogeography Palaeoclimatology Palaeoecology* **158**, 371–395, doi: [10.1016/s0031-0182(00)00059-6](https://doi.org/10.1016/s0031-0182(00)00059-6) (2000).
+
+45. Bartlein, P. J. et al. Pollen-based continental climate reconstructions at 6 and 21 ka: a global synthesis. *Climate Dynamics* **37**, 775–802, doi: [10.1007/s00382-010-0904-1](https://doi.org/10.1007/s00382-010-0904-1) (2011).
+
+46. Nicholson, S. E. in *Oxford Research Encyclopedia of Climate Science* (Oxford University Press, 2018).
+
+47. Osborne, A. H. et al. A humid corridor across the Sahara for the migration of early modern humans out of Africa 120,000 years ago. *Proceedings of the National Academy of Sciences of the United States of America* **105**, 16444–16447, doi: [10.1073/pnas.0804472105](https://doi.org/10.1073/pnas.0804472105) (2008).
+
+48. Neugebauer, I. et al. The unexpectedly short Holocene Humid Period in Northern Arabia. *Communications Earth & Environment* **3**, doi: [10.1038/s43247-022-00368-y](https://doi.org/10.1038/s43247-022-00368-y) (2022).
+
+49. Petraglia, M. D., Groucutt, H. S., Guagnin, M., Breeze, P. S. & Boivin, N. Human responses to climate and ecosystem change in ancient Arabia. *Proceedings of the National Academy of Sciences of the United States of America* **117**, 8263–8270, doi: [10.1073/pnas.1920211117](https://doi.org/10.1073/pnas.1920211117) (2020).
+
+50. Breeze, P. S. et al. Palaeohydrological corridors for hominin dispersals in the Middle East similar to 250 – 70,000 years ago. *Quaternary Science Reviews* **144**, 155–185, doi: [10.1016/j.quascirev.2016.05.012](https://doi.org/10.1016/j.quascirev.2016.05.012) (2016).
+
+51. Crassard, R. et al. Middle Palaeolithic and Neolithic Occupations around Mundafan Palaeolake, Saudi Arabia: Implications for Climate Change and Human Dispersals. *Plos One* **8**, doi: [10.1371/journal.pone.0069665](https://doi.org/10.1371/journal.pone.0069665) (2013).
+
+52. Drake, N. A. et al. Sedimentary and geomorphic evidence of Saharan megalakes: A synthesis. *Quaternary Science Reviews* **276**, doi: [10.1016/j.quascirev.2021.107318](https://doi.org/10.1016/j.quascirev.2021.107318) (2022).
+
+53. Drake, N. A., Blench, R. M., Armitage, S. J., Bristow, C. S. & White, K. H. Ancient watercourses and biogeography of the Sahara explain the peopling of the desert. *Proceedings of the National Academy of Sciences* **108**, 458–462, doi: [10.1073/pnas.1012231108](https://doi.org/10.1073/pnas.1012231108) (2011).
+
+54. Rubin, S., Ziv, B. & Paldor, N. Tropical plumes over eastern North Africa as a source of rain in the middle east. *Monthly Weather Review* **135**, 4135–4148, doi: [10.1175/2007mwr1919.1](https://doi.org/10.1175/2007mwr1919.1) (2007).
+
+55. Skinner, C. B. & Poulsen, C. J. The role of fall season tropical plumes in enhancing Saharan rainfall during the African Humid Period. *Geophysical Research Letters* **43**, 349–358, doi: [10.1002/2015gl066318](https://doi.org/10.1002/2015gl066318) (2016).
+
+56. Tubi, A., Dayan, U. & Lensky, I. M. Moisture transport by tropical plumes over the Middle East: a 30-year climatology. *Quarterly Journal of the Royal Meteorological Society* **143**, 3165–3176, doi: [10.1002/qj.3170](https://doi.org/10.1002/qj.3170) (2017).
+
+57. Waldmann, N., Torfstein, A. & Stein, M. Northward intrusions of low- and mid-latitude storms across the Saharo-Arabian belt during past interglacials. *Geology* **38**, 567–570, doi: [10.1130/g30654.1](https://doi.org/10.1130/g30654.1) (2010).
+
+58. Kutzbach, J. E., Chen, G., Cheng, H., Edwards, R. L. & Liu, Z. Potential role of winter rainfall in explaining increased moisture in the Mediterranean and Middle East during periods of maximum orbitally-forced insolation seasonality. *Climate Dynamics* **42**, 1079–1095, doi: [10.1007/s00382-013-1692-1](https://doi.org/10.1007/s00382-013-1692-1) (2014).
+
+59. Flaounas, E. et al. Assessment of an ensemble of ocean-atmosphere coupled and uncoupled regional climate models to reproduce the climatology of Mediterranean cyclones. *Climate Dynamics* **51**, 1023–1040, doi: [10.1007/s00382-016-3398-7](https://doi.org/10.1007/s00382-016-3398-7) (2018).
+
+60. Gaertner, M. et al. Simulation of medicanes over the Mediterranean Sea in a regional climate model ensemble: impact of ocean-atmosphere coupling and increased resolution. *Climate Dynamics* **51**, 1041–1057, doi: [10.1007/s00382-016-3456-1](https://doi.org/10.1007/s00382-016-3456-1) (2018).
+
+61. Jung, T., Gulev, S. K., Rudeva, I. & Soloviov, V. Sensitivity of extratropical cyclone characteristics to horizontal resolution in the ECMWF model. *Quarterly Journal of the Royal Meteorological Society* **132**, 1839–1857, doi: [10.1256/qj.05.212](https://doi.org/10.1256/qj.05.212) (2006).
+
+62. Niang, C. et al. Transport pathways across the West African Monsoon as revealed by Lagrangian Coherent Structures. *Scientific Reports* **10**, doi: [10.1038/s41598-020-69159-9](https://doi.org/10.1038/s41598-020-69159-9) (2020).
+
+63. Moron, V., Philippon, N. & Fontaine, B. Simulation of West African monsoon circulation in four atmospheric general circulation models forced by prescribed sea surface temperature. *Journal of Geophysical Research-Atmospheres* **109**, doi: [10.1029/2004jd004760](https://doi.org/10.1029/2004jd004760) (2004).
+
+64. Sultan, B. & Janicot, S. The West African monsoon dynamics. Part II: The "preonset" and "onset" of the summer monsoon. *Journal of Climate* **16**, 3407–3427, doi: [10.1175/1520-0442(2003)016<3407:Twamdp>2.0.Co;2](https://doi.org/10.1175/1520-0442(2003)016%3C3407:Twamdp%3E2.0.Co;2) (2003).
+
+65. Lavaysse, C., Flamant, C., Evan, A., Janicot, S. & Gaetani, M. Recent climatological trend of the Saharan heat low and its impact on the West African climate. *Climate Dynamics* **47**, 3479–3498, doi: [10.1007/s00382-015-2847-z](https://doi.org/10.1007/s00382-015-2847-z) (2016).
+
+66. Thorncroft, C. D. & Blackburn, M. Maintenance of the African easterly jet. *Quarterly Journal of the Royal Meteorological Society* **125**, 763–786, doi: [10.1256/smsqj.55501](https://doi.org/10.1256/smsqj.55501) (1999).
+
+67. Cook, K. H. Generation of the African easterly jet and its role in determining West African precipitation. *Journal of Climate* **12**, 1165–1184, doi: [10.1175/1520-0442(1999)012<1165:Gotaej>2.0.Co;2](https://doi.org/10.1175/1520-0442(1999)012%3C1165:Gotaej%3E2.0.Co;2) (1999).
+
+68. Nicholson, S. E. A revised picture of the structure of the "monsoon" and land ITCZ over West Africa. *Climate Dynamics* **32**, 1155–1171, doi: [10.1007/s00382-008-0514-3](https://doi.org/10.1007/s00382-008-0514-3) (2009).
+
+69. Armstrong, E., Valdes, P., House, J. & Singarayer, J. The Role of CO2 and Dynamic Vegetation on the Impact of Temperate Land-Use Change in the HadCM3 Coupled Climate Model. *Earth Interactions* **20**, 1–20, doi: [10.1175/ei-d-15-0036.1](https://doi.org/10.1175/ei-d-15-0036.1) (2016).
+
+70. Ruddiman, W. F. Orbital insolation, ice volume, and greenhouse gases. *Quaternary Science Reviews* **22**, 1597–1629, doi: [10.1016/s0277-3791(03)00087-8](https://doi.org/10.1016/s0277-3791(03)00087-8) (2003).
+
+71. Justino, F., Timmermann, A., Merkel, U. & Peltier, W. R. Vol. 19 3–14 (2006).
+
+72. Cheng, H. et al. Ice Age Terminations. *Science* **326**, 248–252, doi: [10.1126/science.1177840](https://doi.org/10.1126/science.1177840) (2009).
+
+73. Gao, Y., Liu, Z. & Lu, Z. Dynamic Effect of Last Glacial Maximum Ice Sheet Topography on the East Asian Summer Monsoon. *Journal of Climate* **33**, 6929–6944, doi: [10.1175/jcli-d-19-0562.1](https://doi.org/10.1175/jcli-d-19-0562.1) (2020).
+
+74. Wen, X. Y., Liu, Z. Y., Wang, S. W., Cheng, J. & Zhu, J. Correlation and anti-correlation of the East Asian summer and winter monsoons during the last 21,000 years. *Nature Communications* **7**, doi: [10.1038/ncomms11999](https://doi.org/10.1038/ncomms11999) (2016).
+
+75. Dixon, R. D., Peyrillé, P. & Guichard, F. Sahelian Precipitation Change Induced by SST Increase: The Contrasting Roles of Regional and Larger-Scale Drivers. *Geophysical Research Letters* **46**, 11378–11387, doi: [10.1029/2019gl084872](https://doi.org/10.1029/2019gl084872) (2019).
+
+76. Folland, C. K., Palmer, T. N. & Parker, D. E. SAHEL RAINFALL AND WORLDWIDE SEA TEMPERATURES, 1901-85. *Nature* **320**, 602–607, doi: [10.1038/320602a0](https://doi.org/10.1038/320602a0) (1986).
+
+77. Giannini, A., Biasutti, M. & Verstraete, M. M. Vol. 64 119–128 (2008).
+
+78. Zhang, R. & Delworth, T. L. Vol. 33 L17712 (2006).
+
+79. Pope, V. D., Gallani, M. L., Rowntree, P. R. & Stratton, R. A. The impact of new physical parametrizations in the Hadley Centre climate model: HadAM3. *Climate Dynamics* **16**, 123–146, doi: [10.1007/s003820050009](https://doi.org/10.1007/s003820050009) (2000).
+
+80. Gordon, C. et al. The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. *Climate Dynamics* **16**, 147–168, doi: [10.1007/s003820050010](https://doi.org/10.1007/s003820050010) (2000).
+
+81. Essery, R., Best, M. & Cox, P. M. Moses 2.2 technical doumentation. *Hadley Centre technical note* **30**, 1–30 (2001).
+
+82. Hoperoft, P. O. & Valdes, P. J. Green Sahara tipping points in transient climate model simulations of the Holocene. *Environmental Research Letters* **17**, doi: [10.1088/1748-9326/ac7c2b](https://doi.org/10.1088/1748-9326/ac7c2b) (2022).
+
+83. Berger, A., Loutre, M. F. & Gallee, H. Sensitivity of the LLN climate model to the astronomical and CO2 forcings over the last 200 ky. *Climate Dynamics* **14**, 615–629, doi: [10.1007/s003820050245](https://doi.org/10.1007/s003820050245) (1998).
+
+84. Loulergue, L. et al. Orbital and millennial-scale features of atmospheric CH4 over the past 800,000 years. *Nature* **453**, 383–386, doi: [10.1038/nature06950](https://doi.org/10.1038/nature06950) (2008).
+
+85. Spahni, R. et al. Atmospheric methane and nitrous oxide of the late Pleistocene from Antarctic ice cores. *Science* **310**, 1317–1321, doi: [10.1126/science.1120132](https://doi.org/10.1126/science.1120132) (2005).
+
+86. de Boer, B., Lourens, L. J. & van de Wal, R. S. W. Persistent 400,000-year variability of Antarctic ice volume and the carbon cycle is revealed throughout the Plio-Pleistocene. *Nature Communications* **5**, doi: [10.1038/ncomms3999](https://doi.org/10.1038/ncomms3999) (2014).
+
+87. The NCAR Command Language (Version 6.6.2) [Software]. (UCAR/NCAR/CISL/TDD, Boulder, Colorado., 2019).
+
+# Supplementary Files
+
+- [SupplementaryFinalDraft.docx](https://assets-eu.researchsquare.com/files/rs-2375224/v1/868c7e439c1d7f26d1d7486b.docx)

34f7befbd20dae5c695338ff41653b8cd42a39e97aca4a39019001bb2d49d423/preprint/images_list.json

@@ -0,0 +1,50 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Morphological heterogeneity in mixed-halide perovskite thin films. a Bandgap from UV-vis-NIR spectroscopy as a function of (FA1\u2212xMAx)Pb(I1\u2212yBry)3 perovskite composition. Grey areas represent compositions that were not considered due to phase instabilities in ambient measurement environments. Shaded regions mark perovskite compositions compatible for top-cells in tandem and triple-junction solar cells. b \u2013 g Surface SEM images of perovskite thin films for different compositions. b {x/y} = 0.25/0.40. c {x/y} = 0.50/0.40. d {x/y} = 0.75/0.40. e {x/y} = 0.25/0.60. f {x/y} = 0.50/0.60. g {x/y} = 0.75/0.60. h, i Three-dimensional AFM height profiles of perovskite thin films. h {x/y} = 0.25/0.40. i {x/y} = 0.50/0.60. j \u2013 o Line cuts from AFM height profiles for perovskite thin films for different compositions {x/y}. j {x/y} = 0.25/0.40, k {x/y} = 0.50/0.40. l {x/y} = 0.75/0.40. m {x/y} = 0.25/0.60. n {x/y} = 0.50/0.60. o {x/y} = 0.75/0.60.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Structural characterization of mixed-halide perovskite thin films. GIWAXS patterns of perovskite films with different compositions. a {x/y} = 0.25/0.40. b {x/y} = 0.50/0.40. c {x/y} = 0.75/0.40. d {x/y} = 0.25/0.50. e {x/y} = 0.50/0.50. f {x/y} = 0.75/0.50. g {x/y} = 0.25/0.60. h {x/y} = 0.50/0.60. i {x/y} = 0.75/0.60. The peaks corresponding to unreacted PbI2 (q \u2248 0.9 \u00c5\u22121) and the (100) plane of the perovskite (q \u2248 1.0 \u00c5\u22121) are marked in panel a. Azimuthal intensity profiles of the main Debye-Scherrer ring (100) as a function of \u03c7 angle from GIWAXS for perovskite compositions with varying MA and Br contents. j 25% MA. k 50% MA. l 75% MA. The data in panels j, k and l have been vertically offset for clarity. Smooth films show ring-like features for the (100) plane at q \u2248 1.0 \u00c5\u22121 whereas films with increasing roughness show intense spots corresponding to preferential orientation.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Crystallization dynamics of mixed-halide perovskite thin films. In situ GIWAXS patterns of perovskites with compositions {x/y} = 0.25/0.60 and {x/y} = 0.75/0.60 during spin-coating. Panels mark time stamps during the spin coating process. a, b 21 s. c, d 25 s. e, f 29 s. g, h31 s. The frame at 25 s represents the casting of the antisolvent onto the substrate. Azimuthal intensity profiles of the main Debye-Scherrer ring (100) as a function of \u03c7 angle from GIWAXS for different perovskite compositions acquired in the 20 \u2013 40 s period of spin-coating. i {x/y} = 0.25/0.60. j {x/y} = 0.75/0.60. The data in panels iand j have been vertically offset for clarity.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Compositional heterogeneity in mixed-halide perovskite thin films from nano-XRF. a Normalized elemental maps of Pb, I, and Br of perovskite thin film with {x/y} = 0.25/0.40. b Map of iodide-to-bromide ratio in shaded region of panel a. Sub-panels A, B, C, and D represent the normalized Pb elemental map in regions highlighted (dashed squares) in the iodide-to-bromide map. c Line cuts of iodide-to-bromide ratios overlapped with local Pb content line cuts marked with (1) and (2) in panel b. d Normalized elemental maps of Pb, I, and Br of perovskite thin film with {x/y} = 0.50/0.60. e Map of iodide-to-bromide ratio in shaded region of panel d. Sub-panels E, F, G, and H represent the normalized Pb elemental map in regions highlighted in the iodide-to-bromide map. f, g Line cuts of iodide-to-bromide ratios overlapped with local Pb content line cuts at points marked with (1), (2), (3) and (4) in panel e. Maps show that smooth films yield homogeneous halide distribution across the film thickness and wrinkled films have iodide-rich domains concentrated at peak-like regions. All scale bars are 10 \u00b5m.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Heterogeneity in photoluminescence emission. Hyperspectral luminescence of perovskite thin films with compositions. a \u2013 d {x/y} = 0.25/0.40. e \u2013 h{x/y} = 0.75/0.40. i \u2013 l {x/y} = 0.75/0.50. Here, 2D emission maps in panels a, e, and i represent the wavelength at emission maximum for pristine films. 2D maps in panels b, f, and j show the wavelength change (\u0394\u03bb) upon continuous illumination (450 nm, 5 min.). Spectra in panels c, g, and kare averaged over the scanned area of pristine (red line) and illuminated (blue shaded) films. Panels d, h, and l show histogram of maximum emission wavelengths in pristine (red) and illuminated (blue) thin films. The maps show that emission heterogeneity increases with increasing MA and Br content and that in heterogeneous films, regions of low-energy emission undergo a smaller redshift after continuous illumination.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "Defect dynamics in mixed-halide perovskite solar cells. a Device configuration. b, c Sensitive external quantum efficiency spectra of perovskite solar cells with different absorber layers. b Br content (y) = 0.40 and MA content (x) = 0.25, 0.50, and 0.75. c MA content (x) = 0.75 and Br content (y) = 0.40, 0.50, and 0.60. Shaded region in panel c represents the sub-bandgap EQE spectrum for the {x/y} = 0.25/0.40 composition. d \u2013 h Band-edge and sub-bandgap photocurrent response in the 0.80 \u2013 2.10 eV range for solar cells using absorber layers for different compositions. Lines represent the spectra of pristine solar cells and open circles represent the spectra acquired after 10 min of continuous illumination (532 nm, 1-Sun equivalent intensity). Dashed vertical lines in panels d \u2013 h indicate the respective optical bandgaps. d {x/y} = 0.25/0.40. e {x/y} = 0.50/0.40. f {x/y} = 0.75/0.40. g {x/y} = 0.75/0.50. h {x/y} = 0.75/0.60.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

34f7befbd20dae5c695338ff41653b8cd42a39e97aca4a39019001bb2d49d423/preprint/preprint.md

@@ -0,0 +1,235 @@
+# Abstract
+
+Compositional heterogeneity in wide-bandgap (1.8 − 2.1 eV) mixed-halide perovskites is a key bottleneck in the processing of high-quality solution-processed thin films, and prevents their application in efficient multijunction solar cells. Notably, mixed-cation (formamidinium-methylammonium) wide-bandgap perovskite films are prone to form micrometer-scale wrinkles which can interfere with the smooth surfaces ideal for multijunction devices. Here, we study the formation dynamics of wrinkled mixed-halide perovskite films and its impact on the local composition and optoelectronic properties. We use in situ X-ray scattering during perovskite film formation to show that crystallization of bromide-rich perovskites precedes that of mixed-halide phases in wrinkled films cast using an antisolvent-based process. Using nanoscopic X-ray fluorescence and hyperspectral photoluminescence imaging, we also demonstrate the formation of iodide- and bromide-rich phases in the wrinkled domains. This intrinsic spatial halide segregation results in an increased local bandgap disorder and Urbach energy. Morphological and compositional heterogeneity also aggravate the formation of sub-bandgap electronic defects reducing photostability and accelerating light-induced segregation of iodide and bromide ions in thin films and solar cells.
+
+Physical sciences/Energy science and technology/Renewable energy/Solar energy/Photovoltaics/Solar cells  
+Physical sciences/Materials science/Materials for energy and catalysis/Solar cells
+
+# Introduction
+
+Mixed-halide wide-bandgap APbX₃ (A is a monovalent cation and X is a halide ion (iodide or bromide)) perovskite semiconductors are promising candidates for use in monolithic multijunction photovoltaic devices where the use of complementary absorber layers enables an increase in photovoltaic performance¹. Using compositions with bromide contents of 30% – 40% in the top-cell, monolithic tandem solar cells have been developed using perovskite and c-Si bottom sub-cells, which have exceeded 28% and 33% power conversion efficiency respectively²,³. Wide-bandgap perovskites with even higher bromide contents (~ 60%) can potentially surpass the 40% efficiency mark in monolithic triple-junction devices⁴–⁷.
+
+However, solution-processed mixed-halide perovskites undergo complex crystallization routes that influence their compositional and morphological homogeneity⁸–¹⁰. A key challenge is the presence of distinct wrinkled morphological domains in mixed-halide perovskite thin films¹¹–¹⁹. The formation of these wrinkles, with peak- and valley-like features, has been attributed to compressive stress in the film developed during crystallization¹⁶,²⁰. The morphological disorder and the residual stress created during film formation then reduce device performance and stability²¹,²². The extent of wrinkling can be controlled by steering the crystallization rate and interactions with the substrate and solvent environments¹⁴,¹⁶,¹⁸,²³. Wrinkling is especially challenging in developing efficient perovskite-based multijunction devices where several ancillary thin (often less than 10 nm) films are needed for charge extraction, surface passivation, and as a recombination layer. To ensure conformal coverage and form continuous layers, surface planarity is important²⁴.
+
+Solution-processed perovskites also suffer from stochastic compositional disorder resulting in local richness or deficiency in ionic species²⁵–³¹. This compositional disorder is particularly crucial in mixed-halide compositions where local variations in halide distribution affect the bandgap³². In turn, this variation in local halide concentration may influence the local chemical potential, defect distribution, and charge-carrier dynamics³²,³³. Morphological disorder has previously been associated with compositional heterogeneity in cesium-containing perovskite thin films with bandgaps < 1.60 eV¹³. However, the correspondence between morphological and compositional heterogeneity in wide-bandgap compositions has not been investigated extensively. This correlation is especially critical because most wide-bandgap, mixed-halide compositions currently suffer from higher defect densities, resulting in lower radiative recombination efficiencies and a stronger tendency for ion migration³⁴,³⁵. As a result, the role of such disorder in determining defect behavior and subsequent effect on photostability is also lacking.
+
+Herein, we study perovskite thin films and devices using mixed-cation (formamidinium (FA) - methylammonium (MA)) mixed-halide (iodide - bromide) compositions prepared using an antisolvent-based deposition method. We first identify the role of the [FA]/[MA] and [I]/[Br] ratios in determining morphological disorder and their influence on crystallographic structure and orientation. Thereafter, we study the crystallization dynamics using synchrotron-based in situ structural characterization and identify key stages during processing where such heterogeneity begins to develop. Furthermore, using synchrotron-based nanoscopic X-ray fluorescence mapping, we correlate morphological heterogeneity to the compositional inhomogeneity in local halide distribution. This compositional heterogeneity causes local changes in the bandgap and photoluminescence energy. Finally, using sensitive photocurrent spectroscopy, we associate disordered compositions with a higher defect density and a corresponding instability under illumination due to light-induced halide segregation.
+
+# Results
+
+Mixed-cation lead mixed-halide perovskite (nominal composition (FA<sub>1−<em>x</em></sub>MA<sub><em>x</em></sub>)Pb(I<sub>1−<em>y</em></sub>Br<sub><em>y</em></sub>)<sub>3</sub>) thin films (compositions hereby denoted as <em>x</em>/<em>y</em>) were prepared using an ethyl acetate antisolvent-based spin-coating route as detailed in the Methods section. Independently changing the methylammonium (<em>x</em>) or bromide (<em>y</em>) contents in the precursor solution alters the bandgap between 1.55 eV (25% MA, 0% Br) and 2.38 eV (75% MA, 100% Br), as determined by UV-vis-NIR spectroscopy (Fig. 1a and Supplementary Fig. 1)<sup>19</sup>. Calculations indicate that the ideal perovskite bandgap is 1.80–1.90 eV or 1.90–2.00 eV for the top-cell in all-perovskite tandem or perovskite-based triple-junction solar cells, respectively<sup>1</sup>. This bandgap corresponds to a bromide content of 40% – 60% for the two applications (identified by shaded regions in Fig. 1a).
+
+We used scanning electron microscopy (SEM) to characterize the surface morphology of the resulting perovskite thin films. Figures 1b – 1g show low magnification surface SEM images of films with compositions <em>x</em>/<em>y</em> = 0.25/0.40, 0.50/0.40, 0.75/0.40, 0.25/0.60, 0.50/0.60, and 0.75/0.60. The images show a smooth surface in films with lower MA and Br contents (<em>x</em>/<em>y</em> = 0.25/0.40, 0.50/0.40) whereas other compositions that are richer in MA and/or Br (<em>x</em>/<em>y</em> = 0.75/0.40, 0.25/0.60, 0.50/0.60, and 0.75/0.60) show additional morphological features on the surface. Specifically, we observe a heterogeneous distribution of large peaks and valleys on the surface of the films with <em>x</em>/<em>y</em> = 0.75/0.40, 0.50/0.60, and 0.75/0.60 whereas the film with <em>x</em>/<em>y</em> = 0.25/0.60 exhibits a morphology intermediate to the smooth and rough surfaces observed. We note that the appearance of morphological heterogeneity does not significantly affect SEM features at 1–2 µm or reveal voids in the film surface (Supplementary Fig. 2).
+
+Figures 1h and 1i show three-dimensional atomic force microscopy (AFM) profiles of two films with compositions <em>x</em>/<em>y</em> = 0.25/0.40 (smooth), and with higher MA and Br content <em>x</em>/<em>y</em> = 0.50/0.60 (rough) to better visualize the distinct peak- and valley-like features observed in rough films. Figures 1j – 1o show line profiles of surface atomic force micrographs (Supplementary Figs. 3 and 4). The feature sizes in the smooth films (<em>x</em>/<em>y</em> = 0.25/0.40, 0.50/40, and 0.25/0.60) are on the order of 100 nm whereas compositions that yield heterogeneous films (<em>x</em>/<em>y</em> = 0.75/0.40, 0.50/0.60, and 0.75/0.60) exhibit feature sizes as large as 1.50–2.0 µm. The line cuts also show that the widths of the features in the more heterogeneous films are on the order of 2.0–5.0 µm<sup>14,16</sup>.
+
+We used synchrotron-based grazing-incidence wide-angle X-ray scattering (GIWAXS) measurements to understand the crystallographic properties of smooth and rough perovskite thin films (Fig. <span class="InternalRef">2</span>). A shift to a higher scattering vector (<em>q</em>) for the peak corresponding to the (100) plane (from <em>q</em> ≈ 1.02 to 1.06 Å<sup>−1</sup>) with increasing MA or Br content suggests the lattice contraction upon the substitution of small cation/anion species (circular averages are shown in Supplementary Fig. 5)<sup>19</sup>. The peak corresponding to unreacted PbI<sub>2</sub> (<em>q</em> ≈ 0.9 Å<sup>−1</sup>) remains unchanged across compositions.
+
+Notably, we find that compositions that show roughness > 1 µm (Fig. 1) also show a preferential crystallographic orientation. For example, compositions yielding smooth films (<em>x</em>/<em>y</em> = 0.25/0.40, 0.50/0.40, and 0.25/0.50) show GIWAXS patterns (Fig. <span class="InternalRef">2</span> a, <span class="InternalRef">2</span> b, and <span class="InternalRef">2</span> d) with broad and complete ring-like features, consistent with randomly oriented crystallites (Fig. <span class="InternalRef">2</span> j). In contrast, films with roughness > 1 µm with composition (<em>x</em>/<em>y</em> = 0.75/0.50, 0.50/0.60, and 0.75/60) show GIWAXS patterns indicating preferential orientation of the (100) plane due to the appearance of intense spots on the ring at <em>q</em> ≈ 1.0 Å<sup>−1</sup> (Fig. <span class="InternalRef">2</span> f, <span class="InternalRef">2</span> h, <span class="InternalRef">2</span> i, <span class="InternalRef">2</span> j and <span class="InternalRef">2</span> l), especially in the out-of-plane direction<sup>18</sup>. Here, the inclusion of Br has a stronger influence in driving preferential orientation than MA (Fig. <span class="InternalRef">2</span> j, <span class="InternalRef">2</span> k and <span class="InternalRef">2</span> l). The change in preferential orientation does not arise from compositional changes only. For example, perovskite films of nominally identical compositions deposited using a two-step interdiffusion method show the random orientation of crystallites for compositions with high MA and Br content, as opposed to the preferential orientation observed in films coated using the one-step antisolvent method (Supplementary Fig. 6)<sup>16,18,36</sup>.
+
+We then performed synchrotron-based in situ GIWAXS measurements during film formation to characterize the crystallization dynamics of smooth and rough perovskite thin films. For these measurements, we processed the perovskite layer on an indium tin oxide (ITO)-coated glass substrate at a spin-coater placed in the X-ray beam path. We chose the compositions (MA-poor <em>x</em>/<em>y</em> = 0.25/0.60 and MA-rich <em>x</em>/<em>y</em> = 0.75/0.60) such that they yield different morphological outcomes (<em>x</em>/<em>y</em> = 0.25/0.60 yields a smooth film while <em>x</em>/<em>y</em> = 0.75/0.60 yields a rough film) but are similar in their nominal Br content. Figure <span class="InternalRef">3</span> shows in situ GIWAXS patterns as a function of spin-coating time, focusing on the time range near the antisolvent casting at <em>t</em> = 25 s.
+
+Both compositions show no scattering features in the initial stages (up to 21 s, Fig. <span class="InternalRef">3</span> a and <span class="InternalRef">3</span> b) of the spin-coating process when the wet film is largely disordered. Upon adding the antisolvent (Fig. <span class="InternalRef">3</span> c and <span class="InternalRef">3</span> d), the film with composition <em>x</em>/<em>y</em> = 0.75/60 immediately shows a strong scattering feature in the out-of-plane direction, corresponding to the (100) crystallographic plane of the perovskite phase. In contrast, the (100) feature is absent in the film with composition <em>x</em>/<em>y</em> = 0.25/60, appearing later at 29 s (Fig. <span class="InternalRef">3</span> e and <span class="InternalRef">3</span> f) and increasing in intensity thereafter (Fig. <span class="InternalRef">3</span> g and <span class="InternalRef">3</span> h). Here, as seen in Fig. <span class="InternalRef">2</span> g, the feature resembles a ring which is indicative of a random orientation of crystalline perovskite phases. We also observe similar crystallization dynamics in the characterization of other smooth (<em>x</em>/<em>y</em> = 0.25/0.50) and rough (<em>x</em>/<em>y</em> = 0.75/0.50) perovskite films (Supplementary Figs. 7 and 8).
+
+These results indicate a faster crystallization and an earlier onset of preferential orientation of the MA-rich <em>x</em>/<em>y</em> = 0.75/60 thin film compared to the MA-poor <em>x</em>/<em>y</em> = 0.25/60 perovskite<sup>31,37</sup>. This result agrees with prior observations of faster crystallization of Br-rich phases in a MA-based environment than in a FA-based environment<sup>31</sup>. The faster crystallization of Br-rich phases can also be attributed to weak interactions of bromide precursors and DMF, their corresponding poor solubility, and the poor stability of lead bromide complexes in solution, which collectively favor faster nucleation<sup>9,38,39</sup>. Furthermore, it has been demonstrated that Br-based phases form the perovskite phase directly whereas I-based phases undergo crystallization through intermediate complexes<sup>31</sup>. MA-containing Br-based perovskite phases have also been shown to grow as highly oriented phases<sup>37</sup>. As a result, based on the GIWAXS data showing faster crystallization and more preferential (100) orientation in films with roughness > 1 µm (<em>x</em>/<em>y</em> = 0.75/0.50 and 0.75/0.60), we propose that in an MA-rich environment, heterogeneous crystal nucleation leads to the formation of oriented bromide-rich perovskites immediately after antisolvent casting (Supplementary Fig. 9) followed by the incorporation of iodide-containing phases<sup>18</sup>, and that the heterogeneous crystallization causes the film wrinkling. The larger change in the scattering vector (Δ<em>q</em>) for the perovskite composition <em>x</em>/<em>y</em> = 0.75/0.60 during the film-coating and thermal-annealing stages of the crystallization compared to that for <em>x</em>/<em>y</em> = 0.25/0.60, supports the hypothesis that Br- and I-rich phases crystallize at different rates in rough films (Supplementary Fig. 10).
+
+Subsequently, we used synchrotron-based nanoscopic X-ray fluorescence (nano-XRF) microscopy to study the bulk composition of perovskite layers and observe the influence of morphological heterogeneity on compositional disorder (Fig. <span class="InternalRef">4</span>). The technique relies on elemental X-ray emission signatures to map the microscopic distribution of constituents<sup>25,32</sup>. Here, we used the elemental map of Pb (Supplementary Fig. 11) as an indicator for film thickness since peak-like features increase the local film volume that corresponds to a higher X-ray fluorescence intensity.
+
+We used nano-XRF mapping to analyze the spatial elemental distribution in smooth and rough perovskite thin films. For example, a film with composition <em>x</em>/<em>y</em> = 0.25/0.40, which yields a uniform surface (Fig. 1b), shows a homogeneous distribution of Pb, I, and Br (Fig. <span class="InternalRef">4</span> a). As a result, the iodide-to-bromide ratio in the film also shows a homogeneous distribution (Fig. <span class="InternalRef">4</span> b). We examined four highlighted regions (dashed boxes in Fig. <span class="InternalRef">4</span> b) to confirm that the iodide-to-bromide ratio is largely independent of the Pb content in the region, i.e., the local thickness of the film. Line cuts of iodide-to-bromide ratio overlapped with the Pb content (Fig. <span class="InternalRef">4</span> c) further confirm the lack of correspondence between the two quantities for a smooth perovskite film. Similar homogeneous distribution of ions is observed for other smooth films with compositions <em>x</em>/<em>y</em> = 0.50/0.40, 0.50/0.50, and 0.25/0.60, as shown in Supplementary Fig. 12.
+
+In contrast, compositions that yield rough surfaces <em>x</em>/<em>y</em> = 0.75/0.40, 0.75/0.50, 0.50/0.60, 0.75/0.60, show a heterogeneous distribution of Pb, Br, and I (Fig. <span class="InternalRef">4</span> d and Supplementary Fig. 13). For example, the elemental maps (Pb, I, and Br) for the perovskite film with composition <em>x</em>/<em>y</em> = 0.50/0.60 (Fig. <span class="InternalRef">4</span> d) show the presence of peak- and valley-like features, as observed in the surface SEM characterization (Fig. 1). Furthermore, the iodide-to-bromide ratio map (Fig. <span class="InternalRef">4</span> e) shows regions of iodide richness and deficiencies which follow the peaks and valleys observed through the Pb elemental map (highlighted regions in Fig. <span class="InternalRef">4</span> d). Line cuts at four distinct locations of the film and two-dimensional maps (Fig. <span class="InternalRef">4</span> f, <span class="InternalRef">4</span> g and Supplementary Fig. 14) further confirm the positive correlation between increasing layer thickness and higher iodide concentration. Based on these observations, we hypothesize that compositional differences (MA- and Br-richness) influencing the rates of crystallization of iodide- and bromide-rich phases<sup>31,32,38</sup> drive the development of spatial halide heterogeneity during film crystallization<sup>40</sup>, with iodide-rich phases forming in the peak-like regions of the films whereas bromide-containing phases predominantly crystallize at the valley-like regions. Compositional heterogeneity has previously been discussed in lead halide perovskites<sup>25,32,41</sup>; however, this work is the first to report its association with surface morphology in wide-bandgap compositions relevant for multijunction devices.
+
+Because of the link between perovskite bandgap and stoichiometry, we expect that compositional heterogeneity will also cause spatially heterogeneous bandgap disorder with regions of high iodide content exhibiting lower bandgaps, and high relative bromide content exhibiting wider bandgaps respectively<sup>17,30,42</sup>. To test this hypothesis, we used hyperspectral photoluminescence (PL) imaging to characterize perovskite thin films’ emissive properties. We performed hyperspectral PL mapping of as-prepared films (referred to as “pristine”), excited with a mercury halide lamp in the 350–450 nm range at 130 mW cm<sup>−2</sup> excitation intensity, with an acquisition time of 1 min. We then continuously illuminated the film for another 5 min., and measured again with a 1 min. acquisition time (referred to as “illuminated”). Figure <span class="InternalRef">5</span> displays the PL emission peak wavelength maps from those hyperspectral PL imaging of perovskite films deposited on glass with compositions <em>x</em>/<em>y</em> = 0.25/0.40, 0.75/0.40, and 0.75/0.50 (Supplementary Fig. S15), illuminating the top surface (all films were encapsulated in the glovebox with UV-curable glue).
+
+As we demonstrated with AFM and SEM earlier, films with low Br content <em>y</em> = 0.40 and increasing MA content (<em>x</em>) show more pronounced wrinkles (Fig. 1). Using hyperspectral PL imaging, we also observe that wrinkling increases with MA content in low Br films <em>y</em> = 0.40. In a homogeneous film with low MA and Br content (<em>x</em>/<em>y</em> = 0.25/0.40), the PL emission wavelength is centered at ~ 690 nm (Fig. <span class="InternalRef">5</span> a, <span class="InternalRef">5</span> c, and <span class="InternalRef">5</span> d). The spatially averaged PL spectrum also shows a single peak (Fig. <span class="InternalRef">5</span> c), indicating that the emission is the same across the scanned area. On the other hand, for a rougher film with composition <em>x</em>/<em>y</em> = 0.75/0.40, the PL map and spatially averaged spectrum show two distinct emission peaks, indicative of wide- and narrow-bandgap domains within the scanned region (Fig. <span class="InternalRef">5</span> e, <span class="InternalRef">5</span> g, and <span class="InternalRef">5</span> h). The emission is primarily dominated by wide-bandgap domains, with a peak maximum at ~ 685 nm (Fig. <span class="InternalRef">5</span> g). Sparsely distributed, narrow-bandgap domains contribute to the second PL peak maximum at 730–750 nm. Additionally, we note that narrow-bandgap sites (<em>λ</em><sub>long</sub>) are comparatively brighter than wide-bandgap sites (<em>λ</em><sub>short</sub>) (Supplementary Fig. 16). We attribute the increased brightness of the lower bandgap regions to charge-carrier funneling from wider bandgap sites, as proposed previously<sup>43</sup>. This emission heterogeneity is in good agreement with the compositional heterogeneity we observed through nano-XRF mapping (Supplementary Fig. 13), where we found the raised peak-like regions to be comparatively iodide-rich and correspondingly emit at lower energies/longer wavelengths.
+
+Over the range of compositions studied, we observed that PL heterogeneity in the pristine film increases further with increasing the bromide content (Fig. <span class="InternalRef">5</span> i, <span class="InternalRef">5</span> k and <span class="InternalRef">5</span> l), in <em>x</em>/<em>y</em> = 0.75/0.50, with the maximum emission wavelength distribution broadening from 650 to 670 nm (Fig. <span class="InternalRef">5</span> l). The spatially averaged emission spectrum shows a short- (655 nm) and long-wavelength (745 nm) emission contribution, consistent with halide heterogeneity as a result of morphological disorder (Supplementary Fig. 13). However, the redshifted emission at the peak-like regions could also result from self-absorption or optical interference effects that occur at increased layer thickness<sup>44,45</sup>. To rule out optical effects, we used time-of-flight secondary ion mass spectrometry (TOF-SIMS) mapping in the hyperspectral imaging area (Supplementary Fig. 17) and found that stronger long-wavelength emission regions correspond to local increase in the iodide-to-bromide ratio. This correlation is strong evidence that the local redshifts of emission wavelength and increase in PL intensity are due to the high iodide content in peak-like regions. Together, TOF-SIMS and hyperspectral mapping correlation verified that PL wavelength mapping can provide important qualitative information about local compositional heterogeneity in these films.
+
+Subsequently, we continuously illuminated the same samples for 5 min. with blue (450 nm) light with an intensity of 130 mW cm<sup>−2</sup> (Supplementary Fig. 18). This illumination was done in order to cause ion migration that is known to lead to light-induced halide segregation<sup>34</sup>. In Figs. <span class="InternalRef">5</span> b, <span class="InternalRef">5</span> f and <span class="InternalRef">5</span> j, we show the spatially resolved PL redshift after 5 min. of illumination, calculated from Figs. <span class="InternalRef">5</span> a, <span class="InternalRef">5</span> e, <span class="InternalRef">5</span> i and Supplementary Fig. 18. In a compositionally homogeneous, smooth film with composition <em>x</em>/<em>y</em> = 0.25/0.40, continuous illumination does not affect the PL emission maximum (Fig. <span class="InternalRef">5</span> b and <span class="InternalRef">5</span> d), but only causes a slight broadening of the PL full-width at half-maximum (FWHM) from 35 nm to 45 nm (Fig. <span class="InternalRef">5</span> c). In contrast, films with heterogeneous PL emission prior to illumination (<em>x</em>/<em>y</em> = 0.75/0.40 and 0.75/0.50) show a redshift in the maximum emission wavelength by approximately 50–100 nm (Fig. <span class="InternalRef">5</span> f, <span class="InternalRef">5</span> h, <span class="InternalRef">5</span> j, and <span class="InternalRef">5</span> l). We note that the change in the emission wavelength after illumination is lower in the peak-like regions (Δ<em>λ</em> ~ 10 nm) compared to the valley-like regions (Δ<em>λ</em> ~ 50 nm) (Fig. <span class="InternalRef">5</span> f and <span class="InternalRef">5</span> j). This is because in peak-like areas the iodide-to-bromide ratio is already high before illumination, as we show in Supplementary Fig. 17 via correlated TOF-SIMS. For the perovskite film with composition <em>x</em>/<em>y</em> = 0.75/0.40, the initial emission with two peak features evolves to one broad redshifted emission peak indicating the formation of a broad distribution of emissive iodide-rich species (Fig. <span class="InternalRef">5</span> g). The illuminated <em>x</em>/<em>y</em> = 0.75/0.50 film shows a higher low-energy emission of 745 nm (Fig. <span class="InternalRef">5</span> k), and a weak emission from short-wavelength range of 655 nm.
+
+We performed TOF-SIMS mapping of the perovskite film (<em>x</em>/<em>y</em> = 0.75/0.50) before and after continuous illumination to further understand the effect of ion migration as a function of morphological heterogeneity. By tracking the iodide content before and after illumination, we found that peak-like regions in the film undergo an increase in iodide content on the film surface (Supplementary Fig. 19a). In contrast, a homogeneous film (<em>x</em>/<em>y</em> = 0.25/0.50) shows a small and homogeneous increase in iodide content across the film surface upon continuous illumination (Supplementary Fig. 19b), indicating comparatively reduced halide migration as expected from hyperspectral PL imaging (Fig. <span class="InternalRef">5</span> b).
+
+Finally, we fabricated perovskite solar cells (Fig. <span class="InternalRef">6</span> a) in an inverted (<em>p</em>-<em>i</em>-<em>n</em>) device architecture using [2-(9<em>H</em>-carbazol-9-yl)ethyl]phosphonic acid (2PACz) and C<sub>60</sub> as hole- and electron-transport layers, respectively (Supplementary Fig. 20). The transport layers were especially chosen due to their compatibility with multijunction device architectures and the conformal deposition of C<sub>60</sub> with thermal evaporation<sup>46</sup>. Sensitive photocurrent spectroscopy was used to characterize defect dynamics in solar cells. The increase in defect density as a function of Br content has previously been studied<sup>35,38,47</sup>, but the role of morphological and compositional disorder on electronic defects of wide-bandgap perovskites is relatively poorly understood. For a solar cell with composition <em>x</em>/<em>y</em> = 0.25/0.40 (Fig. <span class="InternalRef">6</span> b), the external quantum efficiency (EQE), calculated from the photocurrent spectrum, shows a flat above-bandgap (> 1.8 eV) EQE profile followed by an exponential drop (Urbach tail) at the bandgap-edge of the active layer. Following that, in the sub-bandgap 0.80–1.60 eV region, a clear EQE contribution of approx. 10<sup>−7</sup> – 10<sup>−6</sup> can be observed (at approx. 1.35 eV). A second, less prominent, feature can also be observed at lower energies (at approx. 1.0 eV). Previous work has shown that such sub-bandgap features originate from electronic defects near the perovskite/C<sub>60</sub> interface<sup>48–50</sup>, and that changes in the sub-bandgap photocurrent intensity and photocurrent contribution at the Urbach tail upon prolonged illumination can be associated with light-induced halide segregation<sup>47,51,52</sup>.
+
+With increasing MA content, the optical bandgap increases, causing a blueshift in the EQE onset. At <em>x</em> = 0.75, additionally, the sub-bandgap contribution (< 1.5 eV) appears less pronounced, manifesting as a broad feature instead of a peak as observed in compositions <em>x</em>/<em>y</em> = 0.25/0.40 and 0.50/0.40. This broadening is likely related to the increased roughness resulting from morphological heterogeneity in such compositions (Fig. 1), which reduces optical interference by increasing light scattering<sup>48</sup>. The morphological heterogeneity as a function of increasing MA content also manifests as an increase in the Urbach energy, indicating an increase in band-edge energetic disorder<sup>32,53,54</sup>.
+
+These effects are much more pronounced upon increasing the bromide content (<em>x</em>/<em>y</em> = 0.75/0.50 and 0.75/0.60). Both compositions, for example, show that an additional shoulder appears on the Urbach tail (1.6–2.0 eV), likely indicating the presence of low-energy iodide-rich domains in the pristine device (Fig. <span class="InternalRef">6</span> c)<sup>47,51,52</sup>. These observations agree with the non-uniform halide distribution, as observed using nano-XRF microscopy (Fig. <span class="InternalRef">4</span>) and hyperspectral PL imaging (Fig. <span class="InternalRef">5</span>). Such an increase in band-edge disorder has been shown to limit the open-circuit voltage in solar cells<sup>55</sup>. Furthermore, the sub-bandgap photocurrent contribution in the 0.8–1.6 eV increases by approx. an order of magnitude for <em>x</em>/<em>y</em> = 0.75/0.60 composition, indicating a higher defect density as compared to the composition <em>x</em>/<em>y</em> = 0.25/0.40. The presence of sub-bandgap defects has been consistently associated with high non-radiative recombination in perovskite solar cells<sup>50,56,57</sup>. Taken together, these observations indicate a higher degree of band-edge disorder and sub-bandgap defect density with increasing compositional/morphological disorder in wide-bandgap perovskites.
+
+We used continuous illumination (532 nm, 1-Sun equivalent intensity, 5 min.) to induce defect migration and drive halide segregation in the solar cells<sup>58</sup>. Halide migration in mixed-halide wide-bandgap perovskites can result in redshift of the band-edge, related to the formation of iodide-rich phases, and also cause an increase in the sub-bandgap defect density<sup>34,47,51,52</sup>. Figures <span class="InternalRef">6</span> d – <span class="InternalRef">6</span> h show EQE spectra of solar cells following the photo-stress. In solar cells with low MA and low Br content (<em>x</em>/<em>y</em> = 0.25/0.40 and 0.50/0.40), that show smooth morphology and minimal compositional heterogeneity (Figs. <span class="InternalRef">6</span> d and <span class="InternalRef">6</span> e), light-induced changes to the EQE spectrum are minimal, consistent with our observations from hyperspectral PL and TOF-SIMS imaging. In contrast, high MA and high Br containing solar cells with rough perovskite layers (<em>x</em>/<em>y</em> = 0.75/0.40) show a change in the band-edge signal (Fig. <span class="InternalRef">6</span> f), indicating an increase in iodide-rich phase concentration, in agreement with changes observed through PL imaging. Furthermore, in compositions such as <em>x</em>/<em>y</em> = 0.75/0.50 and 0.75/0.60 (Figs. <span class="InternalRef">6</span> g and <span class="InternalRef">6</span> h), a large increase in the band-edge EQE contribution indicates the formation of a large density of iodide-rich phases because of light-induced halide segregation. Moreover, in composition <em>x</em>/<em>y</em> = 0.75/0.60, the change in the Urbach tail is accompanied by an increase in the sub-bandgap defect contribution (0.8–1.6 eV) to the EQE spectrum, indicating an increased defect density upon continuous illumination<sup>47</sup>. Similar trends in photocurrent contribution from band-edge states and sub-bandgap electronic defects are observed in other solar cells using smooth (<em>x</em>/<em>y</em> = 0.25/0.50) or rough (<em>x</em>/<em>y</em> = 0.50/0.60) perovskite layers (Supplementary Fig. 21).
+
+# Summary
+
+Controlling morphological and compositional heterogeneity of wide-bandgap perovskite thin films is key to the development of efficient perovskite-based multijunction solar cells. Using the perovskite composition as a variable, we find that MA- and Br-rich compositions are vulnerable to the formation of wrinkled morphology. We use GIWAXS characterization to associate the nominal composition of the film, the morphology of the layer and its structural properties such as preferential orientation. We identify that fast nucleation of Br-rich phases upon antisolvent casting is associated to the development of heterogeneity due to heterogeneous crystallization of bromide- and iodide-containing perovskite phases. The resulting features can be 1–2 µm in height and can span 2–5 µm laterally, making them unsuitable for solution-processed multijunction device stacks.
+
+Using nano-XRF microscopy, we find a correspondence between surface wrinkling and halide heterogeneity wherein iodide-rich phases preferentially form at peak-like regions of rough films, influencing the local bandgap distribution across the film. The resulting local narrow-bandgap sites that are rich in iodide ions cause a broadening of band-edges, a higher Urbach energy, and a larger distribution in photoluminescence emission energies, which are potentially detrimental to device performance. Crucially, we also find that compositional/morphological heterogeneity aggravates sub-bandgap defect density. Finally, the combination of high defect density and local halide heterogeneity further reduces material stability to defect-driven light-induced halide segregation under illumination.
+
+The results show how fast nucleation of bromide-rich regions, mediated by the presence of methylammonium, followed by the slower crystallization of iodide-rich regions leads to compositional and morphological heterogeneity that degrades both the performance and stability of the perovskite semiconductor and the resulting solar cells. This insight emphasizes that controlling crystallization kinetics through processing variables such as precursor solubility, film drying conditions and the use of agents that mediate crystallite nucleation and growth can be used to suppress film wrinkling and improve the compositional homogeneity of bromide-rich perovskite compositions<sup>16,18</sup>. These methods provide potential routes to enhance the performance and stability of wide-bandgap perovskites that currently limit the development of multijunction photovoltaics.
+
+# Methods
+
+## Materials
+
+All materials were used as received and were stored in an inert environment prior to use. FAI, MAI, FABr and MABr were purchased from Greatcell Solar Materials. 2PACz (> 98%), PbI<sub>2</sub> (99.99%, trace metal basis) and PbBr<sub>2</sub> (> 98%) were purchased from TCI Chemicals. DMF (99.8%), DMSO (99.9%, anhydrous) and ethyl acetate (99.8%) were purchased from Sigma Aldrich. C<sub>60</sub> was purchased from SES Research and bathocuproine (BCP) was purchased from Lumtec. Ethanol (< 0.1% H<sub>2</sub>O) was purchased from Merck Millipore.
+
+## Sample Preparation
+
+2PACz was dissolved in anhydrous ethanol at a concentration of 0.33 mg mL<sup>−1</sup> by sonication prior to use. To prepare the perovskite solution, PbI<sub>2</sub> (691.5 mg mL<sup>−1</sup>) and PbBr<sub>2</sub> (550.5 mg mL<sup>−1</sup>) were each dissolved overnight at 60°C in solvent mixtures containing DMF and DMSO in a volumetric ratio of 4:1. The solutions were cooled to room temperature, following which 458 µL PbI<sub>2</sub> was added to each 98.4 mg FAI and 91.0 mg MAI respectively. 458 µL PbBr<sub>2</sub> was added to each 75.1 mg FABr and 64.1 mg MABr respectively. The resulting solutions (FAPbI<sub>3</sub>, MAPbI<sub>3</sub>, FAPbBr<sub>3</sub>, and MAPbBr<sub>3</sub>) were stirred at 60°C for approx. 1 h. Following that, the solutions were mixed based on the desired FA:MA and I:Br ratios and stirred at 60°C for approx. 1 h. The solutions were then cooled to room temperature prior to use.
+
+Glass and glass/ITO (Naranjo Substrates, 15–17 Ω sq<sup>−1</sup>) substrates were cleaned by sequential cleaning in acetone, followed by scrubbing with sodium dodecyl sulphate (Acros, 99%) soap solution in deionized water, sonication in soap solution, rinsing in deionized water followed by sonication in 2-propanol. Prior to use, the substrates were exposed to UV-ozone treatment for 30 min after which they were transferred to a N<sub>2</sub>-filled glove-box.
+
+Perovskite thin films were deposited by spin-coating 150 µL of the precursor at 4000 rpm (5 s to ramp to 4000 rpm) for 35 s. At approx. 25 s from the beginning of the spin-coating, 300 µL of ethyl acetate was cast onto the spinning substrate. The substrates were immediately annealed at 100°C for 30 min.
+
+For solar cells, 2PACz was spin-coated at 3000 rpm for 30 s followed by thermal annealing at 100°C for 10 min. 20 nm C<sub>60</sub> and 8 nm BCP were sequentially evaporated at a rate of 0.5 Å s<sup>−1</sup>. 100 nm Ag electrode was thermally evaporated to complete the device. The nominal area of the solar cells determined by the overlap of the ITO and Ag areas is 9 mm<sup>2</sup>.
+
+## Thin film and solar cell characterization
+
+UV-vis-NIR spectra of perovskite thin films were measured using PerkinElmer Lambda 1050 UV-vis-NIR spectrophotometer. SEM images were acquired using FEI Quanta 3D FEG microscope, operated with a 5 kV electron beam and a secondary electron detector. AFM measurements were conducted using a Dimension 3100 AFM in tapping mode. For determining the *J*−*V* characteristics a Keithley 2400 SMU was used. A tungsten halogen lamp, filtered by a Schott GG385 UV filter and a Hoya LB120 daylight filter, was used to simulate 100 mW cm<sup>−2</sup> of visible light. A shadow mask with 0.0676 cm<sup>2</sup> aperture was used to define the illuminated cell area. *J*−*V* scans involved sweeping the applied voltage (with no pre-biasing) from +1.5 to −0.5 V at a rate of 0.25 V s<sup>−1</sup>. EQE measurement was performed in a nitrogen atmosphere. The probe light source was generated by a 50 W tungsten-halogen lamp (Philips focusline), which was modulated at 160 Hz with a mechanical chopper (Stanford Research, SR 540) before passing into a monochromator (Oriel, Cornerstone 130). The spectral response of the device was recorded as a voltage from a pre-amplifier (Stanford Research, SR 570) using a lock-in amplifier (Stanford Research, SR 830), and was calibrated by a reference silicon cell. To accurately determine the current density, a green LED (530 nm, Thorlabs M530L3, driven by a DC4104 driver) was used as a light bias to provide the solar cell with approximately one sun illumination intensity.
+
+## GIWAXS characterization
+
+GIWAXS measurements were done at beamline 11-BM at Brookhaven National Laboratory on perovskite films deposited on glass substrates cut to a size of approx. 0.5 cm × 0.5 cm. The samples were measured at an incident angle (*α*<sub>i</sub>) 0.5° with a 10 s exposure time. The X-ray beam had an energy of 13.5 keV, 0.2 mm (height) × 0.05 mm (width) size, 1 mrad divergence and an energy resolution of 0.7%. Diffraction images were recorded using a Pilatus 800k detector. Data was analyzed using the SciAnalysis package provided by the beamline. In situ GIWAXS was performed during spin coating and thermal annealing in a custom-made spin-coater attached to beamline 12.3.2 at the Advanced Light Source (ALS), Lawrence Berkeley National Laboratory. The incoming X-ray beam was at an angle of 0.5° with a beam energy of 10 keV. A DECTRIS Pilatus1 M X-ray detector at an angle of 35° to the sample plane and a sample−detector distance of ∼ 186 mm was used. Measurements were carried out on an area of 0.1 mm<sup>2</sup> (10 mm × 0.01 mm) under an overpressure of N<sub>2</sub> in the spin-coater chamber. Samples were heated using a ramp rate of 4°C s<sup>−1</sup>. The diffraction data was collected with a framerate of approximately 1.875 s<sup>−1</sup>.
+
+## X-ray fluorescence microscopy
+
+X-ray fluorescence (XRF) microscopy measurements were conducted using the Advanced Photon Source (APS) at beamline 2-ID-D at Argonne National Laboratory. A synchrotron X-ray energy of 14 keV was used and a step size of 0.15 µm and 50 ms dwell time. Data were analyzed using the MAPS software and spectrum fitting was used to deconvolute overlapping peaks and background from fluorescence data. In addition, after a standard calibration, it was possible to quantify the mass concentration in the sample to calculate molar ratios. The NIST thin-film standards SRM 1832 and 1833 were used for calibration.
+
+### Hyperspectral PL
+
+Hyperspectral measurements were performed using a Photon etc. IMA upright microscope fitted with a transmitted darkfield condenser and a 60X objective (Nikon Plan RT, NA 0.7, CC 0-1.2). The excitation was done using a mercury halide lamp (Nikon ultrahigh pressure 130 W mercury lamp) passing through a 450 nm short-pass filter and emission was collected through a 500 dichroic filter and 550 nm long-pass filter. The lamp has six levels of light intensity, and all the measurements were taken using the lowest intensity (ND32) with total incident power on the sample of 130 mW cm<sup>−2</sup>. The scan duration was 1 min 15 s per frame. Post-processing was done in the proprietary Photon etc. PHySpec software.
+
+## Time-of-Flight Secondary Ion Mass Spectrometry
+
+TOF-SIMS negative ion data were acquired on an IONTOF TOF.SIMS5 spectrometer using a 25 keV Bi<sup>3+</sup> cluster ion source in the delayed extraction mode. The ion source was operated at a current of 0.04 pA that was rastered over a 100 µm × 100 µm area at 256 pixels × 256 pixels for a primary ion dose of 2.5 × 10<sup>11</sup> ions cm<sup>−2</sup>. A low energy flood gun was used for charge neutralization. Spectra were acquired over a mass range of *m*/*z* = 0 to 800 amu. The data was calibrated using the C<sub>4</sub>H-, Br-, I- and PbI<sub>2</sub>- peaks. Mass resolution of the *m*/*z* = 49.005 (C<sub>4</sub>H-) peak was around 3500.
+
+## Sensitive EQE spectroscopy
+
+Sensitive EQE measurements to characterize sub-bandgap states were conducted using an Osram 64655 HLX 250 W halogen lamp as the illumination source. The light was mechanically chopped at 333 Hz using an Oriel 3502 chopper and was subsequently passed through a monochromator (Oriel, Cornerstone 260) and appropriate sorting filters. The solar cell was mounted into a nitrogen-filled sample holder and its response was recorded as a voltage from a pre-amplifier (Stanford Research, SR 570) using a lock-in amplifier (Stanford Research, SR 830). The measurements were calibrated using Si and InGaAs reference cells. To illuminate the solar cells, a 532 nm CW laser (B&W Tek Inc., BWN-532-20E/56486) was used and its intensity and illumination area was adjusted to match 1-Sun equivalent intensity and the solar cell area using a set of neutral density filters and fish-eye lenses. The data was normalized to the drop in photocurrent that marks the band edge in the spectrum of the pristine solar cell. In photostability studies, spectra were scaled for the signal produced from pristine solar cells in order to estimate the change in the above-bandgap EQE.
+
+# References
+
+1. Hörantner, M. T. et al. The Potential of Multijunction Perovskite Solar Cells. *ACS Energy Lett*. **2**, 2506–2513 (2017).
+
+2. Lin, R. et al. All-perovskite tandem solar cells with 3D/3D bilayer perovskite heterojunction. *Nature* **620**, 994–1000 (2023).
+
+3. Mariotti, S. et al. Interface engineering for high-performance, triple-halide perovskite-silicon tandem solar cells. *Science* **381**, 63–69 (2023).
+
+4. Eperon, G. E., Hörantner, M. T. & Snaith, H. J. Metal halide perovskite tandem and multiple-junction photovoltaics. *Nat. Rev. Chem.* **1**, 95 (2017).
+
+5. Wang, Z. et al. Suppressed phase segregation for triple-junction perovskite solar cells. *Nature* **618**, 74–79 (2023).
+
+6. Wang, J. et al. Halide homogenization for low energy loss in 2-eV-bandgap perovskites and increased efficiency in all-perovskite triple-junction solar cells. *Nat. Energy* **9**, 70 – 80 (2024).
+
+7. Choi, Y. J., Lim, S. Y., Park, J. H., Ji, S. G. & Kim, J. Y. Atomic Layer Deposition-Free Monolithic Perovskite/Perovskite/Silicon Triple-Junction Solar Cells. *ACS Energy Lett.* **8**, 3141–3146 (2023).
+
+8. Xu, F., Zhang, M., Li, Z., Yang, X. & Zhu, R. Challenges and Perspectives toward Future Wide‐Bandgap Mixed‐Halide Perovskite Photovoltaics. *Adv. Energy Mater.* **13**, 2203911 (2023).
+
+9. Rehermann, C. et al. Origin of Ionic Inhomogeneity in MAPb(I<sub>x</sub>Br<sub>1-x</sub>)<sub>3</sub> Perovskite Thin Films Revealed by In-Situ Spectroscopy during Spin Coating and Annealing. *ACS Appl. Mater. Interfaces* **12**, 30343–30352 (2020).
+
+10. Gratia, P. et al. The Many Faces of Mixed Ion Perovskites: Unraveling and Understanding the Crystallization Process. *ACS Energy Lett* **2**, 2686–2693 (2017).
+
+11. Guo, B. et al. In situ Stress Monitoring Reveals Tension and Wrinkling Evolutions during Halide Perovskite Film Formation. *ACS Energy Lett.* **9**, 75–84 (2023).
+
+12. Tejada, A. et al. Optical characterization and bandgap engineering of flat and wrinkle-textured FA<sub>0.83</sub>Cs<sub>0.17</sub>Pb(I<sub>1-x</sub>Br<sub>x</sub>)<sub>3</sub> perovskite thin films. *J. Appl. Phys.* **123**, 175302 (2018).
+
+13. Bercegol, A. et al. Spatial Inhomogeneity Analysis of Cesium-Rich Wrinkles in Triple-Cation Perovskite. *J. Phys. Chem. C* **122**, 23345–23351 (2018).
+
+14. Braunger, S. et al. Cs<sub>x</sub>FA<sub>1−x</sub>Pb(I<sub>1−y</sub>Br<sub>y</sub>)<sub>3</sub> Perovskite Compositions: The Appearance of Wrinkled Morphology and its Impact on Solar Cell Performance. *J. Phys. Chem. C* **122**, 17123–17135 (2018).
+
+15. Beal, R. E. et al. Structural Origins of Light-Induced Phase Segregation in Organic-Inorganic Halide Perovskite Photovoltaic Materials. *Matter* **2**, 207–219 (2020).
+
+16. Bush, K. A. et al. Controlling Thin-Film Stress and Wrinkling during Perovskite Film Formation. *ACS Energy Lett.* **3**, 1225–1232 (2018).
+
+17. Isikgor, F. H. et al. Concurrent cationic and anionic perovskite defect passivation enables 27.4% perovskite/silicon tandems with suppression of halide segregation. *Joule* **5**, 1566–1586 (2021).
+
+18. Kim, S. G. et al. How antisolvent miscibility affects perovskite film wrinkling and photovoltaic properties. *Nat. Commun* **12**, 1554 (2021).
+
+19. Jesper Jacobsson, T. et al. Exploration of the compositional space for mixed lead halogen perovskites for high efficiency solar cells. *Energy Environ. Sci.* **9**, 1706–1724 (2016).
+
+20. Dailey, M., Li, Y. & Printz, A. D. Residual Film Stresses in Perovskite Solar Cells: Origins, Effects, and Mitigation Strategies. *ACS Omega* **6**, 30214–30223 (2021).
+
+21. Tong, J. et al. High-performance methylammonium-free ideal-band-gap perovskite solar cells. *Matter* **4**, 1365–1376 (2021).
+
+22. Fu, J. et al. Synergistic Effects of Interfacial Energy Level Regulation and Stress Relaxation via a Buried Interface for Highly Efficient Perovskite Solar Cells. *ACS Nano* **17**, 2802–2812 (2023).
+
+23. Rolston, N. et al. Engineering Stress in Perovskite Solar Cells to Improve Stability. *Adv. Energy Mater.* **8**, 1802139 (2018).
+
+24. Xiao, K. et al. Solution-processed monolithic all-perovskite triple-junction solar cells with efficiency exceeding 20%. *ACS Energy Lett.* **5**, 2819–2826 (2020).
+
+25. Correa-Baena, J.-P. et al. Homogenized halides and alkali cation segregation in alloyed organic-inorganic perovskites. *Science* **363**, 627–631 (2019).
+
+26. Gratia, P. et al. Intrinsic Halide Segregation at Nanometer Scale Determines the High Efficiency of Mixed Cation/Mixed Halide Perovskite Solar Cells. *J. Am. Chem. Soc.* **138**, 15821–15824 (2016).
+
+27. Dang, H. X. et al. Multi-cation Synergy Suppresses Phase Segregation in Mixed-Halide Perovskites. *Joule* **3**, 1746–1764 (2019).
+
+28. Mundt, L. E. et al. Mixing Matters: Nanoscale Heterogeneity and Stability in Metal Halide Perovskite Solar Cells. *ACS Energy Lett.* **7**, 471–480 (2022).
+
+29. Brivio, F., Caetano, C. & Walsh, A. Thermodynamic Origin of Photoinstability in the CH<sub>3</sub>NH<sub>3</sub>Pb(I<sub>1-x</sub>Br<sub>x</sub>)<sub>3</sub> Hybrid Halide Perovskite Alloy. *J. Phys. Chem. Lett.* **7**, 1083–1087 (2016).
+
+30. Taddei, M. et al. Ethylenediamine Addition Improves Performance and Suppresses Phase Instabilities in Mixed-Halide Perovskites. *ACS Energy Lett.* **7**, 4265–4273 (2022).
+
+31. Petrov, A. A. et al. Formamidinium Haloplumbate Intermediates: The Missing Link in a Chain of Hybrid Perovskites Crystallization. *Chem. Mater.* **32**, 7739–7745 (2020).
+
+32. Frohna, K. et al. Nanoscale chemical heterogeneity dominates the optoelectronic response of alloyed perovskite solar cells. *Nat. Nanotechnol.* **17**, 190–196 (2022).
+
+33. Chen, H. et al. Regulating surface potential maximizes voltage in all-perovskite tandems. *Nature* **613**, 676–681 (2022).
+
+34. Hoke, E. T. et al. Reversible photo-induced trap formation in mixed-halide hybrid perovskites for photovoltaics. *Chem. Sci.* **6**, 613–617 (2015).
+
+35. Sutter-Fella, C. M. et al. Band tailing and deep defect states in CH<sub>3</sub>NH<sub>3</sub>Pb(I<sub>1−x</sub>Br<sub>x</sub>)<sub>3</sub> perovskites as revealed by sub-bandgap photocurrent. *ACS Energy Lett.* **2**, 709–715 (2017).
+
+36. Wang, J., Datta, K., Weijtens, C. H. L., Wienk, M. M. & Janssen, R. A. J. Insights into Fullerene Passivation of SnO<sub>2</sub> Electron Transport Layers in Perovskite Solar Cells. *Adv. Funct. Mater.* **29**, 1905883 (2019).
+
+37. Hidalgo, J. et al. Solvent and A-Site Cation Control Preferred Crystallographic Orientation in Bromine-Based Perovskite Thin Films. *Chem. Mater.* **35**, 4181–4191 (2023).
+
+38. Huang, T. et al. Performance-limiting formation dynamics in mixed-halide perovskites. *Sci. Adv.* **7**, eabj1799 (2021).
+
+39. Gu, E. et al. Robot-Based High-Throughput Screening of Antisolvents for Lead Halide Perovskites. *Joule* **4**, 1806–1822 (2020).
+
+40. Babbe, F., Masquelier, E., Zheng, Z. & Sutter-Fella, C. M. Flash Formation of I-Rich Clusters during Multistage Halide Segregation Studied in MAPbI<sub>1.5</sub>Br<sub>1.5</sub>. *J. Phys. Chem. C* **124**, 24608–24615 (2020).
+
+41. Huang, Z. et al. Local A-Site Phase Segregation Leads to Cs-Rich Regions Showing Accelerated Photodegradation in Mixed-Cation Perovskite Semiconductor Films. *ACS Energy Lett.* **9**, 3066–3073 (2024).
+
+42. Feldmann, S. et al. Tailored Local Bandgap Modulation as a Strategy to Maximize Luminescence Yields in Mixed-Halide Perovskites. *Adv. Opt. Mater.* **9**, 2100635 (2021).
+
+43. Motti, S. G. et al. Phase segregation in mixed-halide perovskites affects charge-carrier dynamics while preserving mobility. *Nat. Commun.* **12**, 6955 (2021).
+
+44. van der Pol, T. P. A., Datta, K., Wienk, M. M. & Janssen, R. A. J. The Intrinsic Photoluminescence Spectrum of Perovskite Films. *Adv. Opt. Mater.* **10**, 2102557 (2022).
+
+45. Fassl, P. et al. Revealing the internal luminescence quantum efficiency of perovskite films via accurate quantification of photon recycling. *Matter* **4**, 1391–1412 (2021).
+
+46. Al-Ashouri, A. et al. Conformal monolayer contacts with lossless interfaces for perovskite single junction and monolithic tandem solar cells. *Energy Environ. Sci.* **12**, 3356–3369 (2019).
+
+47. Datta, K. et al. Effect of Light-Induced Halide Segregation on the Performance of Mixed-Halide Perovskite Solar Cells. *ACS Appl. Energy. Mater.* **4**, 6650–6658 (2021).
+
+48. van Gorkom, B. T., van der Pol, T. P. A., Datta, K., Wienk, M. M. & Janssen, R. A. J. Revealing defective interfaces in perovskite solar cells from highly sensitive sub-bandgap photocurrent spectroscopy using optical cavities. *Nat. Commun.* **13**, 349 (2022).
+
+49. Zeiske, S. et al. Static Disorder in Lead Halide Perovskites. *J. Phys. Chem. Lett.* **13**, 7280–7285 (2022).
+
+50. Aalbers, G. J. W. et al. Effect of sub-bandgap defects on radiative and non-radiative open-circuit voltage losses in perovskite solar cells. *Nat. Commun.* **15**, 1276 (2024).
+
+51. Mahesh, S. et al. Revealing the origin of voltage loss in mixed-halide perovskite solar cells. *Energy Environ. Sci.* **13**, 258–267 (2020).
+
+52. Tang, X. et al. Local Observation of Phase Segregation in Mixed-Halide Perovskite. *Nano. Lett.* **18**, 2172–2178 (2018).
+
+53. Subedi, B. et al. Urbach energy and open-circuit voltage deficit for mixed anion− cation perovskite solar cells. *ACS Appl Mater. Interfaces* **14**, 7796–7804 (2022).
+
+54. Liu, Y. et al. The Electronic Disorder Landscape of Mixed Halide Perovskites. *ACS Energy Lett.* **8**, 250–258 (2023).
+
+55. Krückemeier, L., Rau, U., Stolterfoht, M. & Kirchartz, T. How to Report Record Open-Circuit Voltages in Lead-Halide Perovskite Solar Cells. *Adv. Energy Mater.* **10**, 1902573 (2020).
+
+56. Kosar, S. et al. Unraveling the varied nature and roles of defects in hybrid halide perovskites with time-resolved photoemission electron microscopy. *Energy Environ. Sci.* **14**, 6320–6328 (2021).
+
+57. Doherty, T. A. S. et al. Performance-limiting nanoscale trap clusters at grain junctions in halide perovskites. *Nature* **580**, 360–366 (2020).
+
+58. Barker, A. J. et al. Defect-Assisted Photoinduced Halide Segregation in Mixed-Halide Perovskite Thin Films. *ACS Energy Lett.* **2**, 1416–1424 (2017).
+
+# Supplementary Files
+
+- [SIWrinklingsubmitted.docx](https://assets-eu.researchsquare.com/files/rs-4814295/v1/ca121f43f97006ec05582cc4.docx)
+- [nrphotovoltaicreportingNC.pdf](https://assets-eu.researchsquare.com/files/rs-4814295/v1/24f2e008025f5bb4c73a8d9e.pdf)
+
+  Dataset 1
+
+- [RawDataWrinklingsubmitted.xlsx](https://assets-eu.researchsquare.com/files/rs-4814295/v1/7f9b23c8670d423bd358160c.xlsx)
+
+  Dataset 1

37e8877d5b5098982432781805618e85f11bb260d5bcd5d6e5df7d2127800b5e/preprint/images_list.json

@@ -0,0 +1,58 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "(a) Chemical structures of 2EH-PDI and PE-PDI. (b) Schematic representation of the formation of dormant monomers by 2EH-PDIand their transformation to thermodynamically stable supramolecular polymers (SPs) via seeded living supramolecular polymerization. (c) Schematic representation providing an overview of three noncovalent synthesis approaches, demonstrating how the activation of dormant monomers of 2EH-PDI leads to the formation of 1D SPs via homo-seeding, 3D spherical spherulites via self-seeding and scarf like SP heterostructures via hetero-seeding.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "(a) Synthesis of dormant monomers of 2EH-PDI. Heating the supramolecular polymers (SPs) formed at 303 K to 353 K leads to the formation of monomers in MCH*. Later by rapid cooling (10 K/min.) to 303 K it led to the formation of dormant monomers. When these dormant monomers were left undisturbed at 303 K, again, SPs were formed after 90 minutes. (b) Temperature-dependent Uv-Vis absorption changes were observed for the 2EH-PDI monitored at monomer wavelength (515 nm) at various volume percentages of DCE in MCH (c = 50 mM). (c) Time-dependent Uv-Vis absorption variation of the 2EH-PDI at 515 nm at various cooling rates in MCH* (c = 50 mM) at 303 K.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "(a) Potential energy surface of 2EH-PDI molecule calculated using the semi-empirical PM6 method. The energy values, in kJ/mol, are shown on the color bar, and the two axes correspond to the two dihedral angles between the \u03c0-plane and the two side chains (C-N-C-C). All subsequent results were obtained using density functional theory. (b) Structure of the 2EH-PDImolecule in three chair-type conformations labeled as 1c, 2c, and 3c. (c) Probability of finding each conformation at 353 K, calculated using the Boltzmann factor. (d) Relative energies (in KJ/mol) of the six conformers (shown in black) and the associated interconversion energy barriers (displayed in red), were both computed using density functional theory.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "(a) Schematic illustration of the homo seeding approach of dormant monomers. (b) Time dependent absorption changes observed for the 2EH-PDI dormant monomers monitored at 515 nm after adding different mol% of seed.\u00a0 FE-SEM images of 2EH-PDI obtained by spin coating the solutions on a silicon wafer (c) after sonication of prefabricated fibres, (d), (e) and (f) after performing LSP to 2EH-PDI dormant monomers at varying [2EH-PDImonomer] / [ 2EH-PDI seed] ratios of 19:1, 3:1, and 1:1 respectively (c = 50 mM, l = 10 mm, solvent = MCH*).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "(a) Schematic illustration of stirring-induced formation of 3D spherical spherulites by 2EH-PDI via self-seeding of its dormant monomers followed by secondary nucleation elongation. (b) Time-dependent variation in the absorbance of 2EH-PDI dormant monomers at 515 nm under stirring at various RPM. (c) log-log plot of the half-times of stir-induced (60 RPM) supramolecular polymerization versus the original concentration of 2EH-PDI. Symbols represent the experimental data; solid line is a power law fit. This fit shows a linear trend with a slope of -3.006 \u00b1 0.374 referred to as the exponent coefficient (g), indicating a monomer dependent secondary nucleated supramolecular polymerization process. (d) Kinetic profiles of the concentration-dependent self-seeding experiments at 60 RPM conditions were obtained by monitoring the absorbance at 515 nm while varying the dormant monomer (2EH-PDI) concentrations to 40 mM, 45 mM, and 50 mM. The sigmoidal growth of the 2EH-PDI dormant monomer indicates the presence of the secondary nucleation process. FE-SEM images of 2EH-PDI, obtained by spin coating the solutions on a silicon wafer after stirring the dormant monomers at (e) 60 RPM, (f) 120 RPM and (g) 300 RPM for 60 minutes at 303 K (c= 50 mM, solvent = MCH*).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "(a) UV-vis absorption spectra of 2EH-PDI dormant monomers (red), one hour after applying the slow mechanical agitation via repetitive pipetting (green), LSP after adding 10 mol % seed (blue) in MCH*. (b) Time-dependent variation in the absorbance of 2EH-PDI at 515 nm after applying the mechanical agitation via repetitive pipetting to the dormant monomer (green). The control experiment by adding the 10 mol% seed (blue) and 0 mol% seed (red) \u00a0to the dormant monomers of the 2EH-PDI is shown for comparison. (c) Optical microscopy image of the spherulites obtained in MCH* after applying the slow mechanical agitation via repetitive pipetting. (d-g) FE-SEM images of 2EH-PDI, obtained by spin coating the solutions on a silicon wafer after (d) 5 min., (e) 15 min., (f) 30 min. (g) 50 min. of applying the mechanical agitation via repetitive pipetting to the dormant monomers of 2EH-PDI (c = 50 mM, solvent MCH*)",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_7.png",
+        "caption": "(a) Schematic illustration for formation SP heterostructures obtained by adding PE-PDI 2D platelets to 2EH-PDI dormant monomers via secondary nucleation. (b) Time-dependent variation in the absorbance of 2EH-PDI dormant monomers at 515 nm after adding the 50 mol% PE-PDI seeds (blue) and spontaneous polymerization of 2EH-PDI (red) is shown for comparison. (c) FE-SEM image of SP heterostructures synthesized via hetero seeding approach by adding various 50 mol% PE-PDI 2D platelets seeds to 2EH-PDI dormant monomers. (d) log-log plot of the half-times of hetero-seeding supramolecular polymerization versus original concentration of 2EH-PDI. Symbols represent the experimental data; solid line is a power law fit. This fit shows a linear trend with a slope of -3.23 \u00b1 0.374 called exponent coefficient (g), which suggests a monomer dependent, seed-induced surface catalysed secondary nucleated supramolecular polymerization process. (e) Kinetic profiles of the concentration-dependent hetero-seeding experiments, obtained by monitoring the absorbance at 515 nm of 2EH-PDI, at constant seed concentration ((PE-PDI)concentration: 25 mM) while varying the dormant monomer concentrations of 2EH-PDI to 40 mM, 50 mM, and 60 mM. The sigmoidal growth of the 2EH-PDI dormant monomers indicates a secondary nucleation-elongation process. (f) Bright-field microscopy images showing time-dependent growth of SP heterostructures after mixing the 50 mol% of PE-PDI seeds to the 2EH-PDI dormant monomers in MCH* (encircled with dashed lines of blue, pink and magenta). (g) Kinetics of growth of SP heterostructures from the PE-PDI seeds (encircled in blue, pink and magenta dashed lines in panel (f) was tracked via image analysis to follow the increase in pixel area corresponding to the growth of new fibrils from the seeds.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

37e8877d5b5098982432781805618e85f11bb260d5bcd5d6e5df7d2127800b5e/preprint/preprint.md

@@ -0,0 +1,135 @@
+# Abstract
+
+The synthesis of supramolecular polymers (SPs) with controlled architecture is a grand challenge in supramolecular chemistry. Although living supramolecular polymerization (LSP) via primary nucleation has been extensively studied for controlling the supramolecular polymerization of small molecules, the resulting SPs have typically exhibited one-dimensional (1D) morphology. In this report, we present the synthesis of intriguing SP architectures through a secondary nucleation event, a mechanism well-established in protein aggregation and the crystallization of small molecules. To achieve this, we selected perylene diimide with 2-ethylhexyl chains (2EH-PDI) at the imide position and stabilized its dormant monomers in solution. Activating these dormant monomers via mechanical stimuli (self-seeding) and hetero-seeding using propoxyethyl PDI (PE-PDI) seeds, secondary nucleation event takes over, leading to the formation of 3D spherical spherulites and scarf-like SP heterostructures, respectively. Therefore, the results presented in this study propose a simple molecular design for synthesizing well-defined SP architectures via secondary nucleation.
+
+Physical sciences/Chemistry/Supramolecular chemistry/Supramolecular polymers  
+Physical sciences/Materials science/Soft materials/Self-assembly  
+Living supramolecular polymerization  
+Secondary nucleation  
+Architectural control  
+Spherulites  
+Heterostructures
+
+# Introduction
+
+In recent years, living supramolecular polymerization (LSP) has emerged as a promising tool for synthesising complex supramolecular homo- and heterostructures, offering meticulous control over their shapes and dimensions.<sup>1–4</sup> The pioneering work in this field began with the research groups led by Winnik and Manners et al., who first reported the living crystallization driven self-assembly (LCDSA).<sup>5</sup> They used poly(ferrocenyldimethylsilane) cored block-copolymers to synthesise various supramolecular nanoarchitectures such as one-dimensional (1D) cylinders,<sup>6–8</sup> 2D platelets,<sup>9,10</sup> dendritic micelles,<sup>11,12</sup> rod-coil micelles,<sup>13</sup> and 3D spherulites<sup>14,15</sup> with excellent precision to name a few. Despite the significant progress in the supramolecular polymerization of small molecules, realizing LSP was challenging as no rational design principles were available to create dormant monomers or aggregates. To this end, Sugiyasu et al. demonstrated LSP in small molecules using hydrogen-bonded (H-bonded) porphyrins, which form metastable J-aggregates.<sup>16</sup> Shortly after, Aida et al. also achieved kinetically trapped monomers using H-bonded corannulenes via intramolecular H-bonding.<sup>17</sup> Both these trapped states are activated using thermodynamically stable SPs as seeds to form SPs with controlled lengths. These inspiring outcomes further led to synthesising block SPs with controlled sequences.<sup>18–23</sup>
+
+Over the past few years, a few researchers have reported the phenomenon of molecular self-assembly operating at higher hierarchical levels, thereby modulating the topology of functional SPs through a secondary nucleation mechanism.<sup>24–29</sup> While this phenomenon has been comprehensively studied theoretically and experimentally in the context of amyloid self-assembly,<sup>30–33</sup> it remains a relatively unexplored frontier within the domain of functional SPs. Recently, Yagai's group has reported the creation of molecularly interlocked nanostructures with a remarkably high level of organization through the utilization of a secondary nucleation strategy.<sup>28</sup> Similarly, Sugiyasu et al. have described the formation of double-stranded Archimedean spirals through a secondary nucleation event.<sup>25</sup> George et al. have recently reported the emergence of secondary nucleation-mediated surface-anchored self-sorted SPs, secondary SPs, and the production of seed-induced, non-covalently cross-linked hydrogels.<sup>24,27,29</sup> Despite these interesting results, achieving fine control over the topology of functional SPs requires a much-needed understanding of the molecular structure, kinetics and underlying mechanisms that govern secondary nucleation.
+
+Currently, LSP and secondary nucleation pathways are explored in small molecule-based H-bonded π-systems mainly because of two reasons: (i) growth via the nucleation elongation process and (ii) the ease of achieving kinetically trapped monomers/aggregates.<sup>16,17,34–43</sup> However, H-bonding interactions typically result in 1D morphology owing to the strong directionality of H-bonding functional groups.<sup>34–39</sup> Hence, the synthesis of complex supramolecular architectures using H-bonded π-systems is limited to a few sets of molecules.<sup>25,28,44,45</sup> On the other hand, Che et al. demonstrated LSP using non-H-bonded π-systems by activating their off-pathway aggregates via homo and hetero-seeding methods leading to the synthesis of interesting homo and block supramolecular architectures.<sup>46–48</sup> However, rational molecular design principles to realize LSP and secondary nucleation in other types of π-systems that form SPs without the aid from H-bonding interactions need to be formulated for further exploration. In this context, we have recently reported that non-H-bonded simple π-systems such as alkyl perylene diimides (PDIs) show cooperative supramolecular polymerization guided solely by dispersive interactions with cooperative factors on par with H-bonded π-systems.<sup>49</sup> Hence, exploring alkyl PDIs to realize LSP and secondary nucleation pathways can broaden the scope of small molecules and expand the non-covalent synthesis toolbox to create complex supramolecular architectures. However, the generation of dormant monomers or off-pathway aggregates from alkyl PDIs must be addressed.
+
+Here, we present a striking improvement in the field of SPs by expanding the structural diversity of monomers that display LSP and secondary nucleation to synthesise SPs with interesting topologies. For this purpose, we have selected a simple π-system, PDI substituted with 2-ethylhexyl side chains (<b>2EH-PDI</b>), as it is known to form SPs via nucleation elongation mechanism (Fig. <span class="InternalRef" refid="Fig1">1</span> a).<sup>49</sup> Despite the lack of any H-bonding functional groups,<b>2EH-PDI</b> forms dormant monomers in solution from the monomeric states upon rapid cooling due to the existence of multiple conformations at high temperatures. By optimising the solvent conditions, we successfully stabilized the <b>2EH-PDI</b> molecules as dormant monomers in solution at room temperature (Fig. <span class="InternalRef" refid="Fig1">1</span> b). Next, these dormant monomers were converted into complex supramolecular architectures via seed-induced LSP and secondary nucleation. To accomplish this, we have employed three distinct seeding methods: (i) homo-seeding, (ii) self-seeding, and (iii) hetero-seeding (Fig. <span class="InternalRef" refid="Fig1">1</span> b,c). By adding various mol% of prefabricated <b>2EH-PDI</b> SP seeds (homo-seeding), we could synthesize 1D SPs with controlled lengths, with the primary nucleation-elongation event taking precedence. Interestingly, when we mechanically agitated (through stirring or pipetting) the dormant <b>2EH-PDI</b> monomers in a controlled manner, inducing seed formation within the solution (self-seeding), we observed the formation of SPs with 3D spherical spherulite architectures, where the secondary nucleation-elongation promoted the growth of SPs around the seed. While stir-induced secondary nucleation is commonly observed in amyloid systems<sup>53</sup> and the crystallization of inorganic crystals,<sup>54,55</sup> it is a relatively rare phenomenon in functional SPs. Similarly, for hetero-seeding experiments, we have utilized 2D platelets formed by PDI substituted with propoxyethyl chains (<b>PE-PDI</b>) (Fig. <span class="InternalRef" refid="Fig1">1</span> a). In this case, the secondary nucleation event promotes the growth of 1D SPs of <b>2EH-PDI</b> above and below the surface of 2D platelets of <b>PE-PDI</b> across their length, resulting in scarf-like SP heterostructures. Such a controlled fabrication of SP architectures using simple π-systems via secondary nucleation at an elevated hierarchical level represents an unprecedented achievement. Moreover, we also provided the direct visualisation of the formation of SP heterostructures in solution using optical microscopy.
+
+# Results and discussion
+
+## Formation of dormant monomers
+
+Previously, we reported that 2EH-PDI forms SPs in methylcyclohexane (MCH) and dimethyl sulfoxide (DMSO) via a cooperative mechanism.⁴⁹ 2EH-PDI (50 µm) is molecularly soluble in chlorinated solvents such as chloroform (CHCl₃) and dichloroethane (DCE), as evidenced by UV-vis absorption spectra which shows characteristic monomeric peaks at 490 nm and 515 nm (Figure S1). In MCH, 2EH-PDI (50 µm) forms SPs, as evidenced by relatively broad absorption spectra and a new redshifted band at 575 nm (Figure S1). These SPs could be reversibly transformed into monomers at high temperatures (~ 363 K), as evident from the UV-Vis spectra (Figure S2a). Interestingly, thermal hysteresis was observed in MCH during the supramolecular polymerization and depolymerization process. While heating (SPs → monomers), the elongation temperature (Tₑ) of 2EH-PDI (50 µm) was observed at 353 K. However, in the cooling process (monomers → SPs), the critical elongation temperature (Tₑ) was reduced to 330 K (Figure S2b). These results revealed the presence of pathway complexity, where the 2EH-PDI monomers get trapped in a higher energy metastable state upon cooling the solution from monomers to form SPs (Figs. 1a and 2a). Interestingly, by increasing the percentage of good solvent, such as DCE, the temperature region of the hysteresis loop enhanced (Fig. 2b and Figures S3-S6), and at 10 vol% of DCE in MCH (MCH*), we obtained the kinetically trapped monomeric state (dormant monomer) at room temperature (Figs. 2 and S6). Here, the cooling rate also dictates the stability of the dormant monomer. In MCH*, the dormant monomer of 2EH-PDI obtained by the cooling at a rate of 10 K min⁻¹ is more stable (~ 25 to 30 minutes) at 303 K than cooling rates of 1, 3 and 5 K min⁻¹ (Fig. 2a,c).
+
+In the case of H-bonded π-systems, intramolecular H-bonding is shown to be responsible for the formation of dormant monomers.¹⁷,¹⁹,³⁶,³⁹ However, in 2EH-PDI it is surprising to see the formation of dormant monomers despite the lack of any H-bonding interactions. To gain insights into the kinetically trapped states of 2EH-PDI, gas phase semi-empirical and quantum chemical calculations of the 2EH-PDI monomer were performed using GAUSSIAN16.⁵⁰ Note that the reported energies are computed in the gas phase and will likely be altered in the presence of an explicit solvent. The details of the computational methods were provided in the Supporting Information. Since all molecules exist in the monomeric state above the critical temperature, as a first step, we used quantum calculations to explore all possible conformations of the 2EH-PDI molecule. We computed the two-dimensional potential energy surface (PES) by varying the two dihedral angles, each belonging to a side chain, defined between the π-plane and the side chains (C-N-C-C), using the semi-empirical PM6 method.⁵¹ Fig. 3a depicts the PES of the 2EH-PDI molecule, revealing several minima regions (highlighted as yellow colored dots). In the next step, we optimized each structure within the density function theory formalism using the ωB97xD/6-31g(d) level of theory.⁵² We identified six stable conformers of 2EH-PDI with energy differences ranging from 0 to 5 kJ/mol. These conformations mainly fall into two types: chair and boat. In both cases, the alkyl moieties are positioned at angles of approximately 30 or 80° with respect to the π-plane. Figure 3b illustrates the three chair-type conformations (labelled as 1c, 2c and 3c) arranged according to their stability. The analogous boat-type conformations (labelled as 1b, 2b and 3b, respectively) are provided in Supporting Information (Figure S7). Each boat conformer has the same energy as its analogous chair conformation. The 1c conformer is the most stable among all chair-type conformers, with an energy lower by 2.5 kJ/mol and 4.9 kJ/mol compared to 2c and 3c, respectively.
+
+The spatial arrangement of the 2-ethylhexyl side chains of 1b, 2b, 3b, 2c, and 3c conformations of 2EH-PDI molecules constrains the π-π stacking interactions due to steric hindrance or defects. However, due to its favorable side-chains spatial arrangement, the stack formed by the 1c conformer results in maximum π- π stacking interactions and lower steric hindrance, facilitating the formation of the stack.⁴⁹ This implies that the 1c conformer serves as the "active" species for the growth of a supramolecular polymer, and its higher concentration in the solution is necessary to observe stack formation. We calculated the probabilities of finding each conformation in the solution at 353 K using the Boltzmann factor, as shown in Fig. 3c. The results indicate that approximately 60% of 2EH-PDI molecules adopt conformations 1c and 1b, while two-thirds of the remaining molecules are in conformations 2c and 2b. As a result, it is expected that in the solution, conformation 1c and 1b dominates, while other conformers are also significantly present. This disparity in the conformational population can lead to a decrease in the concentration of "active" monomers, resulting in the observed hysteresis during the cooling process. This finding aligns with the hypothesis proposed by Würthner and his team, who suggested that the kinetic inactivation of monomers depends on the energetic interplay between kinetically trapped, inactive species and active species for supramolecular polymerization.³⁶
+
+To further explore the likelihood of transitions between these conformers, we conducted calculations to determine the energy barriers involved, as depicted in Fig. 3d. The results reveal that the interconversion among either chair-type or boat-type conformations requires approximately 4 kBT of energy, while the conversion from the 1c to 1b conformer requires approximately 14 kBT of energy. These energy barriers are sufficiently high to hinder spontaneous nucleation, thereby governing the critical temperature during cooling. This observation is further supported by the dependence of the critical temperature on the cooling rates (Fig. 2c). Therefore, our computational findings suggest that the thermal hysteresis primarily arises from the coexistence of multiple conformations at high temperature. The elevated energy barriers inhibit the transition of kinetically trapped species into the "active" species essential for supramolecular polymerization.
+
+## Homo seeding
+
+The formation of dormant monomers by 2EH-PDI (50 µM) in MCH* encouraged us to carry out LSP by adding the prefabricated SP seed of 2EH-PDI (Fig. 4). When a DCE solution of 2EH-PDI (50 µM) is injected into MCH, to result in the final solvent composition as MCH*, it spontaneously assembles into SPs of various lengths (Figure S8). Upon sonication of these long polydisperse SPs for 1 hour at 303 K, they fragmented into many short seed-like fibres with active chain ends (Fig. 4c). Uv-Vis absorption and fluorescence spectra revealed that sonication didn’t alter the aggregation behaviour 2EH-PDI (Figure S9), and the average length of seed particles is around 700 to 800 nm (Fig. 4c). Seeded growth polymerization was attempted immediately by adding the seed solution to the dormant monomers of 2EH-PDI, and the final concentration of the solution was kept constant at 50 µM (Fig. 4a). As previously stated, the final length of SP always depends upon the mass ratio of added seed to monomer solution.¹⁶ In this perspective, we have added different mol% of seed to dormant monomers and monitored through UV-vis spectroscopy and FE-SEM. The Uv-Vis absorption spectra of dormant monomers solution after adding the 10 mol% sonicated seed showed a sudden decrease in the monomeric absorption peaks at 515 nm and 495 nm, with an increase in the absorbance of the band at 575 nm (Fig. 4b and S10). To further investigate the seed-induced living growth of 2EH-PDI dormant monomers, we have added different mol% of seeds and monitored the kinetics of LSP (Fig. 4d-f). As expected, spontaneous elongation without any lag time was observed (Fig. 4b), and the length of resultant fibres analysed by FE-SEM experiments has shown a systematic increase in the length of SPs with a decrease in seed mol%. From FE-SEM studies, we found that the average length of the fibres prepared from [2EH-PDI<sub>monomer</sub>] / [2EH-PDI<sub>seed</sub>] ratios of 19:1, 3:1, and 1:1 are 6 µm, 2 µm and 1 µm, respectively (Fig. 4d-f).
+
+To obtain a clear understanding of the molecular mechanism, we conducted kinetic analyses at various concentrations of dormant monomer, specifically 40 µM, 55 µM, and 60 µM, while keeping seed concentration constant at 12.5 µM. We monitored the growth kinetics by measuring changes in the absorbance at 515 nm. The scaling exponent, which characterizes how the reaction's lag time or half-time (t₅₀) varies with the initial monomer concentration, was determined by the double-logarithmic plot of t₅₀ against monomer concentration, as depicted in Figure S11a under seeded conditions.²⁴–³⁰ The obtained value for the scaling exponent is γ = -1.40 ± 0.19, corresponding to reaction order n₁ = 2.8. The growth kinetics were modelled using a seed-induced nucleation-elongation framework from amyloid software (http://www.amylofit.ch.cam.ac.uk)³⁰ fit well with the experimental data (Section S3, Figure S11, Table S1). The above observations clearly indicate that the conversion of dormant monomers of 2EH-PDI into SPs via homo-seeding takes place through the primary nucleation-elongation mechanism.
+
+## 3D spherical spherulites via secondary nucleation (Self-seeding)
+
+Given prior publications on how mechanical agitation, such as stirring, aids the rate-determining nucleation process,³⁵,³⁷,⁵⁶ we conducted time-dependent kinetic investigations of MCH* containing dormant monomers of 2EH-PDI with stirring (Fig. 5). As expected, the nucleation process is expedited at 60 RPM, as seen by decreased monomeric absorbance after 12 minutes (Fig. 5b). Surprisingly, FE-SEM studies revealed that resultant self-assembled structures have 3D spherical spherulite structures (Figs. 5a,e and S12).¹⁴,¹⁵ In most circumstances, the morphology obtained in LSP is the same as that obtained in thermodynamically controlled SPs. Although mechanical agitation, such as stirring or pipetting, speeds up the nucleation process, previous reports indicate that it has little effect on the morphology of SPs.³⁷ In the present case, mechanical agitating creates 3D spherical spherulites with micrometres length in all directions (Fig. 5e-g and S12). We noticed that up to 300 RPM, we could observe the formation of 3D spherical spherulites. However, after stirring at 600 RPM, we could observe mostly 1D SPs. This could be due to the detachment of 1D SPs from spherulite structures, probably due to strong mechanical agitation at 600 RPM (Figure S12). We conducted concentration-dependent kinetic analysis while maintaining a constant stirring rate of 60 RPM to further understand the growth mechanism. All the kinetic profiles exhibited a sigmoidal-like transition, which included a lag phase followed by an exponential phase (Fig. 5d). These characteristics are typical of the secondary nucleation-elongation process.²⁴–³⁰ We determined the scaling exponent as γ = -3.0 ± 0.37 (Fig. 5c), which is higher than the value obtained in homo-seeding experiments. This further indicates the presence of secondary nucleation events with a secondary reaction order of n₂ = 5. Furthermore, we comprehensively analysed all the kinetics and found that all the data fit well with the secondary nucleation model, with an average mean squared error (MSE) value less than 0.0005 (Fig. 5d, Section S3 and Table S2). These kinetic analyses strongly suggest that the growth of 3D spherulites via a self-seeding approach takes place through a typical secondary nucleation elongation mechanism.
+
+We have also noticed that mechanical agitation via repetitive pipetting can also induce the formation of 3D spherical spherulites like stirring (Figs. 6 and S13). To monitor the temporal evolution of the spherulites, we conducted a series of FE-SEM examinations at different intervals after applying weak mechanical stimuli such as pipetting. For this purpose, we have applied 5 to 6 times repetitive pipetting to the MCH* solution containing dormant monomers and left it undisturbed at 303 K. In this case, the assembly process began with a 5-minute lag period, as indicated in Fig. 6a,b, and we observed the hedrite morphology after 10 minutes (Fig. 6d). With the increase in time, we have observed that fibril growth was observed in all directions, resulting in the formation of star-shaped morphology (Fig. 6e,f). We got larger 3D spherulite structures with increasing branches after 50 minutes (Figs. 6g and S13). To avoid any ambiguity on the formation of 3D spherulites in solution, the dormant monomers solution of 2EH-PDI in MCH* was mechanically agitated via repetitive pipetting (5 to 6 times), and the resultant solution was observed under the optical microscope. Interestingly, after some time, we see the formation of 3D spherulite structures in solution (Fig. 6c). This unambiguously proves that the formation of 3D spherulites via slow mechanical agitation occurs in the solution.
+
+The above results vividly illustrate the remarkable sensitivity of 2EH-PDI supramolecular polymerization to even minor alterations in reaction conditions. The self-assembly behaviour of 2EH-PDI is primarily governed by the dominant primary nucleation event in the homo-seeding approach. However, by introducing shear forces through actions like pipetting and stirring, we could bias its self-assembly pathway, initiating secondary nucleation events that result in the formation of 3D spherical spherulite structures. In this context, the anticipated mechanism involves the agitation of dormant monomers, which facilitates generating a limited number of small nuclei (Fig. 5a). This, in turn, promotes secondary nucleation of high-energy dormant monomers on the small nuclei followed by elongation in all directions, leading to the formation of low-energy SPs with spherulite architecture. Notably, stir-induced secondary nucleation events have been well-documented in processes such as crystallization⁵⁴,⁵⁵ and protein aggregation.⁵³ However, it is worth mentioning that, to the best of our knowledge, no previous reports exist regarding their occurrence in functional SPs. Since no seed was added externally and the seed was generated from dormant monomers in the same solution, we call this process as ``self-seeding`` (Fig. 5a). Here, we demonstrated a unique example where dormant 2EH-PDI monomeric building blocks would lead to the formation of 3D spherulites via mechanical agitation, which is not in the case of homo-seeding experiments.
+
+## Supramolecular polymer heterostructures via secondary nucleation (Hetero-seeding)
+
+The results obtained from homo and self-seeding further encouraged us to explore the hetero-seeding approach (Fig. 7). It has already been demonstrated that supramolecular block copolymers can be synthesized using LSP via a hetero-seeding process, which can provide superior structural and sequence control.¹⁰–¹³,¹⁸−²² For the hetero-seeding approach, we prepared a PE-PDI having a propoxy ethyl side chain at the imide positions (Fig. 1a). Like 2EH-PDI, PE-PDI also self-assembles in MCH* via J-aggregation and forms 2D platelets (Figure S14). PE-PDI seeds were prepared in MCH* via sonication for 1 hour at 303 K and 50 mol% of these were added to the dormant monomers of 2EH-PDI. We found a drop in 2EH-PDI monomeric wavelength absorbances (λ<sub>max</sub> = 515 nm) after a 5-minute lag period (Fig. 7a,b). FE-SEM images revealed that 1D SPs of 2EH-PDI were grown on the surface of 2D platelets of PE-PDI and oriented across their length (Fig. 7c and S15). Contrary to homo-seeding, which demonstrated spontaneous growth without any lag phase (Fig. 4b), the presence of a lag phase lasting for 5 minutes and the growth of 2EH-PDI fibres on the surface of PE-PDI seeds suggests the presence of secondary nucleation elongation mechanism.
+
+Like homo-seeding experiments, we conducted hetero-seeding experiments using various dormant monomer concentrations (40, 50, and 60 µM) of 2EH-PDI, while maintaining a constant seed concentration of 25 µM for PE-PDI. In all cases, the kinetic profiles exhibited a sigmoidal-like transition characterized by a lag phase followed by an exponential phase, which indicates the occurrence of a secondary nucleation event induced by seeding (Fig. 7e). As the monomer concentration increased from 40 µM to 60 µM, there was a noticeable decrease in the half-time. Plotting the half-time against the monomer concentration on a double logarithmic scale yielded a scaling coefficient γ of -3.23 ± 0.03 (Fig. 7d), demonstrating a linear relationship and sign of a secondary nucleation event. Additionally, we derived a secondary nucleation reaction order of n₂ = 5 from this analysis. Furthermore, we conducted a comprehensive analysis of all the kinetics and found that they fit well with the secondary nucleation model, with an average mean squared error (MSE) value less than 0.0007 (Fig. 7e and Table S3). These kinetic analyses strongly support the growth of SPs of 2EH-PDI on the surface of PE-PDI seeds via a hetero-seeding approach triggered by a typical secondary nucleation process.
+
+Since the PDI is common in both 2EH-PDI and PE-PDI, we expected that SPs of 2EH-PDI would grow from the ends of PE-PDI.²⁰ However, FE-SEM studies demonstrated that SPs of 2EH-PDI are grafted on the surface of the PE-PDI nanoplatelets (Figs. 7c and S15). To gain further insights into this phenomenon, we have conducted powder X-ray diffraction (PXRD) studies on SPs of 2EH-PDI, PE-PDI and their heterostructures (Figure S16). PXRD revealed that 2D platelets of PE-PDI are more crystalline than 2EH-PDI SPs. This observation further indicates that the side chain variation results in disparate crystal packing for both PDI derivatives. Due to these dissimilar lattice structures, 2EH-PDI monomers do not grow at the chain ends of PE-PDI and prefer to grow on the surface of PE-PDI via secondary nucleation (Fig. 7a). Hence, the crystalline nature of individual structures is also retained in the heterostructures (Figure S16). We have further controlled the grafting density of 2EH-PDI SPs on the surface of PE-PDI 2D platelets by varying the mol% of PE-PDI (Figure S17). By decreasing the mol% of PE-PDI seeds from 50–30%, we have observed increased covering of the PE-PDI surface by the SPs of 2EH-PDI compared to 50 mol% of the seed (Figs. 7c, S15 and S17). By further decreasing the mol% of PE-PDI seeds to 15%, their surface is fully grafted by the SPs of 2EH-PDI (Figs. 7c, S15 and S17).
+
+We have also explored possibilities to provide the direct visualization of the growth of SPs of 2EH-PDI on PE-PDI in solution via hetero-seeding (Fig. 7f,g). For this purpose, we have taken the MCH* solution containing dormant monomers of 2EH-PDI and 50 mol% of PE-PDI seeds in rectangular long capillary cells. Initially, we observed only 2D platelets of 2EH-PDI seeds in the solution. As time proceeds, we observe the emergence of SPs of 2EH-PDI from 2EH-PDI 2D platelet seeds across their length (Fig. 7f, Movie: Hetero-seeding). This is further supported by the increase in pixel area due to the growth of new fibrils from the seeds (Fig. 7g). These observations unambiguously prove PE-PDI seeds induce the supramolecular polymerization of 2EH-PDI dormant monomers via secondary nucleation in solution. Notably, such a direct visualization of the growth of SPs allows us to understand the temporal evolution of complex supramolecular structures from small molecular building blocks. The kinetics of growth probed through UV-Vis experiments in 10 mm cuvette are slower than the optical microscopy experiments done in rectangular capillaries due to the mass transport limitations in narrow channels of capillaries (Fig. 7b and g). The growth of SPs of 2EH-PDI from PE-PDI seeds within 200 µm rectangular capillaries followed by brightfield microscopy showed quick saturation of the growth within 3 min due to rapid depletion of local concentration of dormant monomers (Fig. 7g). On the other hand, in UV-Vis absorption experiments performed in 10 mm cuvette, fresh monomers are continuously available to the growing SPs on the seeds. As a result, the process takes longer time (> 20 min.) for the saturation.
+
+In conclusion, we have presented a simple molecular design (2EH-PDI) to synthesize dormant monomers, which can be explored for architectural control of SPs via LSP and secondary nucleation. Theoretical studies indicate that dormant monomers of 2EH-PDI are formed due to the statistical distribution of various conformers of 2EH-PDI monomers in the solution at high temperatures. We found the distant nucleation events during the supramolecular polymerization of dormant monomers through homo-seeding, self-seeding and hetero-seeding, resulting in the creation of elegant supramolecular architectures such as 3D spherulites and scarf-like SP heterostructures. Interestingly, mechanical agitation of dormant monomers of 2EH-PDI not only biases the mechanism of supramolecular polymerization from primary nucleation to secondary nucleation but also significantly alters the topology of SPs (1D SPs to 3D spherical spherulites). Our results further indicate that in addition to well-explored H-bonded π-systems for LSP and architectural control, simple π-systems such as alkylated PDIs would also be worth exploring in the same direction. This would be useful to expand the toolbox for non-covalent synthesis and provide access to complex supramolecular architectures.
+
+# Methods
+
+**Preparation of dormant monomers:** A solution of 2EH-PDI (0.3 ml) in DCE was injected into 2.7 ml of methylcyclohexane (MCH), promptly leading to the formation of SPs (Final concentration: 50 mM, final solvent composition: 10% DCE in MCH). Subsequently, this SPs solution was heated to 353 K, resulting in the formation of a monomeric state in solution. Upon cooling this hot monomeric solution to 303 K at a rate of 10 K/min, kinetically trapped dormant monomers were formed and exhibited stability for over 30 minutes. Utilizing these kinetically trapped dormant monomers, three distinct seeding experiments were conducted, as outlined below.
+
+**Homo-seeding:** In this experiment, a prefabricated SPs of 2EH-PDI in 10% DCE in MCH was sonicated for 1 hour at 303 K. Subsequently, 20 vol% of the seed solution (0.6 ml) was introduced into the dormant monomers of 2EH-PDI (40, 55, and 60 mM) at 303 K.
+
+**Self-seeding:** Following the formation of dormant monomers 2EH-PDI (40, 45, and 50 mM) at 303 K in 10% DCE in MCH, the solution was agitated at various RPM levels (60, 120, 300, and 600 RPM). We have also used reparative pipetting (five to six times) of dormant monomers 2EH-PDI in 10% DCE in MCH for self-seeding experiments.
+
+**Hetero-seeding:** Initially, PE-PDI SPs were generated by adding a DCE solution of PE-PDI (0.3 ml) to 2.7 ml of MCH in a vial, resulting in the formation of PE-PDI SPs (Final concentration: 125 mM, final solvent composition: 10% DCE in MCH). These SPs were sonicated for one hour at 303 K. Subsequently, this seed solution (20 vol%) was added to the dormant monomers of 2EH-PDI (40, 50, and 60 mM) at a temperature of 303 K.
+
+# References
+
+1. Jung, S. H., Takeuchi, M. & Sugiyasu, K. Molecular self-assembly under kinetic control. Kinetic control in synthesis and self assembly, Elsevier, 205–229 (2019).
+2. Wehner, M. & Würthner, F. Supramolecular polymerization through kinetic pathway control and living chain growth. Nat. Rev. Chem. 4, 38–53 (2020).
+3. Vanderzwaag, D., DeGreef, T. F. A. & Meijer, E. W. Programmable supramolecular polymerizations. Angew. Chemie - Int. Ed. 54, 8334–8336 (2015).
+4. MacFarlane, L. Zhao, C., Cai, J. Qiu, H. & Manners, I. Emerging applications for living crystallization-driven self-assembly. Chem. Sci. 12, 4661–4682 (2021).
+5. Wang, X. et al. Cylindrical block copolymer micelles and co-micelles of controlled length and architecture. Science 317, 644−647 (2007).
+6. Gilroy, J. B. et al. Monodisperse cylindrical micelles by crystallization-driven living self-assembly. Nat. Chem. 2, 566–570 (2010).
+7. Li, X. et al. Monodisperse cylindrical micelles of controlled length with a liquid-crystalline perfluorinated core by 1D “self-seeding.” Angew. Chemie - Int. Ed. 55, 11392–11396 (2016).
+8. Patra, S. K. et al. Cylindrical micelles of controlled length with a π-conjugated polythiophene core via crystallization-driven self-assembly. J. Am. Chem. Soc. 133, 8842–8845 (2011).
+9. He, X. et al. Two-dimensional assemblies from crystallizable homopolymers with charged termini. Nat. Mater. 16, 481–488 (2017).
+10. Lei, S., Tian, J., Kang, Y., Zhang, Y. & Manners, I. AIE-active, stimuli-responsive fluorescent 2D block copolymer nanoplatelets based on corona chain compression. J. Am. Chem. Soc. 144, 17630–17641 (2022).
+11. Zhang, Y. Dendritic micelles with controlled branching and sensor applications. J. Am. Chem. Soc. 143, 5805–5814 (2021).
+12. Song, S. et al. Solvent effects leading to a variety of different 2D structures in the self-assembly of a crystalline-coil block copolymer with an amphiphilic corona-forming block. Chem. Sci. 11, 4631–4643 (2020).
+13. Song, S., Zhou, H., Manners, I. & Winnik, M. A. Block copolymer self-assembly: Polydisperse corona-forming blocks leading to uniform morphologies. Chem 7, 2800–2821 (2021).
+14. Jiang, J. et al. Polyferrocenylsilane block copolymer spherulites in dilute solution. J. Am. Chem. Soc. 145, 1247–1261 (2023).
+15. Song, S. et al. Spherulite‐like micelles. Angew. Chemie 133, 11045–11051 (2021).
+16. Ogi, S., Sugiyasu, K., Manna, S., Samitsu, S. & Takeuchi, M. Living supramolecular polymerization realized through a biomimetic approach. Nat. Chem. 6, 188–195 (2014).
+17. Kang, J. et al. A rational strategy for the realization of chain-growth supramolecular polymerization. Science 347, 646–651 (2015).
+18. Sarkar, A. et al. Self-sorted, random, and block supramolecular copolymers via sequence controlled, multicomponent self-assembly. J. Am. Chem. Soc. 142, 7606–7617 (2020).
+19. Sarkar, A. et al. Tricomponent supramolecular multiblock copolymers with tunable composition via sequential seeded growth. Angew. Chemie - Int. Ed. 60, 18209–18216 (2021).
+20. Wagner, W., Wehner, M., Stepanenko, V. & Würthner, F. Supramolecular block copolymers by seeded living polymerization of perylene bisimides. J. Am. Chem. Soc. 141, 12044–12054 (2019).
+21. Jung, S. H., Bochicchio, D., Pavan, G. M., Takeuchi, M. & Sugiyasu, K. A Block Supramolecular polymer and its kinetically enhanced stability. J. Am. Chem. Soc. 140, 10570–10577 (2018).
+22. Sasaki, N. et al. Multistep, site-selective noncovalent synthesis of two-dimensional block supramolecular polymers. Nat. Chem. 15, 922-929 (2023).
+23. Khanra, P., Singh, A. K., Roy, L. & Das, A. Pathway complexity in supramolecular copolymerization and blocky star copolymers by a hetero-seeding effect. J. Am. Chem. Soc. 145, 5270–5284 (2023).
+24. Laishram, R. et al. Secondary nucleation-triggered physical cross-links and tunable stiffness in seeded supramolecular hydrogels. J. Am. Chem. Soc. 144, 11306–11315 (2022).
+25. Sasaki, N. et al. Supramolecular double-stranded archimedean apirals and concentric toroids. Nat. Commun. 11, 1–9 (2020).
+26. Singh, N., Lainer, B., Formon, G. J. M., De Piccoli, S. & Hermans, T. M. Re-programming hydrogel properties using a fuel-driven reaction cycle. J. Am. Chem. Soc. 142, 4083–4087 (2020).
+27. Sarkar, S., Sarkar, A., Som, A., Agasti, S. S. & George, S. J. Stereoselective primary and secondary nucleation events in multicomponent seeded supramolecular polymerization. J. Am. Chem. Soc. 143, 11777–11787 (2021).
+28. Datta, S. et al. Self-assembled poly-catenanes from supramolecular toroidal building blocks. Nature 583, 400–405 (2020).
+29. Sarkar, S., Laishram, R., Ded, D. & George, S. J. Controlled noncovalent synthesis of secondary supramolecular polymers. (2023), DOI:10.1021/jacs.3c06844.
+30. Meisl, G. et al. Molecular mechanisms of protein aggregation from global fitting of kinetic models. Nat. Protoc. 11, 252–272 (2016).
+31. Rodriguez Camargo, D. C. et al. Surface-catalyzed secondary nucleation dominates the generation of toxic IAPP aggregates. Front. Mol. Biosci. 8, 1–11 (2021).
+32. Anwar, J., Khan, S. & Lindfors, L. Secondary crystal nucleation: Nuclei breeding factory uncovered. Angew. Chemie - Int. Ed. 54, 14681–14684 (2015).
+33. Törnquist, M. et al. Secondary nucleation in amyloid formation. Chem. Commun. 54, 8667–8684 (2018).
+34. Greciano, E. E., Calbo, J., Ortí, E. & Sánchez, L. N-Annulated perylene bisimides to bias the differentiation of metastable supramolecular assemblies into J- and H-aggregates. Angew. Chemie - Int. Ed. 59, 17517–17524 (2020).
+35. Wagner, W., Wehner, M., Stepanenko, V., Ogi, S. & Würthner, F. Living supramolecular polymerization of a perylene bisimide dye into fluorescent J-aggregates. Angew. Chemie - Int. Ed. 56, 16008–16012 (2017).
+36. Ogi, S., Stepanenko, V., Sugiyasu, K., Takeuchi, M., & Würthner, F. Mechanism of self-assembly process and seeded supramolecular polymerization of perylene bisimide organogelator. J. Am. Chem. Soc. 137, 3300–3307 (2015).
+37. Fukui, T. et al. Control over differentiation of a metastable supramolecular assembly in one and two dimensions. Nat. Chem. 9, 493–499 (2017).
+38. Ogi, S., Matsumoto, K. & Yamaguchi, S. Seeded polymerization through the interplay of folding and aggregation of an amino-acid-based diamide. Angew. Chemie - Int. Ed. 57, 2339–2343 (2018).
+39. Shyshov, O. et al. Living supramolecular polymerization of fluorinated cyclohexanes. Nat. Commun. 12, 1–11 (2021).
+40. Choi, H., Ogi, S., Ando, N. & Yamaguchi, S. Dual trapping of a metastable planarized triarylborane π-system based on folding and lewis acid-base complexation for seeded polymerization. J. Am. Chem. Soc. 143, 2953–2961 (2021).
+41. Wang, H. et al. Living supramolecular polymerization of an aza-BODIPY dye controlled by a hydrogen-bond-accepting triazole unit introduced by click chemistry. Angew. Chemie - Int. Ed. 59, 5185–5192 (2020).
+42. Greciano, E. E., Matarranz, B. & Sánchez, L. Pathway complexity versus hierarchical self‐assembly in N‐Annulated perylenes: structural effects in seeded supramolecular polymerization. Angew. Chemie - Int. ed. 130, 4787–4791 (2018).
+43. Pal, D. S., Kar, H. & Ghosh, S. Controllable supramolecular polymerization: via a chain-growth mechanism. Chem. Commun. 54, 928–931 (2018).
+44. Prabhu, D. D. et al. Self-folding of supramolecular polymers into bioinspired topology. Sci. Adv., 4, eaat8466 (2018).
+45. Yagai, S., Kitamoto, Y., Datta, S. & Adhikari, B. Supramolecular polymers capable of controlling their topology. Acc. Chem. Res. 52, 1325–1335 (2019).
+46. Ma, X. et al. Fabrication of chiral-selective nanotubular heterojunctions through living supramolecular polymerization. Angew. Chemie - Int. Ed. 55, 9539–9543 (2016).
+47. Liu, Y. et al. Emergent self-assembly pathways to multidimensional hierarchical assemblies using a hetero-seeding approach. Chem. - A Eur. J. 25, 13484–13490 (2019).
+48. Hu, K. et al. Fabrication of single-handed nanocoils with controlled length via a living supramolecular self-assembly. Chem. Mater. 31, 1403–1407 (2019).
+49. Kotha, S. et al. Cooperative supramolecular polymerization guided by dispersive interactions. Chem. - An Asian J. 17, 1–9 (2022).
+50. M. J. Frisch et al. Gaussian 16, Revision B.01, Gaussian, Inc., Wallingford CT.
+51. Stewart, J. J. P. Optimization of parameters for semiempirical methods V: modification of NDDO approximations and application to 70 elements. Journal of Molecular Modeling 13, 1173–1213 (2007).
+52. Chai, J.-D. & Head-Gordon, M. Long-range corrected hybrid density functionals with damped atom–atom dispersion corrections. Phys. Chem. Chem. Phys. 10, 6615-6620 (2008).
+53. Cohen, S. I. A. et al. Proliferation of amyloid-Β42 aggregates occurs through a secondary nucleation mechanism. Proc. Natl. Acad. Sci. U. S. A. 110, 9758–9763 (2013).
+54. Dilip, K. K., Rebecca, J. K. & Nolini, S. Chiral symmetry breaking in sodium chlorate crystallization. Science 250, 975-976 (1990).
+55. Waghmare, R. Y., Webb, J. N., Randolph, T. W., Larson, M. A. & Glatz, C. E. Pressure dependence of subtilisin crystallization kinetics. J. Cryst. Growth 208, 678–686 (2000).
+56. Oh, J. S. et al. Chiral supramolecular multiblock copolymerization accompanying chirality transfer in heterostructures via living chain growth. Angew. Chemie Int. Ed. 62, e202300913 (2023).
+
+# Supplementary Files
+
+- [MovieHeteroseeding.mp4](https://assets-eu.researchsquare.com/files/rs-3437857/v1/1cbb249a9575fd737e792c54.mp4)  
+  Movie_Hetero-seeding
+
+- [Supportinginformation.pdf](https://assets-eu.researchsquare.com/files/rs-3437857/v1/b38cf5b8ebc024546f5da9da.pdf)