Add files using upload-large-folder tool

.gitattributes

@@ -57,3 +57,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+dd9fa622529abdee9160c93e7c3ad7399577e19fb9862692ba9a7f882cc52c8a/preprint/images/Figure_4.tif filter=lfs diff=lfs merge=lfs -text
+dd9fa622529abdee9160c93e7c3ad7399577e19fb9862692ba9a7f882cc52c8a/preprint/images/Figure_1.tif filter=lfs diff=lfs merge=lfs -text
+dd9fa622529abdee9160c93e7c3ad7399577e19fb9862692ba9a7f882cc52c8a/preprint/images/Figure_2.tif filter=lfs diff=lfs merge=lfs -text

004d1f4504b43a2431f650a15ed682995331187cad45e266b93bfc3edf062886/preprint/images_list.json

@@ -0,0 +1,50 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Ocean heat storage pattern and the evolution in key regions. a, b, Patterns of 0-2000 m ocean heat content (OHC) trend (W m-2) of 1958-2021 based on the multi-model mean (MMM) of 20 CMIP6 historical and SSP5-8.5 simulations (a) and the Institute of Atmospheric Physics ocean analysis (IAP)35 (b). Stippling indicates significant trends at the 95% confidence level based on a Mann-Kendall test. The black boxes remark the regions with enhanced heat uptake: the mid-latitude North Pacific (MNP; 155\u00b0E-150\u00b0W, 40\u00b0-55\u00b0N) and the northwest tropical Pacific (NWTP; 125\u00b0-180\u00b0E, 8\u00b0-18\u00b0N). c, d, Regional OHCs (GJ m-2, 1 GJ = 109 J) of MNP (c) and NWTP (d) derived from observations and CMIP6 simulations. The observations include IAP, WOA, Ishii, EN4, and ORA-S4. Shadings denote the one standard deviation range.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Effect of wind forcing on heat storage. a-c, Linear OHC trends of 1958-2019 derived from CTRL (a), HTFL (b), and WND (c) runs of the HYCOM 0.5\u00b0 simulation. Stippling indicates significant trends at the 95% confidence level. d, e, Regional OHCs of the MNP (d) and NWTP (e) derived from CTRL, HTFL, and WND of HYCOM 0.5\u00b0 simulation, referenced to those derived from IAP.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Changes in surface winds and impacts of the Pacific Decadal Oscillation (PDO). a, b, Trends of wind stress (arrows; Pa yr-1) and Ekman pumping velocity \u03c9E (color shading; 10-7 m s-1 yr-1) for 1958-2021 based on the ensemble average of four atmospheric reanalysis products (JRA55, ERA5, NCEP-NCAR, and ORA-S4; addressed as \u201cobserved\u201d) (a) and CMIP6 MMM (b). c, Regressions of observed wind stress (Pa) and \u03c9E (10-6 m s-1) on the negative PDO index (-PDO) based on HadISST for 1958-2021. d, Regression map of OHC anomalies derived from IAP on -PDO for 1958-2021. Stippling indicates the fields exceeding the 95% confidence level based on an F-test. The black boxes denote NWTP and MNP. e, f, Same as d, but with the OHC anomalies lagging -PDO by 4 (e) and 9 years (f).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Impacts of the PDO on the Kuroshio Extension (KE). a, The normalized sea surface height (SSH) anomaly of the southern recirculation region (140\u00b0-165\u00b0E, 31\u00b0-36\u00b0N). b, Time-longitude plot of SSH anomaly along the zonal band of 32\u00b0-34\u00b0N. a, b are based on monthly CMEMS data. The black arrow in (b) indicates the propagation of Rossby waves. c, The normalized monthly PDO index based on HadISST data.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Impacts of the PDO on OHC. a, b, Lagged regression map of OHC on -PDO derived from HYCOM-0.1\u00b0 simulation of 1979-2021, with the former lagging the latter by 4 (a) and 9 years (b). c, d, Same as a, b, but based on the MMM of 20 CMIP6 models. Stippling indicates the fields exceeding the 95% confidence level.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "Time of emergence (ToE) for anthropogenic heat storage. a, Multi-model median ToE of the anthropogenic OHC trend under the SSP2-4.5 scenario. The ToE is the year when the trend exceeds the range of the twofold natural interannual standard deviation (Methods). Grey shading denotes the signal that has not emerged by 2099. b-f, Box-and-whisker plots of ToEs of regional OHCs in the Bering Sea (160\u00b0E-170\u00b0W, 52\u00b0-60\u00b0N) (b), Eastern boundary (132\u00b0-117\u00b0W, 30\u00b0-55\u00b0N) (c), central MNP (170\u00b0E-155\u00b0W, 40\u00b0-48\u00b0N) (d), Kuroshio Extension (KE; 141\u00b0-165\u00b0E, 33\u00b0-40\u00b0N) (e), and NWTP (f). The ToEs > 2099 are not shown.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

004d1f4504b43a2431f650a15ed682995331187cad45e266b93bfc3edf062886/preprint/preprint.md

@@ -0,0 +1,172 @@
+# Abstract
+
+Storage of anthropogenic heat in the oceans is spatially inhomogeneous, impacting regional climates and human societies. Climate models project enhanced heat storage in the mid-latitude North Pacific (MNP) and much weaker storage in the tropical Pacific. However, the observed heat storage during the past half-century shows a more complex pattern, with limited warming in the central MNP and enhanced warming in the northwest tropical Pacific. Based on observational datasets, ocean model experiments, and climate models, we show that emergence of human-induced heat storage is likely postponed in the North Pacific by natural variability to the late-21st century. Specifically, phase shifts of the Pacific Decadal Oscillation (PDO) have vitally contributed to trends in the North Pacific winds during recent decades. Changes in surface winds drove meridional heat redistribution via Rossby wave dynamics, leading to regional warming and cooling structures and a more complex historical heat storage than models project. Despite this, enhanced anthropogenic warming has already been emerging in marginal seas along the North Pacific basin rim, for which we shall prepare for the pressing consequences such as increasing marine heatwaves.
+
+Earth and environmental sciences/Ocean sciences/Physical oceanography  
+Earth and environmental sciences/Climate sciences/Ocean sciences/Physical oceanography  
+Earth and environmental sciences/Climate sciences/Climate change/Attribution  
+Earth and environmental sciences/Climate sciences/Climate change/Climate and Earth system modelling
+
+# Introduction
+
+The storage of excess heat caused by anthropogenic greenhouse warming in oceans is geographically inhomogeneous. Climate models suggest that the Southern Ocean, the North Atlantic, and the mid-latitude North Pacific (MNP) warm up more quickly than other oceans (Fig. 1a). These ocean warming features accompany accelerated ocean currents, poleward shift of storm tracks and westerlies, increased extreme atmospheric rivers, and altered marine biodiversity patterns. While the enhanced heat storage in the Southern Ocean and North Atlantic has been witnessed in nature, that in the MNP has hardly emerged (Fig. 1b). Indeed, despite significant warming trends in marginal seas along the basin rim, there were weak warming or cooling trends in the central MNP since the 1950s.
+
+The tropical Pacific is among the regions with the weakest heat storage in CMIP6 multi-model mean (MMM) (Fig. 1a). However, the Northwest Tropical Pacific (NWTP) has exhibited the strongest warming rate of the Pacific Ocean during the past decades (Fig. 1b). Increased ocean heat content (OHC) in the NWTP has led to rapid sea-level rise, increasing marine heatwave and coral bleaching events, and altered tropical cyclone behaviors. It also exerted impacts on the downstream marginal seas through western boundary currents and the Indian Ocean through the Indonesian Throughflow. The Pacific is projected to be the leading reservoir of anthropogenic heat among all oceans by the latter half of the 21st century. Correctly interpreting the model-observation discrepancies there represents a vital scientific issue.
+
+The contrasts between observations and models indicate either systematic model biases in simulating the externally forced heat storage or substantial impacts by natural variability. Although anthropogenic fingerprints have emerged in many aspects of the ocean, natural variability remains influential in the observed OHC changes. For instance, the persistent negative phase of the Pacific Decadal Oscillation (PDO) since 1998 led to heat pile-up in the western tropical Pacific via the strengthened Pacific Walker Circulation. If the model-observation discrepancies can be successfully attributed to natural variability, we shall expect an acceleration of warming in the MNP within the coming decades to catch up with the projected rate.
+
+Here, we set out to quantify the contributions of various processes to the observed North Pacific heat storage pattern, addressing in particular whether natural variability can largely explain the model-observation discrepancy. Our analysis is based on 1) five observational datasets to characterize the historical heat storage, 2) experiments of a forced ocean model to isolate effects of surface heat fluxes and wind-driven ocean dynamics, and 3) 20 CMIP6 models to estimate anthropogenic fingerprints and when they may emerge (Methods). We demonstrate that surface wind changes arising from phase shifts of the PDO have driven basin-scale heat redistribution through Rossby waves and modulations in western boundary currents. This effect complicates the observed heat storage pattern by creating regional warming/cooling structures that conceal anthropogenic fingerprints. According to model projections, the human-induced heat storage pattern will likely hide until the late-21st century. These results provide useful implications for climate prediction in the North Pacific and marginal seas.
+
+## Heat storage in the North Pacific
+
+We begin with more details of the simulated and observed heat storage patterns in the North Pacific during the 1958–2021 period. In the multi-model mean (MMM) of 20 CMIP6 models, consisting of the historical and Shared Socioeconomic Pathway (SSP) 5-8.5 simulations before and after 2014, respectively, the entire MNP is characterized by enhanced 0-2000 m heat storage, in stark contrast to the warming minimum in the western tropical Pacific (Fig. 1a). The observation-based heat storage shows more regional structures (Fig. 1b and Extended Data Fig. 1); for example, there are alternating warming and cooling regions in the North Pacific, particularly in the western basin. The NWTP (e.g., 125°-180°E, 8°-18°N) stands out with a warming maximum, whereas a “warming hole” occurs in the MNP interior with insignificant trends and is surrounded by significant warming trends in marginal seas, such as the Bering Sea and the Gulf of Alaska. There is another cooling region near the western boundary of the 18°-30°N, sandwiched by the warming areas of the NWTP and the Kuroshio extension (KE) between 30°-40°N. Cooling trends are also seen beyond the North Pacific, such as the subpolar North Atlantic, southwestern subtropical Pacific, and southwestern subtropical Indian Ocean. CMIP6 MMM also show cooling or slackened warming in these regions (Fig. 1a). Albeit with weaker intensity and smaller spatial range, the subpolar North Atlantic warming hole is clearly discernible in CMIP6 MMM. Therefore, models and observations are broadly consonant in all major ocean basins except for the North Pacific. The model-observation discrepancies in the North Pacific heat storage are worthy of in-depth investigation.
+
+We further examine the temporal evolution of OHC in key regions (Fig. 1c, d). CMIP6 models suggest an average heat storage rate of 0.72 ± 0.62 W m⁻² (± indicates the one standard deviation range of 20 models) in the MNP of 155°E-150°W, 40°-55°N, close in magnitude to that of the Southern Ocean (0.68 ± 0.27 W m⁻² in MMM for 55°-33°S). In comparison, a much weaker rate of 0.17 ± 0.08 W m⁻² in the MNP is obtained from observation-based datasets (Fig. 1c; ± indicates the one standard deviation range of 5 datasets). Meanwhile, the NWTP shows enhanced warming of 0.58 ± 0.26 W m⁻² in observations, one order stronger than the simulated rate of 0.06 ± 0.32 W m⁻² in CMIP6 models (Fig. 1d). Interestingly, CMIP6 agrees with observation in the OHC change of the entire North Pacific (120°E-80°W, 0°-60°N), which are 0.33 ± 0.07 W m⁻² and 0.33 ± 0.17 W m⁻², respectively, implying a heat redistribution over the North Pacific in the observed realization relative to the simulated pattern.
+
+The observed OHCs show prominent interannual and decadal variabilities in the MNP and NWTP (Fig. 1c, d). Meanwhile, we notice a large inter-model spread relative to the MMM change in the historical simulation of CMIP6 before 2014, which also indicates significant influence from natural variability. The effect of decadal natural variability is particularly notable from the 1990s through the mid-2010s. During this period, the observed OHC of the NWTP was substantially elevated and exceeded the +1 standard deviation range of CMIP6 models, while that of the MNP did not rise significantly as models expected. In the following, we examine how decadal natural variability affects the North Pacific heat storage.
+
+## Heat redistribution driven by wind changes
+
+To understand how the heat storage pattern is formed, we performed experiments using the Hybrid Coordinate Ocean Model (HYCOM) with a coarse horizontal resolution of 0.5° (Methods) and prescribed reanalysis of atmospheric fields as the surface forcing. Despite regional simulation errors, the control run (CTRL) of HYCOM captured broad-scale features of the observed heat storage – the cooling trends in the central MNP and the enhanced warming of the NWTP (Fig. 2a). CTRL also well reproduced the prominent interannual and decadal variabilities (Fig. 2d, e). During the simulation period of 1958–2019, the correlation coefficients between CTRL and IAP are 0.57 and 0.81 for OHCs in the MNP and NWTP, respectively, both significant at the 99% confidence level.
+
+With the aid of HYCOM experiments, we can separate the effects of heat redistribution induced by wind-driven ocean circulation changes and heat uptake through surface heat fluxes in heat storage. The heat-flux run (HTFL) retains changes in surface heat fluxes and keeps wind stress and precipitation invariant (Methods), representing the effect of heat uptake. HTFL produces substantially stronger warming in the MNP than in the tropical Pacific (Fig. 2b) – a pattern resembling CMIP6 MMM (Fig. 1a). This highlights the dominance of heat uptake in shaping the model-projected heat storage pattern. Alternatively, the wind run (WND), retaining changes only in surface wind stress and keeping other forcing fields unchanged (Methods), produces basin-wide cooling trends in the MNP and enhanced warming of the NWTP (Fig. 2c). These wind-driven changes greatly modify the pattern shaped by heat uptake and vitally contribute to the total storage in CTRL. Checking the temporal evolution clearly suggests that while heat fluxes drive quasi-monotonic warming trends in both regions (stronger in MNP), the interannual and decadal fluctuations arise mainly from winds (stronger in NWTP) (Fig. 2d, e). Critically, winds have induced an OHC decrease in the MNP and an abrupt OHC increase in the NWTP since the late 1990s, which greatly altered the overall trends of 1958–2019.
+
+Then, we explore changes in surface winds. Reanalysis datasets suggest basin-scale trends of anti-cyclonic winds over the North Pacific since 1958 (Fig. 3a). Correspondingly, there are easterly winds and negative Ekman pumping velocity ω_E (indicating downwelling; Methods) in the NWTP, which causes convergence of the upper-layer warm water and enhances the heat storage there. Meanwhile, we also see westerly winds and positive w_E (upwelling) in the eastern tropical Pacific, which might also affect the NWTP through Rossby waves. However, this effect fails to dampen the NWTP warming induced by local wind changes, probably owing to the dissipation of Rossby waves during their transition across the Pacific basin. The anti-cyclone also involves strengthening westerlies north of 40°N, which slackens the heat storage in the subpolar North Pacific through Ekman upwelling and enhances heat pile-up in the 30°-40°N band through Ekman downwelling.
+
+In CMIP6 MMM, trends of anti-cyclonic winds are confined north of 30°N and westerly trends occupy the entire tropical Pacific (Fig. 3b). Therefore, some key features of the observed wind trend are missed, including the easterlies in the NWTP and westerlies in the subpolar North Pacific. These differences in surface winds, critically accounting for the model-observation discrepancies in heat storage, likely arise from natural variability, given that the CMIP6 MMM is assumed to represent externally forced changes. The PDO, as the leading mode of decadal natural variability in the North Pacific, was in its positive phase during 1977–1997 and then shifted to its negative phase of 1998–2014 (Extended Data Fig. 2). A regression onto the negative PDO index (-PDO) (Fig. 3c) shows easterlies in the NWTP and westerlies in the northeastern subpolar Pacific – key features seen in observation but missed in CMIP6 MMM. The negative PDO also induces positive w_E anomalies in the western and central parts of the 18°-30°N band, causing the cooling trends observed near the western boundary (Fig. 1b).
+
+The regression of OHC onto the negative PDO shows some features resembling the observed heat storage, particularly the warming in the NWTP and the 30°-40°N band and cooling in the MNP and the 18°-30°N band (Fig. 3d). The NWTP heat content shows correlation coefficients of -0.42 and −0.66 with the 8-year low-passed PDO index and the unfiltered December-January-February Oceanic Niño Index (ONI), respectively (Extended Data Fig. 2a). This suggests a strong modulation of natural variability on the NWTP on interannual and decadal timescales, with heat pile-up in the NWTP under negative PDO and La Niña conditions. In the mid-latitudes, the regression cannot explain the observed heat storage. For example, cooling trends occur mainly in the marginal seas in the regression (Fig. 3d) rather than in the MNP interior as in the trend pattern (Fig. 1b); the meridional dipole-like structure observed east of Japan, linked to the strengthening geostrophic transport of the KE jet (Fig. 1b), is replaced by prevailing warming in Fig. 3d. These discrepancies can be reconciled by considering the time lag of oceanic response to wind forcing through planetary wave adjustments. Lagged regressions show that the strengthened KE east of Japan is established ~4 years after the negative peak of PDO, and the central MNP cooling takes ~9 years (Fig. 3e, f). The wind-driven OHC anomaly of the MNP, measured by the WND run of HYCOM, lags the PDO index (Extended Data Fig. 2b): the PDO shifted from the negative to the positive phase during the 1970s, and correspondingly there was a warming trend of MNP throughout the 1980s; subsequently, the opposite transition from the late 1990s through ~2010 led to a cooling trend of the MNP persisting through the late 2010s.
+
+## Regional ocean dynamics
+
+The above analysis points to the vital role of PDO phase shifts in shaping the historical heat storage pattern through wind-driven redistribution. One question arises as to whether models can correctly simulate the PDO-induced variability, which is critical for predicting regional OHC changes and their climatic and environmental impacts in the upcoming decades. This remains a challenging task for the state-of-the-art models. Even with prescribed reanalysis winds, our 0.5° simulation of HYCOM fails to fully reproduce the observed heat storage in the MNP (Fig. 2a); for instance, the simulated cooling in the central MNP is stronger in amplitude, broader in spatial extent, and shifted to lower latitudes in CTRL compared to that in observation. These discrepancies probably arise from complex regional ocean dynamics that are not properly represented by coarse-resolution models - such as the “bimodal” variability of the KE.
+
+In the mid-latitudes, a negative phase of the PDO drives downwelling Rossby waves through negative w_E anomalies between 150°-140°W (Fig. 3c). Along the latitudinal band of the KE (e.g., 31°-36°N), these Rossby waves propagate across the Pacific basin to the KE region east of Japan by ~4 years, as manifested in satellite-based sea surface height (SSH) anomalies (Fig. 4). These downwelling waves lead to a “stable” state of the KE system characterized by a strengthened KE jet and weakened mesoscale eddy variability, as indicated by the enhanced anticyclonic “southern recirculation” locating south of the jet (Fig. 4a). The strengthened geostrophic KE jet manifests as a meridional dipole in OHC east of Japan (Fig. 3e). In addition to PDO, the KE is also modulated by the upstream Kuroshio path south of Japan. The PDO shifted to a positive phase in ~2014, and correspondingly, an unstable dynamical state of the KE was established in early 2017, as indicated by the weakened southern recirculation (Fig. 4a). However, the occurrence of the Kuroshio large meander in August 2017 - with the Kuroshio following an offshore meandering path in the Shikoku basin south of Japan – compelled the KE to switch back to the stable state in late 2017 and remain stable since then. Albeit interrupted in 2017, there has been a persistently stable KE since ~2009. This led to an overall trend of the KE toward its stable state during the past decades, given the more unstable KE during the 1980s and 1990s due to the positive PDO phase (Extended Data Fig. 3). This trend manifests as a dipole-like structure in heat storage east of Japan (Fig. 1b). By affecting the overlying storm tracks and large-scale winds, the stable KE also contributed to the cooling in the central MNP and the warming in the eastern basin.
+
+The response of KE to Rossby waves is successfully reproduced by a 0.1° simulation of HYCOM (Methods; Extended Data Fig. 3) that can better resolve the Kuroshio path and mesoscale eddies. As a result, this 0.1° simulation can realistically represent the heat redistribution driven by PDO winds, including the meridional dipole east of Japan with a lag time of ~4 years (Fig. 5a) and the cooling in the central MNP with a lag time of ~9 years (Fig. 5b) to the negative peak of PDO. This ~9-year lag time reflects the slow propagation of upwelling Rossby waves between 40°-55°N (Extended Data Fig. 4) generated by positive w_E anomalies near the west coasts of North America (Fig. 3c). There are also detailed discrepancies between the 0.1° simulation and observation in the KE region (Fig. 5a, b compared to Fig. 3e, f), which may result from the lack of feedback between mesoscale eddies and the atmosphere in our standard-alone HYCOM simulation. The eddy-atmosphere feedback has been demonstrated to efficiently damp mesoscale eddies and amend the simulated strength and pathway of the KE.
+
+Most CMIP6 models are of horizontal resolutions of ~50 or ~100 km in their ocean components, close to our 0.5° HYCOM simulation. We examine how the PDO-induced heat redistribution is represented in these models. The regression using individual simulations of CMIP6 models (Methods) suggests that CMIP6 models can capture some large-scale features of the PDO-induced changes in surface winds and OHC, particularly in the mid-latitudes (Fig. 5c, d and Extended Data Fig. 5). The cooling of the central MNP is seen at a lag of ~9 years, albeit with extended spatial range (Fig. 5d). The PDO’s signatures in the NWTP are weaker in models than in observations, linked to the uncertain tropical Pacific wind changes (Extended Data Fig. 5). This indicates a weaker coupling between the PDO and the tropical Pacific climate in models. By prescribing the observed (reanalysis) surface winds, the 0.5° HYCOM simulations can well reproduce the response of NWTP changes associated with the PDO winds (Extended Data Fig. 6). Furthermore, the improved fidelity of the 0.1° HYCOM (Fig. 5a, b) relative to the 0.5° HYCOM and CMIP6 models highlights the necessity of fine resolution to account for the complex ocean dynamics involved in heat redistribution.
+
+To further confirm the PDO’s impact, we select 5 model realizations with the simulated PDO showing the highest positive correlations with the observed PDO (+5 models) and 5 model realizations showing the highest negative correlation with the observation (-5 models). As such, contrasting the +5 and −5 models mimics impacts of the observed PDO phase shifts on the OHC trends since 1958 (Extended Data Fig. 7). The composite of +5 minus −5 models reassures that the PDO phase shifts in observation can dampen the heat storage in the MNP and enhance the warming of the NWTP (Extended Data Fig. 7c) through wind-driven heat redistribution (Extended Data Fig. 7d). Therefore, despite unresolved regional structures due to coarse model resolutions, discrepancies between CMIP6 models and observations in the North Pacific heat storage pattern, particularly the contrasts in the NWTP and the central MNP, are primarily explained by natural variability represented by the PDO.
+
+## Emergence of anthropogenic heat storage
+
+The above analysis indicates that the emergence of anthropogenic heat storage has been delayed by natural variability in some key regions of the North Pacific and will eventually take place in the future as the emission of greenhouse gases continues. Therefore, it is instructive to estimate the time of emergence (ToE) of anthropogenic heat storage based on CMIP6 simulations (Fig. 6), although we fully acknowledge that such estimation is subject to potential influence from inaccurate ocean dynamics as discussed earlier. Here, we utilize the piControl simulation, historical simulation, and projections under three emission scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) of 20 CMIP6 models to determine the “signal-to-noise” ratio and compute the ToE for the 0-2000 m OHC change (Methods).
+
+Under the “middle-of-the-road” scenario of SSP2-4.5, a scenario viewed as where social and economic trends do not shift markedly from historical patterns, the anthropogenic heat storage (Fig. 6a) has already emerged (e.g., ToE earlier than the 2020s) in the Southern Oceans between 55°-33°S and the tropical and northern Atlantic. In the North Pacific, the ToE is as early as the 2010s in marginal seas along the basin rim, including the western coasts of North America, the Bering Sea, the Okhotsk Sea (most of which is shallower than 2000 m and not estimated for OHC and ToE), and the Japan Sea (Fig. 6a). By contrast, anthropogenic signals have not emerged by present (the 2020s) in the basin interior and even until the 2050s in the central MNP, the KE, and the NWTP where we see obvious model-observation discrepancies. The ToEs under SSP1-2.6 and SSP5-8.5 scenarios are similar in spatial distribution to those under SSP2-4.5 but postponed and advanced in average time (Extended Data Fig. 8), respectively.
+
+In marginal seas of the North Pacific, such as the Bering Sea and the eastern boundary coasts, anthropogenic heat storage has already been emerging. The multi-model median ToE is during the 2010s and 2020s in these regions regardless of the emission scenario (Fig. 6b, c). By contrast, the central MNP, KE, and NWTP regions show delayed ToEs and relatively high sensitivities to the emission scenario (Fig. 6d-f). Under the SSP1-2.6 and SSP2-4.5 scenarios, anthropogenic heat storage is projected to emerge in the latter half of this century in these regions, such as the 2080s in the KE region. Under the SSP5-8.5 scenario, the ToE is as early as the 2040s in the central MNP, the 2050s in the KE, and the 2060s in the NWTP. To summarize, despite uncertainties arising from inter-model spread, the North Pacific heat storage is projected to remain under a significant impact of natural variability until the late 21st century, except for marginal seas along the basin rim. Therefore, we expect that the heat storage pattern of the North Pacific will alter but probably still differ from that projected by climate models in the upcoming several decades.
+
+# Summary and Implications
+
+Based on observational datasets, ocean model experiments, and CMIP6 model simulations, we demonstrate a strong modification effect by the PDO on the observed heat storage pattern in the North Pacific since the mid-20th century. Specifically, phase shifts of the PDO in recent decades have altered the trends in surface winds over the North Pacific. Surface wind changes induced meridional heat redistribution through Rossby waves and variability of the KE system, effectively erasing the warming in the central MNP and fueling the heat pile-up in the NWTP. These effects led to a more complex heat storage pattern in observations than in models by creating regional warming/cooling structures that mask the human-induced fingerprints (Extended Data Fig. 9).
+
+Recognition of the strong influence of PDO on heat storage provides critical insights into near-term climate prediction. The persistent negative phase of PDO since 1998, critical in shaping the multi-decadal trends of surface winds and heat storage, terminated in the mid-2010s (Fig. 4c). If the PDO switches to a positive phase in the upcoming decade, the central MNP “warming hole” shall vanish soon and the basin-scale heat storage will evolve toward the pattern projected by models. However, the recent triple La Niña events in 2020–2023 and the possible developing La Niña condition in 2024 and their extratropical impacts indicate a likelihood of a lengthened negative phase, which acts to maintain the existing heat storage pattern. The broad impacts of the PDO on surface atmospheric and ocean conditions have been well established, such as storm tracks, atmospheric rivers, and Pacific salmon production. Here, we add that the PDO is also critical for the transient response of the North Pacific to anthropogenic forcing. Through modulating wind-driven ocean circulations, the PDO fundamentally regulates the distributions of anthropogenic heat, CO₂, and the Fukushima nuclear effluents in the North Pacific and their spread into the Indian and Arctic Oceans via the Indonesian through-flow and Bering Strait, respectively. These implications highlight the urgent need for an accurate initialized prediction of the PDO and its far-reaching impacts.
+
+Climate models tend to underestimate the PDO variability and its impacts over the historical period. As a result, none of the 20 CMIP6 models has produced a historical heat storage pattern resembling the observation (not shown). Increasing model resolution seems promising. The representation of natural variability in the North Pacific is considerably improved in ~10 km simulations compared to ~100 km simulations; our analysis based on the 0.1° HYCOM simulation also supports this notion. In addition, increasing ensemble members can better account for the extreme cases of natural variability, such as the 1998–2014 negative phase, and improve the ToE estimate.
+
+The human-induced heat storage pattern in the Pacific Ocean is projected to eventually emerge by the late 21st century, which is much later than in the Atlantic and Southern Oceans. Apart from the strong impacts of natural variability, the postponed ToEs in the Pacific are linked to the lack of a deep-reaching meridional overturning circulation, which also holds true for the Indian Ocean. In the North Atlantic and Southern Oceans, overturning circulations can efficiently communicate anthropogenic heat to the deep ocean, enhancing the average heat uptake and the signal-to-noise ratio in heat storage. Nevertheless, human-induced warming has already been emerging in marginal seas along the North Pacific basin rim, such as the Bering Sea, Okhotsk Sea, Japan Sea, Gulf of Alaska, and western coasts of North America. Increasing frequency and severity of marine heatwaves have been extensively reported recently, with adverse influences on marine ecosystems and regional socioeconomics. Anthropogenic fingerprints in these marginal seas could be enhanced by influence from lands where the air temperature warms quicker than in oceans or coastal processes that are insufficiently resolved by climate models. This calls for extensive investigations to reveal the mechanisms underlying the warming of marginal seas and multi-disciplinary implications, which shall aid the prediction and decision-making.
+
+# Methods
+
+**Datasets.** To estimate the 0-2000 m ocean heat content (OHC) trend since the mid-20th century, ocean temperature data since 1958 from four observational analyses and one reanalysis product are utilized. The four observational analyses are the Institute of Atmospheric Physics ocean analysis (IAP) for 1958–2021 provided by the Chinese Academy of Sciences<sup>35</sup>, the World Ocean Atlas (WOA) for 1958–2018 provided by the National Oceanic and Atmospheric Administration’s (NOAA’s) National Centers for Environmental Information of the United States<sup>37</sup>; the Ishii analysis for 1958-2021<sup>36</sup>; and the version 4.2.2 of the Met Office Hadley Centre ‘‘EN’’ series of datasets (EN4) for 1958–2021 provided by the Met Office of the United Kingdom<sup>38</sup>. The one ocean reanalysis product is the Ocean Reanalysis System 4 (ORA-S4) for 1958-2017<sup>39</sup> from the European Centre for Medium-Range Weather Forecasts (ECMWF). All five datasets provide 1°×1° ocean temperature fields. WOA provides pentad-mean data in annual intervals, and monthly fields of IAP, Ishii, EN4, and ORA-S4 are averaged into annual-mean data. Surface winds of four atmospheric reanalysis products are used, including the 1.25°×1.25° Japanese 55-year Reanalysis (JRA55) product for 1958-2021<sup>41</sup>, the 1°×1° ORA-S4 product as a combination of the 40-year ECMWF Re-Analysis (ERA-40) of 1958-1988<sup>42</sup> and the ECMWF Interim Re-Analysis (ERA-Interim) of 1989-2017<sup>43</sup>, the 2.5°x2.5° National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) reanalysis for 1958-2021<sup>44</sup>, and the 0.25°x0.25° fifth generation ECMWF reanalysis (ERA5) for 1958-2021<sup>75</sup>. To examine the changes in the Kuroshio Extension (KE) jet, the 0.25°x0.25° sea surface height data distributed by the Copernicus Marine and Environment Monitoring Service (CMEMS) for 1993–2021 is used.
+
+**The coupled model simulations.** Monthly outputs of the pre-industrial control (piControl) (the last 100 years), historical (from 1958 to 2014), Shared Socioeconomic Pathways 126 (SSP1-2.6), SSP2-4.5, and SSP5-8.5 (from 2015 to 2100) simulations of twenty models of the Coupled Model Intercomparison Project Phase 6 (CMIP6) are analyzed. These models include the following: ACCESS-CM2, ACCESS-ESM1-5, BCC-CSM2-MR, CAMS-CSM1-0, CESM2-WACCM, CNRM-CM6-1, CanESM5, EC-Earth3, EC-Earth3-Veg, FGOALS-g3, FIO-ESM-2-0, GFDL-ESM4, GISS-E2-1-G, INM-CM4-8, INM-CM5-0, MPI-ESM1-2-LR, MRI-ESM2-0, NESM3, NorESM2-LM, NorESM2-MM. The multi-model mean (MMM) of 20 CMIP6 models represents the externally forced trend, and the inter-model spread represents the natural climate variability and difference in model physics. Monthly fields of CMIP6 simulations are interpolated onto the same set of 1°×1° grids and averaged into annual-mean data to match observational datasets.
+
+**HYCOM experiments.** We use the Hybrid Coordinate Ocean Model (HYCOM) version-2.3.01<sup>76</sup> to simulate and understand the ocean heat storage pattern. The HYCOM 0.5° simulation adopts a quasi-global ocean domain (78°S-75°N, 30°-389.5°E) with resolutions of 0.5° × 0.5° × 50 levels. The model is spun up from a state of rest for 600 years under the daily JRA55 atmospheric forcing of 1958, and the atmospheric surface forcing fields include the wind stress, net shortwave and longwave radiations (SWR and LWR), wind speed, precipitation, and air temperature and humidity. The control run (CTRL) is forced with realistic daily atmospheric forcings of JRA55 reanalysis from 1958 to 2019. CTRL contains complete physical processes of forming ocean heat storage pattern and is compared with observational data to evaluate the model performance. Two sensitivity runs are conducted to explore the effects of different drivers on the ocean heat storage pattern. Only wind stress adopts realistic daily fields in wind stress run (WND), while all the other forcings, such as wind speed and radiations, are fixed to the 1958 fields as in spin-up. In the WND, OHC trends primarily arise from heat redistribution due to wind stress-driven changes in ocean circulation processes. The heat flux run (HTFL) only adopts realistic SWR and LWR radiations, 2-m air temperature, specific humidity, and wind speed from 1958 to 2019 but fixes wind stress and precipitation to the 1958 fields. As such, changes in these fields of HTFL affect OHC trends mainly by altering the heat fluxes (SWR, LWR, sensible, and latent heat fluxes), representing the role of heat-driven processes or ocean heat uptake. To examine the impacts of decadally varying KE jet and its associated mesoscale eddies, we also use the HYCOM version-2.3.34<sup>77</sup>. The HYCOM 0.1° simulation uses the same domain as the HYCOM 0.5° simulation but with resolutions of 0.1° × 0.1° × 50 levels. After a 20-year spin-up, the HYCOM 0.1° simulation is integrated forwards from 1979 to 2021 under daily ERA5 atmospheric forcing. The comparison of the HYCOM 0.1° simulation and the HYCOM 0.5° CTRL allows us to test the possible impact of mesoscale oceanic processes on heat storage pattern in the North Pacific.
+
+**Definitions**
+
+The OHC is calculated by integrating ocean temperature *T* within the upper 2000 m as  
+OHC = $\:{\int\:}_{0}^{2000}\rho\:{c}_{p}Tdz$, (1)  
+where *c*<sub>p</sub> = 4096 J (kg °C)<sup>−1</sup> is thermal capacity, and *ρ* = 1025 kg m<sup>− 3</sup> is seawater density. The Ekman pumping velocity is calculated as *ω*<sub><em>E</em></sub> = *curl(**τ**/f)ρ*<sup><em>−1</em></sup>, where **τ** is the wind stress vector, and *f* is the Coriolis parameter. The least square method is used to estimate the linear trend. Unless otherwise stated, the statistical significance of the trend is defined as > 95% confidence level based on a modified Mann-Kendall test. The Pacific decadal oscillation (PDO) index is calculated as the leading principle component from the empirical orthogonal function (EOF) analysis of North Pacific (poleward of 20°N) SST anomalies<sup>61</sup>, using the Met Office’s Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST)<sup>78</sup> for observation and historical and SSP5-8.5 simulations of the 20 CMIP6 models. The long-term trend is removed at each grid point before conducting the EOF analysis. The regressions of OHC and wind anomalies on the negative PDO index (-PDO) are first calculated from each of the 20 CMIP6 models, and then the MMM regression maps are derived from the ensemble mean of their coefficients. The Oceanic Niño Index (ONI) is the 3-month running-mean detrended temperature anomalies in the Niño-3.4 region (5°N-5°S, 120°-170°W) based on version 5 of the Extended Reconstructed Sea Surface Temperature<sup>79</sup>, and the monthly ONI is averaged to the annual mean.
+
+**Time of Emergence (ToE).** A “signal-to-noise” method<sup>28,57</sup> is adopted to compute the time of emergence (ToE) of the OHC trend. We use the piControl simulation, historical simulation, and projections under three emission scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) of 20 CMIP6 models to determine the “signal-to-noise” ratio (SNR). The noise is defined as the interannual standard deviation of the quadratically detrended<sup>80</sup> OHC variability of piControl simulation in the last 100 years, representing the natural noise envelope. The anthropogenic signals from 1958 to 2100 are computed as the OHC differences between the historical plus three emission scenario simulations and the time-mean piControl value. The ToE is defined as the time when the signal last exceeds and remains above a threshold of two times of noise (i.e., SNR > 2.0). The threshold for SNR is chosen as 2.0 because it represents a 95% confidence level of signal emergence<sup>81</sup>. For the regional ToEs, the noise and signal of selected regions are first averaged and then the time with SNR > 2.0 is searched.
+
+# References
+
+1. IPCC. *Climate Change 2021: The Physical Science Basis*. (Cambridge University Press, 2021).
+2. Fox-Kemper, B. *et al.* Ocean, cryosphere and sea level change. Climate Change 2021: The Physical Science Basis. (Cambridge University Press, 2021).
+3. Cheng, L. *et al.* Past and future ocean warming. Nat. Rev. Earth Environ. 3, 776–794 (2022).
+4. Cheng, L. *et al.* Another Year of Record Heat for the Oceans. Adv. Atmos. Sci. 40, 963–974 (2023).
+5. Eyring, V. *et al.* Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev. 9, 1937–1958 (2016).
+6. Kuhlbrodt, T., Voldoire, A., Palmer, M. D., Geoffroy, O. & Killick, R. E. Historical Ocean Heat Uptake in Two Pairs of CMIP6 Models: Global and Regional Perspectives. J. Clim. 36, 2183–2203 (2023).
+7. Gille, S. T. Warming of the Southern Ocean Since the 1950s. Science (80-.). 295, 1275–1277 (2002).
+8. Shi, J. R., Xie, S. P. & Talley, L. D. Evolving relative importance of the Southern Ocean and North Atlantic in anthropogenic ocean heat uptake. J. Clim. 31, 7459–7479 (2018).
+9. Cai, W. *et al.* Southern Ocean warming and its climatic impacts. Sci. Bull. 68, 946–960 (2023).
+10. Peng, Q. *et al.* Surface warming–induced global acceleration of upper ocean currents. Sci. Adv. 8, 1–12 (2022).
+11. Tamarin-Brodsky, T. & Kaspi, Y. Enhanced poleward propagation of storms under climate change. Nat. Geosci. 10, 908–913 (2017).
+12. Swart, N. C. & Fyfe, J. C. Observed and simulated changes in the Southern Hemisphere surface westerly wind-stress. *Geophys. Res. Lett.* 39, n/a-n/a (2012).
+13. Wang, S. *et al.* Extreme atmospheric rivers in a warming climate. Nat. Commun. 14, 3219 (2023).
+14. Ibarbalz, F. M. *et al.* Global Trends in Marine Plankton Diversity across Kingdoms of Life. Cell 179, 1084–1097.e21 (2019).
+15. Han, W. *et al.* Intensification of decadal and multi-decadal sea level variability in the western tropical Pacific during recent decades. Clim. Dyn. 43, 1357–1379 (2014).
+16. Hamlington, B. D. *et al.* Uncovering an anthropogenic sea-level rise signal in the Pacific Ocean. Nat. Clim. Chang. 4, 782–785 (2014).
+17. Hughes, T. P. *et al.* Spatial and temporal patterns of mass bleaching of corals in the Anthropocene. Science (80-.). 359, 80–83 (2018).
+18. Smale, D. A. *et al.* Marine heatwaves threaten global biodiversity and the provision of ecosystem services. Nat. Clim. Chang. 9, 306–312 (2019).
+19. Oliver, E. C. J. *et al.* Longer and more frequent marine heatwaves over the past century. Nat. Commun. 9, 1324 (2018).
+20. Kossin, J. P., Emanuel, K. A. & Vecchi, G. A. The poleward migration of the location of tropical cyclone maximum intensity. Nature 509, 349–352 (2014).
+21. Shan, K., Lin, Y., Chu, P., Yu, X. & Song, F. Seasonal advance of intense tropical cyclones in a warming climate. Nature (2023) doi: 10.1038/s41586-023-06544-0.
+22. Mei, W., Xie, S. P., Primeau, F., McWilliams, J. C. & Pasquero, C. Northwestern Pacific typhoon intensity controlled by changes in ocean temperatures. Sci. Adv. 1, (2015).
+23. Wu, L. *et al.* Enhanced warming over the global subtropical western boundary currents. Nat. Clim. Chang. 2, 161–166 (2012).
+24. Wang, F. *et al.* The seas around China in a warming climate. Nat. Rev. Earth Environ. (2023) doi: 10.1038/s43017-023-00453-6.
+25. Hu, D. *et al.* Pacific western boundary currents and their roles in climate. Nature 522, 299–308 (2015).
+26. Lee, S. K. *et al.* Pacific origin of the abrupt increase in Indian Ocean heat content during the warming hiatus. Nat. Geosci. 8, 445–449 (2015).
+27. Li, Y., Han, W. & Zhang, L. Enhanced Decadal Warming of the Southeast Indian Ocean During the Recent Global Surface Warming Slowdown. Geophys. Res. Lett. 44, 9876–9884 (2017).
+28. Lyu, K., Zhang, X., Church, J. A., Slangen, A. B. A. & Hu, J. Time of emergence for regional sea-level change. Nat. Clim. Chang. 4, 1006–1010 (2014).
+29. Silvy, Y., Sallée, J. B., Guilyardi, E., Mignot, J. & Rousset, C. What Causes Anthropogenic Ocean Warming to Emerge from Internal Variability in a Coupled Model? J. Clim. 35, 3835–3854 (2022).
+30. Watanabe, M., Dufresne, J. L., Kosaka, Y., Mauritsen, T. & Tatebe, H. Enhanced warming constrained by past trends in equatorial Pacific sea surface temperature gradient. Nat. Clim. Chang. 11, 3–9 (2020).
+31. Wu, B., Lin, X. & Yu, L. North Pacific subtropical mode water is controlled by the Atlantic Multidecadal Variability. Nat. Clim. Chang. 10, 238–243 (2020).
+32. England, M. H. *et al.* Recent intensification of wind-driven circulation in the Pacific and the ongoing warming hiatus. Nat. Clim. Chang. 4, 222–227 (2014).
+33. Nieves, V., Willis, J. K. & Patzert, W. C. Recent hiatus caused by decadal shift in Indo-Pacific heating. Science (80-.). 349, 532–535 (2015).
+34. Liu, W., Xie, S. P. & Lu, J. Tracking ocean heat uptake during the surface warming hiatus. Nat. Commun. 7, 1–9 (2016).
+35. Cheng, L. *et al.* Improved estimates of ocean heat content from 1960 to 2015. Sci. Adv. 3, (2017).
+36. Ishii, M. *et al.* Accuracy of Global Upper Ocean Heat Content Estimation Expected from Present Observational Data Sets. SOLA 13, 163–167 (2017).
+37. Levitus, S. *et al.* World ocean heat content and thermosteric sea level change (0-2000 m), 1955–2010. *Geophys. Res. Lett.* 39, n/a-n/a (2012).
+38. Good, S. A., Martin, M. J. & Rayner, N. A. EN4: Quality controlled ocean temperature and salinity profiles and monthly objective analyses with uncertainty estimates. J. Geophys. Res. Ocean. 118, 6704–6716 (2013).
+39. Balmaseda, M. A., Mogensen, K. & Weaver, A. T. Evaluation of the ECMWF ocean reanalysis system ORAS4. Q. J. R. Meteorol. Soc. 139, 1132–1161 (2013).
+40. Chemke, R., Zanna, L. & Polvani, L. M. Identifying a human signal in the North Atlantic warming hole. Nat. Commun. 11, 1–7 (2020).
+41. Harada, Y. *et al.* The JRA-55 reanalysis: Representation of atmospheric circulation and climate variability. J. Meteorol. Soc. Japan 94, 269–302 (2016).
+42. Uppala, S. M. *et al.* The ERA-40 re-analysis. Q. J. R. Meteorol. Soc. 131, 2961–3012 (2005).
+43. Dee, D. P. *et al.* The ERA-Interim reanalysis: configuration and performance of the data assimilation system. Q. J. R. Meteorol. Soc. 137, 553–597 (2011).
+44. Kalnay, E. *et al.* The NCEP/NCAR 40-Year Reanalysis Project. Bull. Am. Meteorol. Soc. 77, 437–472 (1996).
+45. Duan, J., Li, Y., Cheng, L., Lin, P. & Wang, F. Heat Storage in the Upper Indian Ocean: The Role of Wind-Driven Redistribution. J. Clim. 36, 2221–2242 (2023).
+46. Qiu, B. & Lukas, R. Seasonal and interannual variability of the North Equatorial Current, the Mindanao Current, and the Kuroshio along the Pacific western boundary. J. Geophys. Res. Ocean. 101, 12315–12330 (1996).
+47. Qiu, B. & Chen, S. Variability of the Kuroshio Extension Jet, Recirculation Gyre, and Mesoscale Eddies on Decadal Time Scales. J. Phys. Oceanogr. 35, 2090–2103 (2005).
+48. Qiu, B., Chen, S., Schneider, N., Oka, E. & Sugimoto, S. On the Reset of the Wind-Forced Decadal Kuroshio Extension Variability in Late 2017. J. Clim. 33, 10813–10828 (2020).
+49. Qiu, B. & Chen, S. Effect of Decadal Kuroshio Extension Jet and Eddy Variability on the Modification of North Pacific Intermediate Water. J. Phys. Oceanogr. 41, 503–515 (2011).
+50. Taguchi, B. *et al.* Decadal Variability of the Kuroshio Extension: Observations and an Eddy-Resolving Model Hindcast*. J. Clim. 20, 2357–2377 (2007).
+51. Qiu, B. & Chen, S. Interannual-to-Decadal Variability in the Bifurcation of the North Equatorial Current off the Philippines. J. Phys. Oceanogr. 40, 2525–2538 (2010).
+52. Kawabe, M. Variations of Current Path, Velocity, and Volume Transport of the Kuroshio in Relation with the Large Meander. J. Phys. Oceanogr. 25, 3103–3117 (1995).
+53. Qiu, B., Chen, S. & Oka, E. Why Did the 2017 Kuroshio Large Meander Event Become the Longest in the Past 70 Years? Geophys. Res. Lett. 50, 1–9 (2023).
+54. Silva, E. N. S. & Anderson, B. T. Northeast Pacific marine heatwaves linked to Kuroshio Extension variability. Commun. Earth Environ. 4, 367 (2023).
+55. Ma, X. *et al.* Western boundary currents regulated by interaction between ocean eddies and the atmosphere. Nature 535, 533–537 (2016).
+56. Newman, M. *et al.* The Pacific Decadal Oscillation, Revisited. J. Clim. 29, 4399–4427 (2016).
+57. Silvy, Y., Guilyardi, E., Sallée, J. B. & Durack, P. J. Human-induced changes to the global ocean water masses and their time of emergence. Nat. Clim. Chang. 10, 1030–1036 (2020).
+58. Jones, N. Rare ‘triple’ La Niña climate event looks likely - what does the future hold? *Nature* vol. 607 21 (2022).
+59. Geng, T. *et al.* Emergence of changing Central-Pacific and Eastern-Pacific El Niño-Southern Oscillation in a warming climate. Nat. Commun. 13, 6616 (2022).
+60. Gan, B. & Wu, L. Seasonal and Long-Term Coupling between Wintertime Storm Tracks and Sea Surface Temperature in the North Pacific. J. Clim. 26, 6123–6136 (2013).
+61. Mantua, N. J., Hare, S. R., Zhang, Y., Wallace, J. M. & Francis, R. C. A Pacific Interdecadal Climate Oscillation with Impacts on Salmon Production. Bull. Am. Meteorol. Soc. 78, 1069–1079 (1997).
+62. Payne, A. E. *et al.* Responses and impacts of atmospheric rivers to climate change. Nature Reviews Earth and Environment vol. 1 143–157 (2020).
+63. Xu, G. *et al.* High-resolution modelling identifies the Bering Strait’s role in amplified Arctic warming. Nat. Clim. Chang. (2024) doi: 10.1038/s41558-024-02008-z.
+64. Farneti, R. Modelling interdecadal climate variability and the role of the ocean. WIREs Clim. Chang. 8, (2017).
+65. Chang, P. *et al.* An Unprecedented Set of High-Resolution Earth System Simulations for Understanding Multiscale Interactions in Climate Variability and Change. J. Adv. Model. Earth Syst. 12, (2020).
+66. An, B. *et al.* The benefits of high-resolution models in simulating the Kuroshio Extension and its long-term changes. Clim. Dyn. 61, 5407–5427 (2023).
+67. Chung, E. S. *et al.* Reconciling opposing Walker circulation trends in observations and model projections. Nat. Clim. Chang. 9, 405–412 (2019).
+68. Ren, Q. *et al.* Substantial Warming of the Atlantic Ocean in CMIP6 Models. J. Clim. (2024) doi: 10.1175/JCLI-D-23-0418.1.
+69. Di Lorenzo, E. & Mantua, N. Multi-year persistence of the 2014/15 North Pacific marine heatwave. Nat. Clim. Chang. 6, 1042–1047 (2016).
+70. Amaya, D. J., Miller, A. J., Xie, S. & Kosaka, Y. Physical drivers of the summer 2019 North Pacific marine heatwave. Nat. Commun. 11, 1903 (2020).
+71. Song, Q., Wang, C., Yao, Y. & Fan, H. Unraveling the Indian monsoon’s role in fueling the unprecedented 2022 Marine Heatwave in the Western North Pacific. npj Clim. Atmos. Sci. 7, 90 (2024).
+72. Cavole, L. *et al.* Biological Impacts of the 2013–2015 Warm-Water Anomaly in the Northeast Pacific: Winners, Losers, and the Future. *Oceanography* 29, (2016).
+73. Nielsen, J. M. *et al.* Responses of ichthyoplankton assemblages to the recent marine heatwave and previous climate fluctuations in several Northeast Pacific marine ecosystems. Glob. Chang. Biol. 27, 506–520 (2021).
+74. Lima, F. P. & Wethey, D. S. Three decades of high-resolution coastal sea surface temperatures reveal more than warming. Nat. Commun. 3, 704 (2012).
+75. Hersbach, H. *et al.* ERA5 monthly averaged data on single levels from 1940 to present. Copernicus Clim. Chang. Serv. Clim. Data Store (2023) doi: 10.24381/cds.f17050d7.
+76. Chassignet, E. P., Smith, L. T., Halliwell, G. R. & Bleck, R. North Atlantic Simulations with the Hybrid Coordinate Ocean Model (HYCOM): Impact of the Vertical Coordinate Choice, Reference Pressure, and Thermobaricity. J. Phys. Oceanogr. 33, 2504–2526 (2003).
+77. Wallcraft, a. J., Metzger, E. J. & Carrol, S. N. Software Design Description for the HYbrid Coordinate Ocean Model (HYCOM). 149 (2009).
+78. Rayner, N. A. Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res. 108, 4407 (2003).
+79. Huang, B. *et al.* Extended Reconstructed Sea Surface Temperature, Version 5 (ERSSTv5): Upgrades, Validations, and Intercomparisons. J. Clim. 30, 8179–8205 (2017).
+80. Cai, W. *et al.* Butterfly effect and a self-modulating El Niño response to global warming. Nature 585, 68–73 (2020).
+81. Hawkins, E. & Sutton, R. Time of emergence of climate signals. *Geophys. Res. Lett.* 39, n/a-n/a (2012).
+
+# Supplementary Files
+
+- [ExtendedDataFig.docx](https://assets-eu.researchsquare.com/files/rs-4905116/v1/0578f47807872655dc583377.docx)

00e60208cf2b9b6bda3f4e587570f84a7a278bfd5878e116520a3364b52f4292/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "HIV Gag p24 persists in germinal centers (GCs) despite early ART initiated in Fiebig stages I/II. (A) IF images of Gag p24 antigen (yellow) and BCL-6 (green) staining in lymph node (LN) sections and a scatter plot showing CD4+ T cell counts, plasma viral loads and the time of LN excision for a Fiebig I treated (Tx) participant. Nuclei are counterstained with DAPI (blue) and scale bar is 20 \u03bcm. (B) Gag p24 density per GC computed from TissueQuest^TM analysis of IF LN sections for 2 HIV negative and 14 Fiebig I/II Tx donors and (C) correlation analyses of average density of Gag p24 per donor and the treatment duration prior to LN excision. (D) Representative IF images of Gag p24 antigen (yellow) and BCL-6 (green) staining (scale bar is 50 \u03bcm) and (E) aggregate data from TissueQuest^TM analysis of IF LN sections for 14 Fiebig I/II Tx, 3 Fiebig III-V Tx, 8 late Tx and 13 untreated (unTx) HIV-infected individuals. Each dot represents the density of Gag p24 per GC and the total number of GCs analysed per group is displayed. (F) Comparison of the total percentage area staining for Gag p24 within GCs and outside the GCs (EF) for all donors. The median and interquartile range adjusted P values from Dunn\u2019s test (E) or a Mann U-Whitney P value (F) are shown. Spearman rho (r) values and P values are reported for correlation analyses (C). Dotted line denotes threshold of detection (B).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "HIV-RNA detection in lymph nodes (LN) of Fiebig I/II treated individuals using RNAscope (A-E) and Cobas\u00ae AmpliPrep HIV-1 test (F-I). (A) RNAscope hybridization for HIV gag-pol RNA was detected using 3, 3\u2019-diaminobenzidene (DAB, brown) or (B) fluorescent Opal polymers (green). Representative images for HIV negative, Fiebig I/II treated (Tx), late Tx and untreated (unTx) HIV-infected LN sections are shown. Single RNA transcripts are seen as punctate dots; clusters of transcripts are also observed. Red arrowheads identify HIV RNA+ cells and green arrowheads identify virions on follicular dendritic cells. (C) Images showing multiplexed RNAscope gag-pol hybridization (green) coupled with IF staining for CD4+ cells (red). Scale bars are 50 \u03bcm (A) or 20 \u03bcm (B & C). (D) RNA signals quantified using Fiji was compared for 12 Fiebig I/II Tx, 4 Fiebig III-V Tx, 2 late Tx and 4 unTx LNs. (E) A correlation analysis of area staining of gag-pol RNA and Gag p24 density for Fiebig I/II Tx LNs. (F) Viral RNA loads are quantified in lymph node mononuclear cells (LNMCs) isolated from 7 Fiebig I/II Tx, 4 Fiebig III-V Tx, 7 late Tx and 9 unTx LNs. Viral loads below the limits of detection of the assay are assigned a value of 20. (G-I) Correlation analysis of LNMCs\u2019 viral loads with (G) peak plasma viral loads, (H) treatment duration, and (I) time to suppression for 7 Fiebig I/II Tx and 4 Fiebig III-V donors. Mann U-Whitney P and Spearman rho (r) and P values are reported. Dotted line denotes threshold of viral load detection.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Expansion of Germinal center T follicular helper (GCTfh) cells during HIV-1 infection. (A-D) (A) Representative flow cytometry plots, (B) pie chart and (C) summary plots comparing the proportions of GCTfh (CXCR5hiPD 1hi), and (D) nonGCTfh (PD-1+CXCR5+) cells in HIV negative (HIVneg), Fiebig I/II treated (Tx), Fiebig III-V Tx, Late Tx and untreated HIV-infected (unTx) groups. (E) Representative images showing lymph node sections stained with BCL-6 (green) to define germinal centers and CD4 (red), or PD-1 (red) to localize GCTfh cells and aggregate results for the area density of GCTfh cells computed using TissueQuest (TissueGnostics, Vienna) software. (F) Representative flow cytometry plot showing gating for Tfh subsets and (G) aggregate data across the groups. (H) Correlation analysis of Gag p24 density measured using image cytometry with GCTfh1 and (I) GCTfh17 subsets distribution measured using flow cytometry. (J) Gating strategy for identifying double-positive tetramer specific (Tet++) CD4+ T cells and overlay plots of Tet++ CD4+ T cells (red dots) on Tfh subsets with (K) aggregate data of Tet++ Tfh subsets across the groups. Mann U-Whitney P and Spearman rho (r) and P values are reported.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "CXCR3+ Tfh cells contribute to HIV persistence in treated hyperacute HIV-1 infection. (A) The co-localization of Gag p24 antigens with cells expressing (i) PD1+, (ii) CD4+, (iii) CXCR3+ and (iv) CCR6+ surface markers and (v) follicular dendritic cells (FDC) are assessed by immunofluorescence microscopy. Representative IF images characterizing HIV Gag p24+ cells in the germinal centers (BCL-6+). Green, red, and orange signals are from Opal fluorophores 520, 570 and 690 (PerkinElmer) and nuclei are counterstained with DAPI (blue)]. (B) Representative flow cytometry plots showing gating for FACs sorted Tfh subsets. (C) HIV mRNA quantified in FACs sorted Tfh subsets using digital droplet PCR. Absolute numbers of quantified HIV transcripts are equated to absolute cell numbers determined using the expression of 2M. Amounts of HIV mRNA within CXCR3+ and CXCR3- subsets are also compared. (D) LNMCs surface stained with 3BNC117 monoclonal antibody, flow plots and aggregate data show proportion of HIV-infected (3BNC117) CD4+ T cells, (E) LNMCs intracellularly stained with KC57 (Gag p24), flow plots and aggregate data show proportion of HIV-infected (KC57+) CD4+ T cells. (F) Representative flow plot and aggregate data show proportion of 3BNC117+ CD4+ T cells that either co-express or do not express CXCR3. Statistical differences are calculated using Mann U-Whitney and Kruskal-Wallis tests.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "HIV specific CD8+ T cell responses limit the amount of persistent HIV antigens in lymph nodes during ART. (A-B) Intracellular cytokine staining was conducted after stimulating PBMCs and lymph node mononuclear cells (LNMCs) with HIV-Gag. (A) Representative flow cytometry plots and aggregate data of 15 donors showing IFN-\u03b3+ CD8+ T cells, and (B) IFN-\u03b3+ CD4+ T cells after stimulation with HIV-1 clade C Gag peptide pools. (C-D) Correlation analysis of average Gag p24 density; measured from image analysis, with the frequency of (C) IFN-\u03b3+CD8+ T cells and (D) IFN-\u03b3+CD4+ T cells in LNMCs and PBMCs. (E) Representative flow cytometry plots of CFSE-labelled CD4+ and CD8+ T cells after 7-days of stimulation of LNMCs with HIV-1 clade C Gag peptide pools. (F) Aggregate data correlating IFN-\u03b3+CD8+ T cell responses and Gag p24 density. (G) Aggregate data correlating IFN-\u03b3+CD4+ T cell responses and Gag p24 density. (H) Representative flow plot and aggregate data showing frequency of HIV Gag-specific CXCR5+ CD8+ T cells. (I) SEB-specific (IFN \u03b3+) CD8+ T cells expressing CXCR5. Mann U-Whitney P and Spearman rho (r) and P values are reported.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

00e60208cf2b9b6bda3f4e587570f84a7a278bfd5878e116520a3364b52f4292/preprint/preprint.md

@@ -0,0 +1,258 @@
+# Abstract
+
+HIV persistence in tissue sites despite ART is a major barrier to HIV cure. Detailed studies of HIV infected cells and immune responses in native lymph node (LN) tissue environment is critical for gaining insight into immune mechanisms impacting HIV persistence and clearance in tissue sanctuary sites. We compared HIV persistence and HIV-specific T cell responses in LN biopsies obtained from 14 individuals who initiated therapy in Fiebig stages I/II, 5 persons treated (Tx) in Fiebig stages III-V and 17 late Tx individuals who initiated ART in Fiebig VI and beyond. Using multicolor immunofluorescence staining and in situ hybridization, HIV RNA and/or protein was detected in 12 of 14 Fiebig I/II Tx persons who were on suppressive therapy for 1 to 55 months, while all late Tx persons had persistent antigens. CXCR3+T follicular helper T cells harbored the greatest amounts of gag mRNA transcripts. Notably, HIV-specific CD8+ T cells responses associated with lower HIV antigen burden in LNs, suggesting that these responses may contribute to HIV suppression in LNs during therapy. These results reveal HIV persistence despite the initiation of ART in hyperacute infection and highlight the contribution of virus-specific responses to HIV suppression in tissue sanctuaries during suppressive ART.
+
+**Immunology**  **Infectious Diseases**  **HIV persistence**  **T cell responses**  **HIV-1 infection**
+
+# Introduction
+
+Antiretroviral therapy (ART) does not eradicate HIV infection due in large part to early establishment and persistence of integrated proviruses in quiescent circulating and tissue reservoirs<sup>1-3</sup>, which are resistant to drug or immune-mediated clearance. Additional mechanisms of persistence that involve ongoing replication have been suggested<sup>4</sup>, including replication in germinal centers (GCs) within secondary lymphoid tissues, which have been described as sanctuary sites due to low CD8<sup>+</sup> T cell infiltration<sup>5</sup> and suboptimal penetration of antiretroviral drugs<sup>6</sup>. Ongoing virus replication, and to a lesser extent viral gene expression in the face of ART are contested concepts, with studies of individuals initiated on ART during chronic infection yielding conflicting results<sup>6-12</sup>. T follicular helper (Tfh) cells have been identified as a major source of persistent virus<sup>13</sup>, but the precise subset of these cells enriched for HIV transcription on therapy within sanctuary sites is unknown. HIV persistence on ART is underscored by the nearly inevitable rebound of plasma viremia when treatment is interrupted, even after years of suppression, and is the major barrier to HIV cure<sup>14</sup>.
+
+It has previously been shown that very early initiation of ART can lead to prolonged remission when treatment is interrupted<sup>15,16</sup>, but this is an infrequent occurrence. In most individuals, virus rebound occurs within weeks to months even in individuals initiated on ART during Fiebig stage I (hyperacute) HIV infection<sup>14,17</sup>, despite rapid suppression of viremia and dramatically lower numbers of latently infected cells in peripheral blood<sup>18-21</sup>. Intriguingly, virus rebound kinetics following treatment interruption are heterogenous, sometimes taking a year or more<sup>14,22</sup>. The underlying immunological and virologic mechanisms responsible for the diverse viral rebound kinetics remain unknown. For instance, prolonged therapy in SIV infected macaques initiating therapy within 6 days of infection, prior to detectable plasma viremia, led to apparent elimination of infection after 600 days of suppressive ART in some animals<sup>23</sup>. However, such clearance of infection has not been observed in acute HIV infection, even in persons in whom therapy was initiated before detection of plasma viremia<sup>20</sup>. We previously showed that early treatment initiation enhances T cell functions in peripheral blood and limits viral diversity<sup>24</sup>, but it is not clear if functional responses occur in tissues and whether such responses play a significant role in HIV suppression during ART. Limited access to tissue samples from persons initiating therapy before peak viremia has impeded a better understanding of the impact of early therapy on the lymphoid reservoir.
+
+Here we analyzed 64 excisional lymph node (LN) biopsies and paired peripheral blood (PB) samples obtained from a well pedigreed cohort of individuals, where some individuals initiated ART during hyperacute HIV infection, to investigate the impact of blunting peak viremia on the microanatomical location, cellular source and role of T cell responses on HIV persistence in LNs. Study participants were drawn from a unique hyperacute HIV infection cohort termed FRESH (Females Rising through Education, Support and Health). FRESH is a prospective study of uninfected 18-23-year-old women at high risk of HIV infection established at the epicenter of the HIV epidemic in South Africa, where yearly incidence rates approach 10%. Despite vigorous prevention efforts, twice weekly monitoring for viral RNA has identified and treated (Tx) persons at the onset of plasma viremia, allowing for immediate institution of ART in many cases resulting in peak plasma viral loads that are sometimes <1,000 RNA copies/ml and the preservation of CD4<sup>+</sup> T cell numbers<sup>18</sup>. Our results show that despite ART-induced blunting of peak viremia<sup>18</sup> and augmentation of functional HIV-specific T cell responses<sup>24</sup>, HIV Gag p24 protein and viral RNA can persist in the LNs of Fiebig I/II Tx donors even after 4.5 years of fully suppressive ART, and these viral antigens are enriched in LN CXCR3<sup>+</sup> Tfh cells. We also show that superior functioning T cell responses were associated with lower HIV antigen persistence in the LNs.
+
+# Results
+
+Hyperacute HIV infection as a model to interrogate antigen persistence in lymph nodes  
+To determine the impact of immediate initiation of ART in hyperacute HIV infection (before peak viremia) on HIV clearance from sanctuary sites, we studied 14 women aged 18–26 who initiated ART during hyperacute HIV infection (Fiebig I/II Tx) and achieved full suppression of plasma viremia within a median of 15 days (range, 6 to 33). LNs were obtained by excisional biopsy after treatment for a median of 370 days [range, 19 to 1647]. All remained fully suppressed except for one donor who had a transient viral load blip prior to LN excision. Five additional individuals identified in Fiebig stages III–V of infection and started on ART one day after diagnosis were also included. Three additional control groups were included: 13 HIV negative (HIVneg) donors; 17 individuals who initiated treatment in Fiebig VI and beyond (late Tx); and 15 untreated individuals whose duration on infection is unknown (unTx). Detailed characteristics of the cohorts are in Tables 1 and S1. 95% of the study participants were females.
+
+Long-term persistence of HIV Gag p24 antigen in germinal centers (GCs) of individuals initiating antiretroviral therapy during hyperacute HIV-infection  
+To investigate HIV persistence in LNs of individuals initiating ART in Fiebig stages I/II, we measured HIV Gag p24 antigen in excisional LN biopsies by multicolor immunofluorescence staining of formalin-fixed paraffin-embedded LNs and imaging of tissue sections (Fig. S1). The transcription factor BCL‑6 was used to identify active GCs (Fig. S1) and images were quantified for Gag p24 content using the algorithm for area measurements in TissueQuest (TissueGnostics). Fig. 1A shows a representative image of HIV Gag p24 LN staining for a participant who was diagnosed in Fiebig stage I, initiated ART within 48 hours and achieved persistent plasma viremia suppression within 33 days. The LN sample shown was obtained after 479 days of uninterrupted ART treatment with undetectable viremia, and depicts HIV Gag p24 antigen within a GC, which was present in 5 of the 7 GCs examined in this LN (Fig. S1E).  
+Gag p24 staining and imaging were conducted on a total of 14 LNs from fully suppressed Fiebig I/II Tx donors obtained at a median of 370 days (range, 19 to 1647 days) post-ART initiation. 12 of 14 (86%) of these Fiebig I/II Tx donors had detectable HIV Gag p24 in at least one GC, and overall 42 of 55 GCs evaluated (76%) were positive for Gag p24. In two Fiebig I/II Tx donors there was no detectable HIV Gag p24 despite examining more than 4 GCs (Fig. 1B). In those with detectable HIV Gag p24, quantitative image analysis revealed no correlation between the amount of HIV Gag p24 present in the LN tissue section and treatment duration prior to LN excision (Fig. 1C) or peak plasma viral load (data not shown). Notably, regardless of treatment duration, Fiebig I/II Tx donors had significantly less detectable HIV Gag p24 compared to Fiebig III–V Tx (P < 0.0001), late Tx (P < 0.0001) and unTx (P < 0.0001) donors, though there was considerable overlap (Figs. 1D–E, S1F–G). This result was also consistent in a subset of donors Tx beyond 1 year (Fig. S1H). Quantitative image analysis of all treated LNs revealed greater area percent of Gag p24 staining in GCs compared to extrafollicular areas of the tissue (P = 0.04, Fig. 1F). Together, these data demonstrate that early ART initiation in Fiebig stage I/II limits the magnitude of HIV Gag p24 antigen in LNs, but that Gag p24 can persist predominantly in follicular areas even after 4.5 years of fully suppressive treatment.
+
+Lymph nodes of Fiebig I/II treated individuals harbor HIV-1 RNA  
+To determine if viral RNA transcription was occurring, which is required to produce infectious virions, we used an in-situ hybridization (ISH) assay called RNAscope to probe for HIV-1 gag-pol RNA within LN sections. 12 Fiebig I/II Tx, 4 Fiebig III–V Tx, 2 late Tx, 4 unTx and 3 HIVneg LN samples were analyzed based on sample availability. Viral RNA was detected as punctate dots in LNs from all HIV infected persons and there were no signals in the HIVneg controls (Figs. 2A, 2B and S2). Productively infected viral RNA+ cells were identified as a dense spherical signal, whereas follicular dendritic cell (FDC)-bound virus particles were defined by a diffuse lattice-like pattern consistent with previous reports. Combined RNAscope ISH gag-pol staining with IF staining for CD4+ T cells confirmed viral RNA (green) within CD4+ T cells (red, Fig. 2C). RNAscope staining was quantified using Fiji. 10 of 12 Fiebig I/II Tx donors had detectable but significantly lower amounts of HIV RNA compared to late Tx (P = 0.04) and unTx (P = 0.001) donors (Fig. 2D). However, there was no difference in RNA density between Fiebig I/II Tx and Fiebig III–V Tx donors. Notably, there was a positive correlation between Gag p24 density measured by IF and gag-pol RNA measured by in situ hybridization in Fiebig I/II Tx donors (P = 0.008; r = 0.8, Fig. 2E). The results are consistent with the persistence of viral RNA despite very early ART initiation in hyperacute infection and durable plasma virus suppression.
+
+Discordant HIV-1 RNA loads in plasma and lymph nodes  
+To better define active virus transcription within LN mononuclear (LNMCs) cells and to determine the viral loads in the LNs of aviremic individuals initiated on treatment either very early or later in infection. We measured cell-associated viral loads in LNMCs using a commercial viral load assay Cobas® Ampliprep HIV-1 test. We found a hierarchy of LNMC viral loads with the values lowest in patients that initiated therapy in Fiebig stages I/II (Fig. 2F). Interestingly, neither the peak plasma viral load (Fig. 2G), treatment duration before LN excision (Fig. 2H), nor the time to suppression (Fig. 2I) impacted viral RNA persistence in the LN. Overall, quantifiable amounts HIV RNA persists in the LNs of most Fiebig I/II Tx individuals and the magnitude of LN viral loads was not dependent on the duration of treatment.
+
+Expansion of GCTfh cells in early ART-treated individuals  
+Identifying the cellular phenotypes of persistent HIV-1 protein and transcripts during therapy will be critical for future anti-HIV interventions. While follicular T helper (Tfh) cells are a key component of the adaptive immune response to HIV-1 infection and provide cognate help to B cells and CD8+ T cells, these cells also serve as a major HIV reservoir. Moreover, HIV antigen can be trapped in the follicular dendritic lymphoreticular network within LNs and persist for years, thus we interrogated persistence within these cell subsets.  
+Given that Tfh are major targets of HIV infection, we first sought to determine the extent to which early ART mitigates HIV induced Tfh expansion. We defined GCTfh as CD4+ CD45RA− CXCR5hi PD-1hi and nonGCTfh cells as CD4+ CD45RA− CXCR5+ PD‑1+ in LNMCs (Fig. 3A) consistent with previous Tfh studies. HIV negatives (n=9) had very low frequencies of GCTfh cells (median 1.3%, IQR; 0.6% to 1.5%) of antigen experienced (CD45RA−) CD4+ T cells whereas nonGCTfh cells were 11% (IQR; 8.5% to 13%, Fig. 3B). HIV infection resulted in significant expansion of GCTfh (Fig. 3C). Timing of treatment initiation impacted the extent of GCTfh expansion. Immediate therapy was associated with significant diminution of GCTfh expansion (Fiebig I/II Tx vs healthy controls P = 0.07), whereas a slight delay in treatment initiation (Fiebig III or later) was associated with significant GCTfh expansion comparable to untreated HIV infection (Fiebig III–V Tx vs healthy controls P = 0.03) (Fig. 3C). Notably, HIV induced Tfh expansion was restricted to GCTfh, as no significant expansion of nonGCTfh cells were observed between the groups (Fig. 3D). To verify these observations, we quantified the area densities of GCTfh in situ using FFPE LNs. Consistent with flow cytometry data GCTfh cell densities were significantly greater in delayed therapy and untreated infection compared to Fiebig I/II Tx and HIV negative controls (Fig. 3E). Together, these data show that early treatment initiated in Fiebig I/II mitigates HIV-induced GCTfh expansion. Reduced HIV targets in GCs might partly explain reduced HIV persistence in LN of individuals who initiate therapy very early.  
+To gain more insight on cellular targets of HIV infection in LNs, next, we investigated if there was a particular subcellular GCTfh that was selectively expanded. We quantified previously described GCTfh subsets namely; GCTfh1 defined as CXCR3+ CCR6−, GCTfh2 as CXCR3− CCR6−, GCTfh1-17 as CXCR3+ CCR6+, and GCTfh17 defined as CXCR3− CCR6+ (Fig. 3F) among our study groups (Fig. 3G) and determined their relationship with HIV Gag p24 densities. While subsets had varying frequencies (Fig. 3G), within the early Tx donors, higher frequency of GCTfh1 was associated with greater degrees of Gag p24 positivity (Fig. 3H), whereas GCTfh17 displayed a reverse trend (Fig. 3I). Overall, these results show that while early treatment mitigates GCTfh responses, subset distribution of Tfh cells might impact virus persistence in early treated LNs.  
+Lastly, given the notion that HIV-specific CD4+ T cells might be more susceptible to infection and contribute to viral persistence, we used class II tetramers to characterize HIV-specific Tfh responses in LN tissues. DRB1*11:01 and DRB1*13:01 class II tetramers previously described were used to identify HIV-specific CD4+ T cells (Fig. 3J). Summary data from the analysis of 3 Fiebig I/II Tx, 2 late Tx and 4 unTx donors within our cohort expressing the class II DRB1*11:01 and DRB1*13:01 alleles, revealed that tetramer-specific GCTfh and nonGCTfh cells were detected in our chronic unTx and Fiebig I/II Tx participants at similar frequencies (Fig. 3K). Phenotypic analysis revealed both GC and nonGC HIV-specific Tfh. These results demonstrate that HIV-specific Tfh responses are induced during early treated HIV infection. However, we did not have sufficient quantities of tetramer+ cells to determine if antigen specific CD4+ T cells are preferentially infected.
+
+CXCR3+ Tfh cells harbour a greater burden of persistent HIV RNA in lymph nodes obtained from treated individuals with sustained plasma viral suppression.  
+Since image analysis for Gag p24 indicated that most of the HIV antigen was confined within discrete regions of GCs, and flow data showed differential correlation between levels of Gag p24 and GCTfh subsets, we next stained for other markers shown to be highly expressed on human Tfh. We also used FDC markers to identify residual Gag p24 that has been reported to persist on FDCs. Immunofluorescence imaging of serial sections stained with different combinations of antibodies and detected with Opal fluorophores revealed that Gag p24 co-localized with several phenotypic markers (Figs. 4A and S3), including PD1 (Fig. 4A, panel i), CD4 (panel ii), CXCR3 (panel iii–iv), CCR6 (panel iv) and FDC (panel v). To more definitively identify the Tfh subset that harbored the most HIV infection burden, we quantified HIV RNA in LNMCs isolated from LN tissue of 3 Fiebig I/II Tx, 2 Fiebig III–V Tx and 3 late Tx donors and FACS-sorted into the 4 different Tfh subsets (Fig. 4B). HIV mRNA was detectable using digital droplet PCR in all the subsets (Fig. 4C). Importantly, when we analyzed the cells based on expression of chemokine receptors, CXCR3 and CCR6, we found that CXCR3+ Tfh subsets harbored significantly greater amounts of HIV RNA than other subsets (P = 0.005, Fig. 4C).  
+To further interrogate preferential infection of CXCR3+ Tfh cells, we used a broadly neutralizing antibody (bNAb) called 3BNC117 to stain HIV infected cells while simultaneously staining for CXCR3. 3BNC117 targets the CD4 binding site on the surface of HIV-1 Envelope (Env) glycoprotein. LNMC and paired PBMC samples obtained from 7 Fiebig I/II treated donors were analyzed without prior manipulation. Representative flow plots for one donor and aggregate data for 7 donors showed detection of HIV‑1 Env (3BNC117) positive LNMCs at significantly greater frequency compared to paired PBMC samples (P = 0.03, Fig. 4D). To confirm detection of low frequency HIV-1 positive cells ex vivo, we intracellularly stained aliquots of the same samples with anti-Gag p24 antibody. Similarly, Gag p24+ CD4+ T cells were readily detectable in LNMCs compared to PBMCs (P = 0.01, Fig. 4E). We phenotyped infected cells by dual staining of 3BNC117 and CXCR3 and observed a trend towards more Env+ CD4+ T cells co-expressing CXCR3 (P = 0.06, Fig. 4F) than those not expressing CXCR3. Together, these data suggest that CD4+ CXCR3+ expressing Tfh cells may be preferentially infected in vivo compared to other subsets.
+
+Impact of HIV-specific CD4+ and CD8+ T cell responses on HIV persistence in the lymph node during ART  
+We previously showed that immediate ART initiation augments HIV-specific T cell function in peripheral blood. To investigate the effects of early ART on LN responses, we begun by investigating if there were compartmental differences in the frequency of HIV-specific responses between LN and PB. We used intracellular cytokine staining (ICS) to measure the proportions of HIV-specific CD4+ and CD8+ T cells in LNs and paired blood samples using 15 fully suppressed Fiebig I/II Tx donors on uninterrupted therapy for greater than a year. Representative flow plots for one donor and aggregate data show significantly higher frequency of Gag-specific CD8+ T cells (P = 0.05) compared to PB responses (Fig. 5A). HIV-specific CD4+ T cell frequencies also trended towards greater frequencies in LN relative to PB (P = 0.06; Fig. 5B). Next, we investigated whether HIV-specific CD8+ T cell responses limit HIV persistence in the LN, and found a negative correlation between the frequency of HIV-specific CD8+ T cell responses and HIV Gag p24 density (r = –0.7, P = 0.02; Fig. 5C). No correlation was observed between peripheral responses and the amount of persistent Gagp24 antigen in the LN, indicating the peripheral responses do not accurately depict HIV persistence in LNs. Notably, there was no correlation between LN or peripheral CD4+ T cell responses and persistent HIV Gag p24 in the LN (Fig. 5D).  
+Considering that proliferative CD8+ T cell responses are often associated with protection, we next measured virus-specific responses by carboxyfluorescein succinimidyl ester (CFSE) dilution. Representative flow plots for a donor with low and high Gagp24 density are shown (Fig. 5E). Aggregate data show proliferative Gag-specific CD8+ T cell responses negatively correlated with HIV Gag p24 burden (r = -0.7, P = 0.04; Fig. 5F). A similar result was obtained for Gag-specific CD4+ T cell responses (r = -0.7, P = 0.04: Fig. 5G). Together, these data show an association between maintenance of functional cellular responses and reduced HIV viral antigens in LNs.  
+Given most of the residual virus was concentrated within GCs, we next assessed the capacity of HIV-specific T cell responses to traffic into the GCs by enumerating the frequencies of CXCR5+ HIV-specific responses in LN, which denote capacity to migrate into GCs. CXCR5+ HIV-specific CD8+ T cells were significantly lower compared to CXCR5+ CD8– T cells (P = 0.001), suggesting reduced capacity to migrate into GCs, which may partly explain the observed greater HIV antigen burden in GCs relative to extrafollicular areas (Fig 5H). Moreover, SEB stimulation showed significantly lower proportion of IFN-γ+ CXCR5+ CD8+ T cells relative to CXCR5− CD8+ T cells (Fig 5I), indicating an inherent deficiency of CXCR5+ CD8+ T cells to secrete cytokines. Additionally, there was no correlation between plasma CXCL-13, (part of the CXCR5-CXCL-19 axis crucial for recruitment of immune cells into GCs) and density of HIV antigens in LNs (Fig. S4). Together, these data show that reduced functional HIV-specific CD8+ T cell responses within GCs might contribute to HIV persistence in this tissue microenvironment.
+
+# Discussion
+
+Comprehensive understanding of ART-mediated HIV suppression in tissue sanctuary sites is critical to the design, optimization, and evaluation of curative strategies. Moreover, a therapeutic vaccine for HIV-1 infection would need to induce robust anti-HIV immune responses in ART suppressed individuals to mediate post-treatment viral control. Here, we used a very well characterized cohort of persons with hyperacute HIV infection to conduct a comprehensive analysis of HIV persistence in LNs following ART initiation in Fiebig stage I/II and to elucidate Tfh cell responses which are critical for robust B cell and CD8<sup>+</sup> T cell functions.
+
+Most donors exhibited persistent HIV antigens in LN despite prompt blunting of initial peak viremia and sustained plasma viral suppression for as long as 55 months, suggesting that early therapy initiation may not fully eradicate persistent virus in lymphoid tissue sites. Immediate therapy reduced GCTfh expansion which is typically associated with dysregulation of B cell responses due to excessive GC reactions in untreated HIV infection<sup>36,44</sup>. Moreover, mitigated GCTfh responses decreased the number of cellular targets of HIV infection. Importantly, the association between functional immune responses and reduced viral burden in LNs indicates that T cell responses contribute towards elimination of infected cells during therapy. Combined, these data highlight the need to prioritize elimination of active HIV persistence in LNs as a critical step to achieving a cure or prolonged HIV remission off therapy.
+
+Our unique ability to obtain excisional lymph node biopsies in the FRESH cohort allowed for characterization of sites of virus persistence within the LN architecture in persons in whom peak viremia is blunted. The topological analysis of persistent HIV antigens within intact LN tissues identified greater HIV protein antigen burden within B cell follicles. Notably, onset and duration on therapy did not significantly affect the amount of detectable Gag p24 protein, consistent with the notion of rapid HIV reservoir establishment followed by very slow decay rate<sup>45</sup>. Importantly, our data highlight LN GCs as major sites of HIV persistence, with the potential to be a major source of rebound viremia upon treatment interruption.
+
+Using a highly specific <em>in-situ</em> hybridization assay, RNAscope<sup>46,47</sup>, which, in addition to HIV antigen detection provided further evidence in support of persistent HIV transcription in LNs in the face of ART in 86% of very early treated donors. We identified densely spherical signals by RNAscope staining in some early treated individuals suggestive of productively infected viral RNA<sup>+</sup> cells, consistent with a previous report in which active HIV RNA transcription was detected in the LNs of patients who initiated therapy in the chronic phase of illness<sup>13</sup>. Moreover, our data reveal heterogeneity in the amount of persistent HIV transcripts in very early treated aviremic individuals despite similar levels of peak viremia and rapid plasma viral suppression kinetics following ART initiation. Viral antigen persistence and ongoing transcription could indicate that HIV continues to cause immune damage in anatomical sites despite full suppression in peripheral blood, but it may also suggest that even in early treated individuals, priming, stimulation and harnessing of HIV-specific immunity for curative strategies will not be insurmountable because functional HIV-specific immunity is preserved.
+
+Identifying cellular reservoirs of HIV in tissues has been a major area of research, (reviewed in<sup>48</sup>). These studies describe GCTfh subset compositions anatomically and phenotypically during HIV infection and their contributions to persistent virus. The observed positive correlation between the proportions of GCTfh1 cells (which are CXCR3<sup>+</sup> CCR6<sup>-</sup> GCTfh) and detection of greater amounts of HIV RNA relative to CXCR3<sup>+</sup> Tfh cells, indicates that CXCR3 may yet be another phenotypic marker of Tfh cells that have greater HIV transcription activity on ART. These data are consistent with a study that reported greater amounts of SIV DNA in CXCR3<sup>+</sup> GCTfh compared to CXCR3<sup>-</sup> GCTfh in macaques<sup>49</sup>. It is reasonable to attribute increased HIV burden in the CXCR3<sup>+</sup> Tfh subset to CXCR3 being used as an alternative co-receptor for HIV entry, which has previously been reported<sup>50</sup>. However, confirmatory work is needed. In any case, we have identified a marker for HIV infected cells during ART that could be targeted for elimination as part of an HIV eradication strategy. Whether or not the CXCR3<sup>+</sup> Tfh population identified in this study represents the same population as the PD-1<sup>+</sup> subset that was recently implicated in HIV persistence is intriguing and warrants further investigation<sup>13,51</sup>.
+
+Low CD8<sup>+</sup> T cell density in GCs is thought to be a major reason for persistently high HIV antigen burden in this anatomical niche<sup>52,53</sup>. Interestingly, we detected significantly greater proliferative responses in individuals with little to no detectable HIV antigens compared to those with greater LN HIV antigen burden. These findings are consistent with our recent study showing that very early ART is associated with functionally superior cellular responses<sup>24</sup>. Together, these data suggest that HIV-specific T cell responses contribute to HIV suppression in LN during therapy<sup>54</sup>. However, longitudinal studies using serial biopsies from the same donor are needed to confirm these findings.
+
+A notable limitation of the study is that we could only obtain one LN sample per study participant, thus we could not conduct intra-individual longitudinal HIV decay kinetics, or longitudinally define the immune responses in tissues associated with control. This limitation was partly overcome by sampling LNs from many participants over a very wide time range. Serial LN studies have been reported in the absence of complications<sup>55</sup>. Alternatively, future studies could attempt serial fine needle aspirates, though cell yields may be limiting for some of the studies described here and the architectural structure would be disrupted. An additional limitation is the lack of gender diversity. Although we studied only women, this is the group at disproportionate risk of infection and women are underrepresented in most studies to date<sup>56</sup>. These issues can and should be addressed in future studies focused by our findings.
+
+In conclusion, our results demonstrate HIV persistence in LNs despite prompt and durable ART-mediated plasma viral suppression. HIV structural proteins and HIV RNA persist in LNs of in 12 of 14 individuals, albeit at lower levels compared to treatment in chronic infection. Given sample limitations, one cannot rule out persistent infection in the other two individuals. Among those with detectable infection, GCs serve as the primary anatomical site of HIV persistence and the major cellular source is CXCR3 expressing Tfh cells. Together, our results emphasize the importance of very early initiation of ART in Fiebig stage I/II to reduce the amounts of persistent virus in the LN and highlight the need for interventions to completely eradicate residual viremia in immune privileged and anatomically compartmentalized tissue sites.
+
+# Methods
+
+## Study population, samples, and performance site
+
+Study participants were drawn from the HIV Pathogenesis Programme (HPP) lymph node study (LNS) cohort, Durban, South Africa. Recruitment into the HPP LNS cohort were from the FRESH cohort described in <sup>16</sup> (n=41) and another primary HIV infection cohort in Durban, South Africa where participant’s time of infection is less defined (n=35). Axillary, cervical or inguinal LNs were surgically excised at Prince Mshiyeni Hospital in Umlazi, and 120 ml paired peripheral blood was also obtained from each participant. Viral load measurements were performed by HIV‑1 RNA testing using the NucliSens EasyQ v2.0 assay (BioMérieux Clinical Diagnostics, Marcy-l’Étoile, France), through a certified commercial laboratory. CD4<sup>+</sup> T cell counts were enumerated by Tru-Count technology and analyzed on a FACSCalibur flow cytometer [Becton Dickinson (BD) New Jersey, USA]. Sample processing and laboratory studies were performed at the Africa Health research Institute in Durban, South Africa.
+
+## Study approval
+
+All study participants provided written informed consent prior to inclusion in the study. Ethical approval for the study was granted by the University of KwaZulu-Natal Biomedical Research Ethics Committee (protocol number BF298/14) and the Institutional Review Board of Massachusetts General Hospital (protocol number 2015-P001018).
+
+## Lymph node and blood sample processing
+
+Excised LNs were sectioned into two, one section was fixed in 10% formal-saline (Sigma-Aldrich, St. Louis, Missouri, USA) for immunofluorescence microscopy (IF) studies, while the second section was macerated to release LN mononuclear cells (LNMCs) according to the method of Schacker, et al. <sup>57</sup>. The cells were passed through a mesh screen and harvested by centrifugation [625 x <em>g</em>, 6 min, room temperature (RT)].
+
+Peripheral blood mononuclear cells (PBMCs) were isolated from patient’s blood samples by density-gradient centrifugation using Histopaque-1077 (Sigma-Aldrich) and cryopreserved in liquid nitrogen <sup>58</sup>.
+
+## Viral RNA quantification in lymph node mononuclear cells (LNMCs)
+
+Cryopreserved LNMCs (10 million cells) were lysed, and viral RNA was quantified using the Cobas<sup>®</sup> AmpliPrep HIV-1 test (Roche, Mannheim, Germany) at an accredited clinical laboratory using standardized protocols.
+
+## Immunofluorescence (IF) microscopy
+
+IF microscopy staining was performed on 4 µM sections of formalin fixed paraffin embedded (FFPE) LNs using the Opal 4‑color fluorescent IHC kit (PerkinElmer, Waltham, MA, USA). Briefly, following antigen retrieval, two blocking steps (2 x 10 min, RT) were performed using the Dako peroxidase-blocking reagent (Agilent Technologies, Glostrup, Denmark) and Bloxall block (Vector Laboratories, Burlingame, CA, USA). The slides were washed with 0.05% Tween 20 in Tris-buffered saline (TBS‑T) for 5 min, sequentially probed with the primary antibody (30 min, RT), and Opal polymer HRP [20 min, RT (PerkinElmer)] and detected using the Opal polymer 520 (10 min, RT). This protocol was repeated for the second and third antibodies with Opal polymers 570 and 690 respectively, followed by counterstaining with spectral DAPI (PerkinElmer) to make a total of 4 different fluorochromes. Primary antibodies used in these combinations include anti-human BCL-6 [(clone PG-B6p) Dako/Agilent Technologies], CCR6 [(R6H1) Thermofisher Scientific, Waltham MA, USA], CD4 [(clone 4B12) Dako/Agilent Technologies], CXCR3 [(clone 6H1L8) Thermo Fisher Scientific], follicular dendritic cells [(cloneCNA.42), Dako/Agilent Technologies] p24 [(clone Kal-1), Dako/Agilent Technologies], and PD-1 [clone NAT105) Abcam, Cambridge, MA, USA]. Slides were mounted with Dako fluorescence mounting medium (Agilent Technologies) and imaged with the Axio Observer, 20x objective lenses, a Hamamatsu C13440-20C camera and TissueFAXS imaging software (TissueGnostics, Vienna, Austria).
+
+## RNAscope<sup>â</sup><strong><em>in situ</em></strong> hybridization (ISH)
+
+RNAscope<sup>â</sup> ISH was conducted using the RNAscope<sup>®</sup> 2.5 HD assay kit [Advanced Cell Diagnostics (ACD), Newark, CA, USA, Cat No: 322300] and the RNAscope<sup>®</sup> multiplex fluorescent kit v2.0 (ACD, Cat No: 323100) as per manufacturer’s instructions. Briefly, pre-treated samples were hybridized with the clade C HIV-1<em>gag-pol</em> probe (Cat No: 317691) at 40<sup>o</sup>C for 16 hours. Next, the samples were incubated with signal amplification probes and horseradish peroxidase conjugated secondary antibodies. The signal was detected with either diaminobenzidine for the RNAscope<sup>®</sup> 2.5 HD assay (ACD) or with Opal fluorophores (PerkinElmer) for the multiplex fluorescent assay. Slides were imaged with Axio Observer and TissueFAXS imaging software (TissueGnostics).
+
+## Quantitative Image analysis
+
+Quantitative image analysis of Gag p24 in IF images of whole tissue section scans was conducted with TissueQuest software (TissueGnostics). Two independent experiments of total area measurements and nuclear segmentation analyses were performed on each whole tissue scan. The numerical data generated from the analyses are displayed in scattergrams. Greyscale images were analysed and each channel was processed separately by the software using DAPI as a master marker. In cases where images were stained with another nuclear marker such as BCL-6, then the FITC channel was used as a virtual channel for nuclei identification. Negative control slides were used to set the threshold values in the scattergrams and to distinguish specific staining signals from non-specific or background fluorescence signals. Although HIV Gag p24 staining was generally intense, there was no notable spillover of the signal to other channels (Fig. S1A-B). Also, p24 co-staining was only observed with FDCs and CD4 markers but not CD8 cells (Fig. S1C-D).
+
+Analysis of<em>gag-pol</em> RNA signals was done using Fiji, an open-source software based on ImageJ which is optimized for biological image analysis <sup>29</sup>. Briefly, images were segmented using the color segmentation plugin with the algorithm for Hidden Markov Model. Thresholding was applied to the segmented image and the total area of brown RNA signals was measured and recorded. Five images were analysed per sample and averaged. Pixel measurements were converted to mm using the scale bar.
+
+## Flow cytometry analysis
+
+Freshly isolated or frozen LNMCs and PBMCs were characterized using flow cytometry analysis with standardized protocols <sup>59</sup>. Cells were stained with LIVE/DEAD Fixable Blue dead cell stain kit (Thermo Fisher Scientific), CD3 Brilliant Violet (BV) 711 (BioLegend, San Diego, CA, USA), CD8 BV786 (BD Biosciences, San Jose, CA), CD4 BV650 (BD Biosciences), CXCR5 Alexa Fluor (AF) 488 (BD Biosciences), PD-1 BV421 (BioLegend), CCR6 Phycoerythrin [(PE) BioLegend], CXCR3 BV605 (BioLegend) and CD45RA PE-Cyanine (Cy)-7 (BioLegend), for 30 mins at RT.
+
+For intracellular cytokine staining, PBMCs or LNMCs were either left unstimulated or stimulated with HIV clade C overlapping peptide (OLP) pools spanning Gag, Nef, or Env proteins or Staphylococcal enterotoxin B (SEB, 0.5 μg/ml) in the presence of GolgiStop and GolgiPlug protein transport inhibitors (BD Biosciences) for 16 hours at 37°C, prior to surface staining with the panel of antibodies comprising LIVE/DEAD fixable Aqua dead cell stain (Thermo Fisher Scientific), CD3 BV711, CD4 BV650 and CD8 BV786. After fixation and permeabilization with the BD Cytofix/Cytoperm kit (BD Biosciences), cells were again stained using TNF-α A700 (BD Biosciences) and IFN‑γ PE-Cy7 (BioLegend) antibodies.
+
+T cell proliferation was measured by labelling LNMCs with carboxyfluorescein succinimidyl ester (CFSE), stimulating cells with HIV clade C OLP pools for 7 days and staining with CD3 BV711, CD4 BV650 and CD8 BV786. Stained cells were acquired using an LSRFortessa (BD Biosciences) with FACSDiva<sup>™</sup> software or sorted using the FACS Aria Fusion (BD Biosciences). Data were analysed using the FlowJo version 10.0.8 (Flowjo, LLC, Ashland, Oregon).
+
+HIV infected cells were identified by surface staining with biotinylated 3BNC117 antibody followed by streptavidin PE (Thermofisher Scientific) and/or intracellular staining with HIV Gag p24 RD1 [(clone KC57), Beckman Coulter, Indianapolis, USA] after fixation and permeabilization with Cytofix/Cytoperm (BD Biosciences).
+
+## HLA Class II tetramer studies
+
+HIV-specific Tfh responses were defined using fluorochrome conjugated HLA class II tetramers. Briefly, cells were stained for 1 hour at 37<sup>o</sup>C with APC and PE conjugated HLA Class II tetramer complexes, washed in 2% FCS-PBS and then stained with these antibodies: LIVE/DEAD Fixable Blue dead cell stain kit (Thermofisher Scientific), CD3 BV711 (Biolegend), CD4 BV650 (BD Biosciences), CD8 BV786 (BD Biosciences), CXCR5 AF488 (BD Biosciences), CXCR3 BV605 (Biolegend), PD-1 BV421 (Biolegend) and CD45RA AF700 (Biolegend); for 20 mins at RT. Cells were washed and acquired on the LSRFortessa (BD Biosciences).
+
+## Digital droplet PCR
+
+Total RNA was extracted from FACS-sorted LNMC Tfh subsets using Qiagen RNeasy kit (Qiagen) after lysing cells with QIAzol lysis reagent (Qiagen, Hilden, Germany) according to manufacturer’s instructions, and used for cDNA synthesis using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA). The cDNA was used as a template for HIV<em>gag</em> mRNA quantification by TaqMan digital droplet PCR assay using custom probes (Assay ID: APCE4R6, Thermo Fisher Scientific) in a two-step digital droplet PCR reaction. PCR thermal cycling was conducted following optimized cycling conditions: an initial denaturation at 95°C for 10 min, 40 cycles of 30 seconds at 94°C, 1 min at 60°C, followed by a final incubation at 98°C for 10 min and holding at 4°C until reading time. After PCR amplification, droplets were measured in the QX200 ddPCR Droplet Reader (Bio-Rad), and target gene copy number was analyzed using QuantaSoft analysis software (Bio-Rad) and recorded as mRNA copies/20μL. Absolute<em>gag</em> mRNA counts were normalized to the expression of the housekeeping gene b2M.
+
+## Statistical analyses
+
+All statistical analyses were conducted with GraphPad Prism 7.0 (GraphPad Software, La Jolla, California, USA) and <em>P</em> values were considered significant if less than 0.05. Specifically, the Mann-Whitney U and Kruskal-Wallis H tests were used for group comparisons. Additional post hoc analyses were performed using the Dunn’s multiple comparisons test. Correlations between variables were defined by the Spearman’s rank correlation test.
+
+# References
+
+1. Murray, A.J., Kwon, K.J., Farber, D.L. & Siliciano, R.F. The Latent Reservoir for HIV-1: How Immunologic Memory and Clonal Expansion Contribute to HIV-1 Persistence. *J Immunol* **197**, 407-417 (2016).
+
+2. Puertas, M.C., et al. Lack of concordance between residual viremia and viral variants driving de novo infection of CD4(+) T cells on ART. *Retrovirology* **13**, 51 (2016).
+
+3. Chun, T.W. Tracking replication-competent HIV reservoirs in infected individuals. *Curr Opin HIV AIDS* **8**, 111-116 (2013).
+
+4. Martinez-Picado, J. & Deeks, S.G. Persistent HIV-1 replication during antiretroviral therapy. *Curr Opin HIV AIDS* **11**, 417-423 (2016).
+
+5. Connick, E., et al. CTL fail to accumulate at sites of HIV-1 replication in lymphoid tissue. *J Immunol* **178**, 6975-6983 (2007).
+
+6. Fletcher, C.V., et al. Persistent HIV-1 replication is associated with lower antiretroviral drug concentrations in lymphatic tissues. *Proceedings of the National Academy of Sciences* **111**, 2307-2312 (2014).
+
+7. Lorenzo-Redondo, R., et al. Persistent HIV-1 replication maintains the tissue reservoir during therapy. *Nature* **530**, 51-56 (2016).
+
+8. McManus, W.R., et al. HIV-1 in lymph nodes is maintained by cellular proliferation during antiretroviral therapy. *The Journal of Clinical Investigation* **129**, 4629-4642 (2019).
+
+9. Musick, A., et al. HIV Infected T Cells Can Proliferate in vivo Without Inducing Expression of the Integrated Provirus. *Front Microbiol* **10**, 2204 (2019).
+
+10. Rasmussen, T.A., et al. The effect of antiretroviral intensification with dolutegravir on residual virus replication in HIV-infected individuals: a randomised, placebo-controlled, double-blind trial. *Lancet HIV* **5**, e221-e230 (2018).
+
+11. Hatano, H., et al. Increase in 2-long terminal repeat circles and decrease in D-dimer after raltegravir intensification in patients with treated HIV infection: a randomized, placebo-controlled trial. *J Infect Dis* **208**, 1436-1442 (2013).
+
+12. Real, F., et al. Platelets from HIV-infected individuals on antiretroviral drug therapy with poor CD4(+) T cell recovery can harbor replication-competent HIV despite viral suppression. *Sci Transl Med* **12**, eaat6263 (2020).
+
+13. Banga, R., et al. PD-1(+) and follicular helper T cells are responsible for persistent HIV-1 transcription in treated aviremic individuals. *Nature medicine* **22**, 754-761 (2016).
+
+14. Colby, D.J., et al. Rapid HIV RNA rebound after antiretroviral treatment interruption in persons durably suppressed in Fiebig I acute HIV infection. *Nature medicine* **24**, 923-926 (2018).
+
+15. Saez-Cirion, A., et al. Post-treatment HIV-1 controllers with a long-term virological remission after the interruption of early initiated antiretroviral therapy ANRS VISCONTI Study. *PLoS Pathog* **9**, e1003211 (2013).
+
+16. Namazi, G., et al. The Control of HIV After Antiretroviral Medication Pause (CHAMP) Study: Posttreatment Controllers Identified From 14 Clinical Studies. *The Journal of Infectious Diseases* **218**, 1954-1963 (2018).
+
+17. Deeks, S.G., et al. International AIDS Society global scientific strategy: towards an HIV cure 2016. *Nature medicine* **22**, 839-850 (2016).
+
+18. Dong, K.L., et al. Detection and treatment of Fiebig stage I HIV-1 infection in young at-risk women in South Africa: a prospective cohort study. *Lancet HIV* **5**, e35-e44 (2018).
+
+19. Buzon, M.J., et al. Long-term antiretroviral treatment initiated at primary HIV-1 infection affects the size, composition, and decay kinetics of the reservoir of HIV-1-infected CD4 T cells. *J Virol* **88**, 10056-10065 (2014).
+
+20. Henrich, T.J., et al. HIV-1 persistence following extremely early initiation of antiretroviral therapy (ART) during acute HIV-1 infection: An observational study. *PLOS Medicine* **14**, e1002417 (2017).
+
+21. Hatano, H., et al. Prospective Antiretroviral Treatment of Asymptomatic, HIV-1 Infected Controllers. *PLoS pathogens* **9**, e1003691 (2013).
+
+22. Frange, P., et al. HIV-1 virological remission lasting more than 12 years after interruption of early antiretroviral therapy in a perinatally infected teenager enrolled in the French ANRS EPF-CO10 paediatric cohort: a case report. *Lancet HIV* **3**, e49-54 (2016).
+
+23. Okoye, A.A., et al. Early antiretroviral therapy limits SIV reservoir establishment to delay or prevent post-treatment viral rebound. *Nature medicine* **24**, 1430-1440 (2018).
+
+24. Ndhlovu, Z.M., et al. Augmentation of HIV-specific T cell function by immediate treatment of hyperacute HIV-1 infection. *Sci Transl Med* **11**, eaau0528 (2019).
+
+25. Goteri, G., et al. Comparison of germinal center markers CD10, BCL6 and human germinal center-associated lymphoma (HGAL) in follicular lymphomas. *Diagnostic Pathology* **6**, 97-97 (2011).
+
+26. Schmid, M., et al. An Emerging Approach for Parallel Quantification of Intracellular Protozoan Parasites and Host Cell Characterization Using TissueFAXS Cytometry. *PLoS One* **10**, e0139866 (2015).
+
+27. Wang, F., et al. RNAscope: a novel in situ RNA analysis platform for formalin-fixed, paraffin-embedded tissues. *J Mol Diagn* **14**, 22-29 (2012).
+
+28. Pantaleo, G., et al. HIV infection is active and progressive in lymphoid tissue during the clinically latent stage of disease. *Nature* **362**, 355-358 (1993).
+
+29. Schindelin, J., et al. Fiji: an open-source platform for biological-image analysis. *Nat Methods* **9**, 676-682 (2012).
+
+30. Crotty, S. Follicular helper CD4 T cells (TFH). *Annual review of immunology* **29**, 621-663 (2011).
+
+31. Crotty, S. T follicular helper cell differentiation, function, and roles in disease. *Immunity* **41**, 529-542 (2014).
+
+32. Chevalier, M.F., et al. HIV-1-specific interleukin-21+ CD4+ T cell responses contribute to durable viral control through the modulation of HIV-specific CD8+ T cell function. *J Virol* **85**, 733-741 (2011).
+
+33. Schultz, B.T., et al. Circulating HIV-Specific Interleukin-21(+)CD4(+) T Cells Represent Peripheral Tfh Cells with Antigen-Dependent Helper Functions. *Immunity* **44**, 167-178 (2016).
+
+34. Perreau, M., et al. Follicular helper T cells serve as the major CD4 T cell compartment for HIV-1 infection, replication, and production. *J Exp Med* **210**, 143-156 (2013).
+
+35. Popovic, M., et al. Persistence of HIV-1 structural proteins and glycoproteins in lymph nodes of patients under highly active antiretroviral therapy. *Proceedings of the National Academy of Sciences of the United States of America* **102**, 14807-14812 (2005).
+
+36. Lindqvist, M., et al. Expansion of HIV-specific T follicular helper cells in chronic HIV infection. *The Journal of clinical investigation* **122**, 3271-3280 (2012).
+
+37. Schmitt, N., Bentebibel, S.E. & Ueno, H. Phenotype and functions of memory Tfh cells in human blood. *Trends Immunol* **35**, 436-442 (2014).
+
+38. Baiyegunhi, O., et al. Frequencies of circulating Th1-biased T follicular helper cells in acute HIV-1 infection correlate with the development of HIV-specific antibody responses and lower set point viral load. *J Virology* **92**, e00659-00618 (2018).
+
+39. Hunt, P.W., et al. HIV-Specific CD4+ T Cells May Contribute to Viral Persistence in HIV Controllers. *Clinical Infectious Diseases* **52**, 681-687 (2011).
+
+40. Laher, F., et al. HIV Controllers Exhibit Enhanced Frequencies of Major Histocompatibility Complex Class II Tetramer+ Gag-Specific CD4+ T Cells in Chronic Clade C HIV-1 Infection. *J Virol* **91** (2017).
+
+41. Scheid, J.F., et al. HIV-1 antibody 3BNC117 suppresses viral rebound in humans during treatment interruption. *Nature* **535**, 556-560 (2016).
+
+42. Gaiha, G.D., et al. Structural topology defines protective CD8<sup>+</sup> T cell epitopes in the HIV proteome. *Science* **364**, 480-484 (2019).
+
+43. Ndhlovu, Z.M., et al. High-dimensional immunomonitoring models of HIV-1–specific CD8 T-cell responses accurately identify subjects achieving spontaneous viral control. *Blood* **121**, 801-811 (2013).
+
+44. Cubas, R.A., et al. Inadequate T follicular cell help impairs B cell immunity during HIV infection. *Nature Medicine* **19** (2013).
+
+45. Whitney, J.B., et al. Rapid seeding of the viral reservoir prior to SIV viraemia in rhesus monkeys. *Nature* **512**, 74-77 (2014).
+
+46. Deleage, C., et al. Defining HIV and SIV Reservoirs in Lymphoid Tissues. *Pathog Immun* **1**, 68-106 (2016).
+
+47. Barton, K.M. & Palmer, S.E. How to Define the Latent Reservoir: Tools of the Trade. *Current HIV/AIDS reports* **13**, 77-84 (2016).
+
+48. Wong, J.K. & Yukl, S.A. Tissue reservoirs of HIV. *Current opinion in HIV and AIDS* **11**, 362-370 (2016).
+
+49. Velu, V., et al. Induction of Th1-Biased T Follicular Helper (Tfh) Cells in Lymphoid Tissues during Chronic Simian Immunodeficiency Virus Infection Defines Functionally Distinct Germinal Center Tfh Cells. *J Immunol* **197**, 1832-1842 (2016).
+
+50. Hatse, S., et al. Modest human immunodeficiency virus coreceptor function of CXCR3 is strongly enhanced by mimicking the CXCR4 ligand binding pocket in the CXCR3 receptor. *J Virol* **81**, 3632-3639 (2007).
+
+51. Banga, R., et al. Blood CXCR3(+) CD4 T Cells Are Enriched in Inducible Replication Competent HIV in Aviremic Antiretroviral Therapy-Treated Individuals. *Front Immunol* **9**, 144 (2018).
+
+52. Fukazawa, Y., et al. B cell follicle sanctuary permits persistent productive simian immunodeficiency virus infection in elite controllers. *Nature medicine* **21**, 132-139 (2015).
+
+53. Li, S., et al. Low levels of SIV-specific CD8+ T cells in germinal centers characterizes acute SIV infection. *PLoS pathogens* **15**, e1007311-e1007311 (2019).
+
+54. Miles, B. & Connick, E. TFH in HIV Latency and as Sources of Replication-Competent Virus. *Trends in Microbiology* **24**, 338-344 (2016).
+
+55. Chintanaphol, M., et al. Brief Report: Safety and Tolerability of Inguinal Lymph Node Biopsy in Individuals With Acute HIV Infection in Thailand. *J Acquir Immune Defic Syndr* **79**, 244-248 (2018).
+
+56. Karim, S.S.A. & Baxter, C. HIV incidence rates in adolescent girls and young women in sub-Saharan Africa. *Lancet Glob Health* **7**, e1470-e1471 (2019).
+
+57. Schacker, T.W., et al. Lymphatic Tissue Fibrosis Is Associated with Reduced Numbers of Naı¨ve CD4 T Cells in Human Immunodeficiency Virus Type 1 Infection. *Clinical and Vaccine Immunology* **13**, 556-560 (2006).
+
+58. McCoy, J.P., Jr. Handling, storage, and preparation of human blood cells. *Curr Protoc Cytom* **Chapter 5**, Unit 5 1 (2001).
+
+59. Ndhlovu, Z.M., et al. Magnitude and Kinetics of CD8+ T Cell Activation during Hyperacute HIV Infection Impact Viral Set Point. *Immunity* **43**, 591-604 (2015).
+
+# Tables
+
+**Table 1.** Characteristics of study participants at the time of lymph node excision
+
+| Characteristic | Fiebig stage I/II treated | Fiebig stage III-V treated | Late treated<sup>A</sup> | Untreated | HIV negative |
+|---|---|---|---|---|---|
+| No. of participants (% female) | 14 (100%) | 5 (100%) | 17 (94%) | 15 (80%) | 13 (100%) |
+| Median (IQR)<sup>C</sup> age of participants (years) | 22<br>(20-24) | 22<br>(19-24) | 24<br>(22-27) | 24<br>(24-29) | 22<br>(21-23) |
+| Median (IQR) CD4 Count (no. of cells/ml) | 911<br>(717-1120) | 942<br>(696-1086) | 589<br>(454-792) | 595<br>(401-720) | 976<br>(792-1180) |
+| Median (IQR) plasma viral load copies/ml | &lt;20<br>(&lt;20-&lt;20) | &lt;20<br>(&lt;20-&lt;20) | &lt;20<br>(&lt;20-&lt;20) | 11000<br>(1900-22000) | NA<sup>B</sup> |
+| Median (IQR) days on treatment | 370<br>(31-550) | 120<br>(46-677) | 571<br>(100-762) | NA | NA |
+
+<sup>A</sup> Donors whose Fiebig stage of infection was either Fiebig VI or unknown are defined as late treated.  
+<sup>B</sup> NA, not applicable.  
+<sup>C</sup> IQR: Interquartile range.
+
+# Supplementary Files
+
+- [SupplementaryfiguresandtablesOct2021.pdf](https://assets-eu.researchsquare.com/files/rs-1061787/v1/d494250c6fe171b58907c0ce.pdf)  
+  Figures S1 to S4 and Table S1

02274220d86210f87fc3660fc1d45c0654f1e7fef07a0cc63208315e58a6d6bf/preprint/images_list.json

@@ -0,0 +1,82 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "PXRD patterns of (a) as prepared, calcined, reduced and used Fe/CBEA, (b) as prepared and used T-Fe/MIL-101, and (c) as prepared and used T-MIL-88B catalysts after 48 h reaction in presence of CH3I and 5 cycles of 21 h each in presence of CH3OH and LiI.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Proposed mechanism of thermal transformation of MIL-88B(Fe) structure. (a) Shows the metal cluster topology, (b) shows MOF supercell structure, (c) desorbed water molecules within the pores of MOF, (d) shows the detachment of organic linkers, and (e) is the thermally transformed MIL-88B(Fe).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpeg",
+        "caption": "TEM micrographs of various studied catalysts, (a) MIL-101, (b) Fe/MIL-101, (c) T-Fe/MIL-101, (d) MIL-88B, (e) T-MIL-88B, and (f) used T-MIL-88B after 48 h of aqueous phase CO2 hydrogenation reaction in the vicinity of CH3OH and LiI additives; and particle size distribution of (g) T-MIL-88B, and (h) used T-MIL-88B.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpeg",
+        "caption": "Narrow scan XPS spectra of (a) Iron 2p, (b) Chromium 2p, and (c) Carbon 1s for the studied catalysts.  ",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Thermogravimetric analysis (TGA) of Fe/MIL-101 and MIL-88B under Argon atmosphere.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.jpg",
+        "caption": "Activity of various Fe based catalysts during aqueous phase CO2 hydrogenation in the presence of CH3I additive at different pressure, (a) Fe/CBEA, (b) T-Fe/MIL-101, and (c) T-MIL-88B. Reaction conditions: T= 150 \u02daC, H2/CO2= 1, tR= 21 h, H2O= 40 mL and stirring speed= 200 RPM.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_7.png",
+        "caption": "Effect of reaction time on carboxylic acids yield and selectivity via aqueous phase CO2 hydrogenation over T-MIL-88B in the presence of various additives, (a) CH3I, and (b) CH3OH and LiI. Reaction conditions: T= 150 \u02daC, H2/CO2= 1, Ptotal= 70 bar at room temperature and stirring speed= 200 RPM.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_8.png",
+        "caption": "Recycling study of T-MIL-88B via aqueous phase CO2 hydrogenation in the presence of CH3OH and LiI additives. Reaction conditions: T= 150 \u02daC, H2/CO2= 1, tR= 21 h, Ptotal= 70 bar at room temperature and stirring speed= 200 RPM.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_9.png",
+        "caption": "Acetic acid production through HCOOH and CH3I reaction in water over T-MIL-88B in the presence of hydrogenation. Reaction conditions: T= 150 \u02daC, nHCOOH=5 mmol, nCH3I=10 mmol, VH2O=40 mL, PH2=35 bar at room temperature and stirring speed= 200 RPM.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_10.png",
+        "caption": "Possible reaction route for acetic acid production via aqueous phase CO2 hydrogenation in the vicinity of methanol and LiI additives over T-MIL-88B catalyst.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

02274220d86210f87fc3660fc1d45c0654f1e7fef07a0cc63208315e58a6d6bf/preprint/preprint.md

@@ -0,0 +1,218 @@
+# Abstract
+
+Sustainable production of acetic acid (AA) is a high priority due to its high global manufacturing capacity and numerous applications. Currently it is predominantly synthesized via carbonylation of methanol, in which both the reactants are fossil-derived. CO₂ transformation into AA is highly desirable to achieve net zero carbon emissions, but significant challenges remain to achieve this efficiently. Herein, we report a heterogeneous catalyst, thermally transformed MIL-88B with Fe⁰ and Fe₃O₄ dual active sites, for highly selective AA formation via methanol hydrocarboxylation. This efficient catalyst showed high AA yield (590.1 mmol/gₐₜ·L) with 81.7% selectivity at 150°C in aqueous phase using LiI as a co-catalyst. The reaction is believed to proceed via formic acid intermediate. No significant difference in AA yield and selectivity was noticed during catalyst recycling study up to five cycles. This work scalable and industrially relevant for CO₂ utilisation to reduce carbon emissions, especially if green methanol and green hydrogen are used.
+
+Chemical Engineering | Energy Engineering | Methanol hydrocarboxylation | CO₂ transformation | Acetic acid | Formic acid | Thermally transformed MOFs
+
+# 1. Introduction
+
+Fixation of overabundant atmospheric carbon dioxide is an urgent and essential research area, which may lead towards climate mitigation. Several routes for carbon dioxide conversion have been investigated, but thermocatalytic CO₂ hydrogenation pathway is one of the major focus due to its fast kinetics, high productivity, scalability and selectivity¹. Synthesis of chemicals such as methane²,³, methanol⁴, formaldehyde⁵,⁶, dimethyl ether⁷, gasoline-range hydrocarbons⁸, oxymethylene dimethyl ethers⁹,¹⁰, methyl formate¹¹, formic acid¹², and acetic acid¹³,¹⁴ have been investigated in recent years. A CO₂ based chemicals industry has the potential to lower the CO₂ concentration in atmosphere, while simultaneously provide revenue for offsetting the capture costs. The production of acetic acid (AA) via CO₂ hydrogenation is one such route which is receiving attention of the researchers recently.
+
+Acetic acid is extensively used in several industrial applications, including food, chemicals, pharmaceuticals, textile, cosmetics and polymers¹⁵. It is a well-known food preservative and traditionally named as vinegar in food industry. Commercially, two major production processes are used for the synthesis of acetic acid – chemical and fermentative¹⁵,¹⁶. Among various chemical routes, the most common industrial processes are carbonylation of methanol (MeOH) developed by BASF, Cativa and Monsanto, in the presence of homogeneous Cobalt, Iridium and Rhodium catalysts, respectively. In Monsanto process, AA is produced from CH₃OH and fossil fuel derived CO in the presence of CH₃I and homogeneous rhodium-based catalyst¹³,¹⁵. The main reaction of acetic acid production from methanol and CO is summarized in Eq. 1.
+
+$$C{H}_{3}OH+CO \leftrightarrows C{H}_{3}COOH$$
+
+Qian et al. recently reported AA production via hydrocarboxylation of MeOH with carbon dioxide and hydrogen in 1,3-dimethyl-2-imidazolidinone (DMI) solvent over homogeneous Rh and Ru based homogeneous co-catalysts with a combination of LiI promoter and imidazole ligand. While the authors report that imidazole played critical role in inhibiting the reverse water gas shift reaction, but the exact role of imidazole in the reaction mechanism was not clear¹³. The same group also showed AA synthesis via the above described reaction system in the presence of Rh₂(CO)₄Cl₂ homogeneous catalyst, LiCl as a co-catalyst, 4-methyl imidazole ligand and LiI as a promoter¹⁷. This reaction system is highly complex due to the presence of multiple catalysts, stabilizing ligands and organic solvents. In many cases, the authors report a black precipitate, which is not explained but is likely to be the Ru or Rh catalyst, which demonstrates that the system is not stable. Hasan et al. reported low yield of acetic acid (1.58 mmol/L) over NiO-C/Al₂O₃, heterogeneous catalyst at 130°C and 35 bar total pressure of CO₂ and H₂ in 1,4 dioxane solvent after 6h of reaction. Instead a higher amount of formic acid (FA, 4.08 mmol/L) was generated¹⁸. Therefore, there is an urgent need to develop a stable and active heterogeneous catalyst based on low cost metals for AA synthesis which can efficient for industrialisation and scaleup.
+
+He et. al. report FA and AA production via hydrothermal CO₂ reduction with Fe nanoparticles as stoichiometric reagent in which they are converted into ferrous carbonate¹⁹. To the best of our knowledge, Fe-based heterogeneous catalysts have not been reported CO₂ conversion in aqueous phase. Heterogeneous catalysts have advantages in scale-up, and compares favourably against homogeneous catalysts which require large downstream separation processes.
+
+Here we present a Fe-based thermally transformed metal organic framework catalyst (MIL-88B) for hydrocarboxylation of MeOH to produced AA. Recently, Metal Organic Framework (MOFs) derived carbonaceous materials have been reported for their remarkable catalytic properties²⁰,²¹,²². Thermal transformation of MOFs results in a carbonaceous material with embedded metal or metal-oxide nanoparticles²². As these particles are embedded in the matrix of decomposed organic linkers, they show greater resistance to sintering at higher temperatures. Depending on the thermal treatment, the thermally transformed MOFs have features such as high surface area, porosity, and fine dispersion of metal nanoparticles that are desired in an ideal heterogeneous catalyst. Moreover, the porous carbon framework provides better mass transfer to enhance the reaction rate. In this work, thermally transformed MIL-88B, called T-MIL-88B, consisted of dual active sites – Fe⁰ and Fe₃O₄, accelerating the conversion of CO₂ into AA, compared with other Fe-based catalysts tested which contained only Fe₃O₄ or Fe⁰ and Fe₂O₃. In this process, AA is produced in a series of reactions (eq. 2 - 4) –
+
+$$C{O}_{2 \\left(aq\\right)}+{H}_{2 \\left(aq\\right)}\\underleftrightarrow{\\text{Fe}} HCOOH$$
+
+$$C{H}_{3}O{H}_{\\left(l\\right)}+LiI\\leftrightarrow C{H}_{3}I+LiOH$$
+
+$$C{H}_{3}I+HCOOH+LiOH\\leftrightarrow C{H}_{3}COOH+LiI+{H}_{2}O$$
+
+**Overall Reaction**
+
+$$C{O}_{2 \\left(aq\\right)}+{H}_{2 \\left(aq\\right)}+ C{H}_{3}O{H}_{\\left(l\\right)}\\underleftrightarrow{\\text{Fe, LiI, 150°C}}{H}_{3}CCOOH+{H}_{2}O$$
+
+# 2. Methods
+
+## 2.1. Materials
+
+Iodomethane (CH₃I, 99.5%), formic acid (HCOOH, ≥ 95%), lithium Iodide (LiI, 99.9%), terephthalic acid (H₂BDC, 98%), chromium chloride hexahydrate (CrCl₃·6H₂O, 98%), and iron nitrate nonahydrate (Fe(NO₃)₃·9H₂O, 98%) were purchased from the Sigma Aldrich. Commercial zeolite-beta (CBEA, SiO₂/Al₂O₃ = 38) was received from Zeolyst International. Methanol (HPLC grade) was obtained from the Scharlau Chemicals. Milli-Q water was used for catalysts synthesis (MIL-101 and Fe/CBEA) and acetic acid production experiments.
+
+## 2.2. Catalysts synthesis
+
+### 2.2.1. Fe/CBEA
+
+Wet impregnation process was used for Fe/CBEA synthesis as described in our previous publication <sup>9</sup>. The loading of Fe was fixed as 10 wt% in this catalyst. Typically, Fe(NO₃)₃·9H₂O (7.2 g) was dissolved in Milli-Q water (30 mL) by using 100 mL Schott bottle and stirred for 15 min at 65°C to prepare a homogeneous mixture of Fe solution. Thereafter, 9.0 g of CBEA support was immersed in this solution under stirring and maintained it for 6 h at the same temperature to achieve an even dispersion of Fe particles on CBEA support. The mixture was dried in oven at 100°C followed by calcination at 550°C with 5°C/min for 5 h in muffle furnace. The synthesised catalyst was reduced in the environment of H₂/Ar (1:1 v/v) gas mixture at 400°C for 5 h with heating rate of 5°C/min prior to carbon dioxide conversion experiment.
+
+### 2.2.2. Thermally transformed Fe/MIL-101
+
+10 mmol of H₂BDC and 10 mmol CrCl₃·6H₂O were poured into a Teflon-lined autoclave. Subsequently, Milli-Q water (72 ml) was added to it. The reaction mixture was sonicated for 30 minutes followed by stirring for another 30 minutes at 500 rpm. Thereafter, the autoclave was kept in the oven at 205°C for 24 h and allowed to cool to room temperature. The resulting solid suspension was transferred into a centrifuge tube. Initially the centrifugation was performed at 1000 rpm for 3-4 min to remove the unreacted H₂BDC present in the reaction mixture. Thereafter, the centrifugation was carried out at 5000 rpm for 10 minutes. The solid sample was then washed with dimethylformamide (DMF) three times and then dried in an oven at 70°C for 12 h. The synthesised material was named as MIL-101.
+
+For Fe/MIL-101 synthesis, 2.7 g of MIL-101 was suspended in 70 ml ethanol in a Schott bottle and sonicated for 30 minutes. Separately, 2.17 g Fe(NO₃)₃·9H₂O was dissolved in 20 ml ethanol in a different Schott bottle and stirred for 15 minutes. The latter solution was poured into the former suspension of MIL-101 in ethanol. Then the Schott bottle which contained Fe(NO₃)₃ solution was washed with 10 ml ethanol three times and poured into MIL-101 suspension to ensure complete transfer of the Fe precursor. The resultant mixture was sonicated for 30 min followed by stirring at 50°C at 500 rpm for 5-6 h. Finally, the resulting reaction mixture was dried in an oven at 80°C for 2-3 days. The synthesized catalyst was named as Fe/MIL-101. The Fe loading was fixed as 10 wt% in the synthesized catalyst. Prior to catalytic activity test, this catalyst was thermally transformed under 100 ml/min H₂/Ar (1:1) gas mixture at 500°C for 5 h with a heating rate of 5°C/min and allowed to cool in 50 ml/min Ar atmosphere and denoted as T-Fe/MIL-101.
+
+### 2.2.3. Thermally transformed MIL-88B
+
+A modified hydrothermal method as described in the literature <sup>23</sup> was adopted for synthesis of MIL-88B. In a typical procedure, 12.12 g of Fe salt (Fe(NO₃)₃·9H₂O) was dissolved in 75 ml DMF under stirring (500 RPM) in a Schott bottle. Separately, H₂BDC (4.98 g) and DMF (75 ml) were added in a 250 ml Teflon-liner under stirring (500 RPM). Both Fe and H₂BDC solutions were stirred further for 15 min at room temperature. The Fe solution was then poured into H₂BDC precursor solution. 12 ml NaOH solution (4.0 M) was slowly transferred into Fe and H₂BDC solution mixture and stirred again for 30 min at room temperature. Thereafter, the Teflon-liner was sealed in an autoclave and heated to 100°C for 24 h. After cooling to room temperature, MIL-88B particles were collected from this mixture via centrifugation at 7000 RPM for 10 min and washed three times with DMF and methanol, respectively. Finally, the as synthesized MIL-88B was dried overnight in the oven at 80°C and denoted as MIL-88B. Thermal transformation of MIL-88B (2 g) was conducted at 500°C for 5 h with a ramp of 5°C/min under 100 ml/min H₂/Ar (1:1 v/v) environment followed by cooling to room temperature under Ar at 50 ml/min atmosphere and denoted as T-MIL-88B.
+
+## 2.3. Catalyst characterisation
+
+The crystal structure of the materials was investigated with Powder X-ray diffraction (PXRD) by using a Rigaku MiniFlex device. The powder catalysts were loaded in a zero-background sample holder and scanned between 2–80° 2θ with 4°/min scan speed at 15 mA and 40 kV. Nitrogen physisorption analysis was conducted with Micromeritics 3Flex 3500 machine to find the type of adsorption isotherm, Brunauer-Emmett-Teller (BET) surface area and Barrett-Joyner-Halenda (BJH) pore distribution. Tecani T20 was used to capture the transmission electron microscopy (TEM) images of the catalysts. All the samples were dispersed in ethanol and immobilised onto the surface of a holy carbon grid followed by drying in air prior to analysis. ThermoScientific K-Alpha machine was utilized for X-ray photoelectron spectroscopy (XPS) at 1486.6 eV Ephoton and coupled with monochromatic Al Kα radiations. The binding energy (B.E.) baseline correction was conducted by adjusting the C 1 s peaks at 284.8 eV. Thermally transformed samples were prepared *ex situ* prior to the XPS characterization. Shimadzu DTG-60H thermogravimetric analyser was used to check the thermal stability of Fe/MIL-101 and MIL-88B. Both samples were analysed in the temperature range of 100–800°C with a ramp of 5°C/min under Ar atmosphere.
+
+## 2.4. Aqueous phase CO₂ conversion
+
+All the aqueous phase CO₂ conversion experiments were performed in a 100 mL Teflon-lined autoclave batch reactor (Amar Equipment, M4). Typically, 0.4 g of thermally transformed catalyst (T-MIL-88B) and 40 mL water was added to the reactor and CH₃I (10 mmol) was carefully poured into it and sealed. It was purged with hydrogen three times to eliminate air from the headspace. The reactor was then pressurised with CO₂ up to 35 bar, followed by H₂ up to a total pressure of 70 bar at room temperature to achieve CO₂:H₂ ratio of 1:1. The reactor was heated to 150°C under continuous stirring at 200 RPM for 21 h. After 21 h of reaction, the reactor was allowed to cool to room temperature and the remaining gases were carefully vented from it before dissembling it. The catalyst was recovered from the liquid product mixture by centrifugation at 8500 RPM for 1 h. The same procedure was repeated for different total pressures at equimolar CO₂:H₂ ratio and different catalysts (T-Fe/MIL-101 and Fe/CBEA). The liquid sample was analysed at intervals for the best catalyst to check the extent of reaction against time at 150°C, equimolar CO₂:H₂ at 70 bar with 200 RPM stirring speed. The liquid samples were analysed using an HPLC (Agilent 1220 Infinity) equipped with a C18 column and a refractive index detector (RID), using 0.5 mM H₂SO₄ aqueous solution as the mobile phase. The product yields (mmol/gₐₜ·L) and selectivity (%) were calculated using equations 6 and 7, respectively.
+
+$$
+\text{Product}_{i} \text{Yield}= \frac{{n}_{i}}{{m}_{cat} . {V}_{{H}_{2}O}}
+$$
+
+6
+
+$$
+\text{Product} \text{selectivity}= \frac{{n}_{i}}{{\sum }_{i}{n}_{i}}\times 100
+$$
+
+7
+
+Where *nᵢ* = moles of product, *i* = HCOOH or CH₃COOH, *mₐₜ* = mass of catalyst (g) and *V*ₕ₂ₒ = volume of water (L)
+
+The best catalyst was also evaluated for aqueous phase conversion using CO₂, H₂ and methanol (10 mmol) as reactants and lithium iodide (10 mmol) as the promoter. All other reaction conditions were identical to the above described procedure.
+
+## 2.5. Catalyst recycling study
+
+The catalyst recyclability was investigated using CO₂, H₂ and CH₃OH (10 mmol) as reactants and lithium iodide (10 mmol) as the promoter at 150°C, equimolar H₂/CO₂ with 70 bar pressure at room temperature and 200 RPM stirring speed. After each cycle, the catalyst was recovered from the product mixture via centrifugation at 8500 RPM for 1 h and without any intermediate treatment, resuspended into a fresh reaction mixture at the same initial conditions. After five cycles, the centrifuged catalyst was dried overnight in oven at 70°C and stored in air tight glass vial for its characterisation.
+
+## 2.6. Reaction mechanism investigation
+
+Reaction mechanism was explored by designing two different experiments – (1) using FA and CH₃I as reactants and experiment was conducted in water by using T-MIL-88B catalyst at 150°C under 35 bar hydrogen and 200 RPM stirring speed. Typically, 40 mL H₂O, 0.4 g of T-MIL-88B, 5 mmol (312.5 mmol/gₐₜ·L) of HCOOH and 10 mmol (625 mmol/gₐₜ·L) of CH₃I were added in Teflon-liner and reactor was sealed. After achieving the above described conditions, 2 mL liquid sample was withdrawn from the reactor after regular intervals (1, 2, 4, 8, 12 and 24 h) for HPLC analysis. In the 2nd reaction system, aqueous phase CO₂ hydrogenation with CH₃OH (10 mmol) and LiI (10 mmol) was performed over MIL-88B (0.4 g) for 48 h at 150 ˚C, 40 mL H₂O, equimolar H₂/CO₂ under 70 bar at room temperature and 200 RPM stirring speed. After 48 h, the reactor was cooled to room temperature. Both liquid and gas samples were collected for product analysis, where, gas sample was analysed through Shimadzu 2014 GC coupled with TCD and FID detectors, respectively.
+
+# 3. Results And Discussion
+
+## 3.1. Characterisation
+
+Figure 1.a-c illustrates the PXRD diffractograms of the catalysts, before and after catalytic tests. Calcined Fe/CBEA catalyst showed characteristic peaks of α-Fe₂O₃, most of which were not observed in the reduced catalyst. Instead, the reduced catalyst showed Fe⁰ peaks at 2θ = 44.7° and 65° and residual α-Fe₂O₃ peaks at 35.98° and 62.83°. However, there were no Fe⁰ or α-Fe₂O₃ peaks detected in the used catalyst which indicated leaching of Fe from the catalyst support. The residual reaction solution slowly turned to red colour over a period of few days, indicating presence of iron oxides in the solution. Therefore, Fe/CBEA catalyst was not considered further.
+
+Both the fresh and the used T-Fe/MIL-101 catalyst showed peaks corresponding to Fe₃O₄, suggesting that the catalyst was stable after the reaction. However, the α-Fe₂O₃ peaks observed in Fe/MIL-101 (Figure S1, ESI) which did not reduce to Fe⁰ in T-Fe/MIL-101.
+
+T-MIL-88B catalyst showed peaks corresponding to both Fe₃O₄ and Fe⁰, which remained steady after a single run of 48 h reaction time and 5 cycles of 21 h each. Only Fe₃O₄ peaks have been reported after the thermal treatment of MIL-88B at 500°C under nitrogen atmosphere<sup>23</sup>. However, due to the reducing atmosphere used in this study, some of iron oxide nanoparticles reduced to Fe⁰. No evidence of iron carbide was found in the PXRD results.
+
+During the thermal transformation of MOFs, first, the linkers break from the metal oxide clusters. After that, the metal oxide clusters agglomerate and reduce depending upon the chemical environment. MIL-88B consists Fe₃O clusters coordinated by six carboxylate ligands and three adsorbed water molecules, depending on the synthesis method (Figure 2 a). Based on our earlier computational study of thermal transformations in Zr-based MOFs<sup>24</sup>, we expect the following physiochemical transformations in MIL-88B upon thermal treatment. First, the adsorbed water molecules desorb, and at c.a. 100°C the MOF is expected to change the morphology<sup>25</sup>. Near the decomposition temperature, some of the linkers start detaching from the cluster. Unlike Fe oxide nanoparticles encapsulated in MIL-101(Cr), where the movement of nanoparticles is less hindered and can easily agglomerate, in MIL-88B, the Fe₃O metal clusters are part of the framework and hence remain less mobile. After detachment of the organic linkers, the linkers go through thermolysis, resulting in formation of small gaseous molecules such as CO and CO₂. At high temperature, hydrogen is expected to dissociate on iron and likely to catalyse the decarboxylation of linkers, reducing the Fe-O coordination. Without the oxygen from carboxylate groups, formation of single Fe₃O₄ phase is stoichiometrically not possible in Fe₃O. Hence, promoted decarboxylation in H₂ environment is likely to increase the abundance of a mixed Fe/Fe₃O₄ metal nanoparticles. Figure 2 shows the proposed mechanism of thermal evolution of MIL-88B(Fe).
+
+Figure 3.a-f shows the TEM images of MIL-101, Fe/MIL-101, T-Fe/MIL-101, MIL-88B, T-MIL-88B, and used T-MIL-88B, respectively. MIL-101 shows the characteristic octahedral shape of ca. 200-300 nm size (Figure 3 a and Figure S2a of ESI)<sup>26</sup>. After impregnation of Fe over MIL-101, agglomerates of Fe nanoparticles were observed on MIL-101 (Fe/MIL-101) with approximately 50-100 nm in size (Figure 3 b), whereas after thermal transformation, T-Fe/MIL-101 exhibited approximately 5-30 nm particles (Figure 3 c). The emergence of these smaller nanoparticles is likely due to the thermal transformation of Fe/MIL-101 in reductive atmosphere, where the deconstruction of linkers leads to breakage of the Fe agglomerates. Figure 3 d and Figure S2b (ESI) show the characteristic fusiform rod shaped morphology of MIL-88B with ~360 nm length and 90 nm width<sup>23</sup>. After thermal transformation, T-MIL-88B shows a narrow range of Fe⁰/Fe₃O₄ nanoparticle which are well-dispersed over the carbonaceous support (Figure 2 e). The amount of Fe on T-MIL-88B is 49.3%, with 13.7% C and negligible amount of H, N and S (Table S1, ESI), which indicates that original MOF structure is completely transformed into porous carbon. Figure 3 f shows that the T-MIL-88B catalyst retains its structure after 48h of reaction. Figure 3.g-h illustrates the particle size distribution (PSD) for T-MIL-88B and used T-MIL-88B, respectively. 525 and 476 particles were measured from multiple images which showed most of the particles in 4-16 nm for both fresh and used T-MIL-88B, respectively. The peaks were observed at 8 nm with average particle sizes of 9.7 and 9.1 nm for fresh and used T-MIL-88B, respectively which suggested that the studied catalyst is stable and potentially reusable for this reaction.
+
+The surface oxidation state of Fe in the different catalysts was evaluated by X-Ray photoelectron spectroscopy (XPS) study, as shown in Figure 4. For T-MIL-88B (Figure 4.a), Fe 2p₃/₂ XPS spectrum exhibited three peaks, including a peak at 706.9 eV corresponding to metallic iron<sup>27</sup>. Moreover, the other two peaks at 710.1 and 712.3 eV which are correlated to Fe⁺² and Fe⁺³ oxidation state of iron and the satellite peaks for these aforementioned oxidation state appeared at 716.6 and 719.8 eV<sup>28</sup>. In the Fe 2p region of T-MIL-88B, Fe2p₁/₂ and Fe2p₃/₂ peaks are situated 710.1 and 723.8 eV, where, the spin orbital splitting is 13.7 eV that indicated the presence of Fe₃O₄ in T-MIL-88B<sup>29</sup>. Fe₃O₄ may exist as mixed FeO and Fe₂O₃ states, which appears from Fe⁺² and Fe⁺³ oxidation states<sup>30</sup>. The present XPS study shows that Fe₃O₄ is the dominant species on the surface, where the amount of Fe⁺² was 60.4% and Fe⁺³ was 21.0%, whereas Fe⁰ was 18.6%. Therefore, the ratio of Fe⁰ to Fe₃O₄ was accounted as 1/4.38 in T-MIL-88B.
+
+The XPS spectra of Fe 2p₃/₂ in T-Fe/MIL-101 exhibited two peaks at 711.7 and 712.4 eV which is related to Fe⁺² and Fe⁺³ along with two satellite peaks at 718.1 and 722.4 eV. Furthermore, Fe2p₁/₂ and Fe2p₃/₂ of Fe⁺² appeared at 711.7 and 725.4 eV and the spin orbital splitting is 13.7 eV which interpreted the existence of Fe₃O₄ in T-Fe/MIL-101. Metallic Fe peak is absent in this catalyst which is in good agreement with PXRD results. For Fe/MIL-101 catalyst, Fe 2p₃/₂ XPS spectra also contained both Fe⁺² and Fe⁺³ at 711.7 and 713.4 eV, respectively. However, the spin orbit splitting for Fe2p₁/₂ and Fe2p₃/₂ is 14.1 eV (711.7 and 725.8 eV) which suggested the absence of Fe₃O₄ phase.
+
+Figure 4 b represented the Cr XPS spectra of MIL-101, Fe/MIL-101 and T-Fe/MIL-101 catalysts. In MIL-101, Cr 2p XPS spectra contained only one peak at 577.6 eV which is corresponds to Cr⁺³ oxidation state<sup>31</sup>. For Fe/MIL-101, Cr XPS spectra attributed to two peaks at 577.2 and 578.8 eV which are mainly resembles with Cr⁺³ and CrO₃<sup>32</sup>. The negative binding energy shift (0.4 eV) of Cr⁺³ as compared to Cr⁺³ present in MIL-101 is most likely due to the interfacial electronic interaction (charge transfer) between Cr and Fe after the inclusion of Fe in MIL-101<sup>28</sup>. The Cr spectra for T-Fe/MIL-101, Cr XPS spectra mainly consisted with Cr⁺³ peak at 577.1 eV and the amount of CrO₃ is very less as compared to Fe/MIL-101 which may be due to the thermal transformation of Fe/MIL-101 under hydrogen atmosphere that reduces the oxidised Cr species on catalyst surface.
+
+The C 1s XPS spectra for Fe/MIL-101 (Figure 3 c) shows three different types of C peak at 285, 286.4 and 288.5 which belongs to C-C, C-O-C and O-C=O<sup>33</sup>. The C 1s XPS spectra of both T-Fe/MIL-101 and T-MIL-88B contains only two peaks corresponding to C-C and C-O-C, whereas, the O-C=O peak is absent, which may be due to the thermal transformation of both Fe/MIL-101 and MIL-88B under hydrogen atmosphere reducing the oxygen content in the catalyst.
+
+A thermogravimetric analysis of Fe/MIL-101 and MIL-88B has been represented in Figure 5. For Fe/MIL-101, the weight loss in the range of 50-250°C is because of the evaporation of water and removal of free terephthalates inside the pore of MOF<sup>34</sup>. Thereafter, the main weight loss in the temperature range of 270 to 670°C is due to the degradation of organic ligand in the framework of MOF which is attributed to the collapse of the framework<sup>34</sup>. The weight loss of MIL-88B before 250°C corresponds to the removal of water and excess DMF from the framework<sup>35</sup>. For MIL-88B, the weight loss occurs in the temperature ranges of 300 to 500°C due to the degradation of H₂BDC and the breakdown of the framework. The step in the TGA profile of between 550-650°C is most likely due to the carbonization of the framework and the formation of Fe₃O₄–carbon composites<sup>35</sup>.
+
+## 3.2. Catalyst Activities
+
+### 3.2.1. Role of Fe based zeolite and MOF catalysts
+
+Figure 6.a-c illustrates the yield and selectivity of AA via aqueous phase CO₂ reduction with iodomethane at various pressures. All the catalysts showed some activity for AA production; however, T-MIL-88B was clearly the most active and selective catalyst with best yield of 504 mmol/gₐₜ.L and AA selectivity of 92.4%. Based on stoichiometric calculation, it is equivalent to 80.6% conversion of CH₃I into AA. Both Fe/CBEA and T-Fe/MIL-101 provide lower activity for CO₂ hydrogenation and >90% selectivity for FA production. With increasing pressure, the yield increased initially but the AA selectivity peaked at 60 bar for both Fe/CBEA and T-Fe/MIL-101. However, the AA yield and selectivity increases with increasing pressure for T-MIL-88B. Since Fe was present in the structural framework of T-MIL-88B, the thermally transformed catalyst consists of - embedded active metal sites dispersed evenly in a carbon matrix<sup>23</sup>. The high AA activity and the selectivity over T-MIL-88B catalyst is most likely due to the presence of both Fe⁰ and Fe₃O₄ which assist the hydrogenation and C-C coupling reactions, respectively<sup>36</sup>, <sup>37</sup>.
+
+### 3.2.2. Extent of reaction with time
+
+Figure 7.a illustrates the extent of reaction over T-MIL-88B to produce AA and FA via CO₂ hydrogenation with CH₃I as the starting material in the aqueous media. The reaction proceeds via formation of FA as the initial product, whereas AA was not detected until after 8h of reaction. The AA yield and selectively sharply increased between 12 to 24 h, thereafter gradually increasing to 657.6mmol/gₐₜ.L and 98.8%, respectively, at 48h as the reaction approached equilibrium conversion. Based on the initial CH₃I concentration (10 mmol), 100% conversion at 100% selectivity for AA was achieved, within the range of measurement errors. However, as discussed later, CO₂ first converts into FA and after reaching the maximum yield (377.4 mmol/gₐₜ.L) at 8h, the FA yield decreases sharply until the end of reaction at 48 h when the FA yield was measured at 8.1 mmol/gₐₜ.L. However, since CH₃I is consumed by this time, the residual FA cannot convert into AA. Therefore, for the CO₂ hydrogenated into carboxylic acids, the selectivity of AA is 98.8%.
+
+When CH₃OH (10 mmol) was used as a reactant with LiI as a co-catalyst (Figure 7 b), in otherwise identical reaction conditions, the reaction generates *in situ* CH₃I and hence the peak of FA is broader than Figure 7 a. The AA yield and selectivity increased more gradually and achieved a similar yield of 590.1 mmol/gₐₜ.L at 81.7% selectivity after 48 h, which is equivalent to 94% conversion of CH₃OH into AA. The *in-situ* production of CH₃I slowed down the conversion of FA into AA, which may be due to mass transfer limitation.
+
+### 3.2.3. Catalyst reusability
+
+Figure 8 shows that the catalytic activity dropped initially but after three cycles, there was no significant decline in AA yield and selectivity. The PXRD of the used catalyst after five cycles (Figure 1.c), and the TEM image (Figure 3.f) and PSD (Figure 3.h) of used catalyst after 48 h confirmed that the structure is stable and there was no sintering or agglomeration of Fe and Fe₃O₄ nanoparticles in T-MIL-88B. The initial loss in activity is likely due to the loss of small particles of the catalyst which could not be recollected in centrifuge.
+
+### 3.2.4. Proposed Reaction Pathway
+
+Reaction mechanism of hydrocarboxylation of methanol in an organic solvent proceeds via reaction of CH₃OH with LiI to produce CH₃I and LiOH which is similar to the carbonylation of methanol (Monsanto processes) followed by formation of CH₃Rh*I due to the insertion of CH₃I into a Rh* complexing catalyst<sup>13</sup>. Further, CO₂ is inserted into CH₃-Rh bond to produce CH₃COORh*I. Finally, CH₃COOH is formed via reduction of CH₃COORh*I with H₂ molecule in the presence of Ru* to produce HI as an intermediate. Whereas, LiI is regenerated in situ via HI formation which reacts with LiOH to produce H₂O and LiI. However, here we show aqueous phase methanol hydrocarboxylation in which the reaction pathway deviates from the published works and FA is formed as an intermediate.
+
+First, we show that FA can react with CH₃I in water over T-MIL-88B in H₂ atmosphere (Figure 9). The conversion of FA closely follows AA yield and after 24 h of the reaction FA conversion of 91.5% is achieved with 100% AA selectivity.
+
+Next, we show aqueous phase hydrocarboxylation of CH₃OH using T-MIL-88B as catalyst and LiI as co-catalyst. Here both liquid and gas samples were collected after 48 of reaction. The liquid sample showed only the presence of HCOOH and CH₃COOH with 81.7% acetic acid selectivity (Figure 7.b). Whereas gas analysis did not detect any carbonaceous molecules apart from CO₂ (ESI, Figure S4), which eliminates the methanol carbonylation route for AA production.
+
+Figure 10 shows the proposed reaction pathway for acetic acid production via hydrocarboxylation of CH₃OH over T-MIL-88B. CO₂ and H₂ adsorbed over the catalyst and converted into FA, which may desorb. Subsequently, the adsorbed formate species reacts with iodomethane (CH₃I) to allow C-C coupling reaction to take place which generates an acetate species and HI as the by-product. Finally, acetate species is converted into acetic acid, whilst LiI might be regenerated from LiOH and HI (step 8).
+
+# 4. Conclusions
+
+We show that thermally transformed Fe-based metal organic framework-based catalyst (T-MIL-88B) exhibited high catalytic activity and stability for aqueous phase CO₂ transformation into acetic acid. Here, the catalytic activity and the structural property of T-MIL-88B was compared with Fe/CBEA and thermally transformed Fe deposited on MIL-101 (T-Fe/MIL-101). The T-MIL-88B consisted both Fe⁰ and Fe₃O₄ phases, which catalyse hydrogenation and C-C coupling reactions, respectively, making this catalyst superior to the others tested here. Using CH₃OH, CO₂ and H₂ as reactants in aqueous phase, and LiI the promoter, a maximum acetic acid yield of 590.1 mmol/gcat.L, with 81.7% selectivity was achieved after 48 h at 150 ˚C. We propose that the hydrocarboxylation of methanol to make acetic acid is mediated by formate route, which is evidenced by formic acid as an intermediate. The T-MIL-88B catalyst was active for at least five cycles for acetic acid production without showing any signs of deactivation via sintering, oxidation or phase change.
+
+# References
+
+1. Li W, et al. A short review of recent advances in CO₂ hydrogenation to hydrocarbons over heterogeneous catalysts. *RSC advances* **8**, 7651–7669 (2018).
+
+2. Ahmad W, Younis MN, Shawabkeh R, Ahmed S. Synthesis of lanthanide series (La, Ce, Pr, Eu & Gd) promoted Ni/γ-Al₂O₃ catalysts for methanation of CO₂ at low temperature under atmospheric pressure. *Catalysis Communications* **100**, 121–126 (2017).
+
+3. Ahmad W, Al-Matar A, Shawabkeh R, Rana A. An experimental and thermodynamic study for conversion of CO₂ to CO and methane over Cu-K/Al₂O₃. *Journal of environmental chemical engineering* **4**, 2725–2735 (2016).
+
+4. Rui N, Wang Z, Sun K, Ye J, Ge Q, Liu C-j. CO₂ hydrogenation to methanol over Pd/In₂O₃: effects of Pd and oxygen vacancy. *Applied Catalysis B: Environmental* **218**, 488–497 (2017).
+
+5. Lee DK, Kim DS, Kim SW. Selective formation of formaldehyde from carbon dioxide and hydrogen over PtCu/SiO₂. *Applied organometallic chemistry* **15**, 148–150 (2001).
+
+6. Chan FL, Altinkaya G, Fung N, Tanksale A. Low temperature hydrogenation of carbon dioxide into formaldehyde in liquid media. *Catalysis Today* **309**, 242–247 (2018).
+
+7. Frusteri F, Cordaro M, Cannilla C, Bonura G. Multifunctionality of Cu–ZnO–ZrO₂/H-ZSM5 catalysts for the one-step CO₂-to-DME hydrogenation reaction. *Applied Catalysis B: Environmental* **162**, 57–65 (2015).
+
+8. Gao P, et al. Direct conversion of CO₂ into liquid fuels with high selectivity over a bifunctional catalyst. *Nature Chemistry* **9**, 1019–1024 (2017).
+
+9. Ahmad W, Chan FL, Hoadley A, Wang H, Tanksale A. Synthesis of oxymethylene dimethyl ethers (OMEn) via methanol mediated COx hydrogenation over Ru/BEA catalysts. *Applied Catalysis B: Environmental* **269**, 118765 (2020).
+
+10. Siebert M, Krennrich G, Seibicke M, Siegle AF, Trapp O. Identifying high-performance catalytic conditions for carbon dioxide reduction to dimethoxymethane by multivariate modelling. *Chemical science* **10**, 10466–10474 (2019).
+
+11. Corral-Pérez JJ, et al. Decisive Role of Perimeter Sites in Silica-Supported Ag Nanoparticles in Selective Hydrogenation of CO₂ to Methyl Formate in the Presence of Methanol. *Journal of the American Chemical Society* **140**, 13884–13891 (2018).
+
+12. Preti D, Resta C, Squarcialupi S, Fachinetti G. Carbon dioxide hydrogenation to formic acid by using a heterogeneous gold catalyst. *Angewandte Chemie* **123**, 12759–12762 (2011).
+
+13. Qian Q, Zhang J, Cui M, Han B. Synthesis of acetic acid via methanol hydrocarboxylation with CO₂ and H₂. *Nature communications* **7**, 1–7 (2016).
+
+14. Hasan S, Ahmad K, Isahak W, Masdar M, Jahim J. Synthesis of low-cost catalyst NiO (111) for CO₂ hydrogenation into short-chain carboxylic acids. *International Journal of Hydrogen Energy* **45**, 22281–22290 (2020).
+
+15. Pal P, Nayak J. Acetic acid production and purification: critical review towards process intensification. *Separation & Purification Reviews* **46**, 44–61 (2017).
+
+16. Ndoye B, Lebecque S, Destain J, Guiro AT, Thonart P. A new pilot plant scale acetifier designed for vinegar production in Sub-Saharan Africa. *Process Biochemistry* **42**, 1561–1565 (2007).
+
+17. Cui M, Qian Q, Zhang J, Chen C, Han B. Efficient synthesis of acetic acid via Rh catalyzed methanol hydrocarboxylation with CO₂ and H₂ under milder conditions. *Green Chemistry* **19**, 3558–3565 (2017).
+
+18. Hasan S, Ahmad K, Isahak W, Pudukudy M, Masdar M, Jahim J. Synthesis, Characterisation and Catalytic Activity of NiO supported Al₂O₃ for CO₂ Hydrogenation to Carboxylic Acids: Influence of Catalyst Structure. In: *IOP Conference Series: Earth and Environmental Science). IOP Publishing (2019).
+
+19. He C, Tian G, Liu Z, Feng S. A mild hydrothermal route to fix carbon dioxide to simple carboxylic acids. *Organic letters* **12**, 649–651 (2010).
+
+20. Lippi R, et al. Highly active catalyst for CO₂ methanation derived from a metal organic framework template. *Journal of Materials Chemistry A* **5**, 12990–12997 (2017).
+
+21. Lippi R, et al. Unveiling the structural transitions during activation of a CO₂ methanation catalyst Ru₀/ZrO₂ synthesised from a MOF precursor. *Catalysis Today* **368**, 66–77 (2021).
+
+22. Alqarni DS, et al. Ru-zirconia catalyst derived from MIL140C for carbon dioxide conversion to methane. *Catalysis Today* **371**, 120–133 (2021).
+
+23. Liu J, et al. Fe-MOF-derived highly active catalysts for carbon dioxide hydrogenation to valuable hydrocarbons. *Journal of CO₂ Utilization* **21**, 100–107 (2017).
+
+24. Dwivedi S, et al. Atomistic Mechanisms of Thermal Transformation in a Zr-Metal Organic Framework, MIL-140C. *The Journal of Physical Chemistry Letters* **12**, 177–184 (2021).
+
+25. Horcajada P, et al. How Linker’s Modification Controls Swelling Properties of Highly Flexible Iron(III) Dicarboxylates MIL-88. *Journal of the American Chemical Society* **133**, 17839–17847 (2011).
+
+26. Zhao M, et al. Metal–organic frameworks as selectivity regulators for hydrogenation reactions. *Nature* **539**, 76–80 (2016).
+
+27. Ma Z, Song T, Yuan Y, Yang Y. Synergistic catalysis on Fe–Nx sites and Fe nanoparticles for efficient synthesis of quinolines and quinazolinones via oxidative coupling of amines and aldehydes. *Chemical science* **10**, 10283–10289 (2019).
+
+28. Koley P, et al. Leveraging Cu/CuFe₂O₄-Catalyzed Biomass-Derived Furfural Hydrodeoxygenation: A Nanoscale Metal–Organic-Framework Template Is the Prime Key. *ACS applied materials & interfaces* **12**, 21682–21700 (2020).
+
+29. Lee H, Lee W-J, Park Y-K, Ki SJ, Kim B-J, Jung S-C. Liquid phase plasma synthesis of iron oxide nanoparticles on nitrogen-doped activated carbon resulting in nanocomposite for supercapacitor applications. *Nanomaterials* **8**, 190 (2018).
+
+30. Madhuvilakku R, Alagar S, Mariappan R, Piraman S. Green one-pot synthesis of flowers-like Fe₃O₄/rGO hybrid nanocomposites for effective electrochemical detection of riboflavin and low-cost supercapacitor applications. *Sensors and Actuators B: Chemical* **253**, 879–892 (2017).
+
+31. Huo Q, et al. Preparation of a direct Z-scheme α-Fe₂O₃/MIL-101 (Cr) hybrid for degradation of carbamazepine under visible light irradiation. *Applied Catalysis B: Environmental* **255**, 117751 (2019).
+
+32. Guo L, Qin S, Yang B, Liang D, Qiao L. Effect of hydrogen on semiconductive properties of passive film on ferrite and austenite phases in a duplex stainless steel. *Scientific reports* **7**, 1–6 (2017).
+
+33. Fen-rong L, Wen L, Hui-qing G, Bao-qing L, Zong-qing B, Rui-sheng H. XPS study on the change of carbon-containing groups and sulfur transformation on coal surface. *Journal of Fuel Chemistry and Technology* **39**, 81–84 (2011).
+
+34. Liu Z, He W, Zhang Q, Shapour H, Bakhtari MF. Preparation of a GO/MIL-101 (Fe) Composite for the Removal of Methyl Orange from Aqueous Solution. *ACS omega* (2021).
+
+35. Hou S, et al. Green synthesis and evaluation of an iron-based metal–organic framework MIL-88B for efficient decontamination of arsenate from water. *Dalton transactions* **47**, 2222–2231 (2018).
+
+36. Zhang L, Chen X, Peng Z, Liang C. Chemoselective hydrogenation of cinnamaldehyde over MOFs-derived M₂Si@C (M= Fe, Co, Ni) silicides catalysts. *Molecular Catalysis* **449**, 14–24 (2018).
+
+37. Zeng T, Chen W-W, Cirtiu CM, Moores A, Song G, Li C-J. Fe₃O₄ nanoparticles: a robust and magnetically recoverable catalyst for three-component coupling of aldehyde, alkyne and amine. *Green Chemistry* **12**, 570–573 (2010).
+
+# Supplementary Files
+
+- [20211025AceticAcidManuscriptESI.docx](https://assets-eu.researchsquare.com/files/rs-1029433/v1/bb4c24b881c8800e38c27d1e.docx)  
+  Electronic Supplementary Information

032dad3d59669c324774b0aa1256f76c8f34876110dc4463b65122b1b415c729/preprint/images_list.json

@@ -0,0 +1,50 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Deletion of the KDM2B-CxxC causes agenesis of hippocampus.\n(A) Representative images showing Nissl staining on sagittal sections of adult control and Kdm2bEmx1-\u0394CxxC brains.\n(B, C) Immunofluorescent (IF) staining of Calbindin (B) and ZBTB20 (C) on sagittal sections of adult control (left) and Kdm2bEmx1-\u0394CxxC (right) hippocampi. Nuclei were labeled with DAPI (blue). Boxed CA1 and dentate gyri (DG) were enlarged on the right.\n(D, H) Immunohistochemical (IHC) staining of NeuN (D) and DCX (H) on sagittal sections of adult control (left) and Kdm2bEmx1-\u0394CxxC (right) dentate gyri (DG).\n(F) Immunofluorescence of GFAP (green) and SOX2 (red) on sagittal sections of adult control (left) and Kdm2bEmx1-\u0394CxxC (right) dentate gyri (DG). Nuclei were labeled with DAPI (blue). Boxed regions were enlarged on bottom-left corners. White arrows denote GFAP+SOX2+ signals in the subgranule zone (SGZ).\n(E, G) Quantification of NeuN+ cells in the DG (E) and GFAP+SOX2+ cells in SVZ (G). n = 3 for control brains and n = 4 for Kdm2bEmx1-\u0394CxxC brains.\nData are represented as means \u00b1 SEM. Statistical significance was determined using an unpaired two-tailed Student\u2019s t-test (E, G). ***P<0.001; ****P<0.0001. Scale bars, 2mm (A), 100 \u03bcm (B-D, F, H), 50 \u03bcm (C CA1 and DG). DG, dentate gyrus.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Kdm2bEmx1-\u0394CxxC mice exhibit defects in spatial memory, contextual fear conditioning, and motor learning.\n(A) Diagram of Morris water maze.Latency to find the hidden platform across training period in the Morris water maze test.\n(B) An overhead view of the Morris water maze, and representative swim paths of control mice and Kdm2bEmx1-\u0394CxxC mice during the probe trial. The platform was set in the SE quadrant.\n(D, E) Distance moved (D) and velocity (E) during the probe trial (platform removed).\n(F) Frequencies of platform crossing during the probe trial.\n(G) Time spent in each quadrant during the probe trial.\n(H, I) The proportion of freezing time in context before training (Baseline) and after training (Contextual).\n(J, K) The proportion of freezing time in a new context before tone (Pre-Tone) and after tone (Tone).\n(L) Latency to fall during the rotarod test.\n(M) Representative traces of control mice and Kdm2bEmx1-\u0394CxxC mice in the open-field arena.\n(N) Quantification of number of entries in center, and time spent in the center in the open-field test.\nData are represented as means \u00b1 SEM. Statistical significance was determined using two-way ANOVA followed by Sidak\u2019s multiple comparisons test (B, G, I, K, L), or using an unpaired two-tailed Student\u2019s t-test (D-F, N). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; NS, not significant. n = 10 in (A-G), n = 14 in (H-K), n = 11 mice in (L-N) for control and n = 10 mice for Kdm2bEmx1-\u0394CxxC.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Ablation of the KDM2B-CxxC impedes the migration of intermediate progenitors and production of granular cells.\n(A)\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 Double immunofluorescence of TBR2 (green) and GFAP (red) on P0 wild-type and Kdm2bEmx1-\u0394CxxC hippocampi. Nuclei were labeled with DAPI (blue). Boxed regions of FDJ and DG were enlarged on the right.\n(B, C) The schematic of P0 wild-type and Kdm2bEmx1-\u0394CxxC hippocampi. Green dots represent migrating TBR2+ intermediate progenitors, and red lines represent GFAP+ glial scaffold.\n(D, E) Distribution of TBR2+ cells along the three matrices, where dashed lines demarcate regions considered as 1ry, 2ry, and 3ry matrix (D). n = 6 for control brains and n = 4 for Kdm2bEmx1-\u0394CxxC brains.\n(F) EdU was administrated at E15.5 and double labeling of PROX1 and EdU was performed on P2 coronal sections. Boxed regions were enlarged on the right, and single channel fluorescence staining of PROX1 and EdU were shown respectively. Dashed lines indicate DG and FDJ.\n(G) Quantification of PROX1+ cells in FDJ and DG.\n(H) Quantification of PROX1+EdU+ cells in FDJ and DG.\n(I) Quantification of total PROX1+EdU+ cells in hippocampi.\n(J) Analysis of the distribution of PROX1+ granule neurons within the upper and lower blade of the forming DG at P2: the 3ry matrix was divided into 10 ventral- -to-dorsal bins spanning the lower to upper blade domain.\n(K) PROX1+ cells were counted within each bin. The percentage of PROX1+ cells in each bin is represented.\n(L) Quantification of the percentage of PROX1+ granule neurons positioned in the DG lower blade (bins 1\u20135), versus the DG upper blade (bins 6\u201310) in P2 controls and Kdm2bEmx1-\u0394CxxC.\nn = 5 for control brains and n = 4 for Kdm2bEmx1-\u0394CxxC brains (F-L). Data are represented as means \u00b1 SEM. Statistical significance was determined using an unpaired two-tailed Student\u2019s t-test (E left, I), or using two-way ANOVA followed by Sidak\u2019s multiple comparisons test (E middle and right, G, H, K, L). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Scale bars, 100 \u03bcm (A), 500 \u03bcm (F). DG, dentate gyrus; DMS, dentate migratory stream; FDJ, fimbriodentate junction; HF, hippocampal fissure; 1ry, primary matrix; 2ry, secondary matrix; 3ry, tertiary matrix.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Blocked differentiation of neural progenitors on loss of KDM2B-CxxC\n(A) Triple-labeling of PAX6 (green), TBR2 (violet) and EdU (red) on E16.5 control and Kdm2bEmx1-\u0394CxxC brain sections. Boxed regions were enlarged on the right, and single channel fluorescence staining of PAX6, TBR2 and EdU were shown respectively. Dashed lines outline the hippocampi and distinguish DNE, FDJ and DG.\n(B) Experimental analysis scheme: total number of cells, the proportion of proliferative cells, and the distribution of cells along the DMS were quantified.\n(C, D) The schematic of E16.5 wild-type hippocampi. Green dots represent PAX6+ progenitors and blue dots represent migrating TBR2+ intermediate progenitors. Red dashed lines distinguish DNE, FDJ and DG.\n(E, F) Quantification of PAX6+ and TBR2+ cells.\n(G-N) Quantification of the distribution of PAX6+ (G), PAX6+EdU+ (H), TBR2+ (I), TBR2+EdU+ (J) and PAX6+TBR2+ cells (M), and quantification of the proportion of PAX6+EdU+/PAX6+ (K), TBR2+EdU+/TBR2+ (L) and PAX6+TBR2+/PAX6+ (N) along DMS. The distribution patterns of PAX6+ (G) and TBR2+ (I) cells in control and cKO hippocampi were shown as line graphs on the upper right corners. n = 5 for control brains and n = 5 for Kdm2bEmx1-\u0394CxxC brains.\n(O) Double-labeling of TBR2 (red) and Ki67 (green) on P7 coronal section of control and Kdm2bEmx1-\u0394CxxC SVZ, FDJ and DG. Dashed lines outline the hippocampi. Immunofluorescence staining of whole brain sections were shown in Fig. S6A.\n(P-S) Quantification of the distribution of TBR2+ (P), Ki67+ (Q) and TBR2+Ki67+ cells (R), and the proportion of TBR2+Ki67+/TBR2+ (S) at SVZ, FDJ and DG. n = 4 for control brains and n = 4 for Kdm2bEmx1-\u0394CxxC brains.\nData are represented as means \u00b1 SEM. Statistical significance was determined using an unpaired two-tailed Student\u2019s t-test (E, F), or using two-way ANOVA followed by Sidak\u2019s multiple comparisons test (G-N, P-S). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Scale bars, 200 \u03bcm (A, left), 50 \u03bcm (A, right), 200 \u03bcm (O). DNE, dentate neuroepithelium; FDJ, fimbriodentate junction; DG, Dentate Gyrus; SVZ, subependymal ventricular zone; Cx, Cortex; HP, Hippocampus.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpg",
+        "caption": "Loss of KDM2B-CxxC results in activation of the Wnt signaling pathway in the hippocampus.\n(A) The volcano plot of genes up-regulated (red) and down-regulated (blue) in P0 Kdm2bEmx1-\u0394CxxC hippocampi compared to controls.\n(B) GO analysis of the biological progress of up-regulated genes in P0 Kdm2bEmx1-\u0394CxxC hippocampi revealed terms related to cell proliferation (blue) and Wnt signaling pathways (red).\n(C) GO analysis of the biological progress of down-regulated genes in P0 Kdm2bEmx1-\u0394CxxC hippocampi.\n(D) GSEA analysis of positive regulation of cell population proliferation.\n(E) GSEA analysis of Wnt signaling pathway.\n(F)\u00a0 The heat map of leading genes in the GSEA of Wnt signaling pathway.\n(G-H) X-gal staining on P0 coronal sections of Control_BAT (G) and Kdm2bEmx1-\u0394CxxC_BAT (H) brains. Boxed regions were enlarged on the right (G\u2019 and H\u2019). Red arrows indicate areas where X-gal signals are significantly enhanced.\n(I) Quantification of normalized X-gal signal density on the DG-CA-FDJ-Sub path. n = 4 for Control_BAT brains and n = 3 for Kdm2bEmx1-\u0394CxxC_BAT brains.\n(J-K) In situ hybridization (ISH) of Lef1 on E16.5 Control (J) and Kdm2bEmx1-\u0394CxxC (K) coronal brain sections, with boxed regions magnified on the right (J\u2019 and K\u2019). Red arrows indicate areas where the Lef1 expression is significantly elevated.\n(L) Quantification of normalized ISH signal density of Lef1 on the DNE-HNE-Sub path. n = 4 for Control_BAT brains and n = 4 for Kdm2bEmx1-\u0394CxxC_BAT brains.\n(M-N) ISH of Sfrp2 on E16.5 Control (M) and Kdm2bEmx1-\u0394CxxC (N) coronal brain sections, with boxed regions magnified on the right (M\u2019 and N\u2019).\n(O) Quantification of normalized ISH signal density of Sfrp2 on the DNE-HNE-Sub path. n = 4 for Control_BAT brains and n = 4 for Kdm2bEmx1-\u0394CxxC_BAT brains.\nScale bars, 1 mm (G, H, J, K, M and N), 300 \u03bcm (G\u2019, H\u2019, J\u2019, K\u2019, M\u2019 and N\u2019). HP, Hippocampus; DG, dentate gyrus; FDJ, fimbriodentate junction; CA, Cornu Ammonis; Sub, Subiculum; DNE, dentate neuroepithelium; HNE, hippocampal neuroepithelium; Cx, cortex; SVZ, subependymal ventricular zone.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.jpg",
+        "caption": "KDM2B epigenetically silences components of Wnt signaling genes in developing hippocampi.\n(A) The working diagram of KDM2B: KDM2B-CxxC recognizes and binds to CpG islands (CGI) of DNA, therefore recruiting PRC1 to CpG islands (CGIs). Reciprocal recognition of modifications by PRC1 and PRC2 leads to enrichment of H2AK119ub and H3K27me3, hence stabilizing gene repression.\n(B) Line charts showing average H2AK119ub, H3K27me3 and H3K36me2 signals at CGIs (\u00b1 2 kb flanking regions) in P0 control (black lines) and Kdm2bEmx1-\u0394CxxC (cKO) (red lines) hippocampi.\n(C) Line charts showing average H2AK119ub, H3K27me3, H3K36me2 and ATAC-seq signals at CGI+ TSS (\u00b12 kb flanking regions) in P0 control (black lines) and Kdm2bEmx1-\u0394CxxC (cKO) (red lines) hippocampi. TSS, transcription starting sites.\n(D-G) RT-qPCR showing relative expressions of Wnt7b, Wnt3, Wnt10a and Sfrp2 in control (black lines) and Kdm2bEmx1-\u0394CxxC (red lines) hippocampi of indicated developmental stages (E14.5, E15.5, E16.5, P0 and P7).\n(H) The UCSC genome browser view of HA2K119ub, H3K27me3 and H3K36me2 enrichment and ATAC-seq signal in P0 control and Kdm2bEmx1-\u0394CxxC (cKO) hippocampi at Wnt gene loci [corresponding to (D-G), Wnt7b, Wnt3, Wnt10a and Sfrp2]. CGIs were shown as black columns at the bottom, and signals represent ChIP-seq RPM (reads per million). Colored regions marked enrichment differences between control and cKO.\n(I) The schematic diagram of in utero electroporation (IUE) to target the developing hippocampi.\n(J) The schematic diagram of hippocampal structure, and the hierarchical partition of CA1 region (VZ, IZ, Py).\n(K) E14.5 mouse hippocampi were electroporated with empty or Wnt-mix-expressing vector (Wnt3a, Wnt5a, Wnt5b, Wnt7b, and Wnt8b) or SFRP2-expressing vector, along with the GFP-expressing vector (PCIG) to label transduced cells. Embryos were sacrificed at E18.5 for immunofluorescent analysis. Representative immunofluorescent images showing expression of TBR2+ (red) in GFP+ (green) transduced cells at E18.5 CA1 regions. Nuclei were labeled with DAPI (blue). Arrowheads denote double-labeled cells. White dashed lines distinguish three layers of CA1: VZ, IZ, Py. The VZ layer marked by the yellow dashed box is enlarged below.\n(L) The relative location of GFP+ cells in VZ, IZ and Py were quantified. n = 7 for PCIG, n = 6 for Wnt-mix and n = 10 for SFRP2.\n(M) Quantification of the proportion of TBR2+GFP+/TBR2+ in CA1 of PCIG, Wnt-mix and SFRP2. n = 7 for PCIG, n = 6 for Wnt-mix and n = 10 for SFRP2.\nData are represented as means \u00b1 SEM. Statistical significance was determined using two-way ANOVA followed by Sidak\u2019s multiple comparisons test (D-G), using two-way ANOVA followed by Tukey\u2019s multiple comparisons test (L), or using one-way ANOVA analysis (M). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Scale bars, 50 \u03bcm (K). CA, Cornu Ammonis; VZ, ventricular zone; IZ, intermediated zone; Py, pyramidal cell layer of the hippocampus.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

032dad3d59669c324774b0aa1256f76c8f34876110dc4463b65122b1b415c729/preprint/preprint.md

@@ -0,0 +1,283 @@
+# Abstract
+
+The hippocampus plays major roles in learning and memory, and its formation requires precise coordination of patterning, cell proliferation, differentiation, and migration. Here we removed the chromatin-association capability of KDM2B in the progenitors of developing dorsal telencephalon (Kdm2b<sup>∆CxxC</sup>) to discover that Kdm2b<sup>∆CxxC</sup> hippocampus, particularly the dentate gyrus, became drastically smaller with disorganized cellular components and structure. Kdm2b<sup>∆CxxC</sup> mice displayed prominent defects in spatial memory, motor learning and fear conditioning, resembling patients with KDM2B mutations. The migration and differentiation of neural progenitor cells was greatly impeded in the developing Kdm2b<sup>∆CxxC</sup> hippocampus. Mechanism studies revealed that Wnt signaling genes in developing Kdm2b<sup>∆CxxC</sup> hippocampi were de-repressed due to reduced enrichment of repressive histone marks by polycomb repressive complexes. Activating the Wnt signaling disturbed hippocampal neurogenesis, recapitulating the effect of KDM2B loss. Together, we unveiled a previously unappreciated gene repressive program mediated by KDM2B that controls progressive fate specifications and cell migration, hence morphogenesis of hippocampus.
+
+Biological sciences/Developmental biology/Neurogenesis/Developmental neurogenesis  
+Biological sciences/Neuroscience/Development of the nervous system/Neuronal development  
+Biological sciences/Neuroscience/Diseases of the nervous system/Developmental disorders
+
+# Introduction
+
+The hippocampus of the mammalian brain includes three major compartments: the hippocampus proper, which can be divided into three pyramidal subregions [cornu ammonis (CA) fields], the dentate gyrus (DG), and the subiculum. The hippocampus plays essential roles in spatial memory that enables navigation, in the formation of new memories, as well as in regulating mood and emotions<sup>1</sup>. Constructing a functional hippocampus requires precise production, migration, and assembly of a variety of distinct cell types during embryonic and early postnatal stages. Largely resembling neurogenesis of neocortical pyramidal neurons, the production of pyramidal neurons in CAs and granule cells in dentate gyrus follows the path of indirect neurogenesis<sup>2</sup>, i.e., PAX6-expressing radial glial progenitor cells (RGCs) first give rise to TBR2<sup>+</sup> intermediate progenitor cells (IPCs), which then produce NeuN<sup>+</sup> neurons. Particularly, the formation of the dentate gyrus relies on sequential emergence of germinative foci at different locations, including the dentate notch around embryonic (E) day 13.5 in mice, the fimbriodentate junction (FDJ) around E15.5 and the hilus around birth, which generate granule cells at different parts of DG<sup>2,3</sup>. Cell fate defects and skewed cell migration during hippocampal development underly a cohort of human neurologic and psychiatric diseases<sup>4</sup>. Moreover, the subgranule zone (SGZ) of adult DG contains neural stem cells (NSCs) to support neurogenesis, which might have implications in the formation of new memory.
+
+The cell fate specification during development requires coordinated actions of transcription factors (TFs), epigenetic factors and *cis*-acting elements to ensure precise gene expression and silencing. Notably, repressive histone modifications including mono-ubiquitinated histone H2A at lysine 119 (H2AK119ub1) and trimethylated histone H3 at lysine 27 (H3K27me3), respectively modified by the Polycomb repressive complex 1 (PRC1) and PRC2, are essential fate regulators in embryogenesis, organogenesis and tissue homeostasis<sup>5–9</sup>. The PRC1 can be recruited to the chromatin *via* either one of the CBX proteins – *a.k.a.* canonical PRC1, or other adapter proteins including KDM2B – *a.k.a.* variant PRC1<sup>10–14</sup>. Interestingly, PRC1 and PRC2 can reciprocally recognize repressive histone modifications mediated by each other or itself to stabilize repressive chromatin environments<sup>15–19</sup>. It has been reported that the PRC2 is required for hippocampal development and the maintenance of adult neural stem cell pool of the DG, but the underlying cellular processes and molecular mechanisms were largely elusive<sup>20,21</sup>. Furthermore, very little is known to what extent and how PRC1 is involved in hippocampal development.
+
+KDM2B, previously known as JHDM1B, FBXL10, or NDY1, can recruit PRC1 to non-methylated CpG islands (CGIs), particularly at promoter regions, *via* its CxxC Zinc finger (ZF)<sup>10–12,22,23</sup>. The long isoform of KDM2B – KDM2BLF – contains a Jmjc demethylation domain which removes the di-methylated lysine 36 of histone H3 (H3K36me2) to regulate pluripotency and early embryogenesis<sup>24</sup>. Mutations of human *KDM2B* gene are associated with neurodevelopment defects including intellectual disability (ID), speech delay and behavioral abnormalities<sup>25–27</sup>. A recent study indicated that heterozygosity of *Kdm2b* in mice impaired neural stem cell self-renewal and leads to ASD/ID-like behaviors<sup>28</sup>. However, there is no in-depth analysis to dissect how PRC1- and/or demethylase-dependent roles of KDM2B participates in multiple facets of neural development, including self-renewal, migration, differentiation and localization of neural progenitors and their progeny.
+
+We previously showed that KDM2B controls neocortical neuronal differentiation and the transcription of *Kdm2blf* is *cis*-regulated by a long non-coding RNA which is divergently transcribed from the promoter of *Kdm2blf*<sup>29</sup>. However, questions remain regarding to what extent and how KDM2B regulates neural development. Here we ablated the chromatin association capability of KDM2B in the developing dorsal forebrain to surprisingly find the morphogenesis of hippocampus, especially the DG, was greatly hampered. Moreover, intermediate progenitors could not properly migrate and differentiate upon dissociating KDM2B from the chromatin. The canonical Wnt signaling were aberrantly activated in mutant hippocampi, probably due to decreased enrichment of H2AK119ub and H3K27me3 in CGI promoters of key Wnt pathway genes.
+
+# Results
+
+Removing KDM2B’s chromatin association capability causes agenesis of the hippocampus
+
+KDM2B has two main isoforms: the long isoform KDM2BLF contains a demethylase Jmjc domain while both isoforms share the CxxC zinc finger (ZF), the PHD domain, a F-box and the LRR domain. We generated the conditional knockout allele of *Kdm2b* by flanking exon 13 that encodes the CxxC ZF with two *loxP* sequences (*Fig. S1 A*). These floxed *Kdm2b* mice, *Kdm2b*<sup>*flox(CxxC)*</sup>, were crossed with *Emx1*-Cre and *Nestin*-Cre mice to generate conditional *Kdm2b*<sup>*Emx1−∆CxxC*</sup> and *Kdm2b*<sup>*Nestin−∆CxxC*</sup> conditional knockout (cKO) mice respectively to abolish KDM2B’s association with the chromatin. Although *Kdm2b*<sup>*Nestin−∆CxxC*</sup> mice could not survive past postnatal day 7 (P7), *Kdm2b*<sup>*Emx1−∆CxxC*</sup> mice were born at the mendelian ratio and thrive through adulthood without gross abnormality (*Fig. S1 B-S1C*). RNA-seq of hippocampal neurospheres confirmed the deletion of exon 13 in cKO brains (*Fig. S1 D*). We then performed *in situ* hybridization of neonatal (postnatal day 0, P0) brains using probes targeting exon 13, which showed *Kdm2b* is expressed in the developing hippocampus across the CA region and the DG of control *Kdm2b*<sup>*flox/flox*</sup> brains (*Fig. S1 F*). Importantly, the expression of *Kdm2b-CxxC* was almost gone in the *Kdm2b*<sup>*Emx1−∆CxxC*</sup> hippocampi (*Fig. S1 G*). Moreover, immunoblotting of embryonic day 15.5 *Kdm2b*<sup>*Nestin−∆CxxC*</sup> neocortex revealed truncated long and short isoforms of KDM2B, confirming the selective deletion of the CxxC ZF, which could abolish the CGI association by KDM2B and variant PRC1.1 (*Fig. S1 H*).
+
+Strikingly, the hippocampi, particularly the DGs, of adult *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains were greatly shrunk in size in all examined sections (Fig. *1 A, S2B*). Although Wfs1-expressing pyramidal cells could be detected in the CA1 region of *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains (*Fig. S2A*), Calbindin-expressing cells were mostly diminished (Fig. *1 B*). Both upper and lower blades of cKO DGs were much shorter than controls (Fig. *1 C; S1C*), with numbers of NeuN + granule cells and GFAP + SOX2 + NSCs decreased by 39.7% and 82.1% respectively (Fig. *1 D-*<span class="InternalRef" refid="Fig1">1</span>*G*). Immuno-staining showed irregularity and ectopic dispersion of Calbindin+, ZBTB20 + or NeuN + granule cells in cKO DGs (Fig. *1 B-*<span class="InternalRef" refid="Fig1">1</span>*D*), accompanied with fewer HopX + NSCs, fewer TBR2 + neuroblasts, and fewer PROX1 + or DCX + neurons in the cKO SGZ (Fig. *1 H; S1C*). Interestingly, many HopX + NSCs were found to be ectopically localized inside the granule cell layer of cKO DGs (*Fig. S1 C*, red arrows). The ventricles were also enlarged in cKO brains along with thinner neocortices (Fig. *1 A, S2D*). The enlarged ventricles of cKO brains could already be seen at P0 for unknown reasons without thinning of neocortices (*Fig. S2G-S2J*). No increase of apoptotic cells was detected in P7 cKO brains and hippocampi (*Fig. S2K-S2L*). At P7, numbers of upper-layer (SATB2+) and lower-layer (CTIP2+) neocortical neurons were not significantly altered upon CxxC deletion of KDM2B (*Fig. S2M-S2N*). Consistently, numbers of PAX6-expressing radial glial progenitors and TBR2-expressing intermediate progenitors were not changed in E16.5 cKO neocortices (data not shown). Together, ablation of the chromatin association capability of KDM2B in developing dorsal forebrains causes hippocampal agenesis, while the thinning of adult cKO neocortices might be secondary to ventricle dilation.
+
+*Kdm2b*<sup>*Emx1−∆CxxC*</sup> mice displays defects in spatial memory, motor learning and fear conditioning
+
+Since hippocampus is essential for spatial navigation and memory consolidation, as well as for exploration, anxiety and depression behaviors<sup><span class="CitationRef" citationid="CR30">30</span>–<span class="CitationRef" citationid="CR33">33</span></sup>, we conducted a series of behavior tests. First, Morris water maze tests revealed that *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO mice were defective in spatial learning. It took cKO mice longer time to find the platform in the training stage (Fig. *2 A-*<span class="InternalRef" refid="Fig2">2</span>*B*). Consistently, in the probe trial, *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO mice spent shorter time in quadrant holding the platform (Fig. *2 G*), whereas control and cKO mice showed no differences in swimming distances and velocity (Fig. *2 C-*<span class="InternalRef" refid="Fig2">2</span>*F*). Secondly, in contrast to control mice, *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO mice failed to display prolonged freezing time in both context- and sound-induced fear conditioning tests (Fig. *2 H-*<span class="InternalRef" refid="Fig2">2</span>*K*). Third, the rotarod performance tests indicated that *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO mice have compromised capacity for motor coordination and learning, as they endured shorter time in rotarods than controls (Fig. *2 L*). Interestingly, open field tests showed reluctance of *Kdm2b*<sup>*Emx1−∆CxxC*</sup> mice to explore center regions of open fields (Fig. *2 M-*<span class="InternalRef" refid="Fig2">2</span>*N*), which could be decreased willingness of cKO mice to explore and/or increased anxiety. The distance and velocity traveled, and immobility time in open field tests were not significantly altered in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> mice (*Fig. S3A*). The immobility time of *Kdm2b*<sup>*Emx1−∆CxxC*</sup> mice in forced swimming tests were a tad shorter than control mice (no statistical significance), whereas the immobility time of cKO mice in tail suspension test was the same as controls, suggesting cKO mice did not display depression-related behavior (*Fig. S3B-S3C*). Interestingly, cKO mice spent shorter time in open arms of elevated plus maze (no statistical significance), indicating that cKO mice tend to be hypersensitive to anxiety (*Fig. S3D*). Together, ablation of the chromatin association capability of KDM2B leads to development failure of the hippocampus, which might cause defects in spatial memory, fear conditioning and motor learning.
+
+Deletion of the CxxC ZF of KDM2B has no effect on adult neurogenesis of the DG
+
+The shrunken hippocampi and diminished NSC pool in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO mice prompted us to investigate whether the ablation of KDM2B’s chromatin association capability could hamper adult neurogenesis of the DG. We thus crossed the *Kdm2b*<sup>*flox(CxxC)*</sup> mice with *Nestin-CreERT2* mice to produce *Kdm2b*<sup>*Nestin − CreERT2−∆CxxC*</sup> cKO mice. Reporter analyses indicated that *Nestin-CreERT2* is active in SGZ upon tamoxifen (TAM) induction (*Fig. S3E*). Adult *Kdm2b*<sup>*Nestin − CreERT2−∆CxxC*</sup> cKO and control mice were administered with TAM for six consecutive days to ablate the CxxC ZF in adult NSCs and their progeny. BrdU was also administered for six consecutive days to label NSCs’ progeny (*Fig. S3F*). Brains were collected at day 8 and day 35 of post-TAM injection for immunofluorescent staining of BrdU, along with DCX (day 8) or PROX1 (day 35), two markers for new-born and mature granule neurons respectively. Data showed numbers of BrdU-labelled cells and DCX<sup>+</sup> BrdU<sup>+</sup> double-positive cells, and ratios for DCX<sup>+</sup> BrdU<sup>+</sup>/BrdU<sup>+</sup> cell were comparable between *Kdm2b*<sup>*Nestin − CreERT2−∆CxxC*</sup> cKO and control DGs in all examined sections at day 8 (*Fig. S3F-S3G, S3J-S3L*). Similarly, at day 35, numbers of BrdU-labelled cells and PROX1<sup>+</sup> BrdU<sup>+</sup> double-positive cells, and ratios for PROX1<sup>+</sup> BrdU<sup>+</sup>/BrdU<sup>+</sup> cell were not altered in *Kdm2b*<sup>*Nestin − CreERT2−∆CxxC*</sup> cKO DGs (*Fig. S3H-S3I, S3M-S3O*). Thus, removal of KDM2B’s chromatin association capability in adult NSCs exerts no effect on adult neurogenesis of DGs.
+
+Hampered migration of intermediate progenitors and neurogenesis of granule cells upon loss of KDM2B
+
+In DG development, neural progenitors, especially TBR2-expressing intermediate progenitor cells (IPCs) migrate from the dentate neuroepithelial (DNe) stem zone (the primary −1ry matrix) through the dentate migratory stream (DMS, the secondary −2ry matrix) to the developing DG (the 3ry matrix), while being distributed in multiple transient niches (Fig. *3 D*). The prominent DG defects in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains prompted us to examine distributions of intermediate progenitors and neurons along the migratory path. P0 brain sections were co-stained with TBR2 to label IPCs and with GFAP to label astrocytic scaffold at the fimbriodentate junction (FDJ) of the DMS and the fimbria. Total number of IPCs were increased by 15.3% upon loss of KDM2B-CxxC. Strikingly, in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains, significant more TBR2 + IPCs were accumulated at the DNe and the DMS, along with significant fewer TBR2 + IPCs at the DG (Fig. *3 A-*<span class="InternalRef" refid="Fig3">3</span>*E*). Of note, TBR2 + IPCs were more loosely distributed at the FDJ, with GFAP-labelled astrocytic scaffold scattered at FDJ but constricted at fimbria (Fig. *3 A-*<span class="InternalRef" refid="Fig3">3</span>*C*). We next carried out EdU birthdating experiments by administering EdU at E15.5 followed by EdU and PROX1 co-labelling at P2. Consistently, 31.3% fewer PROX1-labelled granule cells could be detected in cKO DGs but 259.0% more in the cKO FDJ. In cKO brains, 33.3% fewer E15.5-labelled EdU cells reached the DG and co-stained with PROX1, while ectopic PROX1 cells were detected at FDJ (Fig. *3 F-*<span class="InternalRef" refid="Fig3">3</span>*H*). Total number of PROX1 + EdU + cells was also reduced by 27.1% in cKO hippocampi (Fig. *3 I*), indicating hampered neuronal production on loss of KDM2B-CxxC. Furthermore, proportions of PROX1 + cells in lower and upper blades of DGs were switched into a lower-more and upper-fewer status in cKO DGs (Fig. *3 J-*<span class="InternalRef" refid="Fig3">3</span>*L*). Congruently, by P7, a big chunk of PROX1 + cells could be seen at the FDJ of *Kdm2b*<sup>*Emx1−∆CxxC*</sup> brains (*Fig. S4G*). Together, loss of KDM2B-CxxC impedes migration and differentiation of IPCs, hence proper production and localization of granule neurons during hippocampal formation.
+
+The hampered migration of IPCs upon loss of KDM2B-CxxC could be due to defects of the cortical hem (CH) derived astrocytic scaffolds at the fimbria<sup><span class="CitationRef" citationid="CR34">34</span></sup>. To exclude the possibility, *Kdm2b*<sup>*Nestin−∆CxxC*</sup> cKO brains were inspected, because reporter analyses validated that *Nestin* is mostly not expressed in CH-derived astrocytic scaffolds, which are labelled by BLBP (*Fig. S4F*). Data showed that P0 *Kdm2b*<sup>*Nestin−∆CxxC*</sup> brains displayed almost the same phenotypes as those in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> brains, *i.e.*, overproduction of TBR2 + IPCs, significant more IPCs accumulated at the DNe and the FDJ, but fewer IPCs at the DG, suggesting that the migrating defects of IPCs upon loss of KDM2B-CxxC were not due to defects of CH-derived astrocytic scaffolds (*Fig. S4A-S4E*).
+
+Disturbed differentiation of neural progenitors on loss of KDM2B-CxxC
+
+We next investigated how the RGC-IPC-Neuron neurogenesis path of hippocampi were influenced on loss of KDM2B-CxxC. At E13.5, when the hippocampal primordia just emerged, the distribution of BLBP, the marker for astrocytic progenitors at CH, were unaltered in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains (*Fig. S5A-S5B*). Similarly, the expression pattern of SOX2, the marker for neocortical and hippocampal RGCs, were not changed in CH, DNe and hippocampal neuroepithelium (HNe) upon loss of KDM2B-CxxC (*Fig. S5A*, the middle panel). In addition, the distribution of PAX6 and TBR2, markers for RGC and IPCs respectively, at E13.5 were almost the same between cKOs and controls (*Fig. S5A*, the bottom panel). We further examined whether the capacity of proliferation and differentiation of hippocampal progenitors was affected at E14.5 by co-labeling brain sections with PAX6, TBR2 and EdU (2 hours pulse). Data showed that deletion of KDM2B-CxxC had no effect on abundance and proliferation of PAX6 + RGCs and TBR2 + IPCs in DNEs and CHs. Consistently, the differentiation capability from RGCs to IPCs was unchanged in cKO brains, because the number of PAX6 + TBR2 + cells and the ratio of PAX6 + TBR2 + cells among all PAX6 + cells were comparable between cKOs and controls (*Fig. S5C-S5D*). Thus, the hippocampal agenesis in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains was not due to specification, maintenance, and differentiation of early neural progenitors.
+
+We then moved onto E16.5, when migration and neurogenesis of hippocampal progenitors are at the climax. In control brains, most PAX6 + RGCs were localized at the DNe, while TBR2 + IPCs were more evenly distributed along the DNe-FDJ-DG migratory/differentiating path (Fig. *4 A-*<span class="InternalRef" refid="Fig4">4</span>*D*). Although total numbers of PAX6 + RGCs and TBR2 + IPCs were not significantly altered in cKO hippocampi (Fig. *4 E-*<span class="InternalRef" refid="Fig4">4</span>*F*), their distribution along the DNe-FDJ-DG path were dramatically delayed. The cKO DNes were significantly enriched with more PAX6+, PAX6 + EdU+ (dividing) and TBR2 + cells (Fig. *4 G-*<span class="InternalRef" refid="Fig4">4</span>*I*). In contrast, the distribution of total and dividing PAX6 + and TBR2 + cells in cKO FDJs and DGs was drastically reduced, except for that of TBR2 + cells in the FDJ (Fig. *4 G-*<span class="InternalRef" refid="Fig4">4</span>*L*). Furthermore, the differentiating rate from RGCs to IPCs (PAX6 + TBR2+/PAX6+) at FDJ is 60% higher in cKOs (Fig. *4 N*), but PAX6 + TBR2 + cells were 73.9% fewer in cKO DGs (Fig. *4 M*). By P7, numbers of total and dividing TBR2 + IPCs were significantly decreased at cKO DGs, whereas those at hippocampal SVZ and FDJ were greatly increased (Fig. *4 O-*<span class="InternalRef" refid="Fig4">4</span>*S; S6A*). In summary, the migratory and differentiating trajectory of hippocampal progenitors were greatly delayed upon loss of KDM2B-CxxC.
+
+We next asked whether neuronal maturation was impeded on loss of KDM2B-CxxC. To this end, we crossed the *Kdm2b*<sup>*flox(CxxC)*</sup> mice with *Nex-Cre* mice to obtain *Kdm2b*<sup>*Nex−∆CxxC*</sup> cKO mice, where KDM2B-CxxC was specifically ablated in all postmitotic neurons but not progenitors of neocortices and hippocampi. Phenotypic analyses revealed that hippocampal morphology and cellular components, including PROX1 + DG granule cells, of P7 *Kdm2b*<sup>*Nex−∆CxxC*</sup> cKOs did not show any abnormality, suggesting that hippocampal agenesis was not due to defects on postmitotic neuron differentiation (*Fig. S6B*). Together, KDM2B regulates hippocampal morphogenesis by controlling multiple behaviors, including coordinated RGC to IPC differentiation, migration, and divisions of neural progenitors (*Fig. S6C*).
+
+Aberrantly activated Wnt signaling due to alterations of key histone modifications in *Kdm2b* mutant hippocampus
+
+We next sought to unveil molecular events and mechanisms underlying hippocampal agenesis caused by KDM2B mutation. First, hippocampal tissue from P0 control and *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains were harvested and subjected to RNA-seq transcriptome analyses. We chose P0 because the third germinal primordium of the dentate gyrus just appears, and the hippocampal neurogenesis is at peak. Importantly, the blockade of IPC migration is prominent in P0 cKO hippocampi. Data showed 886 genes and 575 genes were activated and repressed respectively in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO hippocampi. In line with aforementioned phenotypic analyses, expression levels of markers for progenitor cells including *Pax6*, *Neurog2* and *TBR2*/*Eomes* were increased in cKO hippocampi (Fig. *5 A*). Gene ontology (GO) analyses revealed that up-regulated genes were involved in pattern formation, morphogenesis, cell proliferation and canonical Wnt signal pathways (Fig. *5 B, S8A*), whereas down-regulated genes encompassing those associated with neuronal structures and functions (Fig. *5 C*). Notably, a series of canonical Wnt pathway components, including ligands, receptors, and signal transducers, were significantly activated upon loss of KDM2B-CxxC (Fig. *5 D-*<span class="InternalRef" refid="Fig5">5</span>*F*). To validate whether the canonical Wnt signaling was enhanced in cKO hippocampi, *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO mice were crossed with the BAT-GAL (B6.Cg-Tg(BAT-lacZ)3Picc/J) Wnt-reporter mice. Beta-galactosidase staining showed that P0 cKO hippocampi had stronger canonical Wnt activity, including the CA region and FDJ. The ventricular surface of hippocampi and neocortices of cKO brains also displayed elevated Wnt signaling (Fig. *5 G-*<span class="InternalRef" refid="Fig5">5</span>*I*). In particular, components of canonical Wnt signaling genes such as *Lef1* and *Sfrp2* were elevated. *In situ* hybridization verified that the expression of *Lef1*, the gene encoding the transcriptional co-factor of β-Catenin to activate Wnt signaling, is significantly up-regulated in E16.5 cKO hippocampi (Fig. *5 J-*<span class="InternalRef" refid="Fig5">5</span>*L*). LEF1 not only has an early role in specifying the hippocampus, but also controls the generation of dentate gyrus granule cells<sup><span class="CitationRef" citationid="CR35">35</span></sup>. Similarly, the expression of *Sfrp2*, another Wnt signaling pathway component was greatly enhanced in E16.5 cKO hippocampi and VZ/SVZ of neocortices (Fig. *5 M-*<span class="InternalRef" refid="Fig5">5</span>*O*). Although members of the SFRP family were first reported as Wnt inhibitors, *Sfrp2* regulates anteroposterior axis elongation, optic nerve development, and cardiovascular and metabolic processes by the promoting or inhibiting Wnt signaling pathway<sup><span class="CitationRef" citationid="CR36">36</span>–<span class="CitationRef" citationid="CR39">39</span></sup>.
+
+Since P0 hippocampi contained multiple cell types ranging from RGCs to IPCs to neurons, we propagated RGCs *in vitro* under the serum-free neurosphere condition (*Fig. S7A*). RNA-seq transcriptome studies revealed that a significant portion of activated and repressed genes in cKO neurospheres overlapped with those in cKO hippocampal tissue (*Fig. S7B-S7D*). GO analyses indicated genes involved in cell division and canonical Wnt signaling were activated in cKO hippocampal neurospheres (*Fig. S7E*). The canonical Wnt pathway genes *Lef1* and *Sfrp2* were also significantly up-regulated in neurospheres derived from cKO hippocampi (*Fig. S7F*).
+
+KDM2B is a key component of variant PRC1 to mediate repressive histone modification H2AK119ub and subsequent H3K27me3 (Fig. *6 A*) and its long isoform KDM2BLF also bears demethylase activity for H3K36me2. We then examined how loss of KDM2B-CxxC affects these modifications in hippocampal tissue and neurospheres, and whether these effects were associated with enhanced Wnt signaling. To this end, chromatin immunoprecipitation sequencing (ChIP-seq) was performed using P0 hippocampal tissue and neurospheres. As expected, the overall levels of H2AK119ub and H3K27me3 were decreased in cKO hippocampal tissues and neurospheres (Fig. *6 B-*<span class="InternalRef" refid="Fig6">6</span>*C*). However, the level of H3K36me2 was not altered in cKO tissue but slightly increased in neurospheres (Fig. *6 B-*<span class="InternalRef" refid="Fig6">6</span>*C; S8H-S8I*). Moreover, assay for transposase-accessible chromatin using sequencing (ATAC-seq) showed that chromatin became more accessible in cKO hippocampi (Fig. *6 C*).
+
+We next verified changes of Wnt signaling pathway and investigated how histone modifications and chromatin accessibilities correlate with changes of gene expression. Since PRC1.1 could be recruited to CGIs *via* KDM2B’s CxxC zinc finger to catalyze H2AK119ub, we paid special attention to chromatin status of CGIs. Quantitative reverse transcription PCR (RT-qPCR) revealed that the transcripts of multiple Wnt ligands and *Sfrp2* in cKO hippocampi were significantly elevated throughout developmental time points (Fig. *6 D-*<span class="InternalRef" refid="Fig6">6</span>*G; S8B-S8G*).
+
+Importantly, peaks for H2AK119ub and H3K27me3 were decreased on CGI-enriched promoters of many activated genes, including *Wnt7b*, *Wnt3*, *Wnt6*, *Wnt10a* and *Sfrp2*, as well as *Pax6*, *Eomes* and *Neurod1*; but peaks for H3K36me2 on these sites were not significantly altered in P0 cKO hippocampi (Fig. *6 H; S8J*). Consistently, ATAC-seq showed these CGI promoters were more accessible in cKO hippocampi (Fig. *6 H; S8J*). For *Lef1*, the enrichment of H2AK119ub and H3K27me3 was diminished around its CGI promoter in cKO neurospheres (*Fig. S8K*). To ask whether activating the Wnt pathway could hamper hippocampal neurogenesis, we electroporated a mix of plasmids expressing Wnt ligands into E14.5 hippocampal primordia. Since *Sfrp2* is one of the most enhanced Wnt signaling components upon loss of KDM2B-CxxC, constructs overexpressing *Sfrp2* were separated transduced into E14.5 hippocampal primordia (Fig. *6 I-*<span class="InternalRef" refid="Fig6">6</span>*J*). Data showed that overexpressing Wnt ligands or *Sfrp2* could significantly block hippocampal neurogenesis, as more transduced cells resided in VZ and IZ but fewer in the pyramidal cell layer (Py) compared to controls (Fig. *6 K-*<span class="InternalRef" refid="Fig6">6</span>*L*), with significantly more transduced cell co-expressing TBR2 (Fig. *6 M*).
+
+KDM2B selectively suppresses the Wnt signaling pathway in developing hippocampus
+
+During early brain development, patterning of hippocampus is majorly controlled by Wnt and BMP signaling from hem while patterning of neocortex is regulated by signals from anti-hem, with the pallial-subpallial boundary (PSPB) being the signal center<sup><span class="CitationRef" citationid="CR40">40</span>, <span class="CitationRef" citationid="CR41">41</span></sup>. We collected hippocampal and neocortical tissues of P0 control and *Kdm2b*<sup>*Emx1−ΔCxxC*</sup> brains for RNA-seq analysis (*Fig. S9A-S9G*). First, Wnt signal genes are more abundant in the hippocampus, while PSPB genes, including *Pax6* and *Sfrp2*, are more abundant in the neocortex (*Fig. S9A*, *S9F*), suggesting that the two signal gradients persist through perinatal stage and the hem signal continues to affect hippocampal neurogenesis but exert little effect on neocortex. Secondly, both hem genes and PSPB genes were significantly up-regulated in *Kdm2b*<sup>*Emx1−ΔCxxC*</sup> hippocampi. In contrast, the change of Wnt signal genes in *Kdm2b*<sup>*Emx1−ΔCxxC*</sup> neocortex was not obvious, while PSPB signal genes in neocortex were mildly up-regulated (*Fig. S9A*, *S9B-S9E*). Hence, the silence of Wnt signaling by KDM2B is more dominant in hippocampus than in neocortex, which could explain why neocortical development is not disturbed on loss of KDM2B-CxxC.
+
+Loss of Ring1B did not cause accumulation of neural progenitors
+
+KDM2B recruits other component of PRC1.1, including the ubiquitin protein ligase Ring1B, to CGIs to initiate and stabilize gene silencing. We then asked whether the impeded migration and differentiation of neural progenitors in *Kdm2b* cKO hippocampi is caused by PRC1’s loss-of-function. We obtained *Rnf2* (the gene encoding Ring1B) cKO mice -*Rnf2*<sup>*Emx1 − cKO*</sup> – by crossing floxed *Rnf2* mice with *Emx1*-Cre mice (*Fig. S9H*). As expected, ablation of *Rnf2* greatly decreased the level of H2AK119ub in P0 neocortical tissues, with levels of H3K27me3 also slightly decreased (*Fig. S9I*). We then stained P0 brains with TBR2 (*Fig. S9J-S9K*) to find *Rnf2*<sup>*Emx1 − cKO*</sup> hippocampi were smaller than controls and the number of TBR2 + progenitors in *Rnf2*<sup>*Emx1 − cKO*</sup> hippocampi was decreased by 9.2%. Moreover, the distribution of TBR2 + progenitors in DGs, but not DNE and FDJ, was significantly decreased in the *Rnf2*<sup>*Emx1 − cKO*</sup> hippocampi. However, to our surprise, there was no accumulation and dispersion of TBR2 + IPCs in the FDJ region (2ry) of the *Rnf2*<sup>*Emx1 − cKO*</sup> hippocampi as found in *Kdm2b*<sup>*Emx1−∆CxxC*</sup> cKO brains (*Fig. S9L*). Therefore, although PRC1’s loss-of-function also causes hippocampal agenesis, it did not lead to buildup of neural progenitors in the migrating path of developing hippocampi. Transcriptome analyses showed that although Wnt pathway and neurogenesis genes were up-regulated in *Kdm2b*<sup>*Emx1−ΔCxxC*</sup> hippocampi, they were not significantly altered in *Rnf2*<sup>*Emx1 − cKO*</sup> hippocampal tissue (*Fig. S9M*). Thus, insufficient PRC1 activity alone does not account for de-repressed Wnt signaling and subsequent defects of IPC fates.
+
+Together, loss of KDM2B-CxxC reduces repressive histone modifications on key Wnt signal genes, hence leading to the block of their attenuation over time, which causes hampered differentiation and migration of hippocampal progenitors (*Graphic abstract*).
+
+# Discussion
+
+The hippocampus is evolutionarily more ancient than the neocortex<sup>42</sup> and the production and localization of CA pyramidal neurons also follows the birthdate-dependent inside-out pattern<sup>43</sup>. In neocortical development, Pax6 + RGCs and TBR2 + IPCs largely reside in the VZ and SVZ respectively, with their nuclei undergoing local oscillation. Although many cellular and epigenetic programs were found to control numbers and differentiation of neocortical RGCs and IPCs, it remains unclear how these mechanisms were applied in hippocampal development, which involves migration and dispersion of neural progenitors. Here we revealed that the chromatin association of KDM2B, an essential component of variant PRC1.1, is required for hippocampal formation. KDM2B mediates silencing of Wnt signaling genes to facilitate proper migration and differentiation of hippocampal progenitors.
+
+Our knowledge on the mammalian Polycomb repressive system has mostly come from studies in pluripotent stem cells and in embryos at early developmental stages<sup>6,15,44−46</sup>, when the establishment of repressive domains is initiated<sup>47</sup>. Nonetheless, how Polycomb controls sequential fate determination in specific tissues and at later developmental stages largely remains elusive. Both PRC1 and PRC2 are essential players in neocortical development<sup>48–52</sup>. The deletion of Ring1B, the core enzymatic component of PRC1, prolonged neocortical neurogenesis at the expense of gliogenesis. PRC1 regulates the chromatin status of neurogenic genes of neural progenitors, hence altering their responsiveness to neurogenic Wnt signals over developmental time<sup>53</sup>. Ring1B was also found to regulate dorsoventral patterning of the forebrain<sup>49</sup> and sequential production of deep and upper-layer neocortical PNs<sup>50</sup>. However, we surprisingly revealed that ablation of Ring1B did not significantly hamper migration and distribution of TBR2 + neural progenitors of developing hippocampi, whereas these defects are prominent in the *Kdm2b*<sup><em>Emx1−∆CxxC</em></sup> hippocampi. In addition, removal of KDM2B from the chromatin only have mild effects on neocortical development and overall H2AK119ub1 and H3K27me3 levels. Therefore, KDM2B likely controls fate determination of hippocampal progenitors by selectively repressing a series of progenitor genes including those in the Wnt pathway *via* PRC1.1.
+
+Knocking out EED, one of the core components of PRC2, results in hampered neurogenesis of postnatal DG. However, unlike *Kdm2b* cKO brains, the EED knockouts did not display any hippocampal malformation at P0<sup>20</sup>. Moreover, deletion of *Ezh2* in adult NSCs leads to disturbed neurogenesis of DG<sup>21</sup>, which was unseen in *Kdm2b* cKO DGs. These discrepancies echoes either distinct roles or spatiotemporal activities of PRC1 and PRC2. It would be worthy of dissecting functions of distinct PRC1/2 variants or their components in neural development and homeostasis<sup>54</sup>. A number of studies indicated that PRC1 and PRC2 can directly or indirectly regulate Wnt signaling in developmental, physiological and disease conditions<sup>55–57</sup>. Wnt signaling governs multiple aspects of neural development including neurulation, pattern formation, and fate choices of neural progenitors<sup>58–60</sup>. Moreover, the strength and gradient of the canonical Wnt signaling in the RGC-IPC-neuron path and through developmental time ensures proper cell fate establishment and transition<sup>61–63</sup>, including those in hippocampal morphogenesis<sup>4,35</sup>. Deletion of KDM2B-CxxC greatly elevated Wnt signaling in hippocampi of multiple developing stages and in hippocampal progenitors, which could lead to impeded migration and differentiation of IPCs. Of note, however, the ablation of KDM2B-CxxC has minimal effects on neocortical development, reflecting regional and timing difference between neocortical and hippocampal progenitors. For instance, at P0 stage, when hippocampal neurogenesis is at the peak while neocortical neurogenesis is turned off, Wnt signaling in hippocampus is stronger than that in neocortex (Fig. <span class="InternalRef" refid="Fig5">5</span> G–<span class="InternalRef" refid="Fig5">5</span> H, S9A). Sustained Wnt activation on loss of KDM2B hence could have greater impact on hippocampal neurogenesis compared to neocortical development.
+
+The deletion of KDM2B-CxxC would totally abolish KDM2B’s association with CGIs, thus disabling KDM2B’s two major functions on chromatin - mediating H2AK119Ub *via* PRC1.1 and demethylating H3K36me2, the latter of which is solely executed by the JmjC-containing KDM2BLF. Previous studies indicated that the demethylase activity of KDM2A/B is required for PRC establishment at CGIs of peri-implantation embryos<sup>24</sup>, but contributes moderately to the H3K36me2 state at CGI-associated promoters and is dispensable for normal gene expression in mouse embryonic stem cells<sup>64</sup>. Interestingly, levels of H3K36me2 were not significantly altered globally or locally in KDM2B-CxxC deleted hippocampi. Moreover, *Kdm2blf*<sup><em>KO</em></sup> mice did not display hippocampal agenesis or malformation<sup>29,65</sup>. Thus, the H3K36me2 demethylase activity of KDM2B is likely dispensable for hippocampal development.
+
+*KDM2B* is implicated in neurological disorders including ID and behavior abnormalities, and the region encoding the CxxC ZF is the mutational hotspot<sup>27</sup>. Consistently, *Kdm2b*<sup><em>Emx1−∆CxxC</em></sup> cKO mice displayed prominent defects in spatial and motor learning and memory, as well as contextual fear conditioning. It would be essential to explore whether patients with *KDM2B* mutations have defects of hippocampal morphogenesis and function, and how KDM2B mediated gene repression is implicated in human brain development.
+
+# MATERIALS AND METHODS
+
+## Mice and genotyping
+
+All animal procedures were approved by the Animal Care and Ethical Committee of Wuhan University. Wild-type CD-1 (ICR) and C57BL/6 mice were obtained from the Hunan SJA Laboratory Animal Company (Changsha, China). Mice were housed in a certified specific-pathogen-free (SPF) facility. The noon of the day when the vaginal plug was found was counted as embryo (E) day 0.5.
+
+Mice with conditional deletion of *Kdm2b-CxxC* were obtained by first crossing *Kdm2b*<sup><em>fl/fl</em></sup> (generated by Applied Stem Cell) females with *Emx1-Cre* (Jackson Laboratories, stock number 005628), *Nestin-Cre* (Jackson Laboratories, stock number 003771) or *Nex-Cre* males [<em>Neurod6</em><sup><em>tm1(cre)Kan</em></sup>, MGI:2668659]. *Emx1-Cre; Kdm2b*<sup><em>fl/+</em></sup>, *Nestin-Cre; Kdm2b*<sup><em>fl/+</em></sup> or *Nex-Cre; Kdm2b*<sup><em>fl/+</em></sup> males were crossed with *Kdm2b*<sup><em>flfl</em></sup> females to obtain conditional knockout mice (*Kdm2b*<sup><em>Emx1−ΔCxxC</em></sup>, *Kdm2b*<sup><em>Nestin−ΔCxxC</em></sup>, *Kdm2b*<sup><em>Nex−ΔCxxC</em></sup>). *Kdm2b*<sup><em>fl/+</em></sup> and *Kdm2b*<sup><em>fl/fl</em></sup> were phenotypically indistinguishable from each other, and used as controls. The primer set forward 5’- cctgtagtccttggtatttcctggc-3’/reverse 5’ - cccaacttgcccttaggccg-3’ was used for mice genotyping, and band sizes for *Kdm2b*<sup><em>fl/+</em></sup> mice are 364 bp (WT allele) and 404 bp (targeted allele with 5’ loxP). Forward 5’- cctgttacgtatagccgaaa-3’/reverse 5’- cttagcgccgtaaatcaatc-3’ was used for *Emx1-Cre*, *Nestin-Cre* and *Nex-Cre* genotyping with band size 319 bp (Cre allele).
+
+To analyse adult neurogenesis, the *Nestin-CreERT2* (Jackson Laboratories, stock number 016261) and *Kdm2b*<sup><em>fl/fl</em></sup> mice were crossed to generate *Nestin-CreERT2; Kdm2b*<sup><em>fl/+</em></sup> animals. *Nestin-CreERT2; Kdm2b*<sup><em>fl/+</em></sup> mice were further crossed with *Kdm2b*<sup><em>fl/fl</em></sup> mice to obtain homozygous *Kdm2b*<sup><em>NestinCreERT2−ΔCxxC</em></sup> animals, which were used for the experiment. Forward 5’- gaccaggttcgttcactca-3’/reverse 5’- caagttaggagcaaacagtagc-3’ was used for *Nestin-CreERT2* genotyping with band size 993 bp (CreERT2 allele).
+
+To verify the activity of *Nestin-CreERT2* in hippocampal SGZ and the expression profile of *Nestin-Cre* in developing hippocampus, we constructed *Nestin-CreERT2;Ai14* (Rosa-CAG-LSL-tdTomato-WPRE) and *Nestin-Cre;Ai14* mice. The primer sets forward 5’- aagggagctgcagtggagta-3’/reverse 5’ - ccgaaaatctgtgggaagtc-3’ and forward 5’- ggcattaaagcagcgtatcc-3’/reverse 5’ - ctgttcctgtacggcatgg-3’ were used for *Ai14* mice genotyping, and band sizes for *Ai14+/-* mice are 297 bp (WT allele) and 196 bp (*Ai14* allele).
+
+BAT-Gal mice were kind gifts from Dr. Junlei Chang (Jackson Lab, stock number 005317). To explore the Wnt signaling pathway in *Kdm2b*<sup><em>Emx1−ΔCxxC</em></sup> mice, the BAT-Gal and *Kdm2b*<sup><em>fl/fl</em></sup> mice were crossed to generate *Kdm2b*<sup><em>fl/+</em></sup>; BAT-Gal animals. *Emx1-Cre; Kdm2b*<sup><em>fl/+</em></sup> mice were further crossed with *Kdm2b*<sup><em>fl/+</em></sup>; BAT-Gal mice to obtain *Kdm2b*<sup><em>Emx1−ΔCxxC</em></sup>; BAT-Gal animals, which were used for the experiment. The primer set forward 5′-atcctctgcatggtcaggtc-3′/reverse 5′-cgtggcctgattcattcc-3′ was used for BAT-Gal mice with band size 315 bp (LacZ allele).
+
+Mice with conditional deletion of *Rnf2* were obtained by first crossing *Rnf2*<sup><em>fl/fl</em></sup> (purchased from GemPharmatech, Strain NO. T014803) females with *Emx1-Cre* males (Jackson Laboratories, stock number 005628). The primer set forward 5’- agctgtggtcctgcgtttcatttc-3’/reverse 5’ - gctcttactgtgttacaaccctagccc-3’ was used for *Rnf2*<sup><em>fl/+</em></sup> mice genotyping, and band sizes for *Rnf2*<sup><em>fl/+</em></sup> mice are 289 bp (WT allele) and 391 bp (targeted allele with 5’ loxP).
+
+## Tamoxifen and BrdU administration
+
+To activate Cre-mediated recombination, tamoxifen (TAM; Sigma-Aldrich) was used, which was made fresh daily and dissolved in sunflower oil solution (Sigma-Aldrich). 8-week-old *Kdm2b*<sup><em>NestinCreERT2−ΔCxxC</em></sup> mice were daily administered with 30 mg/kg prewarmed TAM intraperitoneally for 6 consecutive days (d1-d6). From day 2 to day 7, mice were injected with 50 mg/kg BrdU (Sigma-Aldrich) intraperitoneally for 6 consecutive days and were sacrificed 1 day later (day 8, Short-Term) or 4 weeks later (day 35, Long-Term) to identify BrdU-positive adult-born cells.
+
+## Tissue fixation and sectioning
+
+The pregnant dam was anesthetized with 0.7% w/v pentobarbital sodium (105 mg/kg body weight) in 0.9% sodium chloride. Embryos were sequentially removed from the uterus. Brains of embryos were dissected out in cold PBS and immersed in 4% paraformaldehyde (PFA) overnight at 4°C. For P0, P2, P7 and adult mice, animals were anesthetized with 0.7% w/v pentobarbital sodium solution followed by trans-cardiac perfusion with 4% PFA in PBS (P0, 5 ml; P7, 10 ml; adult, 30 ml). Brains were dissected and post-fixed in 4% PFA overnight at 4°C. Next day, brains were dehydrated in 20% w/v sucrose overnight at 4°C. For sectioning, brains were embedded in OCT (SAKURA) and cut at 20 µm for adult brains and 14 µm for other stages with a cryostat (Leica CM1950).
+
+## Nissl staining
+
+Adult brain sections were stained with 0.25% Cresyl Violet (Sigma-Aldrich) solution for 15 min at 65°C. Sections were then decolorized in ethanol for 0.5-1 min, dehydrated in gradient ethanol solutions for 5 min each and cleared twice in xylene for 5 min. Sections were mounted in the neutral balsam.
+
+## In situ hybridization (ISH)
+
+Sections were dried in a hybridization oven at 50°C for 15 min and fixed in 4% PFA for 20 min at room temperature, followed by permeabilization in 2 µg/ml proteinase K in PBS for 10 min at room temperature. Prior to hybridization, sections were acylated in 0.25% acetic anhydride for 10 min. Then, sections were incubated with a digoxigenin-labeled probe diluted (0.2 ng/µl) in hybridization buffer (50% deionized formamide, 5× SSC, 5× Denhart’s, 250 µg/ml tRNA, and 500 µg/ml Herring sperm DNA) under coverslips in a hybridization oven overnight at 65°C. The next day, sections were washed 4 times for 80 min in 0.1× SSC at 65°C. Subsequently, they were treated with 20 µg/ml RNase A for 20 min at 37°C, then blocked for 3.5 h at room temperature in 10% normal sheep serum. Slides were incubated with 1:5,000 dilution of anti-digoxigenin-AP conjugated antibody (Roche) overnight at 4°C. BCIP/NBT (Roche) was used as a color developing agent. ISH primers used are listed in Table S1.
+
+## Immunofluorescence
+
+Frozen brain sections were mounted onto Superfrost plus slides and then dried at room temperature. For heat-mediated antigen retrieval, slides were incubated for 15 min in 10 mM sodium citrate buffer (pH 6.0) at 95°C. For BrdU staining, sections were treated with 20 µg/ml proteinase K (Sigma) (1:1000 in PBC) for 5 min and 2 N HCl for 30 min at room temperature. Sections were then immersed in blocking buffer (3% normal sheep serum and 0.1% Triton X-100 in PBS; or 5% BSA and 0.5% Triton X-100 in PBS) for 2 h at room temperature. Sections were then incubated in primary antibodies [mouse anti-Calbindin (1:1000; Sigma, C9848), rabbit anti-ZBTB20 (1:1000; Sigma, HPA016815), mouse anti-HopX (1:200; Santa Cruz, sc-398703), rabbit anti-Wfs1 (1:1000; Proteintech, 86995), mouse anti-PROX1 (1:200; Millipore, MAB5654), rabbit anti-GFAP (1:500; DAKO, Z0334), rat anti-CTIP2 (1:500; Abcam, ab18465), rabbit anti-SATB2 (1:500; Abcam, ab92446), rat anti-BrdU (1:500; Abcam, ab6326), mouse-anti-BrdU (1:500; Roche, 11170376001), rabbit-anti-DCX (1:500; Abcam, ab18723), rabbit anti-TBR2 (1:500; Abcam, ab23345), rat anti-TBR2 (1:500; Thermo Fisher, 14-4875-82), rabbit anti-PAX6 (1:500; Millipore, ab2237), and rabbit anti-Ki67 (1:500; Abcam, ab15580] in blocking buffer overnight at 4°C. After three rinses in PBS, sections were incubated in secondary antibodies (Alexa Fluor 488-conjugated anti-mouse, A11029; Alexa Fluor 555-conjugated anti-mouse, A21422; Alexa Fluor 488-conjugated anti-rat, A11006; Alexa Fluor 555-conjugated anti-rat, A21434; Alexa Fluor 647-conjugated anti-rat, A21247; Alexa Fluor 488-conjugated anti-rabbit, A11034; Alexa Fluor 555-conjugated anti-rabbit, A21429; Alexa Fluor 647-conjugated anti-rabbit, A21245; Alexa Fluor 488-conjugated anti-chicken, A11039; Thermo Fisher Scientific; 1:1000) for 1 h at room temperature. Nuclei were labeled by incubation in PBS containing 4′,6-diamidino-2-phenylindole (DAPI) (0.1 µg/ml) (Sigma-Aldrich), and samples were mounted in ProLong Gold Antifade Mountant (Thermo Fisher Scientific).
+
+## 5-Ethynyl-2′-Deoxyuridine (EdU) staining
+
+Proliferation of cells was investigated with BeyoClick™ EdU Cell Proliferation Kit (C0075S, Beyotime, China) according to the manufacturer’s protocols. In brief, frozen brain sections were dried at temperature and permeated with 0.3% Triton X-100 in PBS for 30 min. Sections were then incubated with EdU working solution for 1 h at 37°C in the dark. After incubation, regular immunofluorescence staining can be followed.
+
+## Immunohistochemical staining
+
+Frozen brain sections were dried at room temperature, and then pretreated with 0.3% H<sub>2</sub>O<sub>2</sub> for 15 min to deactivate endogenous peroxidase. Sections were blocked with 3% normal sheep serum with 0.1% Tween 20 at room temperature for 2 h. Sections were then incubated in primary antibodies [rabbit anti-NeuN (1:500; Abcam, ab177487), rabbit-anti-DCX (1:500; Abcam, ab18723), rabbit anti-SOX2 (1:500; Millipore, ab5603), rabbit anti-TBR2 (1:500; Abcam, ab23345), and rabbit anti-BLBP (1:500; Abcam, ab32423)] in blocking buffer overnight at 4°C, followed by addition of the avidin-biotin-peroxidase complex (1:50; VECTASTAIN Elite ABC system, Vector Laboratories). Peroxidase was reacted in 3,3′-diaminobenzidine (5 mg/ml) and 0.075% H<sub>2</sub>O<sub>2</sub> in Tris-HCl (pH 7.2). Sections were dehydrated in gradient ethanol (75% ethanol, 95% ethanol, 100% ethanol and 100% ethanol, each for 5 min), and cleared twice in xylene for 5 min, then mounted in the neutral balsam.
+
+## Behavior tests
+
+We used 12- to 16-week-old age-matched male mice for all behavioral tests. Mice were housed (3–5 animals per cage) in standard filter-top cages with access to water and rodent chow at all times, maintained on a 12:12 h light/dark cycle (09:00–21:00 h lighting) at 22°C, with relative humidity of 50–60%. All behavioral assays were done blind to genotypes.
+
+*Open field test.* The test mouse was gently placed near the wall-side of a length of 50 cm, a width of 50 cm, and a height of 50 cm open-field arena and allowed to explore freely for 20 min. Only the last 10 min of the movement of the mouse was recorded by a video camera and analyzed with Ethovision XT 13 (Noldus).
+
+*Rotarod test.* The test consists of 4 trials per day for 10 days with the rotarod (3 cm in diameter) set to accelerate from 4 rpm to 40 rpm over 5 minutes. The trial started once mice were placed on the rotarod rotating at 4 rpm in partitioned compartments. The time for each mouse spent on the rotarod were recorded. At least 20 min recovery time was allowed between trials. The rotarod apparatus was cleaned with 70% ethanol and wiped with paper towels between each trial.
+
+*Morris water maze.* Mice were introduced into a stainless water-filled circular tank, which is 122 cm in diameter and 51 cm in height with non-reflective interior surfaces and ample visual cues. Two principal axes were drawn on the floor of the tank, each line bisecting the maze perpendicular to one another to create an imaginary ‘+’. The end of each line demarcates four cardinal points: North, South, East and West. To enhance the signal-to-noise ratio, the tank was filled with water colored with powdered milk. A 10-cm circular plexiglass platform was submerged 1 cm below the surface of the water in middle of the southwest quadrant. Mice started the task at fixed points, varied by day of testing<sup><span citationid="CR66" class="CitationRef">66</span></sup>. Four trials were performed per mouse per day with 20 min intervals for 5 days. Each trial lasted 1 min and ended when the mouse climbed onto and remained on the hidden platform for 10 s. The mouse was given 20 s rest on the platform during the inter-trial interval. The time taken by the mouse to reach the platform was recorded as its latency. Times for four trials were averaged and recorded as a result for each mouse. On day 6, the mouse was subjected to a single 60-s probe trial without a platform to test memory retention. The mouse started the trial from northeast, the number of platform crossings was counted, and the swimming path was recorded and analyzed using the Ethovision XT 13 (Noldus).
+
+*Fear conditioning (FC).* The FC apparatus consisted of a conditioning box (18 × 18 × 30 cm), with a grid floor wired to a shock generator surrounded by an acoustic chamber and controlled by Ethovision XT 13 (Noldus). On Day 1, each mouse was placed in the conditioning box for 2 min, and then a pure tone (80 db) was sounded for 30 s followed by a 2 s foot shock (0.4 mA). Two minutes later, this procedure was repeated. After the delivery of the second shock, mice were returned to their home cages. On Day 2, each mouse was first placed in the fear conditioning chamber containing the exact same context, but there was no administration of a tone or foot shock. Freezing was analyzed for 4 min. One hour later, the mice were placed in a new context (containing a different odor, cleaning solution, floor texture, walls and shape) where they were allowed to explore for 3 min before being re-exposed to the fear conditioning tone and freezing was assessed for an additional 3 min. FC was assessed through the continuous measurement of freezing (complete immobility), which is the dominant behavioral fear response. Freezing was measured using the Noldus Ethovision video tracking system (Ethovision XT 13).
+
+*Forced swimming test.* For the forced swimming test, the test mouse was placed into a 20 cm height and 17 cm diameter glass cylinder filled with water to a depth of 10 cm at 22°C. The test continues 6 min and the immobility time of the last 5 minutes was recorded for further processing.
+
+*Tail suspension test.* The test mouse was suspended in the middle of a tail suspension box (55 cm height × 60 cm width × 11.5 cm depth) above the ground by its tail. The mouse tail was adhered securely to the suspension bar using adhesive tapes. After 1 min accommodation, the immobility time was recorded by a video camera and analyzed by Ethovision XT 13 (Noldus).
+
+*Elevated plus maze test.* The elevated plus maze, made of gray polypropylene and elevated about 40 cm above the ground, consists of two open arms and two closed arms (each 9.5 cm wide and 40 cm long). To assess anxiety, the test mouse was placed in the central square facing an open arm and allowed to explore freely for 5 min. The time spent in the open arm was analyzed with the Ethovision XT 13 (Noldus).
+
+## X-Gal staining
+
+Frozen sections were fixed in fresh cold fixative (0.2% PFA) in buffer L0 (0.1M PIPES buffer (pH 6.9), 2mM MgCl<sub>2</sub>, 5mM EGTA) for 10 min. Slides were rinsed in PBS plus 2mM MgCl<sub>2</sub> on ice, followed by a 10 min wash in the same solution. Place slides in detergent rinse [0.1M PBS (pH 7.3), 2mM MgCl<sub>2</sub>, 0.01% sodium deoxycholate, 0.02% Nonidet P-40] on ice for 10 min. Slides were then moved to a freshly made and filtered X-Gal staining solution [0.1M PBS (pH 7.3), 2mM MgCl<sub>2</sub>, 0.01% sodium deoxycholate, 0.02% Nonidet P-40, 5mM K<sub>3</sub>Fe(CN)<sub>6</sub>, 5mM K<sub>4</sub>Fe(CN)<sub>6</sub>·3H<sub>2</sub>O and 1 mg/ml X-Gal]. Sections were incubated at 37°C from a few minutes to overnight in the dark. Sections were rinsed with water to stop the reaction. Sections were dehydrated with gradient ethanol and xylene sequentially, and mounted with the neutral balsam.
+
+## RNA isolation and reverse transcription (RT)
+
+RNA isolation was performed using the RNAiso Plus (TAKARA) according to manufacturer’s instructions. Tissue or cells were homogenized using a glass-Teflon in 1 ml or 500 µl RNAiso Plus reagent on ice and phase separation was achieved with 200 µl or 100 µl chloroform. After centrifugation at 12,000× g for 15 min at 4°C, RNA was precipitated by mixing aqueous phase with equal volumes of isopropyl alcohol and 0.5 µl 20 mg/ml glycogen. Precipitations were dissolved in DNase/RNase-free water (not diethylpyrocarbonate treated, Ambion). 1 µg of total RNA was converted to cDNA using M-MLV reverse transcriptase (TAKARA) under standard conditions with oligo(dT) or random hexamer primers and Recombinant RNase Inhibitor (RRI, TAKARA). Then the cDNA was subjected to quantitative RT-PCR (qRT-PCR) using the SYBR green assay with 2× SYBR green qPCR master mix (Bimake). Thermal profile was 95°C for 5 min and 40 cycles of 95°C for 15 sec and 60°C for 20 sec. *Gapdh* was used as endogenous control genes. Relative expression level for target genes was normalized by the Ct value of *Gapdh* using a 2<sup>−ΔΔCt</sup> relative quantification method. Reactions were run on a CFX Connect™ Real-Time PCR Detection System (Bio-Rad). The primers used are listed in Table S2.
+
+## Neurosphere culture
+
+Mouse hippocampal neural progenitor cells (NPCs) were enriched from P0 mouse hippocampi, cultured on ultra-low-attachment plates (Corning, New York, United States) and maintained in indicated culture media (DMEM/F12, Life Technologies) containing N2 and B27 supplements (1×, Life Technologies), 1mM Na-pyruvate, 1 mM N-acetyl-L-cysteine (NAC), human recombinant FGF2, and EGF (20 ng/mL each; Life Technologies). After cultured *in vitro* for three generations, neurospheres were subjected to RNA-seq and ChIP-seq analyses.
+
+## RNA-seq library construction
+
+Total RNA was extracted as described above. The concentration and quality of RNA was measured with Nanodrop 2000c (Thermo Fisher Scientific) and an Agilent 2100 Bioanalyzer (Agilent Technologies), respectively. RNA-seq libraries were constructed by NEBNext® Ultra™ II RNA Library Prep Kit for Illumina® (NEB #E7775). Briefly, mRNA was extracted by poly-A selected with magnetic beads with poly-T and transformed into cDNA by first and second strand synthesis. Newly synthesized cDNA was purified by AMPure XP beads (1:1) and eluted in 50 µl nucleotide-free water. RNA-seq libraries were sequenced by Illumina NovaSeq 6000 platform with pair-end reads of 150 bp. The sequencing depth was 60 million reads per library.
+
+## Bulk RNA-Seq data analysis
+
+P0 hippocampus and neurosphere RNA-seq data were checked for quality control by FastQC (version 0.11.9). Paired-end reads were trimmed to remove adaptors and low-quality reads and bases using cutadapt (version 3.2). Clean reads were aligned to the mouse UCSC mm10 genome using STAR (version 2.7.10b) with default parameters. The number of covering reads were counted using featureCounts (version v2.0.1). The resulting read counts were processed with R package DESeq2 (version 1.38.1) to identify differential expression genes (log2 fold change > 0.4 and p value < 0.05) between datasets. Cufflinks package (version 2.2.1) assembles individual transcripts from reads that have been aligned to reference genome. The gene expression level was normalized by fragments per kilobase of bin per million mapped reads (FPKM). Gene Ontology (GO) analysis and Gene Set Enrichment Analysis (GSEA) in this study were performed using clusterProfiler (version 4.2.2).
+
+## Chromatin immunoprecipitation (ChIP) assay
+
+For each experiment, single-cell suspensions from P0 hippocampi were collected as described above. The hippocampal tissue was digested into single cells by Papain (20U dissolved in each mL of DMEM/F12 medium, preheated at 37°). Cells were cross-linked with 1% formaldehyde for 10 minutes at room temperature, and quenched with 0.125 M of glycine for 5 minutes. Cross-linked samples were then rinsed twice in PBS, then cells were collected by centrifugation. Next, cells were pretreated with lysis buffer (50 mM of Tris-HCl [pH 8.0], 0.1% SDS, and 5 mM of EDTA) and incubated for 5 minutes with gentle rotation at 4°C. After centrifugation, bottom cells were washed for two times with ice-cold PBS and harvested in ChIP digestion buffer (50 mM of Tris-HCl [pH 8.0], 1 mM of CaCl<sub>2</sub>, and 0.2% Triton X-100). DNA was digested to 150–300 bp by micrococcal nuclease (NEB; M0247S). Sonicate cells in EP tubes with power output 100 W, 5 min, 0.5 s on, 0.5 s off on ice. The resulting lysate was centrifugation and divided into four parts for 10% input, H2AK119Ub, H3K27me3, and H3K36me2 immunoprecipitation. After diluting each sample (in addition to input) to 1 mL with dilution buffer (20 mM of Tris-HCl [pH 8.0], 150 mM of NaCl, 2 mM of EDTA, 1% Triton X-100, and 0.1% SDS), immunoprecipitation was further performed with sheared chromatin and 3 µg rabbit anti-H2AK119Ub antibody (CST, 8240S); or rabbit anti-H3K27me3 antibody (CST, 9733S); or rabbit anti-H3K36me2 antibody (CST, 2901S), then incubated with protein A/G beads overnight at 4°C on a rotating wheel. The next day, beads were wash with Wash Buffer I (20mM Tris-HCl, pH 8.0; 1% Triton X-100; 2mM EDTA; 150mM NaCl; 0.1% SDS), Wash Buffer II (20mM Tris-HCl, pH 8.0; 1% Triton X-100; 2mM EDTA; 500mM NaCl; 0.1% SDS), Wash Buffer III (10mM Tris-HCl, pH 8.0; 1mM EDTA; 0.25M LiCl; 1% NP-40; 1% deoxycholate) and TE buffer. DNA was eluted by ChIP elution buffer (0.1 M of NaHCO<sub>3</sub>, 1% SDS, 20 µg/mL of proteinase K). The elution was incubated at 65°C overnight, and DNA was extracted with a DNA purification kit (DP214-03; TIANGEN).
+
+## ChIP-seq library construction
+
+ChIP-seq libraries were constructed by VAHTS Universal DNA Library Prep Kit for Illumina V3 (Vazyme ND607). Briefly, 50 µL purified ChIP DNA (5 ng) was end-repaired for dA tailing, followed by adaptor ligation. Each adaptor was marked with a barcode of 6 bp which can be recognized after mixing different samples together. Adaptor-ligated ChIP DNA was purified by VAHTS DNA Clean Beads (Vazyme N411) and then amplified by PCR of 10 cycles with primers matching with adaptors’ universal part. Amplified ChIP DNA was purified again using VAHTS DNA Clean Beads in 35-µL EB elution buffer. For multiplexing, libraries with different barcode were mixed together with equal molar quantities by considering appropriate sequencing depth (about 30 million reads per library). Libraries were sequenced by Illumina Nova-seq 6000 platform with pair-end reads of 150 bp.
+
+## ChIP-seq data analysis
+
+DNA libraries were sequenced on Illumina NovaSeq 6000 platform. All P0 hippocampus and neurosphere ChIP-seq data were checked and removed with adaptor sequences same as the RNA-seq data processing. Clean reads were aligned to the mouse UCSC mm10 genome using Bowtie2 (version 2.4.5). Duplicates were removed using the samtools rmdup module. Regions of peaks were called using the SICER software package, with the input genomic DNA as a background control (parameters: -w 200 -rt 1 -f 150 -egf 0.77 -fdr 0.01 -g 600 -e 1000 --significant_reads). The bigwig signal files were visualized using the computeMatrix, plotHeatmap, plotProfile modules in Deeptools (version 3.5.1). Homer was used to identify adjacent genes from the peaks obtained from SICER.
+
+## ATAC-seq library construction
+
+ATAC-seq libraries were constructed by TruePrep DNA Library Prep Kit V2 for Illumina (Vazyme TD501). Briefly, P0 hippocampi were dissected and gently homogenized in cold nuclear isolation buffer (10 mM Tris-HCl, pH 7.4, 10 mM NaCl, 3 mM MgCl<sub>2</sub>, 0.1% Igepal CA-630). Nuclei were collected by centrifugal precipitation. 50,000 nuclei were put into the tagmentation reaction for each sample (performed with 30 min incubation time at 37°C). Immediately following the tagmentation, DNA fragments were purified using VAHTS DNA Clean Beads (2X). Purified DNA fragments were added with Illumina i5 + i7 adapters with unique index to individual samples followed by PCR reaction (PCR program: 72°C for 3 min, 98°C for 30 s, 98°C for 15 s, 60°C for 30 s, 72°C for 30 s, repeat 3–5 for 13 cycles, 72°C for 5 min, and hold at 4°C). Generated libraries were purified using VAHTS DNA Clean Beads (1.2X). For multiplexing, libraries with different barcode were mixed together with equal molar quantities by considering appropriate sequencing depth (about 50 million reads per library). Libraries were sequenced by Illumina Nova-seq 6000 platform with pair-end reads of 150 bp.
+
+## ATAC-seq data analysis
+
+P0 hippocampus ATAC-seq raw data were trimmed by Cutadapt with parameters -u 3 -u -75 -U 3 -U -75 -m 30 and then aligned to mouse mm10 genome using Bowtie2 (-X 2000 --very-sensitive). Subsequently, we downloaded blacklisted regions including a large number of repeat elements in the genome from ENCODE project and then removed these significant background noise. ATAC-seq datasets contained a large percentage of reads that were derived from mitochondrial DNA. We removed mitochondrial reads after alignment using Samtools. Then, we filtered reads to remove exact copies of DNA fragments that arise during PCR using Picard’s MarkDuplicates (version 2.26.4). All reads aligning to the + strand were offset by + 4 bp, and all reads aligning to the – strand were offset − 5 bp, since Tn5 transposase has been shown to bind as a dimer and insert two adaptors separated by 9 bp. We adjusted the shift read alignment using alignmentSieve. Next, peaks calling was finished by Macs2 (version 2.2.7.1) with parameters -f BAMPE --nomodel --keep-dup all --shift − 100 --extsize 200 -g mm --cutoff-analysis -B. We created bigwig files for visualizing using bamCoverage (parameters: --normalizeUsing RPGC –effectiveGenomeSize 2407883318) in deepTools. Homer took narrow peak files as input and checked for the enrichment of both known sequence motifs and de novo motifs.
+
+## Defining genomic features
+
+Mm10 CpG island (CGI) regions were downloaded from the UCSC genome browser database. Promoters were defined as all mouse UCSC mm10 gene TSSs, extended by 2 kb upstream and downstream. CGIs in promoters were defined as ± 2 kb around CGI centers overlap with promoter regions. Overlapped regions between CGIs and promoters were identified using bedtools (version 2.29.2) intersect with parameters -e -f 0.5 -F 0.5. H2AK119ub1 +/- promoter genomic loci were defined as such promoter location with or without H2AK119ub1.
+
+## In utero electroporation (IUE)
+
+*In utero* microinjection and electroporation were performed as followed. Pregnant CD-1 mice with E14.5 embryos were anesthetized by injection of pentobarbital sodium (70 mg/kg), and the uteri were exposed through a 2 cm midline abdominal incision. Embryos were carefully pulled out using ring forceps through the incision and placed on sterile gauze wet with 0.9% sodium chloride. Plasmid DNA (prepared using Endo Free plasmid purification kit, Tiangen) mixed with 0.05% Fast Green (Sigma) was injected through the uterine wall into the telencephalic vesicle using pulled borosilicate needles (WPI). For gain-of-function experiments, pCIG (1 µg/µl) was mixed with pCAGGS-Wnt3a, pCAGGS-Wnt5a, pCAGGS-Wnt5b, pCAGGS-Wnt7b, pCAGGS-Wnt8b, (Wnt-mix) (0.5 µg/µl each), or with pCAGGS-SFRP2 (2 µg/µl). Control mice were injected with pCIG (1 µg/µl). Five electric pulses (33 V, 50 ms duration at 1 s intervals) were generated using CUY21VIVO-SQ (BEX) and delivered across the head of embryos using 5 mm forceps-like electrodes (BEX). The uteri were then carefully put back into the abdominal cavity, and both peritoneum and abdominal skin were sewed with surgical sutures. The whole procedure was completed within 30 min. Mice were warmed on a heating pad until they regained consciousness and were treated with analgesia (ibuprofen in drinking water) until sacrifice at E18.5.
+
+## Plasmid construction
+
+Full-length mouse *Wnt3a* and *Wnt8b* were amplified from cDNAs of E14.5 mouse hippocampi and then cloned into pCAGGS. Full-length *Wnt5a*, *Wnt5b*, *Wnt7b* were amplified from cDNAs of E16.5 mouse cortex, and then cloned into pCAGGS. Full-length mouse *Sfrp2* was amplified from cDNAs of P0 mouse cortex and then cloned into pCAGGS. The primers used are listed in Table S3.
+
+## Quantification and statistical analysis
+
+Sections used for quantification were position-matched for control and experimental brains. Three corresponding brain cross sections of each control and cKO samples were selected for cell count statistics and the total number of these three cross sections was counted. Images were binned against proximal-distal transverse axis to quantify the intensity of ISH or LacZ signals of hippocampi. A plot of normalized average signal intensity with standard error of the mean across those regions was generated using ImageJ.
+
+Statistical tests were performed using GraphPad Prism (version 8.0.2). Data analyzed by unpaired two-tailed t-test were pre-tested for equal variance by F-tests. Unpaired Student’s t-tests (two-tailed) were chosen when the data distributed with equal variance. For normally distributed data with unequal variance, an unpaired t-test with Welch’s correction was used. One-way ANOVA followed by Tukey post hoc test was used for multiple group comparison. Significant difference is indicated by a p value less than 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). No statistical methods were used to pre-determine sample sizes but our sample sizes are similar to those reported in previous publications. Experiments were not randomized. Investigators were blinded as to the animal genotype during tissue section staining, image acquisition, and image analysis.
+
+## Data and code availability
+
+The GEO accession number for the RNA-seq, ChIP-seq and ATAC-seq data reported in this paper is GSE222465. RNA-seq data of P0 hippocampus, P0 neurosphere and P0 cortex have been deposited at GEO: GSE222464. ChIP-seq and ATAC-seq data of P0 hippocampus and P0 neurosphere in this study have been deposited at GEO: GSE222463 and GSE222462. All data are publicly available as of the date of publication. Custom codes were described in detail at methods part. Any additional information required to analyze the data in this paper is available from authors upon reasonable request.
+
+# References
+
+1. Anacker, C. & Hen, R. Adult hippocampal neurogenesis and cognitive flexibility - linking memory and mood. Nat Rev Neurosci 18, 335–346, doi: 10.1038/nrn.2017.45 (2017).
+2. Li, G. & Pleasure, S. J. The development of hippocampal cellular assemblies. Wiley Interdiscip Rev Dev Biol 3, 165–177, doi: 10.1002/wdev.127 (2014).
+3. Nelson, B. R. et al. Intermediate progenitors support migration of neural stem cells into dentate gyrus outer neurogenic niches. eLife 9, e53777, doi: 10.7554/eLife.53777 (2020).
+4. Zhong, S. et al. Decoding the development of the human hippocampus. Nature 577, 531–536, doi: 10.1038/s41586-019-1917-5 (2020).
+5. Blackledge, N. P. & Klose, R. J. The molecular principles of gene regulation by Polycomb repressive complexes. Nat Rev Mol Cell Biol 22, 815–833, doi: 10.1038/s41580-021-00398-y (2021).
+6. Laugesen, A. & Helin, K. Chromatin repressive complexes in stem cells, development, and cancer. Cell Stem Cell 14, 735–751, doi: 10.1016/j.stem.2014.05.006 (2014).
+7. von Schimmelmann, M. et al. Polycomb repressive complex 2 (PRC2) silences genes responsible for neurodegeneration. Nat Neurosci 19, 1321–1330, doi: 10.1038/nn.4360 (2016).
+8. Piunti, A. & Shilatifard, A. Epigenetic balance of gene expression by Polycomb and COMPASS families. Science 352, aad9780, doi: 10.1126/science.aad9780 (2016).
+9. Schuettengruber, B. & Cavalli, G. Recruitment of polycomb group complexes and their role in the dynamic regulation of cell fate choice. Development 136, 3531–3542, doi: 10.1242/dev.033902 (2009).
+10. He, J. et al. Kdm2b maintains murine embryonic stem cell status by recruiting PRC1 complex to CpG islands of developmental genes. Nat Cell Biol 15, 373–384, doi: 10.1038/ncb2702 (2013).
+11. Wu, X., Johansen, J. V. & Helin, K. Fbxl10/Kdm2b Recruits Polycomb Repressive Complex 1 to CpG Islands and Regulates H2A Ubiquitylation. Molecular cell, doi: 10.1016/j.molcel.2013.01.016 (2013).
+12. Farcas, A. M. et al. KDM2B links the Polycomb Repressive Complex 1 (PRC1) to recognition of CpG islands. Elife 1, e00205, doi: 10.7554/eLife.00205 (2012).
+13. Gao, Z. et al. PCGF homologs, CBX proteins, and RYBP define functionally distinct PRC1 family complexes. Mol Cell 45, 344–356, doi: 10.1016/j.molcel.2012.01.002 (2012).
+14. Fursova, N. A. et al. Synergy between Variant PRC1 Complexes Defines Polycomb-Mediated Gene Repression. Mol Cell 74, 1020–1036 e1028, doi: 10.1016/j.molcel.2019.03.024 (2019).
+15. Sugishita, H. et al. Variant PCGF1-PRC1 links PRC2 recruitment with differentiation-associated transcriptional inactivation at target genes. Nat Commun 12, 5341, doi: 10.1038/s41467-021-24894-z (2021).
+16. Blackledge, N. P. et al. Variant PRC1 complex-dependent H2A ubiquitylation drives PRC2 recruitment and polycomb domain formation. Cell 157, 1445–1459, doi: 10.1016/j.cell.2014.05.004 (2014).
+17. Kalb, R. et al. Histone H2A monoubiquitination promotes histone H3 methylation in Polycomb repression. Nature structural & molecular biology 21, 569–571, doi: 10.1038/nsmb.2833 (2014).
+18. Tamburri, S. et al. Histone H2AK119 Mono-Ubiquitination Is Essential for Polycomb-Mediated Transcriptional Repression. Mol Cell 77, 840–856 e845, doi: 10.1016/j.molcel.2019.11.021 (2020).
+19. Margueron, R. et al. Role of the polycomb protein EED in the propagation of repressive histone marks. Nature 461, 762–767, doi: 10.1038/nature08398 (2009).
+20. Liu, P. P. et al. Polycomb Protein EED Regulates Neuronal Differentiation through Targeting SOX11 in Hippocampal Dentate Gyrus. Stem Cell Reports 13, 115–131, doi: 10.1016/j.stemcr.2019.05.010 (2019).
+21. Zhang, J. et al. Ezh2 regulates adult hippocampal neurogenesis and memory. J Neurosci 34, 5184–5199, doi: 10.1523/JNEUROSCI.4129-13.2014 (2014).
+22. Gearhart, M. D., Corcoran, C. M., Wamstad, J. A. & Bardwell, V. J. Polycomb group and SCF ubiquitin ligases are found in a novel BCOR complex that is recruited to BCL6 targets. Molecular and cellular biology 26, 6880–6889, doi: 10.1128/MCB.00630-06 (2006).
+23. Wang, Z. et al. A Non-canonical BCOR-PRC1.1 Complex Represses Differentiation Programs in Human ESCs. Cell Stem Cell 22, 235–251 e239, doi: 10.1016/j.stem.2017.12.002 (2018).
+24. Huo, D. et al. CpG island reconfiguration for the establishment and synchronization of polycomb functions upon exit from naive pluripotency. Molecular cell, doi: 10.1016/j.molcel.2022.01.027 (2022).
+25. Yokotsuka-Ishida, S. et al. Positional cloning and comprehensive mutation analysis identified a novel KDM2B mutation in a Japanese family with minor malformations, intellectual disability, and schizophrenia. J Hum Genet 66, 597–606, doi: 10.1038/s10038-020-00889-4 (2021).
+26. Labonne, J. D. et al. An atypical 12q24.31 microdeletion implicates six genes including a histone demethylase KDM2B and a histone methyltransferase SETD1B in syndromic intellectual disability. Hum Genet 135, 757–771, doi: 10.1007/s00439-016-1668-4 (2016).
+27. van Jaarsveld, R. H. et al. Delineation of a KDM2B-related neurodevelopmental disorder and its associated DNA methylation signature. Genet Med, doi: 10.1016/j.gim.2022.09.006 (2022).
+28. Gao, Y. et al. Impaired KDM2B-mediated PRC1 recruitment to chromatin causes defective neural stem cell self-renewal and ASD/ID-like behaviors. iScience, 103742 (2022).
+29. Li, W. et al. Long non-coding RNA LncKdm2b regulates cortical neuronal differentiation by cis-activating Kdm2b. Protein Cell 11, 161–186, doi: 10.1007/s13238-019-0650-z (2020).
+30. Anderson, M. C. & Floresco, S. B. Prefrontal-hippocampal interactions supporting the extinction of emotional memories: the retrieval stopping model. Neuropsychopharmacology 47, 180–195, doi: 10.1038/s41386-021-01131-1 (2022).
+31. Gruber, M. J. & Ranganath, C. How Curiosity Enhances Hippocampus-Dependent Memory: The Prediction, Appraisal, Curiosity, and Exploration (PACE) Framework. Trends Cogn Sci 23, 1014–1025, doi: 10.1016/j.tics.2019.10.003 (2019).
+32. Bannerman, D. M. et al. Hippocampal synaptic plasticity, spatial memory and anxiety. Nat Rev Neurosci 15, 181–192, doi: 10.1038/nrn3677 (2014).
+33. Maren, S., Phan, K. L. & Liberzon, I. The contextual brain: implications for fear conditioning, extinction and psychopathology. Nat Rev Neurosci 14, 417–428, doi: 10.1038/nrn3492 (2013).
+34. Caramello, A., Galichet, C., Rizzoti, K. & Lovell-Badge, R. Dentate gyrus development requires a cortical hem-derived astrocytic scaffold. Elife 10, doi: 10.7554/eLife.63904 (2021).
+35. Galceran, J., Miyashita-Lin, E. M., Devaney, E., Rubenstein, J. L. & Grosschedl, R. Hippocampus development and generation of dentate gyrus granule cells is regulated by LEF1. Development 127, 469–482, doi: 10.1242/dev.127.3.469 (2000).
+36. Guan, H. et al. Secreted Frizzled Related Proteins in Cardiovascular and Metabolic Diseases. Front Endocrinol (Lausanne) 12, 712217, doi: 10.3389/fendo.2021.712217 (2021).
+37. Marcos, S. et al. Secreted frizzled related proteins modulate pathfinding and fasciculation of mouse retina ganglion cell axons by direct and indirect mechanisms. J Neurosci 35, 4729–4740, doi: 10.1523/JNEUROSCI.3304-13.2015 (2015).
+38. Satoh, W., Gotoh, T., Tsunematsu, Y., Aizawa, S. & Shimono, A. Sfrp1 and Sfrp2 regulate anteroposterior axis elongation and somite segmentation during mouse embryogenesis. Development 133, 989–999, doi: 10.1242/dev.02274 (2006).
+39. von Marschall, Z. & Fisher, L. W. Secreted Frizzled-related protein-2 (sFRP2) augments canonical Wnt3a-induced signaling. Biochem Biophys Res Commun 400, 299–304, doi:10.1016/j.bbrc.2010.08.043 (2010).
+40. Subramanian, L., Remedios, R., Shetty, A. & Tole, S. Signals from the edges: the cortical hem and antihem in telencephalic development. Semin Cell Dev Biol 20, 712–718, doi: 10.1016/j.semcdb.2009.04.001 (2009).
+41. Dias, C. & Guillemot, F. Revealing the inner workings of organoids. EMBO J 36, 1299–1301, doi: 10.15252/embj.201796860 (2017).
+42. Bingman, V. P., Salas, C. & Rodriguez, F. in Encyclopedia of Neuroscience (eds Marc D. Binder, Nobutaka Hirokawa, & Uwe Windhorst) 1356–1360 (Springer Berlin Heidelberg, 2009).
+43. Supèr, H., Soriano, E. & Uylings, H. B. M. The functions of the preplate in development and evolution of the neocortex and hippocampus. Brain Research Reviews 27, 40–64, doi: https://doi.org/10.1016/S0165-0173(98)00005-8 (1998).
+44. Chen, Z., Djekidel, M. N. & Zhang, Y. Distinct dynamics and functions of H2AK119ub1 and H3K27me3 in mouse preimplantation embryos. Nat Genet 53, 551–563, doi: 10.1038/s41588-021-00821-2 (2021).
+45. O’Carroll, D. et al. The polycomb-group gene Ezh2 is required for early mouse development. Mol Cell Biol 21, 4330–4336, doi: 10.1128/MCB.21.13.4330-4336.2001 (2001).
+46. Voncken, J. W. et al. Rnf2 (Ring1b) deficiency causes gastrulation arrest and cell cycle inhibition. Proc Natl Acad Sci U S A 100, 2468–2473, doi: 10.1073/pnas.0434312100 (2003).
+47. Bonnet, J. et al. PR-DUB preserves Polycomb repression by preventing excessive accumulation of H2Aub1, an antagonist of chromatin compaction. Genes Dev 36, 1046–1061, doi: 10.1101/gad.350014.122 (2022).
+48. Pereira, J. D. et al. Ezh2, the histone methyltransferase of PRC2, regulates the balance between self-renewal and differentiation in the cerebral cortex. Proceedings of the National Academy of Sciences of the United States of America 107, 15957–15962, doi: 10.1073/pnas.1002530107 (2010).
+49. Eto, H. et al. The Polycomb group protein Ring1 regulates dorsoventral patterning of the mouse telencephalon. Nat Commun 11, 5709, doi: 10.1038/s41467-020-19556-5 (2020).
+50. Morimoto-Suzki, N. et al. The polycomb component Ring1B regulates the timed termination of subcerebral projection neuron production during mouse neocortical development. Development 141, 4343–4353, doi: 10.1242/dev.112276 (2014).
+51. Piper, M. et al. NFIB-mediated repression of the epigenetic factor Ezh2 regulates cortical development. J Neurosci 34, 2921–2930, doi: 10.1523/JNEUROSCI.2319-13.2014 (2014).
+52. Sun, B., Chang, E., Gerhartl, A. & Szele, F. G. Polycomb Protein Eed is Required for Neurogenesis and Cortical Injury Activation in the Subventricular Zone. Cereb Cortex 28, 1369–1382, doi: 10.1093/cercor/bhx289 (2018).
+53. Hirabayashi, Y. et al. Polycomb limits the neurogenic competence of neural precursor cells to promote astrogenic fate transition. Neuron 63, 600–613, doi: 10.1016/j.neuron.2009.08.021 (2009).
+54. Lan, X. et al. PCGF6 controls neuroectoderm specification of human pluripotent stem cells by activating SOX2 expression. Nature Communications 13, 4601, doi: 10.1038/s41467-022-32295-z (2022).
+55. Chiacchiera, F. et al. Polycomb Complex PRC1 Preserves Intestinal Stem Cell Identity by Sustaining Wnt/β-Catenin Transcriptional Activity. Cell stem cell 18, 91–103, doi: https://doi.org/10.1016/j.stem.2015.09.019 (2016).
+56. Wang, J. et al. EED-mediated histone methylation is critical for CNS myelination and remyelination by inhibiting WNT, BMP, and senescence pathways. Science Advances 6, eaaz6477, doi:doi: 10.1126/sciadv.aaz6477 (2020).
+57. Oittinen, M. et al. Polycomb Repressive Complex 2 Enacts Wnt Signaling in Intestinal Homeostasis and Contributes to the Instigation of Stemness in Diseases Entailing Epithelial Hyperplasia or Neoplasia. Stem Cells 35, 445–457, doi: 10.1002/stem.2479 (2016).
+58. Chenn, A. & Walsh, C. A. Regulation of cerebral cortical size by control of cell cycle exit in neural precursors. Science (New York, N.Y.) 297, 365–369, doi: 10.1126/science.1074192 (2002).
+59. Machon, O., van den Bout, C. J., Backman, M., Kemler, R. & Krauss, S. Role of β-catenin in the developing cortical and hippocampal neuroepithelium. Neuroscience 122, 129–143, doi: https://doi.org/10.1016/S0306-4522(03)00519-0 (2003).
+60. Bengoa-Vergniory, N. & Kypta, R. M. Canonical and noncanonical Wnt signaling in neural stem/progenitor cells. Cellular and Molecular Life Sciences 72, 4157–4172, doi: 10.1007/s00018-015-2028-6 (2015).
+61. Lie, D.-C. et al. Wnt signalling regulates adult hippocampal neurogenesis. Nature 437, 1370–1375, doi: 10.1038/nature04108 (2005).
+62. Wang, J. et al. A Ctnnb1 enhancer regulates neocortical neurogenesis by controlling the abundance of intermediate progenitors. Cell Discovery 8, 74, doi: 10.1038/s41421-022-00421-2 (2022).
+63. Munji, R. N., Choe, Y., Li, G., Siegenthaler, J. A. & Pleasure, S. J. Wnt signaling regulates neuronal differentiation of cortical intermediate progenitors. J Neurosci 31, 1676–1687, doi: 10.1523/JNEUROSCI.5404-10.2011 (2011).
+64. Turberfield, A. H. et al. KDM2 proteins constrain transcription from CpG island gene promoters independently of their histone demethylase activity. Nucleic Acids Res 47, 9005–9023, doi: 10.1093/nar/gkz607 (2019).
+65. Fukuda, T., Tokunaga, A., Sakamoto, R. & Yoshida, N. Fbxl10/Kdm2b deficiency accelerates neural progenitor cell death and leads to exencephaly. Mol Cell Neurosci 46, 614–624, doi: 10.1016/j.mcn.2011.01.001 (2011).
+66. Vorhees, C. V. & Williams, M. T. Morris water maze: procedures for assessing spatial and related forms of learning and memory. Nat Protoc 1, 848–858, doi: 10.1038/nprot.2006.116 (2006).
+
+# Supplementary Files
+
+- [Graphicalabstract.jpg](https://assets-eu.researchsquare.com/files/rs-2867884/v1/17ac4a71fa9356cf64b67926.jpg)  
+  Graphic Abstract  
+  Loss of KDM2B-CxxC reduces repressive histone modifications – H2AK119ub and H3K27me3 - on key Wnt signal genes, hence leading to the block of their attenuation over time. Hampered differentiation and migration of hippocampal progenitors leads to hippocampal agenesis.
+
+- [ZhangKDM2B2023427SupplementaryInformation1.pdf](https://assets-eu.researchsquare.com/files/rs-2867884/v1/09a8c53004401ce5126e8cc4.pdf)  
+  Supplementary information

04d582b40fff4ca72ea92fb6489e1caf504d7821023ce030e4a5dc7af0ffe859/preprint/images_list.json

@@ -0,0 +1,50 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Overview of Glioma Proteomics Analyses\nAn overview of glioma proteomics analyses conducted on a cohort comprising 188 WHO grade glioma patients, involving a comprehensive proteomic profiling of 343 tumor samples and 53 normal-appearing brain samples. Tissue specimens of 0.5*0.5mm from formalin-fixed paraffin-embedded (FFPE) samples were extracted for proteomic analysis. To ensure the quality of sample preparation, mouse liver samples were utilized as controls, while pool samples served as controls for LC-MS/MS analysis. Additionally, 27 batches were randomly assigned for analysis. The study encompassed a thorough integration of glioma genetic information, proteomic data, and clinical information. A total of 8,561 proteins were quantified for subsequent analyses. The focus of the analysis primarily centered on metabolic pathways and biomarkers. The findings were further validated using cellular and mouse models in order to confirm their accuracy and relevance.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Proteomic Features Differ between Gliomas and their Normal-appearing Brain\n(A) Heatmap shows that 916 differentially expressed proteins between tumor and normal-appearing brain samples (Student t-test, BH adjusted p< 0.001, fold change > 5). (B) Pathways are significantly enriched by IPA using the 916 differentially expressed proteins between tumor and benign samples (p<0.01). (C) Volcano plot shows differentially expressed proteins between tumor and normal-appearing brain samples (Student t-test, B-H adjusted p< 0.001, fold change > 5). Immunity and metabolism associated proteins are labeled with red and purple. (D) Manhattan plot shows 2138 differentially expressed proteins among four grades (ANOVA test B-H adjusted p< 0.001). Immunity and metabolism associated proteins are labeled with red and purple. (E) Enriched immune and metabolic processes are displayed, which related to nucleotide synthesis, antigen presentation, interferon pathway, and TGF-\u03b2 pathways.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "The IDH Status of Glioma could influence Immune and Metabolic Proteome\n(A) Heatmap shows that differentially expressed proteins in IDH-mutant and wild-type glioma (Student t-test, B-H adjusted p < 0.001). (B)-(C) The volcano plot and pathway enrichment show differentially expressed proteins between IDH-mutant and wild-type glioma (Student t-test, B-H adjusted p< 0.001). (D)-(E) The volcano plot and pathway enrichment show differentially expressed proteins between Grade 2-3 and Grade 4 astrocytoma (Student t-test, B-H adjusted p< 0.001).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Proteomic Subtypes of Glioma could Better Predict Clinical Outcomes\n(A)Heatmap shows two proteomic subtypes within our study cohort based on differentially expressed proteins. And subgroup 1 enriches metabolism-related proteins and has been designated as the \u201cmetabolism subgroup,\u201d denoted as S-Me. In contrast, subgroup 2 displays a significant upregulation of immune and inflammatory proteins and has been named as the \u201cimmune subgroup,\u201d designated as S-Im (Student t-test, B-H adjusted p< 0.05, fold change > 3). Pathways significantly enriched by IPA using the 299 proteins in S-Me (upper) and 617 proteins in S-Im (lower). The proteins in those pathways have been listed. (B)-(C) K-M survival curve based on proteomics, W TCGA database shows better clinical outcomes of S-Me subtype, while poor of S-Im subtype. (D) PCA analysis of the glioma proteome effectively stratified the two subtypes. (E) TCGA gene expression data validate our proteomic classification.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Pyrimidine Metabolism Associate with Glioma Patients Survival\n(A)Heatmap shows the 18 different expressed proteins related to metabolic processes based on overall survival time. (B) K-M survival curve of a risk-scoring model based on these 18 proteins. (C) Pathway enrichment shows pyrimidine metabolism is predominant involved. (D) K-M survival curve of the risk-scoring model using TCGA data.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "DPYD is Required for GSC Proliferation and Self-renewal\n(A) IHC validation of DPYD expression according to proteomic profiling. (B) DPYD mRNA levels in T4121 and Mes28 cells after knocking-down. (C) Proliferation assay shows targeting DPYD impairs the proliferation ability as assessed by cell number on day 3 and 5. (D) In vitro limiting dilution assays and (E) (F) tumorsphere formation shows the impairment of self-renewal of GSCs. The sphere number and size significantly decreased in DPYD knockdown GSCs. (G)-(H) Immunofluorescence shows increased phospho-\u03b3H2A.X foci in the nucleus of GSCs after knockdown of DPYD leading to severe DNA damage. (I) Western blot shows increased protein levels of phospho-\u03b3H2A.X and CC3 in DPYD. (J) Knockdown of DPYD a significantly increased the survival of mouse. (K) HE staining of mouse brain shows the lack of DPYD inhibit growth of GSCs.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

04d582b40fff4ca72ea92fb6489e1caf504d7821023ce030e4a5dc7af0ffe859/preprint/preprint.md

@@ -0,0 +1,275 @@
+# Abstract
+
+Gliomas exhibit high heterogeneity and poor prognosis. Despite substantial progress has been made in glioma at the genomic and transcriptomic levels, comprehensive proteomic characterizations and their implications remain largely unexplored. Here, we performed proteogenomic characterization of gliomas using 343 FFPE tumor samples and 53 normal-appearing brain samples from 188 patients, which was integrated with genomic panel data and clinical information. Proteomics profiles uncovered two subgroups: Subgroup 1, termed the “metabolism subgroup” (S-Me), characterized by an enrichment of metabolism-related proteins; and Subgroup 2, named the “immune subgroup” (S-Im), showing an upregulation of immune and inflammatory proteins. These proteomic subgroups exhibited significant differences in prognosis, tumorigenesis, microenvironment dysregulation and potential therapeutics, emphasizing the critical roles of metabolism and immune processes in glioma biology and patient outcomes. By delving into metabolic pathways guided by our proteomic findings, DPYD and TYMP were further identified as potential prognostic biomarkers associated with nucleotide metabolic reprogramming. Functional validation using GSCs and animal models highlighted nucleotide metabolism as a promising therapy against gliomas. The integrated multi-omics analysis introduces a novel proteomic classification for gliomas, and also identified two new metabolic biomarkers, DPYD and TYMP, which offer insights into the molecular pathogenesis and identify treatment opportunities.
+
+**Biological sciences/Cancer/CNS cancer**  
+**Biological sciences/Molecular biology/Proteomics**  
+**Proteomic**  
+**Glioma**  
+**Profiling**  
+**Immune**  
+**Nucleotide metabolism**
+
+# Introduction
+
+Gliomas, accounting for around 81% of all primary central nervous system (CNS) malignancies, constitute the most common type of intrinsic brain tumor with high heterogeneity and poor prognosis [1]. Recently, gliomas have been classified into distinct subgroups according to an integrated set of molecular and histological parameters, which was initially introduced in 2016 and subsequently updated in 2021 (WHO CNS5) [2, 3].
+
+High-throughput research has been conducted with the aim of uncovering major oncogenic molecular events, identifying therapeutically actionable alterations, and delineating prognostic and predictive subgroups among glioma patients [4–6]. Notably, these efforts have predominantly concentrated on the genomic, transcriptomic, and epigenetic dimensions [7]. At the genomic level, isocitrate dehydrogenase (IDH) mutation is a widely recognized biomarker [8]. The GLASS consortium has described other genetic alterations during glioma recurrence, such as CDKN2A deletion and hypermutations [9]. At the transcriptomic level, the Cancer Genome Atlas Program (TCGA) classified glioblastoma (GBM) into classical, mesenchymal, proneural, and neural subtypes [10]. The GLIOTRAIN discovery cohort also identified three unique patient transcriptome clusters, characterized by tumor microenvironment compositions. Although continuous efforts have been made in multi-omics research, clinical applications remain confined to a few biomarkers, such as IDH mutation, MGMT promoter methylation, and 1p/19q co-deletion [11–13]. Consequently, the development of targeted therapies and refinement of glioma classification for patient stratification remain ambitious, and are long-term goals necessitating significant progress.
+
+Despite the substantial insights gained from DNA- and RNA-based classifications, it is anticipated that proteomic patterns, as discerned through mass spectrometry (MS), will serve as the primary conduits for elucidating biological functions in glioma [7, 14, 15]. In the meantime, emerging studies have demonstrated the limited correlation between protein abundance and DNA or RNA level (r = 0.23–0.45) in breast, colorectal, and ovarian cancer [16–18]. Intriguingly, proteins have shown superior prognostic predictive capabilities compared to their upstream mRNA counterparts [18]. Hence, the proteome, which reflects the functional consequences of genomic modifications, holds the potential to offer insights into the real state of tumor cells. In intrahepatic cholangiocarcinoma, researchers have successfully identified four subgroups characterized by unique genetic alterations, microenvironment dysregulation, tumor microbiota composition, and potential therapeutics features, with the identification of SLC16A3 and HKDC1 as prognostic biomarkers and therapeutic targets [19]. The proteomic classification of GBM has demonstrated a capacity to better reflect immune and metabolic biological processes associated with survival as compared with transcriptomic profiling [14]. Oh et al. similarly identified two proteomic clusters based on distinct Warburg-like and oxidative phosphorylation-related protein profiles, paving the way for the identification of the mTORC1/2 dual inhibitor AZD2014 as a treatment modality for GBM [7]. Collectively, these results demonstrate the value of proteomes of glioma. However, comprehensive proteomic subtyping across all four WHO-graded gliomas is still lacking, which may enhance the predictive power of clinical outcomes. Furthermore, a growing body of evidence supports the potential of targeting metabolism in cancer treatment, given its critical role as a metabolic dependency in cancer cells [20]. In glioma research, further investigation is required to understand the diverse biological roles of hyperactive nucleotide metabolism for the development of effective combination therapies [20, 21].
+
+In this study, we conducted a comprehensive analysis of data obtained from 188 glioma patients, with integrated mass spectrometry-based proteomic data, genomic panel data, and clinical information. Two proteomic subgroups were identified, S-Im and S-Me, with distinct immune and metabolic features that correlated with survival outcomes. We also discovered two metabolic biomarkers, dihydropyrimidine dehydrogenase (DPYD) and thymidine phosphorylase (TYMP), that had prognostic and therapeutic implications for glioma patients highlighting the potential of nucleotide metabolism. Overall, our findings highlight the promise of effective proteomic classification and targeted metabolic reprogramming strategies as promising approaches for glioma treatment.
+
+# Results
+
+## Overview of Glioma Proteomics Analyses
+
+A proteomic profiling was performed using 343 tumor samples and 53 normal-appearing brain samples from a cohort of 188 glioma patients, including 6 pilocytic astrocytoma, 51 diffuse astrocytomas (27 WHO grade 2, 14 grade 3, 9 grade 4, 1 grade NA), 47 diffuse oligodendrogliomas (28 WHO grade 2, 19 grade 3), 82 glioblastomas (WHO grade 4), and 2 not elsewhere classified (NEC) [3]. Our analysis employed a well-established mass spectrometry workflow, as depicted in Fig. 1. To ensure the purity of samples (tumor or normal-appearing tissue) for proteome analysis, two neuroscientists independently confirmed the punch sites, and specimens of size 0.5*0.5mm were extracted for proteomic analysis, as shown in Figure S1. Clinical information, including gender, age, tumor location, WHO 2021 histology, grade, and survival, is summarized in Table 1. The median age of the patients from whom the tumors were excised was 49.03 ± 14.48 years (range: 13–84 years), with a male-to-female ratio of 1.76:1 (male: N = 120; female: N = 68). Predominantly, the tumors were located in the frontal (99 cases) and temporal (46 cases) lobes. We performed next-generation sequencing on 129 samples to determine the status of IDH mutation, O6-alkylguanine DNA alkyl transferase (MGMT) promoter methylation, 1p/19q codeletion, and telomerase reverse transcriptase (TERT) promoter mutation. The remaining 59 samples were tested by immunohistochemistry only, and some marker statuses were unavailable. As of the last follow-up in September 2019, 58 patients (30.8%) had passed away, 127 (76.6%) were still alive, and 3 (1.6%) did not attend the follow-up. The median follow-up for patients alive at the last follow-up was 26.7 months.
+
+The proteomic analysis of all glioma samples and normal-appearing brain tissues was conducted by randomly assigning them to 27 batches. As part of quality control measures for sample preparation and MS, we conducted correlation analyses involving mouse liver samples (Figure S2A) and pooled samples (Figure S2B), which demonstrated a strong and satisfactory correlation, affirming the reliability and robustness of our experimental procedures. The stability of the MS analysis was confirmed by the Pearson correlation between replicates’ protein abundances, which was over 0.9 (Figure S2C). The t-SNE plots indicated the absence of any batch effect among the MS machines (Figure S2E), and the variance among the 27 batches and sample types were also represented in Figure S2D and E. After removing isoforms, 8,561 proteins were quantified for further analyses in combination with clinical information.
+
+## Proteomic Features Differ between Gliomas and Normal-appearing Brain
+
+Next, we conducted an analysis to identify proteins with significantly different expression levels between glioma and normal-appearing brain samples. Our investigation demonstrated 916 proteins with significant differential abundance (Student’s t-test, B-H adjusted P-value < 0.001, fold change > 5) (Fig. 2 A). Utilizing Ingenuity Pathway Analysis (IPA), we uncovered enrichment in proteins associated with neurotransmission, synaptogenesis, neuroinflammation, tumor necrosis, and the PD-1 signaling pathway within the glioma samples, indicating the vital role of immune and inflammation-associated proteins (Fig. 2 B). Recognizing the significance of altered cellular metabolism as a communication interface between glioma cells and immune cells within the microenvironment [18, 19], our subsequent focus centered on proteins related to immune regulation and metabolic processes. Among these 916 differentially expressed proteins between glioma and normal-appearing brain tissues (Student��s t-test, B-H adjusted P-value < 0.001, fold change > 5), we identified 28 immune-related and 51 metabolism-related proteins (Fig. 2 C). To uncover critical proteins contributing to tumor progression across various glioma grades, we identified 2,138 differentially expressed proteins (ANOVA test B-H adjusted P-value < 0.001) among WHO grades 1–4 glioma. Notably, these proteins associated with immune and metabolic processes displayed considerable abundance (Fig. 2 D). They were predominantly enriched in pathways related to nucleotide synthesis, antigen presentation, interferon pathway, and TGF-β pathways, indicating intricate and complex interactions between metabolic characteristics of glioma cells and the immune microenvironment (Fig. 2 E).
+
+### The IDH Status of Gliomas Could Influence its Immune and Metabolic Proteome.
+
+Numerous outstanding studies have revealed the importance of the IDH status on the metabolic and epigenetic reprogramming of gliomas [22, 23]. The significance of IDH status was officially recognized in the WHO classification of CNS tumors in 2016, where it was established as a biomarker with prognostic outcomes. In our investigation, we conducted a comparative analysis between IDH mutant (N = 87) and IDH wild-type (N = 81) gliomas across varying WHO grades, resulting in the identification of 203 differentially expressed proteins (Fig. 3 A). The volcano plot presented in Fig. 3 B distinctly illustrated differential protein expression profiles between IDH mutant and wild-type gliomas. Notably, our analysis revealed that metabolic processes and cytokine signaling pathways, such as lipid metabolism, fatty acid transport, DNA biosynthetic process, IL-12 signaling pathway, and IL-6 signaling pathway, were among the significantly affected pathways (Fig. 3 C).
+
+Next, we investigated the differential proteins within IDH mutant tumors at different stages. Previous studies have illuminated the substantial prognostic disparities between grade 4 astrocytomas and grade 2/3 tumors [24, 25]. Our findings indicated a significant decrease in CDK2 and IRF3 expression in grade 2/3 astrocytoma compared to grade 4, which is consistent with the established consensus that the loss of CNDK2A/B, accompanied by increased CDK activity, is intimately associated with the malignant progression of gliomas (Fig. 3 D). Finally, the cell cycle control of chromosomal replication pathway was identified as the primary differential pathway indicating the different proliferation characteristics between grade 2/3 and grade 4 astrocytomas (Fig. 3 E).
+
+Furthermore, our analysis revealed conspicuous variations in glioma protein expression across different grades, histological types, and between LGG and GBM. The t-SNE plots visually depicted pronounced distinctions among these various categories (Figure S3A-C). Regarding histological types, the heatmap provided a clear illustration of how the differentially expressed proteins within distinct histological types of gliomas can be categorized into two distinct clusters (Figure S3D). Cluster 1 exhibits significant enrichment in pathways related to protein translation such as eIF2 and eIF4, while Cluster 2 is notably enriched in pathways associated with synaptic synthesis and signal transduction pathways (Figure S3E and F). Additionally, there were obvious distinctions in metabolic pathways, such as amino acid, nucleotide, and lipid metabolism, between these two clusters (Figure S3G and H). We also observed distinct metabolic pathways, such as amino acid, nucleotide, and lipid metabolism, between these two groups (Figure S3G and H).
+
+### Proteomic Metabolism and Immune Subtypes of Glioma Could Better Predict Clinical Outcomes.
+
+Within our glioma cohort, we employed subtype classification based on proteome data, which is known to better reflect the disease state and underlying biology compared to genome data [14]. Through the utilization of consensus clustering, we successfully identified two proteomic subtypes within our study cohort (Figure S4A), designated as subgroup 1 (n = 21) and subgroup 2 (n = 48), based on an analysis of 335 differentially expressed proteins (proteins with missing rates over 20% were removed, B-H adjusted P-value < 0.05, fold change > 3). Intriguingly, subgroup 1 exhibited a remarkable enrichment of metabolism-related proteins, exemplified by the notable presence of proteins such as PDGFRA and GLS, and has been designated as the “metabolism subgroup,” denoted as S-Me. In contrast, subgroup 2 displayed a significant upregulation of immune and inflammatory proteins, including AKT2, CDK2, TGFB1, and IRF3, which has been named as the “immune subgroup,” designated as S-Im (Fig. 4 A, S4). These subtypes exhibited distinct molecular and clinical features. The proteins enriched in the S-Me subgroup were primarily associated with synaptogenesis signaling pathway and signal transduction, among other pathways. While, the S-Im subgroup exhibited an enrichment of proteins mainly involved in the PD-1 cancer immunotherapy pathway and death receptor pathways. In particular, patients in S-Me demonstrated better clinical outcomes than those in S-Im (Fig. 4 B). The principal component analysis of the diffused glioma proteome effectively stratified the two subtypes, with the first principal component (PC1) distinguishing them (Fig. 4 C).
+
+To validate the accuracy of our proteomic classification, we utilized TCGA gene expression data. Our analysis revealed that the subtypes could effectively differentiate between glioma patients (Fig. 4 D), with patients from the S-Me subtype showing a favorable prognosis, while those from the S-Im subtype displayed a poorer prognosis (Fig. 4 E). Considering the different prognoses of patients with different WHO grades (Figure S5A) and IDH status, we also performed subgroup clustering of grade 1–3 (Figure S4B and S5B), grade 4 (Figure S4C and S5C), and IDH wildtype (Figure S4D and S5D) patients. All the optimal grouping of patients was achieved by dividing them into two categories, with those in S-Me demonstrating better clinical survival regardless of their WHO grade and IDH status, indicating the stability and predictive power of our proteomics subtype clustering, suggesting that binary classification is applicable to all diffuse gliomas.
+
+## Pyrimidine Metabolism is Associated with Glioma Patients’ Survival
+
+Metabolic reprogramming stands as a hallmark of cancer and represents an early event in the gliomagenesis [26, 27]. For instance, GBM cells adapt to their distinctive microenvironment by upregulating enzymes associated with pyrimidine metabolism, facilitating rapid cell proliferation and DNA damage repair [28, 29]. In our analysis, we divided glioma patients into two groups based on their overall survival: long-survival ( ≥ 30 months) and short-survival ( ≤ 12 months) (Figure S6A). Our focus then centered on the major proteins related to metabolic processes. We then examined 18 metabolism-related proteins exhibiting significant differences between these two survival groups (Fig. 5 A, S6B). Importantly, the expression levels of these 18 proteins were found to be closely associated with patients’ prognosis (Figure S6C). We proceeded to establish a risk-scoring model based on the expression of these proteins, and Kaplan-Meier curves affirming that patients in the low-risk score group exhibited more favorable clinical outcomes (Fig. 5 B). Upon further analysis of these metabolic pathways, we discovered their predominant involvement in pyrimidine metabolism, a process closely related to DNA and RNA synthesis (Fig. 5 C). To validate the robustness of this model, we utilized TCGA data, yielding consistently the same prognosis conclusion as our own data (Fig. 5 D). Thus, our analysis further confirmed the crucial involvement of nucleotide metabolism in controlling glioma cell behavior and patient survival and how the metabolic reprogramming especially the nucleotide metabolism contributes to tumor malignancy remained to be explored.
+
+### DPYD and TYMP are required for GSCs proliferation and self-renewal.
+
+As illustrated in the comprehensive nucleotide metabolism map depicted in Fig. 5 C highlighting proteomic-identified differential nucleotide enzymes, we found the significant roles played by catabolic enzymes within this process. The homeostasis of nucleotide metabolism requires the balance between nucleotide biosynthesis and degradation and affecting the balance of nucleotide pools could cause genomic instability and cancer progression. Among them, DPYD and TYMP are enzymes that regulate the catabolism of pyrimidine nucleosides. To explore their potential functions and mechanisms in glioma, we focused on their roles in patient-derived glioma stem cells (GSCs), which are at the top of the glioblastoma cell hierarchy and has a self-renewing effect that contributes to treatment resistance and tumor recurrence [30].
+
+Initially, we divided patients into high- and low-expression groups based on the levels of DPYD and TYMP expression derived from our proteomic profiling. To validate the accuracy of our proteomic profiling findings, we conducted immunohistochemical (IHC) analyses on seven randomly selected patient samples from each group and the protein levels obtained from IHC were consistent with our proteomic profiling (Fig. 6 A, S7A). Subsequently, we employed two independent shRNAs to efficiently knock down the DPYD and TYMP mRNA levels in T4121 and Mes28 cells, as confirmed by real-time PCR (Fig. 6 B, S7B).
+
+Knocking down DPYD and TYMP in GSCs markedly impaired their proliferation ability, as evidenced by reduced cell numbers on the third and fifth days following knockdown (Fig. 6 C, S7C). Moreover, the self-renewal capacity of GSCs was impaired, as assessed through in vitro limiting dilution assays (Fig. 6 D, S7D) and neurosphere formation (Fig. 6 E, S7E). Specifically, the number and size of spheres significantly decreased in DPYD and TYMP knockdown GSCs (Fig. 6 F, S7F).
+
+Given the close connection of these two proteins to DNA metabolism, and the well-known DNA damage repair capacity of GSCs, which contributes to treatment resistance (chemo drug and radiation), we evaluated the level of phospho-γH2A.X in GSCs through western blot and immunofluorescence. The knockdown of DPYD resulted in severe DNA damage and apoptosis, as indicated by an increase in phospho-γH2A.X foci within the nucleus (Fig. 6 G and I) and elevated protein levels of phospho-γH2A.X and CC3 (Fig. 6 I). Collectively, these findings support the critical function of DPYD and TYMP in GSC maintenance and DNA metabolism.
+
+To address the potential benefit of therapeutic targeting of DPYD and TYMP in vivo, mice bearing intracranial patient-derived GSCs tumors were transduced with a nontargeting, control shRNA (shCONT) or shRNA targeting DPYD and TYMP. In particular, we aimed to investigate their roles in tumor initiation and growth. Knockdown of DPYD and TYMP significantly prolonged the survival of mice (Fig. 6 J, S7G). Subsequently, the mice were sacrificed, and their brains were subjected to hematoxylin-eosin staining to assess glioma presence, confirming the absence of glioma growth in GSCs lacking DPYD and TYMP (Fig. 6 K, S7G). Together, these findings have revealed the significance of DPYD and TYMP in the initiation and growth of glioma in vivo, there thereby reinforcing the importance of targeting nucleotide metabolism as a potential strategy for treatment of glioma.
+
+# Discussion
+
+In this study, we conducted a large-scale proteomic analysis of gliomas, with a specific focus on uncovering the immune and metabolic heterogeneity within these tumors, while also exploring proteome-based subtyping. Our investigations led to the identification of two distinct proteomic subtypes within diffuse glioma, each characterized by unique immune and metabolic features. Importantly, these subtypes demonstrated a robust association with clinical outcomes. In addition, we validated the effectiveness of our proteomic subtyping approach using data sourced from the TCGA database. Our biological experiments also revealed the critical roles of DPYD and TYMP in promoting the proliferation and self-renewal of GSCs, making them potential targets for precision treatment strategies.
+
+For many years, adult malignant gliomas have been managed using uniformed approaches, regardless of their molecular characteristics, often following the conventional Stupp regimen[29]. While significant progress has been made in understanding glioma development at the DNA and RNA levels, the establishment of targeted therapies tailored to specific molecular profile, such as IDH mutant or VEGF expression gliomas, has remained elusive. We propose that, given the primary role of proteins in executing physiological functions within organisms, a proteome-based approach may uncover signaling pathways that could pave the way for the development of innovative therapeutic strategies. Indeed, there is increasing evidence in the literature that supports the existence of distinct metabolic proteomic subtypes in glioblastoma based on aerobic/anaerobic energy metabolism [7, 31].
+
+We have made several significant discoveries in our study. Firstly, we identified key proteins that distinguish between normal-appearing brain tissue and glioma tissue, with a focus on pathways related to synaptic synthesis, neuroinflammation, and PD1 pathways. These pathways are closely associated with tumor metabolism and immune inflammation. Furthermore, we categorized WHO grade 2–4 gliomas into two subtypes based on proteomic differences, revealing significant disparities in patient survival between these subtypes. Notably, this proteomic classification remained robust across varying tumor WHO grade and IDH status. To further explore the underlying prognostic mechanisms, we selected differentially expressed proteins associated with metabolism for subsequent analysis and investigation.
+
+Metabolic reprogramming stands as a well-established hallmark of cancer, and the growing understanding of the cross-talk between metabolism and cancer epigenetics expands our insights into the contributions of metabolic derangements to tumor malignancy. Within this context, pyrimidine metabolism has emerged as a promising vulnerability in IDH-mutant glioma. For example, BAY 2402234, an inhibitor targeting dihydroorotate dehydrogenase in pyrimidine synthesis, has exhibited the capacity to sensitize DNA damage in response to nucleotide pool imbalance, thus influencing clinical outcomes in patients [32, 33]. However, the maintenance of a dynamic DNA metabolism pool requires not only enzymes involved in anabolism but also those participating in catabolism. In our investigation, we unveiled the critical role played by pyrimidine catabolism enzymes in glioma, and notably, their association with clinical prognosis, which is consistent with the current prevailing understanding in the field. Our study provides a valuable contribution to the field of cancer research by highlighting the critical importance of nucleotide metabolism in cancer cells and the potential of targeting this metabolic dependency as an effective cancer therapy strategy. DPYD, serving as the initial committed step in pyrimidine degradation, has been extensively investigated, particularly in the context of 5-fluorouracil (5-FU) metabolism, a widely used chemotherapy drug. Altered DPYD activity in the context of tumorigenesis may disrupt the balance of nucleotide pools within cells, potentially leading to genomic instability and contributing to cancer development. Notably, in glioma research, studies have demonstrated elevated DPYD expression in diffuse midline gliomas compared to normal astrocytes, indicating resistance to 5-FU treatment and underscoring the malignant phenotype of glioma biology [34]. Furthermore, in pancreatic cancer, a disease exhibiting similar aggressiveness like glioma, the inhibition of STAT3-DYPD expression by chemopreventive agent has demonstrated the capacity to restrain tumor proliferation [35]. Beyond its influence on tumor cell proliferation, DPYD has been implicated in maintaining a mesenchymal-like state in epithelial tumors by directing pyrimidine metabolites towards the catabolic pathway [36]. TYMP encodes the enzyme thymidine phosphorylase, an enzyme that exhibits overexpression in various cancer types, with established roles in tumor growth, invasion, and chemotherapy resistance [37–39]. Beyond its primary role in thymidine metabolism, TYMP serve as a potent angiogenic factor and functions as a cytokine that stimulates endothelial cell proliferation and migration, thereby promoting the formation of new blood vessels [40]. This angiogenic activity of TYMP holds particularly relevant in the context of glioma, given that tumor growth and invasion critically depend on the development of a robust blood supply. Elevated TYMP levels have been associated with increased vascularity and more aggressive tumor behavior. In general, our study provided experimental validation of the roles of DPYD and TYMP, which have traditionally been investigated primarily in the context of chemotherapy response within the malignant progression of glioma. These findings demonstrated that the downregulation of these two enzymes significantly reduces the self-renewal capacity of GSCs, ultimately leading to DNA damage, apoptosis, and impairments in *in vivo* tumorigenesis, which strongly suggest that DPYD and TYMP might contribute to the genesis of glioma through mechanisms distinct from their known involvement in drug metabolism. Moreover, considering the widespread use of drugs targeting these two enzymes in preclinical gastrointestinal and urinary tumor models [41–44], conducting clinical trials to evaluate the efficacy of these drugs in glioma patients represents a promising direction for future research.
+
+Our study has some limitations that should be acknowledged. First, our analysis only measures protein expression, and does not account for post-translational modifications, such as phosphorylation and glycosylation, that may also affect glioma pathogenesis. Second, our proteomic subtyping approach needs to be validated in larger cohorts to increase its robustness. Third, more research is needed to elucidate the exact molecular pathways and downstream effectors of DPYD and TYMP in glioma development.
+
+In conclusion, our study provides new insights into the molecular mechanisms of glioma and identifies potential biomarkers and therapeutic targets. Proteomic subtyping emerges as a more accurate predictor of patient prognosis in glioma. Notably, enzymes involved in pyrimidine catabolism, specifically DPYD and TYMP, emerge as pivotal players in glioma pathogenesis. Further research is warranted to validate and expand upon these findings, with the aim of improving the diagnosis and treatment of this devastating cancer.
+
+# Materials and Methods
+
+## Clinical samples collection and preparation
+
+Formalin-fixed and paraffin-embedded (FFPE) human brain glioma samples were obtained from the Department of Neurosurgery, Huashan Hospital of Fudan University during January to December 2018. The specimens were diagnosed and graded according to WHO guidelines by two neuropathologists [3]. Patients provided informed consent (KY2015-256) before surgery, agreeing to donate tumor samples for research purposes. The acquisition and use of human glioma samples were approved by the Ethics Committee of Fudan University (KY2021-064) and adhered to ethical principles outlined in the Helsinki Declaration (1964, amended in 2013) and the guidelines of the World Medical Association [45].
+
+For the subsequent proteomic analysis, two neuroscientists selected representative histological area of either the tumor or normal-appearing tissue based on hematoxylin-eosin (H&E) staining slides, ensuring sample purity for punching. Areas displaying hemorrhagic, necrotic, or degenerative changes, such as mucous or cystic changes, were deliberately excluded from the punching process. Normal-appearing brain tissue samples (53 samples from 32 cases) were included as controls. All samples were obtained using a 1 mm biopsy punch (Kai medical, #BP-10F) for MS analysis. To minimize potential bias from tumor heterogeneity, portions of FFPE samples were punched at two independent sites.
+
+## Hematoxylin and eosin (H&E) staining
+
+The FFPE tissues were sliced into 4 µm sections, and H&E staining was performed following standard protocols using the fully automated BOND-III stainer system (Leica Biosystems, USA).
+
+## Peptide sample preparation
+
+Protein extraction and peptide digestion were performed as described previously [46, 47]. Briefly, the FFPE tissue samples were dewaxed by heptane and hydrated with a gradient ethanol solution. Proteins were extracted from the tissues and then enzymatically digested into peptide samples using trypsin (Hualishi Scientific, China) and LysC (Hualishi Scientific, China) relying on PCT (Pressure Biosciences Inc., MA, US) [46, 47].
+
+## Liquid chromatography-tandem mass spectrometry (LC-MS/MS)
+
+Clean peptide samples were separated with a nano Elute system (Bruker Daltonics, Germany) at a flow rate of 300 nL/min, at 217.5 bar. The linear liquid phase gradient of buffer B was 5–27% for 90 min during the data-dependent acquisition (DDA) mode and then for 60 min during the PulseDIA acquisition. The mobile phase was mixed with buffer A (2% ACN, 0.1% formic acid) and buffer B (98% ACN, 0.1% formic acid). Peptides were scanned by a Captive Spray nano electrospray ion source on a hybrid trapped ion mobility spectrometer (TIMS) quadrupole time-of-flight mass spectrometer (TIMSTOF Pro, Germany).
+
+To generate a library of ion mobility-enhanced spectra, we performed the DDA of the Parallel Cumulative Sequential Fragmentation (PASEF) mode with 10 PASEF scans every top-N acquisition cycle. A total cycle time of 1.17 s was achieved with an accumulation and ramp time of 100 ms each for the dual TIMS analyzer. The ion mobility was scanned from 0.6 to 1.6 V/cm². The MS1 and MS2 acquisitions were performed in the m/z range of 100 to 1,700 Th. Precursors reaching a target value of 20,000 arbitrary units were dynamically excluded for 0.4 min, and singly charged precursors were excluded at positions in the m/z-plane of ion mobility.
+
+The peptide samples were analyzed using the PulseDIA mode [48] of the diaPASEF [49]. Ion mobility was scanned from 0.7 to 1.3 V/cm². MS1 and MS2 acquisitions were performed in the m/z range of 100 to 1700 Th. We defined two sets of complementary isolation windows and applied them to the two MS methods for two injections. We set a window with a width of 25, and the rest of the parameters were the same as for the DDA.
+
+## High-pH reversed-phase chromatography fractionation
+
+Approximately 100 µg of peptide mix of glioma tumor and normal-appearing brain were fractioned on a chromatographic column (BEH C18, 300 Å, 5 µm, 4.6 mm x 250 mm) coupled to a Thermo Dionex Ultimate 3000 (Thermo Fisher Scientific, USA). A linear gradient of 5–35% buffer B (98% ACN, 0.6% ammonia, pH = 10) was kept for 68 min to generate 60 fractions. The fractions were next combined into 20 samples according to a published strategy [50]. Finally, all samples were dried by a SpeedVac (Thermo Fisher Scientific, USA), reconstituted in 0.1% formic acid, and spiked with standard peptides (iRT, Switzerland).
+
+## Human glioma-specific spectral library generation and proteomic data analysis
+
+The 20 glioma DDA data files and 88 normal brain DDA data files were analyzed using Spectronaut (version 14.5). The background FASTA file was downloaded from SwissProt on 22 January 2020 and contained 20,365 human protein entries. The tolerance for fragment and precursor was set to dynamic, which depends on each file. The digestion enzyme was trypsin, cutting after “KR” but not before “P.” False discovery rates (FDRs) of peptide-spectrum matches, precursors, and proteins were set to 1%. A total of 17,0445 precursors and 10,067 proteins were identified in this spectral library. The self-constructed library was further used for the PulseDIA data analysis by Spectronaut (version 14.5) without cross-run normalization. FDRs of precursors and proteins were set to 1%. Other settings were used as default parameters. Missing values were imputed by 0.8 times the minimum intensity found in the matrix of 8,561 proteins.
+
+## Proteomic data quality control analysis
+
+To minimize the batch effect, the 343 tumor and 53 normal-appearing brain samples were randomly separated into 27 batches each (Dataset 1) for proteomic data acquisition. During sample preparation, from batches 1 to 25, each batch contained 13 tumor samples, two normal-appearing brain samples, and one mouse liver sample for PCT quality control; batch 26 included ten tumor samples, two normal-appearing brain samples, and one mouse liver sample; batch 27 included nine samples contaminated by blood and one mouse liver sample. Additionally, for the LC-MS/MS acquisition, each batch contained one pooled peptide sample and one randomly selected peptide sample as a replicate for the LC-MS/MS technical quality control.
+
+## Cell culture and reagents
+
+Patient-derived GSCs (T4121 and Mes28) were derived by our laboratory and obtained via a material transfer agreement with Duke University. These GSCs were cultured in Neurobasal media (Thermo Fisher Scientific, #21103049) supplemented with 2% B27 supplement without vitamin A (Thermo Fisher Scientific, # 12587010), EGF and bFGF (20 ng/mL each; R&D Systems, #236-EG and #3718-FB), 1% sodium pyruvate (Thermo Fisher Scientific, #11360070), and 1% GlutaMAX (Thermo Fisher Scientific, #35050079). In addition, 293T cells (sourced from ATCC) were cultured in Dulbecco’s modified Eagle’s medium (Thermo Fisher Scientific, #11965092 ) with 10% fetal bovine serum (Thermo Fisher Scientific, #A4766801).
+
+## Plasmids and lentiviral transduction
+
+Two non-overlapping shRNAs targeting human DPYD (target sequences: GCCGTATGATGTAGTGAATTT and GCAATTTGCTACTGAGGTATT), TYMP (target sequences: GCCTCCATTCTCAGTAAGAAA and GCTGGAGTCTATTCCTGGATT), and a non-targeting control shRNA were construed into a PLKO.1 plasmid. The knockdown efficiency of all shRNAs were assessed through qRT-PCR before employed in functional assays. To generate lentiviral particles, we employed 293T cells and conducted co-transfection with the packaging vectors psPAX2 (Addgene, #12259) and pMD2.G (Addgene, #12259) using the polyethylenimine transduction method in the OPTI-MEM medium (Thermo Fisher Scientific, #31985062).
+
+## Quantitative RT-PCR
+
+An RNA purification kit (EZB, #B0004DP) was used to isolate total cellular RNA from cell pellets. The 4x EZscript Reverse Transcription Mix II with gDNA remover (EZB, RT2GQ) was utilized to reverse cDNA transcription. Quantitative real-time PCR was performed on the QuantStudio 6 Flex Real-Time PCR System (Thermo Fisher Scientific) with a 2x S6 universal SYBR qPCR mix (NovaBio, #Q204). The primer pairs for qPCR in this study were as follows: human DPYD forward 5’-GCTGTCCCTGAGGAGATGGA-3’ and reverse 5’- GTCCGAACAAACTGCATAGCAA-3’; TYMP forward 5’- CACCTTGGATAAGCTGGAGTC-3’ and reverse 5’-GGCTGCATATAGGATTCCGTC-3’; ACTIN forward 5’- TCCCTGGAGAAGAGCTACG-3’and reverse 5’- GTAGTTTCGTGGATGCCACA-3’.
+
+## Western blot
+
+Equal cell amount per group were collected and boiled in sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer. Electrophoresis was performed using 12% protein gels, followed by transfer onto polyvinylidene fluoride (PVDF) membrane. TBST supplemented with 3% bovine serum albumin (BSA) was used for blocking for 1 hour at room temperature. Blotting with primary antibody was performed at 4°C overnight. The following antibodies were used: anti-Cleaved Caspase-3 (CST, #9661), anti-phospho-γH2AX (CST, #9718), and anti-α-Tubulin (Proteintech, #66031-1-lg). A secondary antibody (Thermo Fisher Scientific, # SA5-10041) was applied to bind with the primary antibody for 2 hours. Finally, membranes were visualized under an enhanced chemiluminescence detection system (Typhoon FLA 9500, GE Healthcare).
+
+### In vitro limiting dilution assay
+
+In vitro limiting dilution was performed following established protocols [51, 52]. Briefly, decreasing numbers of cells per well (50, 20, 10, 5, and 1) were plated into 96-well plates. Seven days after plating, the presence and count of neurospheres in each well were recorded. An extreme limiting dilution analysis was performed using software available at http://bioinf.wehi.edu.au/software/elda [51, 52].
+
+## Cell proliferation and neurosphere formation assay
+
+Cell proliferation experiments were conducted by seeding the cells of interest at a density of 10,000 cells/well in 12-well plates, with four replicates for each condition. Cell counts were performed on day 3 and day 5 to assess cell proliferation. The neurosphere formation was measured by seeding 5,000 cells into a 12-well plate with eight replicates for each group. The presence and quantily of tumor spheres in each well were recorded seven days after plating.
+
+## Immunohistochemistry (IHC) Staining
+
+The IHC staining procedures followed established protocols [53]. In brief, 3% H₂O₂ was used to quench endogenous peroxidases, followed by antigen retrieval using the citrate buffer. Primary antibodies against DPYD (CST, #27662-1-AP) and TYMP (CST, #12383-1-AP) were incubated overnight at 4℃. Staining was carried out using an LSAB IHC kit (DAKO Cytomation, #VB-6017D), with 3,3-diaminobenzidine as the chromogen and Mayer’s hematoxylin for nuclear staining.
+
+For semi-quantitative analysis of the IHC staining, three randomly selected fields (100 µm² each) were chosen from each sample, and images were captured using a charge-coupled device camera. Staining intensity was analyzed using IMT i-Solution software version 10.1 (IMT i-Solution, Inc., Canada), as previously described [54]. The results of IHC image analysis were presented as mean ± SEM. Based on the IHC staining density, cases were categorized into a high-expression group with staining intensity above the average value and a low-expression group with staining intensity below than the average value.
+
+## Immunofluorescence staining
+
+For immunofluorescence microscopy, cells were digested into single cells and then collected. Cells were fixed with 4% paraformaldehyde twice for 30 minutes at room temperature, followed by permeabilization using 0.01% Triton X-100 (Thermo Fisher Scientific, # HFH10) for 10 minutes, then blocked in 3% BSA at room temperature for 30 minutes. Next, cells were incubated with anti-phospho-γH2AX (CST, #9718) antibodies at 4°C overnight. After washing with PBS, cells were stained with the second antibody conjugated with Alexa Fluor 647 (Invitrogen, # A32733) at 37°C for 1 hour. Nuclei were stained with DAPI (4,6-diamidino-2-phenylindole). After adding the Fluoromount-G (SourthenBiotech, #0100), the cells were examined using a Zeiss LSM900 confocal microscope. All imaging results were processed using the ImageJ software. Phospho-γH2AX foci were quantified in individual nuclei, with 50 nuclei counted for foci quantification in each group.
+
+### In Vivo Tumorigenesis
+
+In the knockdown experiment, GSCs were transduced with lentiviral vectors expressing shDPYD, shTYMP, and a non-targeting control shRNA (shCont). After 48 hours of infection, cells were assessed by qPCR to confirm successful knockdown. Then the cells were counted and implanted intracranially into 4–6 weeks Balbc/nu mice purchased from the GemPharmatech Company in Jiangsu Province. All mice experiments were performed following an animal protocol approved by the Institutional Animal Care and Use Committee of the Shanghai Medical College, Fudan University. The animals were maintained until they displayed neurological symptoms, at which point they were sacrificed. The brains were collected, fixed in 4% formaldehyde, paraffin-embedded, and then sectioned. These sections were stained with H&E for histological analysis. In parallel survival experiments, animals were monitored until they developed neurological signs.
+
+## Statistical analysis
+
+All analyses were performed using R statistical software (R-4.1.1) and SPSS (version 19.0; SPSS Inc., USA). Unless otherwise specified, all tests were two-tailed, and P-values less than 5% were considered statistically significant.
+
+# References
+
+1. Ostrom QT et al (2014) The epidemiology of glioma in adults: a state of the science review. Neuro Oncol 16(7):896–913
+2. Louis DN et al (2016) The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol 131(6):803–820
+3. Louis DN et al (2021) The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol 23(8):1231–1251
+4. Brennan CW et al (2013) The somatic genomic landscape of glioblastoma. Cell 155(2):462–477
+5. Ceccarelli M et al (2016) Molecular Profiling Reveals Biologically Discrete Subsets and Pathways of Progression in Diffuse Glioma. Cell 164(3):550–563
+6. Cancer Genome Atlas Research (2008) Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455(7216):1061–1068
+7. Oh S et al (2020) Integrated pharmaco-proteogenomics defines two subgroups in isocitrate dehydrogenase wild-type glioblastoma with prognostic and therapeutic opportunities. Nat Commun 11(1):3288
+8. Yang K et al (2022) Glioma targeted therapy: insight into future of molecular approaches. Mol Cancer 21(1):39
+9. Varn FS et al (2022) Glioma progression is shaped by genetic evolution and microenvironment interactions. Cell 185(12):2184–2199e16
+10. Verhaak RG et al (2010) Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 17(1):98–110
+11. Sanson M et al (2009) Isocitrate dehydrogenase 1 codon 132 mutation is an important prognostic biomarker in gliomas. J Clin Oncol 27(25):4150–4154
+12. Hegi ME et al (2005) MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 352(10):997–1003
+13. Cairncross G et al (2013) Phase III trial of chemoradiotherapy for anaplastic oligodendroglioma: long-term results of RTOG 9402. J Clin Oncol 31(3):337–343
+14. Yanovich-Arad G et al (2021) Proteogenomics of glioblastoma associates molecular patterns with survival. Cell Rep 34(9):108787
+15. Wang LB et al (2021) Proteogenomic and metabolomic characterization of human glioblastoma. Cancer Cell 39(4):509–528e20
+16. Zhang B et al (2014) Proteogenomic characterization of human colon and rectal cancer. Nature 513(7518):382–387
+17. Mertins P et al (2016) Proteogenomics connects somatic mutations to signalling in breast cancer. Nature 534(7605):55–62
+18. Zhang H et al (2016) Integrated Proteogenomic Characterization of Human High-Grade Serous Ovarian Cancer. Cell 166(3):755–765
+19. Dong L et al (2022) Proteogenomic characterization identifies clinically relevant subgroups of intrahepatic cholangiocarcinoma. Cancer Cell 40(1):70–87e15
+20. Wu HL et al (2022) Targeting nucleotide metabolism: a promising approach to enhance cancer immunotherapy. J Hematol Oncol 15(1):45
+21. Mullen NJ, Singh PK (2023) Nucleotide metabolism: a pan-cancer metabolic dependency. Nat Rev Cancer 23(5):275–294
+22. Waitkus MS, Diplas BH, Yan H (2016) Isocitrate dehydrogenase mutations in gliomas. Neuro Oncol 18(1):16–26
+23. L MG et al (2017) Oncogenic Activities of IDH1/2 Mutations: From Epigenetics to Cellular Signaling. Trends Cell Biol 27(10):738–752
+24. Shirahata M et al (2018) Novel, improved grading system(s) for IDH-mutant astrocytic gliomas. Acta Neuropathol 136(1):153–166
+25. Iuchi T et al (2018) Clinical significance of the 2016 WHO classification in Japanese patients with gliomas. Brain Tumor Pathol 35(2):71–80
+26. Hanahan D, Weinberg RA (2011) Hallmarks of cancer: the next generation. Cell 144(5):646–674
+27. Bjorkblom B et al (2022) Distinct metabolic hallmarks of WHO classified adult glioma subtypes. Neuro Oncol 24(9):1454–1468
+28. Lafita-Navarro MC et al (2020) Inhibition of the de novo pyrimidine biosynthesis pathway limits ribosomal RNA transcription causing nucleolar stress in glioblastoma cells. PLoS Genet 16(11):e1009117
+29. Wang W et al (2021) Targeting Pyrimidine Metabolism in the Era of Precision Cancer Medicine. Front Oncol 11:684961
+30. Richards LM et al (2021) Gradient of Developmental and Injury Response transcriptional states defines functional vulnerabilities underpinning glioblastoma heterogeneity. Nat Cancer 2(2):157–173
+31. Bader JM et al (2023) Proteomics separates adult-type diffuse high-grade gliomas in metabolic subgroups independent of 1p/19q codeletion and across IDH mutational status. Cell Rep Med 4(1):100877
+32. Willson J (2022) Gliomas lean on pyrimidines. Nat Rev Cancer 22(11):606–607
+33. Shi DD et al (2022) De novo pyrimidine synthesis is a targetable vulnerability in IDH mutant glioma. Cancer Cell 40(9):939–956e16
+34. Pal S et al (2022) A druggable addiction to de novo pyrimidine biosynthesis in diffuse midline glioma. Cancer Cell 40(9):957–972e10
+35. Kato H et al (2021) DPYD, down-regulated by the potentially chemopreventive agent luteolin, interacts with STAT3 in pancreatic cancer. Carcinogenesis 42(7):940–950
+36. Siddiqui A, Ceppi P (2020) A non-proliferative role of pyrimidine metabolism in cancer. Mol Metab 35:100962
+37. Marangoni E et al (2018) Capecitabine Efficacy Is Correlated with TYMP and RB1 Expression in PDX Established from Triple-Negative Breast Cancers. Clin Cancer Res 24(11):2605–2615
+38. Gao Y et al (2021) Chemotherapy-induced CDA expression renders resistant non-small cell lung cancer cells sensitive to 5'-deoxy-5-fluorocytidine (5'-DFCR). J Exp Clin Cancer Res 40(1):138
+39. Tarar A, Alyami EM, Peng CA (2021) Mesenchymal stem cells anchored with thymidine phosphorylase for doxifluridine-mediated cancer therapy. RSC Adv 11(3):1394–1403
+40. Li W, Yue H (2018) Thymidine phosphorylase: A potential new target for treating cardiovascular disease. Trends Cardiovasc Med 28(3):157–171
+41. Fukushima M et al (2000) Structure and activity of specific inhibitors of thymidine phosphorylase to potentiate the function of antitumor 2'-deoxyribonucleosides. Biochem Pharmacol, 59(10): p. 1227-36
+42. Liekens S et al (2004) The nucleoside derivative 5'-O-trityl-inosine (KIN59) suppresses thymidine phosphorylase-triggered angiogenesis via a noncompetitive mechanism of action. J Biol Chem 279(28):29598–29605
+43. Hamzic S et al (2020) Fluoropyrimidine chemotherapy: recommendations for DPYD genotyping and therapeutic drug monitoring of the Swiss Group of Pharmacogenomics and Personalised Therapy. Swiss Med Wkly 150:w20375
+44. Paladhi A et al (2022) Targeting thymidine phosphorylase alleviates resistance to dendritic cell immunotherapy in colorectal cancer and promotes antitumor immunity. Front Immunol 13:988071
+45. World Medical A (2013) World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA 310(20):2191–2194
+46. Guo T et al (2015) Rapid mass spectrometric conversion of tissue biopsy samples into permanent quantitative digital proteome maps. Nat Med 21(4):407–413
+47. Cai X et al (2022) High-throughput proteomic sample preparation using pressure cycling technology. Nat Protoc 17(10):2307–2325
+48. Cai X et al (2021) PulseDIA: Data-Independent Acquisition Mass Spectrometry Using Multi-Injection Pulsed Gas-Phase Fractionation. J Proteome Res 20(1):279–288
+49. Meier F et al (2020) diaPASEF: parallel accumulation-serial fragmentation combined with data-independent acquisition. Nat Methods 17(12):1229–1236
+50. Mertins P et al (2018) Reproducible workflow for multiplexed deep-scale proteome and phosphoproteome analysis of tumor tissues by liquid chromatography-mass spectrometry. Nat Protoc 13(7):1632–1661
+51. Flavahan WA et al (2013) Brain tumor initiating cells adapt to restricted nutrition through preferential glucose uptake. Nat Neurosci 16(10):1373–1382
+52. Xie Q et al (2015) Mitochondrial control by DRP1 in brain tumor initiating cells. Nat Neurosci 18(4):501–510
+53. Wang Z et al (2021) The Hippo-TAZ axis mediates vascular endothelial growth factor C in glioblastoma-derived exosomes to promote angiogenesis. Cancer Lett 513:1–13
+54. Xu YP et al (2019) Tumor suppressor TET2 promotes cancer immunity and immunotherapy efficacy. J Clin Invest 129(10):4316–4331
+
+# Tables
+
+**Table 1**
+
+|  | No. (%) of glioma patients |
+|--- | --- |
+|  | No | Percentage |
+| **Age** (year) |  |  |
+| Median (range) | 58 () | - |
+| **Sex** |  |  |
+| Male | 120 | 63.8 |
+| Female | 68 | 36.2 |
+| **Grade** |  |  |
+| 1 | 6 | 3.2 |
+| 2 | 55 | 29.3 |
+| 3 | 33 | 17.6 |
+| 4 | 91 | 48.4 |
+| NA | 3 | 1.5 |
+| **Histopathology** |  |  |
+| Pilocytic astrocytoma | 6 | 3.2 |
+| Astrocytoma | 51 | 27.1 |
+| Oligodendroglioma | 47 | 25.0 |
+| Glioblastoma | 82 | 43.6 |
+| NEC | 2 | 1.1 |
+| **Tumor location** |  |  |
+| Frontal lobe | 99 | 52.7 |
+| Parietal lobe | 21 | 11.2 |
+| Temporal lobe | 46 | 24.5 |
+| Occipital lobe | 3 | 1.6 |
+| Insular lobe | 12 | 6.4 |
+| Midline | 3 | 1.6 |
+| Cerebellum | 4 | 2.1 |
+| **IDH status** |  |  |
+| Mutant | 97 | 51.6 |
+| Wild type | 91 | 48.4 |
+| **1p19q status** |  |  |
+| codel | 45 | 23.9 |
+| Non-codel | 77 | 41.0 |
+| NA | 66 | 35.1 |
+| **TERT status** |  |  |
+| Mutant | 71 | 37.8 |
+| Wild type | 51 | 27.1 |
+| NA | 6 | 35.1 |
+| **MGMT status** |  |  |
+| methylated | 84 | 44.7 |
+| unmethylated | 40 | 21.3 |
+| NA | 64 | 34.0 |
+| **Radiotherapy** |  |  |
+| Yes | 117 | 62.2 |
+| No | 51 | 27.1 |
+| NA | 20 | 10.6 |
+| **Chemotherapy** |  |  |
+| Yes | 117 | 62.2 |
+| No | 51 | 27.1 |
+| NA | 20 | 10.6 |
+
+# Supplementary Files
+
+- [Dataset1.xlsx](https://assets-eu.researchsquare.com/files/rs-3808475/v1/e6919402ef21be4c76102ac4.xlsx)  
+  Dataset 1
+
+- [SupplementaryInformation.pdf](https://assets-eu.researchsquare.com/files/rs-3808475/v1/696763cc180a2bb69266d31a.pdf)

05bf643a9d532f9bf0fa15baf0d9ab972ffa3fc6f3d979b4a60cf0f537b0972d/preprint/images_list.json

@@ -0,0 +1,66 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpeg",
+        "caption": "a) Shaded map of Campi Flegrei with the indication of the horizontal displacement of the GNSS stations in the period 2016-2024 (black arrows) and locations of the earthquakes that occurred between 2005 and 2024 (red dots). The RITE GNSS station, which is located in the center of the city of Pozzuoli, is indicated with a black square. Ellipses A and B mark the zones with the greatest concentration of epicentres, corresponding to the main seismogenic volumes, at crustal level. b) East-West section of the Campi Flegrei caldera with the hypocenters of the earthquakes recorded between 2005 and 2024. c) Temporal evolution of the cumulative earthquake count. d) Uplift measured from the vertical component of the GNSS RITE station. e) Soil CO2 flux in the Solfatara area (\u2018target area\u2019, (36))",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpeg",
+        "caption": "Seismograms and spectrograms of examples of periodical VT sequences (a-c). d) Seismogram and spectrogram of the earthquake of 27 April 2024, 03:44:56 UTC (Md = 3.9) located in the Gulf of Pozzuoli (group B in Figure 1a) reported as an example of a Campi Flegrei earthquake not included in a burst-like swarm. (e\u2013h) Seismograms and spectrograms of examples of burst-like swarms. These sequences also include earthquakes with relatively high magnitudes such as the one with Md 4.2 occurred on 27 September 2023 at 01:35 (UTC) (panel g) and the largest Campi Flegrei earthquake, with Md=4.4, recorded on May 20, 2024 at 18:10 (UTC) (panel h).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpeg",
+        "caption": "Comparison between the regular VT sequence of 12 October 2023 and the burst-like swarm of 14 April 2024. a) 30-minute seismogram of the two sequences. b) Location of the events of the regular VT sequence (blue circles) and the burst-like swarm (semi-transparent red circles). c) Seismogram and spectrogram of the 5-minute window of the regular VT sequence highlighted with a yellow rectangle in panel a); seismogram and spectrogram of the 5-minute window of the burst-like swarm highlighted with a yellow rectangle in panel a).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpeg",
+        "caption": "a) Map of the locations of the events of the sequences with short earthquake intertime (see the square in the inset). Red circles: burst-like swarm events; blue circles: periodical VTs; light gray circles: uncategorized. The dashed line includes the Solfatara-Pisciarelli hydrothermal area. The solid line contains the area of the Mt. Olibano geodetic anomaly. The earthquake locations are included in the supplementary material b) Red scale: residuals from the inversion of DInSAR Sentinel 1 data estimated in (46), highlighting the geodetic anomaly. The residuals are negative and are concentrated in the Monte Olibano area with a minimum of approximately -10 cm. The image of the Solfatara diffuse degassing structure, modified after Cardellini et al. (52), is shown on the map. c) Residuals between observed and expected vertical displacements of the ACAE GNSS station, in the area of the geodetic anomaly. d) CO2 flux in the target area of the Solfatara diffuse degassing structure, measured monthly by the INGV-OV staff.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpeg",
+        "caption": "Map with the locations of earthquakes with Md >1.0, which are distributed on an elliptical ring. The vertical displacement of the ground from 2015 to May 2024 obtained from DInSAR data is shown in blue scale on the map. The blue star indicates the location of the source of the deformation (depth 3800 +/-50 m). 1 focal mechanism of the earthquake of April 14, 2024 at 08:01:44, Md=3.0. 2 focal mechanism of the earthquake of February 5, 2023 at 00:45:36, Md=3.0.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.jpeg",
+        "caption": "a) Comparison between the uplift deficit of the GNSS ACAE station (cyan line), representative of the geodetic anomaly of Mt. Olibano, and the cumulative earthquake count in the period January 2021 - April 2024. The black line indicates the cumulative count of earthquakes recorded in the caldera. The correlation coefficients between earthquakes and ACAE uplift deficit is r = 0.998. b) cumulative earthquake count in the period January 2021 - April 2024 versus uplift deficit of the GNSS ACAE station. c) Comparison between the uplift of the GNSS RITE station (orange line), representative of the general deformation process (see deformation source Figure 5), and the cumulative earthquake count in the period January 2021 - April 2024. The correlation coefficients between all earthquakes and RITE up is r = 0.989. d) cumulative earthquake count in the period January 2021 - April 2024 versus RITE station uplift.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_7.jpeg",
+        "caption": "a) Mean vertical velocity computed for the 2021-2024 period (every color cycle represents a velocity variation of 4 cm/year). Inset shows a closed look to the velocity patterns in the Solfatara - Mt. Olibano area. b) Difference between the actual and the expected displacement time series for the vertical component for a pixel representative of the deformation behavior in the Mt. Olibano area. Dashed red lines correspond to 2s (standard deviation) error bar computed before 2021.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_8.jpeg",
+        "caption": "Solfatara diffuse degassing structure (DDS) and locations of the target area and of the 63 points that are monthly measured. The map is modified from Cardellini et al. (52) and Chiodini et al. (36) and shows the probability that the CO2 flux is > 50 g m-2 d-1, a suitable threshold for a pure biogenic CO2 flux. Above this threshold the CO2 emission is at least partially fed by the hydrothermal-magmatic source.\u00a0 Coordinates are expressed in UTM-WGS84",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

05bf643a9d532f9bf0fa15baf0d9ab972ffa3fc6f3d979b4a60cf0f537b0972d/preprint/preprint.md

@@ -0,0 +1,152 @@
+# Abstract
+
+Since 2021, peculiar seismic sequences became evident and frequent in Campi Flegrei caldera (Italy), while deformation, seismicity and gas emission showed an acceleration. We distinguished burst-like swarms and periodical VT sequences. The earthquakes of both types of sequences resulted located in an area that includes the main hydrothermal field, and a zone affected by a geodetic anomaly, which clearly appeared in 2021. Burst-like swarms (max Md = 4.4) are accompanied by a pseudo-tremor, suggesting a mechanism involving near-continuous brittle failure. The periodical VT sequences are shallow and appear linked to the dynamics of the Mt Olibano lava dome, which deforms non-uniformly compared to the rest of the caldera and coincides with the geodetic anomaly. This peculiar seismicity, described in the Campi Flegrei for the first time in this study, has been associated with phreatic explosions and critical phases of unrest in other volcanoes, and currently characterizes the rapidly evolving state of activity of this high-risk volcano.
+
+Earth and environmental sciences/Solid Earth sciences/Volcanology  
+Earth and environmental sciences/Natural hazards
+
+# Introduction
+
+The volcanic area of Campi Flegrei (Italy) is a subcircular caldera with a diameter of approximately 12 km (1, 2). The last eruption occurred in 1538 (3) and formed the volcanic edifice of Monte Nuovo. Today, the Campi Flegrei area is densely inhabited, with about 500,000 people living in this area (4). The main city is Pozzuoli, which is located in the central part of the caldera (Fig. 1). Ground deformations are typical of this caldera and, in the past, they have resulted in phases of subsidence alternating with phases of uplift. This phenomenon is called bradyseism and has been studied since the 19th century by the major scientists of that time (5, 6). In the first half of the 19th century, relative measurements of sea level compared to ground level (7) demonstrated that the Campi Flegrei caldera was in subsidence. Since then, the first uplift phase began in the 1950s (8). Then, two other bradyseism crises occurred in the 1969–1970 period (9) and during the 1982–1984 time interval (10–14), accompanied by significant uplift (1.7 and 1.8 m respectively) and seismicity. These bradyseism crises were followed by temporary subsidence (15, 16). Finally, the last phase of bradyseism began around 2005 and is still ongoing. The current phase is characterized by increasing seismicity (17–21) and fumarolic tremor amplitude, which is indicative of hydrothermal activity (22–26). Moreover, increasing fluid emission (27–36) and significant ground deformations (14, 37–45), which produced about 1.2 m uplift in the central sector of the caldera, are typical of the current unrest (Fig. 1).
+
+During the bradyseism crises of the second half of the 20th century and in the current one the deformation pattern has always shown a bell-shaped radial pattern with the maximum uplift in the central area of the caldera, where the RITE GNSS station is presently located (Fig. 1). In the 1982-84 crisis, deformation measurements were mainly based on topographic leveling, which provided vertical deformation measurements. For that period the deformation pattern was modeled as a Mogi source (10, 11) or as a horizontal crack (13, 13). The deformation pattern of the current unrest, which is measured through the GNSS network, managed by the Osservatorio Vesuviano of the Istituto Nazionale di Geofisica e Vulcanologia (INGV-OV) and through DInSAR measurements, has often been modeled as a horizontal crack (e.g. 40,41,45,46), or a horizontal thermo-poro-elastic zone (47).
+
+As a consequence of the current unrest, in December 2012 the Italian Civil Protection decided to move to the yellow alert phase, which is the second of four alert levels (green = basic; yellow = attention; orange = pre-alarm; red = alarm). From 2012 onwards, the unrest continued and accelerated (Fig. 1 c,d,e) in terms of increase in the earthquake occurrence rate, earthquake magnitude (17, 18, 25), ground deformation rate (18, 19) and variations in geochemical parameters (36). In particular, Bevilacqua et al. (19) have shown that the increase in the cumulative number of earthquakes (in total or above a given Md) versus the vertical uplift of the caldera is well fitted with two exponential functions and that the “connecting time” between these functions falls in the period between 4/2020 and 9/2022. This transition between the two exponential functions, the second having a larger exponent, marks a clear acceleration of the seismicity in the caldera. Furthermore, analyzing in detail the deformation pattern, Giudicepietro et al. (46) discovered a geodetic anomaly that clearly manifested in 2021, with an uplift deficit of approximately 9 cm in the Mt. Olibano area (East of Pozzuoli) compared to the surrounding areas.
+
+The aforementioned recent studies indicate that during 2021 the unrest of the Campi Flegrei caldera has intensified. For this reason, in this work we analyzed the seismicity that occurred in 2021–2024 with the aim of investigating the relationships between the characteristics of the earthquakes, the seismogenic processes, the evolution of the deformation field and the geochemical variations of the Campi Flegrei caldera. In particular, we focused our attention on several seismic sequences characterized by very short inter-event times, of even a few seconds. Within this category, we distinguished two types of sequences which we call burst-like swarms and periodical Volcano-Tectonic (VT) events, following the classic literature of volcanic seismology (i.e. 48,49). These types of sequences can provide clues on the seismogenic processes in relation to the evolution of the current accelerating unrest, which is controlled by intense ground deformation and escalating hydrothermal activity.
+
+# Results
+
+In Campi Flegrei the INGV-OV seismic network has recorded approximately 18,500 earthquakes (max Md = 4.4; average Md = 0.26, minimum Md= -1.6) from March 2005 to May 2024. Not all of these earthquakes are located because part of them are too small to obtain a reliable location (those located are around 10,200). The locations show a distribution that approximately describes an ellipse in the central area of the caldera, however the majority of earthquakes (about 90%) are located at Pozzuoli and east of Pozzuoli, in an area that includes the Solfatara-Pisciarelli hydrothermal system and the lava domes of Mt. Olibano (see the zone marked as A in Fig. 1a). A minor group of earthquakes (about 2% of the total) is located in the Gulf of Pozzuoli (zone B in Fig. 1a). The cumulative earthquake count highlights the progressive increase in seismicity rate over time (Fig. 1c). We focused on the seismicity in the period from 2021 to 2024, when the current unrest showed an intensification in terms of seismicity rate and uplift velocity. Specifically, we looked at the types of seismic events that show peculiar characteristics.
+
+The seismic events recorded in the Campi Flegrei caldera from 2021 to 2024 are generally of VT type. In the context of VT seismicity, we recognized sequences characterized by a very short time interval between two consecutive events, which we call here intertime (even less than a few seconds). In particular, we distinguished sequences of periodical VT events, which can be considered similar to drumbeat type events (49, 50) as regards the regularity of occurrence, although in the case of the Campi Flegrei these sequences contain a limited number of events. Moreover, we recognized earthquake sequences that we call burst-like swarms, following the definition of Hill et al. (48).
+
+Periodical VT type sequences (Fig. 2a-c) are generally characterized by events of short duration (from a few seconds to a few tens of seconds), with impulsive onset, which occur with earthquake intertimes comparable to each other (e.g. a few seconds or tens of seconds). Generally, no major events are associated with pure periodical VT sequences. Burst-like swarm sequences (Fig. 2e-h) are characterized by widely overlapping earthquakes, with intertimes that are sometimes not easily recognizable, and are generally accompanied by a continuous tremor-like background signal. However, the background signal cannot be defined as volcanic tremor because it does not show distinct frequency peaks as volcanic tremor typically exhibits. Therefore, we define this background signal as pseudo-tremor. The spectrogram analysis, which highlights the wide frequency band that characterizes the pseudo-tremor, suggests that it can be due to signals generated by a mechanism involving near-continuous brittle failure. Examples of this type of background signal are depicted in Fig. 2f, which shows a burst-like swarm recorded on 19 April 2023, starting with a pseudo-tremor signal, with increasing amplitude, and Fig. 2e, which shows the pseudo-tremor type signal recorded on 22 July 2022, with remarkable amplitude, in the second part of the recording. Burst-like swarm sequences can be accompanied by larger events. In particular, the earthquake with the greatest duration magnitude (Md = 4.4), recorded in the Campi Flegrei on 20 May 2024 at 18:10 (UTC), belongs to this type of sequence (Fig. 2h). The spectrogram helps to distinguish the different events in the burst. For comparison, the seismogram and spectrogram of an earthquake not associated with a burst-like swarm is shown in Fig. 2d. The selected event is the largest one (Md = 3.9) of the group of events located offshore in the Gulf of Pozzuoli (zone B in Fig. 1a).
+
+Looking at Fig. 2 it can be noted that the two categories of sequences in certain cases show common characteristics. For example, the burst-like swarm in Fig. 2f shows a regular sequence of VTs in the second half of the recording.
+
+To better investigate these seismic sequences, we selected two well-defined examples of the two types. Figure 3 shows the comparison between the periodical VT sequence recorded on 12 October 2023 and the burst-like swarm recorded on 14 April 2024. The burst-like swarm includes two earthquakes with Md > 3.0 and shows the background signal here defined as pseudo-tremor, well recognizable in the spectrogram (Fig. 3d). In the 5-minute signal windows of the two sequences reported in Fig. 3c-d, we counted 38 events in the regular VT type sequence (Fig. 3c), and 36 events in the burst-like swarm sequence, with an average earthquake intertime of approximately 8 seconds for both sequences.
+
+We compared the locations of the events belonging to these two sequences (Fig. 3b). For each sequence, we selected the locations of the events contained in the half hour of signal shown in Fig. 3a. For the periodical VT sequence of 12 October 2023, we obtained 10 locations with depths between 700 and 900 m and Md ranging between − 0.5 and 1.7. For the burst-like swarm sequence of 14 April 2024, we found 37 locations with depths between 730 and 2870 m and Md between − 0.3 and 3.7. Both sequences fall within the Solfatara-Olibano area (zone A in Fig. 1a).
+
+Finally, we selected all the sequences with very short earthquake intertimes that we recognized in the period between January 2021 and May 2024. In this time interval this type of seismic sequences has become more frequent and more evident than in previous years. In our selection we did not always distinguish between the two types as the distinction is not clear for all the cases. It is appreciable only in the clearest sequences, such as those reported in the examples of Figs. 2 and 3. We show the distribution of event epicenters belonging to the uncategorized very short earthquake intertimes sequences (light gray circles) in Fig. 4a, together with those classified as burst-like swarms (red circles) and those classified as periodical VTs (light blue circles). The burst-like swarms are those shown in Fig. 2e-h, the April 14, 2024 sequence reported in Fig. 3a, and a sequence recorded on June 2, 2023 (total number of events = 70; depth ranging between 450 and 2870 m and Md between − 0.8 and 4.4). The periodical VTs are the events in Fig. 2a-c and the sequence of October 12, 2023, shown in Fig. 3a (total number of events = 47; depth ranging between 210 and 1410 m and Md between − 0.1 and 2.7). Moreover, in Fig. 4a, we delimited the area of the geodetic anomaly discovered in Giudicepietro et al. (46) (solid line) and the Solfatara-Pisciarelli hydrothermal area (dashed line), based on the Solfatara Diffuse Degassing Structure (Solfatara DDS, 51). It is worth noting that the two types of events are closely spatially co-related as already shown in the examples in Fig. 3. They partly share the same seismogenic volume, especially in the border sector between the Solfatara-Pisciarelli hydrothermal system and the area of the Mt. Olibano geodetic anomaly. Actually, as a result of our analysis, it emerges that burst-like swarms often also contain periodical VT events which are generally localized at smaller depths (Fig. 3). However, the locations of burst-like events are generally slightly north of the locations of pure periodical VT sequences. This distribution indicates a greater concentration of burst-like swarms in the Solfatara-Pisciarelli hydrothermal area and the preferential locations of periodical VT events around the area of the geodetic anomaly (Fig. 4a, b). In this zone of the caldera, where the majority of the Campi Flegrei earthquakes are located, the diffuse degassing from the Solfatara crater increased significantly in the period 2021–2024 (Fig. 4b, d), when the burst-like and periodical VT sequences became evident. In the same period, the geodetic anomaly at the lava dome of Mt. Olibano became considerably more pronounced, as evidenced by the residual between the actual uplift of the ACAE GNSS station and the estimated one (Fig. 4c; see Materials and Methods).
+
+To frame this type of sequences in the general context of the ongoing unrest in the Campi Flegrei caldera, we considered the geodetic, GNSS and DInSAR measurements and the geochemical measurements and compared them with the ongoing seismicity.
+
+To show the distribution of earthquakes in the caldera, in relation to ground deformation, we selected earthquakes with Md > 1.0 and plotted them on the uplift map obtained with Sentinel 1 data (Fig. 5). We updated the processing performed in Giudicepietro et al.(46) to obtain the map of the vertical displacement of all the correlated pixels of the Sentinel 1 data from 2015 to May 2024 (maximum uplift approximately equal to 103 cm). We reported on the map the indication of the deformation source obtained in Giudicepietro et al. (46) whose parameters are: latitude = 40.817814 +/- 13 m; longitude = 14.126922 +/- 17 m; depth = 3826 +/- 45 m. The spatial distribution of seismicity in the Campi Flegrei caldera highlights the occurrence of earthquakes in an elliptical ring around the location of the ground deformation source (blue star in Fig. 5). We also added the focal mechanisms of two earthquakes, calculated with the FPFIT program (53), one falling in zone A (2024-04-14 08:01:44; Md = 3.0) and the other in zone B (2023-02-05 00:45:36, Md = 3.0) of Fig. 1a. The focal mechanism of the earthquake in zone A is extensive whereas the focal mechanism of the earthquake in zone B is compressive. This characteristic is also common to other earthquakes recorded in the two zones, which for simplicity of representation we do not show in the figure but which are reported in the INGV-OV surveillance bulletins (https://www.ov.ingv.it/index.php/monitoraggio-e-infrastrutture/bollettini-tutti/bollett-mensili-cf, last accessed 8 May 2024).
+
+However, the concentration of locations in zone A of Fig. 1a suggests the overlap of multiple seismogenic processes. In particular, in zone A there is the Solfatara-Pisciarelli hydrothermal area, with CO₂ diffuse degassing which shows a progressive increase over time and is currently comparable with the average flux of CO₂ in the plume of active volcanoes with continuous degassing (36). In recent years, the diffuse emission of CO₂ from the target area, which is systematically monitored, increased from 200–300 t/d in 2010 to values higher than 1200 t/d in the 2024 measurement campaigns (Fig. 1e). Furthermore, at the southern edge of the Solfatara crater, there is the geodetic anomaly of Mt. Olibano. Taking advantage of the high temporal sampling of the GNSS network, we calculated the vertical displacement deficit (uplift deficit) of the ACAE station, which is located in the geodetic anomaly, compared to the expected one. To do this, we applied a method similar to the one used with the DInSAR Sentinel 1 data to spatially map the geodetic anomaly in Giudicepietro et al. (46). In this case, we used RITE GNSS station, which well represents the uplift of the central area of the caldera, as a reference to derive the correlation coefficient to estimate the expected uplift of the ACAE station, from the time series relevant to the 2004–2024 period, therefore well before the start of the 2021 acceleration. We obtained the residuals by subtracting the expected uplift of the ACAE station from the observed uplift data (Fig. 4c). Finally, we defined the opposite of the residuals as the “uplift deficit" (see Material and Methods for details). This allowed us to retrieve the temporal evolution of the uplift deficit of the ACAE station located in the geodetic anomaly. Thus, we were able to correlate the uplift deficit in the anomaly area with the cumulative earthquake count, which is dominated by events occurring in area A of Fig. 1a. We found a very high correlation between the uplift deficit and the number of earthquakes (r = 0.998) (Fig. 6a, b). We also compared the vertical component of the RITE station with the cumulative earthquake count and obtained a slightly lower correlation between the two parameters (r = 0.988) and a non-linear relationship (Fig. 6c, d); this latter is consistent with the exponential-type trend analyzed in Bevilacqua et al. (55).
+
+# Discussion
+
+The picture that emerges from this study indicates that the current phase of bradyseism taking place in the Campi Flegrei caldera is accelerating. The process is controlled by the inflation of a source at a depth of about 3800 meters in the central area of the caldera (Fig. <span class="InternalRef" refid="Fig5">5</span>), whose evolution seems to justify the seismicity of the elliptical ring at the center of the caldera. This can be interpreted as the effect of a system of faults that border the uplifted block in the center of the caldera and create a ring fault-type structure. The focal mechanisms of the two earthquakes reported in Fig. <span class="InternalRef" refid="Fig5">5</span>, extensive in zone A and compressive in zone B, represent a common characteristic of the earthquakes that occur in these two areas. In particular, extensive mechanisms in zone A are reported in numerous articles (e.g. 24, 54) as well as in the INGV-OV surveillance bulletins. The compressive mechanisms in zone B are also reported in the surveillance bulletins, however some of these are less constrained than those in zone A, that is better covered by the seismic network. In any case, the mutual positions of these focal mechanisms suggest a possible dislocation of the block in the center of the caldera, which in depth could behave like a trapdoor-type structure. However, we remark that the interpretation of the deformation source located in the center of the caldera at a depth of approximately 3800 m is still speculative. It could be due to an accumulation zone of magmatic fluids, the expansion of a porous medium subject to an increase in pressure or the intrusion of magma.
+
+The seismic sequences with short earthquake intertimes, which are the main object of this study, are mainly concentrated in the seismogenic area A. Burst-like swarms has been observed in other volcanoes such as Mammoth Mountain (United States), a hydrothermally active lava dome complex in the Long Valley caldera (<span citationid="CR55" class="CitationRef">55</span>). For them, Hill et al. (<span citationid="CR48" class="CitationRef">48</span>) hypothesized a process driven by a transient increase in local fluid pressure, based on the similarity of these sequences with signals associated with pressure transients generated by shutdown operations on production wells in Japanese geothermal fields (<span citationid="CR56" class="CitationRef">56</span>). Nishi et al. (<span citationid="CR57" class="CitationRef">57</span>) recorded sequences of this type on White Island and observed that they were located in the same seismogenic volume as volcano-tectonic earthquakes. The mechanism that caused them was attributed to brittle failure due to rapid fluid pressure fluctuations. Also, Lin et al. (<span citationid="CR58" class="CitationRef">58</span>) recorded burst-like swarms on the Tatun Volcano Group (Taiwan) and interpreted them as the response of an extended network of cracks to the pressure variations due to fluid injection. Finally, a close association between these signals and hydrothermal activity is also suggested by McCausland et al. (<span citationid="CR59" class="CitationRef">59</span>) who consider them possible precursors of phreatic explosions when recorded in association with low-frequency and hybrid events. This leads us to interpret the burst-like swarms of the Campi Flegrei as the brittle response to the increase in the hydrothermal system fluid pressure. This interpretation is compatible with the model proposed by Fournier (<span citationid="CR60" class="CitationRef">60</span>) for the movement of fluid from a shallow magmatic zone to the hydrothermal environment. Furthermore, we have evidence that this variation in hydrothermal fluid pressure occurs in extensive stress regime. Indeed, the burst-like swarms share the same source region as other earthquakes occurring in zone A, which have extensive focal mechanisms indicating an extensive stress regime. Furthermore, the burst-like swarm locations are close to the Mt Olibano area, where a geodetic anomaly is located, which is a further evidence of the local extensive stress regime.
+
+Periodical VT sequences are also concentrated in the zone A. In particular they are distributed around the lava block of Mt Olibano at depths generally less than 1000 m. The regular occurrence of seismic events in volcanic environments has been associated with a stick-slip mechanism between an extruding lava dome and conduit walls (<span additionalcitationids="CR62" citationid="CR61" class="CitationRef">61</span>–<span citationid="CR63" class="CitationRef">63</span>). This hypothesis could also be suitable to explain the regular VT sequences recorded at Campi Flegrei. In this case there is not an extruding lava dome, but a dome in relative subsidence compared to the general uplift of the surrounding areas, as evidenced by the uplift deficit of Mt Olibano, which correlates closely with seismicity (Fig. <span class="InternalRef" refid="Fig6">6</span>) and degassing (Fig. <span class="InternalRef" refid="Fig4">4</span> c).
+
+In general, in the Campi Flegrei current unrest, the spatial distribution of earthquakes and the characteristics of the seismic sequences suggest the strong interconnection of the geodetic, hydrothermal and seismic phenomena, that are controlled by the deformation source located at the center of the caldera. When the uplift of the center of the caldera (RITE GNSS station) accelerates, the uplift deficit of the geodetic anomaly (ACAE GNSS station) and the cumulative earthquake count also accelerate. However, previous studies (<span citationid="CR18" class="CitationRef">18</span>, <span citationid="CR19" class="CitationRef">19</span>) have shown that seismicity, as represented by the cumulative earthquake count (total or above a given magnitude), increases more rapidly, in particular with two exponential regressions with increasing exponent in time, than GNSS data, such as the vertical component of the RITE station, which shows the greatest uplift among the GNSS stations. This difference in the temporal evolution of the two parameters is also observed in the period 2021–2024 (Fig. <span class="InternalRef" refid="Fig6">6</span> c-d). On the other hand, in this study we calculated the daily time series of the uplift deficit in the Mt Olibano area, identified as a geodetic anomaly (<span citationid="CR46" class="CitationRef">46</span>), exploiting the high acquisition rate of data from the INGV-OV GNSS network. It is noteworthy that this uplift deficit, due to the different uplift rate of the lava block of Mt Olibano compared to the surrounding areas, evolves similarly to seismicity, and shows a linear relationship with the cumulative earthquake count (Fig. <span class="InternalRef" refid="Fig6">6</span> a, b). This linear relationship suggests that the uplift deficit is a good indicator of the inelastic component of the behavior of this localized sector of the caldera in response to the overall deformation process. Therefore, we interpret this finding as the effect of the lithological and mechanical discontinuity of the crustal rock, which in this area begins to deviate from elastic behavior when subjected to deformation. This interpretation aligns with the recent study on the spatial distribution of the b value (<span citationid="CR21" class="CitationRef">21</span>), which highlights that the b value of the Mt. Olibano earthquakes is different from that found for the Solfatara-Piscairelli earthquakes, showing different physical characteristics of the medium.
+
+# Methods
+
+The seismic network of the Campi Flegrei is developed and managed by the INGV-OV. It includes 27 stations distributed within the rim of the caldera. Moreover, four stations were installed in the Gulf of Pozzuoli (64). Most of the stations are equipped with three-component broadband seismometers, with a sampling rate of 100 samples per second and with accelerometers, whose signals are acquired at 200 samples per second. The data is transmitted in real time to the INGV-OV acquisition center. In 2020, the location sensibility in space of the Campi Flegrei seismic network was characterized by a magnitude threshold ranging between 0 and 0.5 (17). In the Solfatara-Pisciarelli area, where a very local seismicity has become increasingly evident since 2010, the seismic network is denser and allows us to locate earthquakes with magnitude (Md) < 0, when the seismic noise is particularly low. To describe the general trend of the current long-term unrest, we used the locations of the earthquakes from 2005 to 2024, and the seismic catalogue of the Campi Flegrei from 2004 to 2024 (Fig. 1 a,b,c). The locations and the seismic catalogue are extracted from the INGV-OV seismological database (https://terremoti.ov.ingv.it/gossip/flegrei/index.html, last accessed 30 May 2024). The velocity model used by INGV-OV is reported in Calò and Tramelli (65).
+
+Since the burst-like and regular VT seismic sequences share the same source region as the other A-zone earthquakes, shown in Fig. 1, we also created a specific A-zone earthquake catalog. This catalog includes only earthquakes with first arrival recorded by a seismic station located in the A-zone (see the seismic station map in Fig. 1 of the supplementary material). In selecting the A-zone earthquakes, we considered both localized and non-localized events, including those recorded by only one or two stations, which are typically very small. Thus, we generated plots similar to those in Fig. 6, both with the cumulative count of all earthquakes recorded in the entire caldera and with the cumulative count of A-zone earthquake catalog. The two counts show negligible differences in their temporal evolution (see Fig. 2 of the supplementary materials), therefore, we no longer took into account the A-zone earthquake catalog, and in Fig. 6, we only showed the count of all earthquakes in the caldera (which is evidently dominated by the seismicity of zone A).
+
+The selection of the burst-like swarms and VT regular sequences was conducted through visual analysis of the seismograms. The spectrogram analysis and time series correlations were performed using ObsPy (66), a system for seismic data analysis, along with the Python packages numpy, scipy, and matplotlib for figure creation. Additionally, since the ground deformations and geochemical changes taking place at Campi Flegrei are significant, and are closely correlated with each other and with seismicity, we also used data from the INGV-OV GNSS network and DInSAR Sentinel-1 data.
+
+Data from the continuous GNSS network of Campi Flegrei were processed using Bernese GNSS software on a daily basis with the IGS final and reprocessing products to obtain homogeneous results. To remove the regional tectonic background from the volcanic deformation pattern, the time series and velocity fields were transformed into a local reference frame including six stations from the INGV RING network, located outside the Neapolitan volcanic area. No correction was applied to the vertical component, as the tectonic contribution was considered negligible. Furthermore, seasonal signals were removed (for more details on the data processing of the Campi Flegrei GNSS network see (67)). To obtain the time series of the vertical residuals of the ACAE GNSS station (Fig. 4), we applied the geodetic anomaly identification method from Giudicepietro et al. (46), based on DInSAR data, to the GNSS data. Considering the RITE station is located in the area of maximum uplift of the radially symmetrical bell-shaped deformation pattern of Campi Flegrei, and the ACAE station is located in Mt Olibano, we used these two stations to derive the uplift deficit in the area characterized by the geodetic anomaly. Specifically, we: 1) assumed RITE as a reference for the uplift trend in the central area of the caldera; 2) calculated a proportionality coefficient (α) between the RITE and ACAE uplift time series using data from 2015–2018, when the ACAE area was not yet affected by the geodetic anomaly; 3) estimated the expected ACAE uplift time series by multiplying the RITE uplift time series by α; 4) subtracted the expected ACAE uplift time series from the actual ACAE uplift time series, obtaining the residual series from 2004 to 2024 as shown in Fig. 4. Then, we calculated the opposite of the residuals to obtain the time series of the “uplift deficit” shown in Fig. 6 a.
+
+The DInSAR data were generated by applying the Parallel Small BAseline Subset (P-SBAS) approach (68, 69). In particular, we separately processed 430 ascending (Track 44) and 429 descending (Track 22) orbits data acquired by Sentinel-1 constellation from March 24, 2015 to May 24, 2024 to obtain the Line of Sight (LOS) displacement time series for each coherent pixel. Then, the vertical displacement component was calculated by combining the LOS P-SBAS time series using the method described in Casu and Manconi (70). The mean velocity of the vertical component computed for the 2021–2024 period represented as DInSAR fringes (every color cycle corresponds to a velocity variation of 4 cm/year) is shown in Fig. 7 a. It clearly highlights that the uplift deficit in the Mt. Olibano area, starting form 2021, makes the fringes to deviate from a radial pattern, which instead is preserved in the whole caldera. The vertical component of the Sentinel-1 data, updated to May 24, 2024, is also shown in blue scale in Fig. 5.
+
+Finally, we used geochemical data from the INGV-OV geochemical monitoring network. The data consists in the CO₂ emission (in t/d) from an area of ~ 90000 m² (‘target area’ in Fig. 8) located inside the Solfatara crater. The CO₂ emission is monthly computed through the measurement of 63 fixed points with the accumulation chamber method (71). This monitoring activity started in 2004 and is still ongoing. The data show a marked increase of the CO₂ emission in the last years (Fig. 4 d). Further details of the technique measurement and data treatment are reported in Chiodini et al. (36) and Cardellini et al. (52).
+
+# References
+
+1. Rosi M, Sbrana A, Principe C (1983) The Phlegraean Fields: Structural evolution, volcanic history and eruptive mechanisms. J Volcanol Geotherm Res 17:273–288
+2. Smith VC, Isaia R, Pearce NJG (2011) Tephrostratigraphy and glass compositions of post-15 kyr Campi Flegrei eruptions: implications for eruption history and chronostratigraphic markers. Quat Sci Rev 30:3638–3660
+3. Di Vito MA et al (2016) Magma transfer at Campi Flegrei caldera (Italy) before the 1538 AD eruption. Sci Rep 6:32245
+4. Turchi A, Lumino R, Gambardella D, Leone MF (2023) Coping capacity, adaptive capacity, and transformative capacity preliminary characterization in a multi-hazar resilience perspective: The Soccavo district case study (City of Naples, Italy). Sustainability 15:10877
+5. Babbage C (1847) Observations on the temple of Serapis, at Pozzuoli, near Naples, with remarks on certain causes which may produce geological cycles of great extent. Q J Geol Soc 3:186–217
+6. Lyell C (1835) Principles of Geology, vol 1. John Murray, London
+7. Niccolini A (1846) Descrizione della gran Terma Puteolana volgarmente detta Tempio di Serapide. Napoli
+8. Del Gaudio C, Aquino I, Ricciardi GP, Ricco C, Scandone R (2010) Unrest episodes at Campi Flegrei: A reconstruction of vertical ground movements during 1905–2009. J Volcanol Geotherm Res 195:48–56
+9. Corrado G, Guerra I, Lo Bascio A, Luongo G, Rampoldi R (1977) Inflation and microearthquake activity of Phlegraean Fields, Italy. Bull Volcanol 40:169–188
+10. De Natale G et al (2006) The Campi Flegrei caldera: unrest mechanisms and hazards. In: Troise C, De Natale G, Kilburn CRJ (eds) Mechanisms of Activity and Unrest at Large Calderas, vol 269. Geological Society, London, pp 25–45
+11. Gottsmann J, Rymer H, Berrino G (2006) Unrest at the Campi Flegrei caldera (Italy): A critical evaluation of source parameters from geodetic data inversion. J Volcanol Geotherm Res 150:132–145
+12. Battaglia M, Troise C, Obrizzo F, Pingue F, De Natale G (2006) Evidence for fluid migration as the source of deformation at Campi Flegrei caldera (Italy). Geophys Res Lett 33:L01307
+13. Woo JYL, Kilburn CRJ (2010) Intrusion and deformation at Campi Flegrei, southern Italy: Sills, dikes, and regional extension. J Geophys Res 115:B12210
+14. Giudicepietro F, Macedonio G, Martini M (2017) A physical model of sill expansion to explain the dynamics of unrest at calderas with application to Campi Flegrei. Front Earth Sci 5:1–11
+15. Manconi A et al (2010) On the effects of 3-D mechanical heterogeneities at Campi Flegrei caldera, southern Italy. J Geophys Res 115:B08405
+16. D’Auria L et al (2011) Repeated fluid-transfer episodes as a mechanism for the recent dynamics of Campi Flegrei caldera (1989–2010). J Geophys Res 116:B04313
+17. Tramelli A et al (2021) Statistics of seismicity to investigate the Campi Flegrei caldera unrest. Sci Rep 11
+18. Bevilacqua A et al (2022) Data analysis of the unsteadily accelerating GPS and seismic records at Campi Flegrei caldera from 2000 to 2020. Sci Rep 12:19175
+19. Bevilacqua A et al (2024) Investigating deformation and seismic data and their relationships during the ongoing unrest of Campi Flegrei caldera (Italy). PREPRINT (Version 1) available. http://doi.org/10.21203/rs.3.rs-4164255/v1. at Research Square
+20. Tramelli A, Giudicepietro F, Ricciolino P, Chiodini G (2022) The seismicity of Campi Flegrei in the contest of an evolving long term unrest. Sci Rep 12:2900
+21. Tramelli A, Convertito V, Godano C (2024) b value enlightens different rheological behaviour in Campi Flegrei caldera. Commun. Earth Environ. 5, 275
+22. Chiodini G et al (2017) Fumarolic tremor and geochemical signals during a volcanic unrest. Geology 45:1131–1134
+23. Giudicepietro F et al (2019) Insight into Campi Flegrei caldera unrest through seismic tremor measurements at Pisciarelli fumarolic field. Geochem Geophys Geosyst 20:5544–5555
+24. Giudicepietro F et al (2020) Tracking episodes of seismicity and gas transport in Campi Flegrei caldera trough seismic, geophysical and geochemical measurements. Seismol Res Lett 92:965–975
+25. Giudicepietro F et al (2021) Campi Flegrei, Vesuvius and Ischia seismicity in the context of the Neapolitan volcanic area. Front Earth Sci 9
+26. Esposito AM, De Cesare W, Macedonio G, Giudicepietro F (2023) Efficient SOM’s application to seismic fumarolic tremor for the detection of anomalous hydrothermal activity in Campi Flegrei volcano (Italy). Appl Sci 13:5505
+27. Chiodini G, Caliro S, De Martino P, Avino R, Gherardi F (2012) Early signals of new volcanic unrest at Campi Flegrei caldera? Insights from geochemical data and physical simulations. Geology 40:943–946
+28. Chiodini G et al (2015) Evidence of thermal-driven processes triggering the 2005–2014 unrest at Campi Flegrei caldera. Earth Planet Sci Lett 414:58–67
+29. Tamburello G et al (2019) Escalating CO2 degassing at the Pisciarelli fumarolic system, and implications for the ongoing Campi Flegrei unrest. J Volcanol Geotherm Res 384:151–157
+30. Siniscalchi A et al (2019) Reservoir structure and hydraulic properties of the Campi Flegrei geothermal system inferred by audiomagnetotelluric, geochemical, and seismicity study. J Geophys Res Sol Ea 124:5336–5356
+31. Ambrosino F, Sabbarese C, Roca V, Giudicepietro F, Chiodini G (2020) Analysis of 7-years Radon time series at Campi Flegrei area (Naples, Italy) using artificial neural network method. Appl Radiat Isot 163:109239
+32. Ambrosino F, Sabbarese C, Roca V, Giudicepietro F, De Cesare W (2020) Connection between Rn emission and geophysical-geochemical parameters recorded during the volcanic unrest at Campi Flegrei caldera (2011–2017). Appl Radiat Isot 166
+33. Ambrosino F, Thinová L, Briestenský M, Sabbarese C (2019) Anomalies identification of Earth’s rotation rate time series (2012–2017) for possible correlation with strong earthquakes occurrence. Geod Geodyn 10:455–459
+34. Sabbarese C et al (2020) Continuous radon monitoring during seven years of volcanic unrest at Campi Flegrei caldera (Italy). Sci Rep 10
+35. Ambrosino F et al (2021) Study of surface emissions of Rn (Thoron) at two sites in the Campi Flegrei caldera (Italy) during volcanic unrest in the period 2011–2017. Appl Sci 11:5809
+36. Chiodini G et al (2021) Hydrothermal pressure-temperature control on co2 emissions and seismicity at Campi Flegrei (Italy). J Volcanol Geotherm Res 414
+37. D’Auria L, Giudicepietro F, Martini M, Lanari R (2012) The 4D imaging of the source of ground deformation at Campi Flegrei caldera (southern Italy). J Geophys Res 117:B08209
+38. Amoruso A et al (2014) Clues to the cause of the 2011–2013 Campi Flegrei caldera unrest, Italy, from continuous GPS data. Geophys Res Lett 41:3081–3088
+39. Macedonio G, Giudicepietro F, D’Auria L, Martini M (2014) Sill intrusion as a source mechanism of unrest at volcanic calderas. J Geophys Res Sol Ea 119:3986–4000
+40. D’Auria L et al (2015) Magma injection beneath the urban area of Naples: a new mechanism for the 2012–2013 volcanic unrest at Campi Flegrei caldera. Sci Rep 5:1–11
+41. Trasatti E, Polcari M, Bonafede M, Stramondo S (2015) Geodetic constraints to the source mechanism of the 2011–2013 unrest at Campi Flegrei (Italy) caldera. Geophys Res Lett 42:3847–3854
+42. Giudicepietro F, Macedonio G, D’Auria L, Martini M (2016) Insight into vent opening probability in volcanic calderas in the light of a sill intrusion model. Pure Appl Geophys 173:1703–1720
+43. Todesco M (2021) Caldera’s breathing: poroelastic ground deformation at Campi Flegrei (Italy). Front Earth Sci 9:702655
+44. Bevilacqua A et al (2020) Radial interpolation of gps and leveling data of ground deformation in a resurgent caldera: application to Campi Flegrei (Italy). J Geodesy 94:24
+45. Amoruso A, Crescentini L (2022) Clues of ongoing deep magma inflation at Campi Flegrei caldera (Italy) from Empirical Orthogonal Function analysis of SAR data. Remote Sens 14:5698
+46. Giudicepietro F et al (2024) First evidence of a geodetic anomaly in the Campi Flegrei caldera (Italy) ground deformation pattern revealed by DInSAR and GNSS measurements during the 2021–2023 escalating unrest phase. Submitted
+47. Nespoli M, Tramelli A, Belardinelli ME, Bonafede M (2023) The effects of hot and pressurized fluid flow across a brittle layer on the recent seismicity and deformation in the Campi Flegrei caldera (Italy). J Volcanol Geotherm Res 443:107930
+48. Hill DP et al (1990) The 1989 earthquake swarm beneath Mammoth Mountain, California: An initial look at the 4 May through 30 September activity. BSSA 80:325–339
+49. Chouet BA, Matoza RS (2013) A multi-decadal view of seismic methods for detecting precursors of magma movement and eruption. J Volcanol Geotherm Res 252:108–175
+50. Waite GP, Chouet BA, Dawson PB (2008) Eruption dynamics at Mount St. Helens imaged from broadband seismic waveforms: Interaction of the shallow magmatic and hydrothermal systems. J Geophys Res Sol Ea 113:B02306
+51. Chiodini G et al (2001) degassing and energy release at Solfatara volcano, Campi Flegrei, Italy. J Geophys Res 106:16213–16221
+52. Cardellini C et al (2017) Monitoring diffuse volcanic degassing during volcanic unrests: the case of Campi Flegrei (Italy). Sci Rep 7:6757
+53. Reasenberg PA, Oppenheimer DFPFIT (1985) FPPLOT, and FPPAGE: Fortran computer programs for calculating and displaying earthquake fault-plane solutions. Open-File Report 85–739, U.S. Geological Survey
+54. di Scotto F et al (2023) Delineation and fine-scale structure of fault zones activated during the 2014–2024 unrest at the Campi Flegrei caldera (Southern Italy) from high-precision earthquake locations. Geophys. Res. Lett. 51, eGL107680 (2024)
+55. Bevilacqua A et al (2018) Late quaternary eruption record and probability of future volcanic eruptions in the Long Valley volcanic region (CA, USA). J Geophys Res Sol Ea 123:5466–5494
+56. Niitsuma H, Nakatsuka K, Chubachi N, Yokoyama H, Takanohashi M (1985) Acoustic emission measurement of geothermal reservoir cracks in Takinoue (Kakkonda) field, Japan. Geothermics 14:525–538
+57. Nishi Y, Sherburn S, Scott BJ, Sugihara M (1996) High-frequency earthquakes at White Island volcano, New Zealand: insights into the shallow structure of a volcano-hydrothermal system. J Volcanol Geotherm Res 72:183–197
+58. Lin CH et al (2005) Preliminary analysis of volcanoseismic signals recorded at the Tatun Volcano Group, northern Taiwan. Geophys Res Lett 32:L10313
+59. McCausland WA et al (2019) Using a process-based model of pre-eruptive seismic patterns to forecast evolving eruptive styles at Sinabung Volcano, Indonesia. J Volcanol Geotherm Res 382:253–266
+60. Fournier RO (1999) Hydrothermal processes related to movement of fluid from plastic into brittle rock in the magmatic-epithermal environment. Econ Geol 94:1193–1211
+61. Iverson RM et al (2006) Dynamics of seismogenic volcanic extrusion at Mount St Helens in 2004-05. Nature 444:439–443
+62. Iverson RM (2008) Dynamics of seismogenic volcanic extrusion resisted by a solid surface plug, Mount St. Helens. In Sherrod, D. R., Scott, W. E. & Stauffer, P. H. (eds.) A volcano rekindled: the renewed eruption of Mount St. Helens, 2004–2006, Professional Paper 1750, Chap. 21, 425–460, USGS Publications Warehouse
+63. Harrington RM, Brodsky EE (2007) Volcanic hybrid earthquakes that are brittle-failure events. Geophys Res Lett 35:L06308
+64. Iannaccone G et al (2018) Measurement of seafloor deformation in the marine sector of the Campi Flegrei caldera (Italy). J Geophys Res Sol Ea 123:66–83
+65. Calò M, Tramelli A (2018) Anatomy of the Campi Flegrei caldera using Enhanced Seismic Tomography Models. Sci Rep 8
+66. Krischer L et al (2015) Obspy: a bridge for seismology into the scientific Python ecosystem. Comput Sci Discov 8:014003
+67. De Martino P, Dolce M, Brandi G, Scarpato G, Tammaro U (2021) The ground deformation history of the Neapolitan volcanic area (Campi Flegrei caldera, Somma-Vesuvius volcano, and Ischia Island) from 20 years of continuous GPS observations (2000–2019). Remote Sens 13:2725
+68. Casu F et al (2014) SBAS-DInSAR parallel processing for deformation time series computation. IEEE JSTARS 7:3285–3296
+69. Manunta M et al (2019) The parallel SBAS approach for Sentinel-1 interferometric wide swath deformation time-series generation: Algorithm description and products quality assessment. IEEE T Geosci Remote 57:6259–6281
+70. Casu F, Manconi A (2016) Four-dimensional surface evolution of active rifting from spaceborne SAR data. Geosphere 12:697–705
+71. Chiodini G, Cioni R, Guidi M, Raco B, Marini L (1998) Soil co2 flux measurements in volcanic and geothermal areas. Appl Geochem 13:543–552
+
+# Supplementary Files
+
+- [dataset1locallsequences.txt](https://assets-eu.researchsquare.com/files/rs-4708123/v1/e360615bafbdd4f6b0762fb6.txt)  
+  Dataset 1
+
+- [dataset2locburstlikeswarms.txt](https://assets-eu.researchsquare.com/files/rs-4708123/v1/50fd5d747bd1239c7a632f7e.txt)  
+  Dataset 2
+
+- [dataset3locperiodicalVT.txt](https://assets-eu.researchsquare.com/files/rs-4708123/v1/45e14458072714c09854d731.txt)  
+  Dataset 3
+
+- [dataset4databaseCO2flux.xls](https://assets-eu.researchsquare.com/files/rs-4708123/v1/8628da599b027a60b29483a0.xls)  
+  Dataset 4
+
+- [supplementaryinfo.docx](https://assets-eu.researchsquare.com/files/rs-4708123/v1/d02f8ce6b6d7f20d9e4c479d.docx)

05d4b9f2850394e9266c31450e9dda48f8747cd9d03b3709c218ddce72f9a01d/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Structural comparison of characteristic cyclic amino acid residues identified in known RiPPs, lanthionine (Lan), labionin (Lab) and avionin (Avi) moieties with the novel amino acid moiety, methyl-amino-bithionin (MAbi) in 1 in this study. B. Other amino acid residues resulting from different post-translational modifications (PTMs) in 1. C. Schematic representation of 1. Residues derived from Ser, Thr and Cys are shown as orange, green and brown balls, respectively. Abbreviation: Dha, dehydroalanine; Dhb, dehydrobutyrine; DAla, D-alanine; Aaa, 3-aminoacrylic acid.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Structure of 1 determined using a combination of NMR, HR-MS, genomic and computational calculation analyses and isotope fine structure analysis of the y6 fragment to confirm elemental formula. A. Structural elucidation of 1 by NMR, key NOE correlations used to solve the structure are indicated. The coloured dash lines represent the estimated NOE distances through the interpretation of NOE spectrum. Use of NOE/NMR restraints17,18 have been used successfully to provide relevant solution conformations of cyclic peptides. The configurations in 1 were confirmed by either Marfey derivatization or comprehensive NMR elucidation except for stereogenic centres in star (*). B. The overall isotopic signal for the y6 ion. C and D. The isotopic fine structure of the M+1 and M+2 signals are shown in more detail with the theoretical isotope signals for [C26H34N7O8S1]+ ion shown overlaid as a scatter plot (green). Individual isotopologues are annotated.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Molecular dynamics snapshots of the computational model of the structure of 1 calculated using the GFN2-xTB-MD method. A. the cage-like C-terminal with possible H-bonds interactions (pink lines) among the back bond where the MAbi residue is buried inside the \u201ccage\u201d as well as the \u03b2-amino acid, Aaa-7, in the bottom. B. the helix-like N-terminal with possible H-bonds (pink lines) between the side chain carboxylate and NH residues of the backbone peptide as well as the \u03b2-amino acid, Aaa-7, on the top.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "(A). The minimal biosynthetic gene cluster of 1. (B). EIC traces of the ions corresponding to the peptide 1 and the biosynthetic intermediates, 2 and 3, in different Streptomyces coelicolor M1152 variants. (i). a variant harbouring an empty plasmid, pCAP03; (ii). a variant harbouring the construct with the kin BGC; (iii). A knockout variant \u0394kinC; (iv). A knockout variant \u0394kinD; (v). A knockout variant \u0394kinE; (vii). A knockout variant \u0394kinF; (viii). A knockout variant \u0394kinH; (ix) A knockout variant \u0394kinI; (x). A knockout variant \u0394kinJ; (xi). A knockout variant \u0394kinO; (vi) and (xi). The isolated 1 as the standard.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpg",
+        "caption": "Phylogenetic analysis of KinA-like proteins and comparison of representative BGCs. KinA homologs containing different conserved residues were classified based on the primary alignment. A maximum likelihood tree (branch lengths removed) was constructed and color-coded to depict different groups.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

05d4b9f2850394e9266c31450e9dda48f8747cd9d03b3709c218ddce72f9a01d/preprint/preprint.md

@@ -0,0 +1,182 @@
+# Abstract
+
+Ribosomally synthesized and post-translationally modified peptides (RiPPs) are structurally complex naturally occurring metabolites across all three domains of life. Despite the structural diversity of RiPPs that stems from the extensive post-translational modifications, only α-amino acid residues have been found in known RiPPs. Here we report discovery of a new 27-mer peptide, kintamdin, using comprehensive MS and NMR structural elucidation and genomic analysis together with computational modelling. The peptide features a β-amino acid residue and a new thioether macrocyclic ring. Heterologous expression and gene inactivation allowed the identification of the minimal biosynthetic gene cluster (BGC). The gene products in *kin* BGC share low homologues compared to other known RiPP pathways, further rendering the novelty of kintamdin. Biochemical analysis indicated that KinO mediate di-methylation reaction to yield kintamidn. Finally, the occurrence of the *kin*-like BGCs in Gram-positive bacteria suggested the biological importance of this new group of RiPPs.
+
+Nanoscience  
+Molecular Biology  
+Organic Chemistry  
+Ribosomally synthesized and post-translationally modified peptides (RiPPs)  
+metabolites
+
+# Introduction
+
+Peptide natural products are a subset of secondary metabolites and can be of ribosomal or non-ribosomal origin. While non-ribosomal peptides (NRPs) have been studied for decades, the chemical diversity of peptides from ribosomal origin, now called ribosomally-synthesized and post-translationally modified peptides (RiPPs), has just been recognised across all three domains of life during the last decade, as a result of recent advances in genome sequencing technology<sup>1</sup>. It is now known that the majority of RiPPs are synthesized as precursor peptides, typically ranging from 20 to 110 amino acid residues in length, encoded by a structural gene<sup>2</sup>.
+
+Although the RiPPs exploit only the 20 proteinogenic amino acids. extensive post-translational modifications (PTMs) provide these peptides with a great spectrum of chemical diversity. The dehydroalanine (Dha) and (<em>Z</em>)-dehydrobutyrine ((<em>Z</em>)-Dhb) residues are commonly found in the RiPPs. Dha and (<em>Z</em>)-Dhb residues in RiPPs are commonly derived from serine/cysteine and threonine respectively<sup>3</sup>. In some occasions, D-amino acid residues, such as D-Ala, have been introduced in lanthipeptides by an F<sub>420</sub>H<sub>2</sub>-dependent reductase<sup>4</sup>. The structural complexity in RiPPs also reflects that they contain a range of cyclic motifs, including noncanonical thioether amino acid residues, lanthionine (Lan), labionin (Lab)<sup>5</sup> and more recently a hybrid aminovinlycysteine (AviCys)-labionin (Avionin)<sup>6</sup> (Fig. <span class="InternalRef">1</span> A), which are enzymatically installed as PTMs. The enzymes responsible for these PTMs catalyse the dehydration of either Ser or Thr residues to generate dehydroalanine (Dha) or dehydrobutyrine (Dhb), respectively, followed by the intramolecular Michael-like addition of the side chain of Cys to Dha or Dhb to form Lan motif. When a second Dha motif is involved, the triamino triacid Lab can be formed. In case of the avionin formation, biochemical analysis demonstrated that a ribosomal peptide is post-translationally modified by oxidative decarboxylation of the C-terminal Cys and subsequent cyclization to yield avioin<sup>6</sup>.
+
+Unlike NRPs that contain a range of β-amino acids, only α-amino acid motifs have been identified in the RiPP natural product inventory thus far. Interestingly, recent conserved genomic survey has discovered the unprecedented function of orphan radical SAM enzymes (PlpX-type spliceases) encoded in several bacterial RiPP biosynthetic gene clusters (BGC). Biochemical analysis demonstrated that PlpX splicease, together with its partner protein pair, PlpY, in the <em>plp</em> BGC from the cyanobacterium <em>Pleurocapsa</em> catalyses an unusual peptide backbone slicing reaction to remove all atoms of a C-terminal Tyr unit except the amide carbonyl in the precursor peptide PlpA3 to generate an α-keto-β-amino residue, suggesting that the corresponding mature RiPP metabolite contains β-amino acid residues. However, the structure of this RiPP directed by the <em>plp</em> BGC has not yet been determined<sup>2, 7</sup>.
+
+Here we report structural and biosynthetic investigation of a new macrocyclic peptide that does not fall into any of the known categories of RiPPs and is thus the founding member of a new family of RiPPs. It is produced by a soil isolate, <em>Streptomyces</em> sp. RK44 and bears unprecedented chemical motifs, including <em>bis</em>-thioether crosslink, methyl-amino-bithionin (MAbi), and an unsaturated (<em>Z</em>)-β-amino acid ((<em>Z</em>)-3-amino-acrylic acid (Aaa) or Δ<sup>Z</sup>βAla) residues, together with motifs that are observed in other classes of RiPPs, such as D-Ala, <em>N</em>, <em>N</em>-dimethylated N-terminal amino acid residues, Dha and Dhb residues. The structure of <strong>1</strong> has been elucidated by comprehensive high-resolution MS and NMR analyses coupled with molecular dynamics studies at the density functional tight-binding level of theory. The identity of the minimal biosynthetic gene cluster (BGC) directing the production of <strong>1</strong> is confirmed through heterologous expression and gene inactivation. Biochemical investigation demonstrated that KinO is responsible for the installation of <em>N</em>, <em>N</em>-methylation at the <em>N</em>-terminal of <strong>1</strong>. The biosynthetic gene cluster display low homology with those of known RiPPs, suggesting a new RiPP biosynthetic pathway. Finally, we provide evidence of the occurrence of these biosynthetic gene clusters in actinobacteria, suggesting an important biological function in this genus.
+
+# Results
+
+## Identification of kintamdin 1 from *Streptomyces* sp RK44
+
+*Streptomyces* sp. RK44 is a new streptomyces strain isolated from a soil sample collected at the Kintampo waterfall in the Bono East of Ghana in 2014. Metabolite profiling of the strain revealed the presence of a high molecular weight metabolite (2507 Da) and other metabolites<sup>8</sup> under laboratory culture conditions. Large scale fermentation (10 L), followed by chemical workup allowed the isolation of pure compound <strong>1</strong> (3 mg).
+
+Analysis of high-resolution electrospray ionization mass spectrometry (HR-ESIMS) gave a [M + 3H]<sup>3+</sup> ion at *m*/*z* of 836.7524, indicating a neutral monoisotopic mass of 2507.2349 Da (Supplementary Fig. 1). Inspection of <sup>1</sup>H NMR indicated the presence of numerous overlapping α protons of amino acid (δ<sub>H</sub> 3.4–4.6 ppm), suggesting that <strong>1</strong> is a peptidic natural product (Supplementary Fig. 2). <strong>1</strong> displays a pair of unusual H signals (5.3 and 6.85 ppm, respectively, 8.4 Hz *j*-coupling), suggesting the presence of a moiety containing *cis* configuration (Fig. 1C and Supplementary Fig. 2). Interpretation of 1- and 2-D NMR spectra enabled identification of the major fragment (19 mers) of the peptide (Supplementary Figs. 3–7). Along with different proteinogenic amino acids, modified residues such as dehydroalaine (Dha) and dehydrobutyrine (Dhb) were found in <strong>1</strong> (Supplementary Figures S8-11). The unit containing *cis* alkene was identified as an unsaturated β amino acid, (*Z*)-3-amino-acrylic acid (Aaa), which to the best of our knowledge, has not been previously reported in any other RiPPs (Supplementary Figures S12-14). The sequence tag was also confirmed through assignment of *de novo* analysis of MS<sup>n</sup> spectra using tandem MS fragmentation and the sequential fragment ions (Supplementary Fig. 15 and Supplementary Table 1). Almost all of the mass shifts generated can be substituted with proteinogenic amino acids except mass shifts of 69 and 83 Da which were assigned to the non-proteinogenic amino acids, Dha and Dhb. It is worth to note that Aaa (Δ<sup>Z</sup> βAla), presumably a rearranged product of dehydrated serine, has the same molecular weight as Dha in the MS analysis. Collectively, the NMR data combined with MS analysis allowed us to connect the major fragments of the <strong>1</strong> sequence while both sequences in both N and C terminals remain undetermined.
+
+## Kintamdin 1 featuring unusual chemical moieties
+
+To assist the structural elucidation of the remaining sequence of <strong>1</strong>, we took advantage of genome mining strategy. Blast search in the annotated RK44 genome in RAST servers<sup>9</sup> using the sequence tag as a probe led to the identification of a 174 bp open reading frame (*orf*) encoding the precursor peptide, KinA (Supplementary Figure 16). With the AA sequence on hands, the molecular formula of <strong>1</strong> was established as C<sub>115</sub>H<sub>174</sub>N<sub>28</sub>O<sub>31</sub>S<sub>2</sub> based on the HR-ESIMS analysis (observed [M+ H]<sup>+</sup> = 2508.2427, calculated [M+ H]<sup>+</sup> = 2508.2414, Δ = 0.518 ppm]. We then revisited the NMR spectra including <sup>1</sup>H, COSY, TOCSY, HSQC, HMBC and NOESY (Supplementary Figures 17-27 and Supplementary Table 2). An *N*,*N*-dimethyl isoleucine was found to be present at the N-terminal of <strong>1</strong>, a typical chemical feature of the rare linaridin family<sup>10</sup> and more recently cacaoidine<sup>11</sup>. Unfortunately, fragmentation within the C-terminal structure was not observed in MS<sup>2</sup> and the *y*<sub>6</sub> ion was resistant to fragmentation in MS<sup>3</sup> experiments. Therefore, HR-MS and isotope fine structure analysis was used to confirm the elemental formula of the *y*<sub>6</sub> ion as [C<sub>26</sub>H<sub>34</sub>N<sub>7</sub>O<sub>8</sub>S<sub>1</sub>]<sup>+</sup> (Fig. 2 B-D). However, inspection of NMR spectra together with the sequence of the precursor peptide allowed us to determine an unprecedented *bis*-thioether crosslink ring system in the planar structure of the C-terminus of <strong>1</strong> (Supplementary Figures S28-30 and Supplementary Table 2). The *bis*-thioether crosslink (MAbi) is unique in natural products and further highlights the structural novelty of <strong>1</strong>. Overall, the combination of these unusual chemical features makes <strong>1</strong> unique among natural products described to date.
+
+The planar structure of <strong>1</strong> (Fig. 2A and Supplementary Fig. 31) confirmed the dehydration of genetically encoded Thr-2, 3, 4, 6, 20, 22, and Ser 8 and the subsequent Michael addition of decarboxylated Cys-27 to Dhb-22 to yield the transient *S*-[(Z)-2-aminovinyl]-(3*S*)-3-methyl-cysteine (AviMeCys) as observed in biochemical precedent<sup><span class="CitationRef">2</span>, <span class="CitationRef">11</span></sup>, followed by the second cyclization of Cys-11 to AviMeCys to generate the *bis*-thioether crosslink. The Aaa-7 residue in <strong>1</strong> is derived from the rearranged dehydration of Ser-7. At the same time, it was shown that three of the serine amino acids in the genomic sequence (Ser-13, 16 and 18) was present in the final structure as Ala residues. It is likely that these L-Ser residues are converted to D-Ala. The biochemical precedent has also been found in a few RiPPs<sup>4,12−14</sup>, suggesting that L-Ser undergoes dehydration, followed by subsequent reduction. To confirm those conversions and determine the absolute configurations for all the amino acids in <strong>1</strong>, we performed advanced Marfey’s analysis (Supplementary Fig. 32, Supplementary Table 3). The unmodified amino acid residues are L-configuration, except Cys-11 and Ala residues. Both L- and D-Ala are present in <strong>1</strong> and the relative peak area suggested a molecular ratio of two L-Ala to three D-Ala. This was consistent with the formation of D-Ala stereoisomer in <strong>1</strong> exclusively from the three genetically encoded serine residues (Supplementary Fig. 33).
+
+## A computational model of 1
+
+The unusual fused macrocylic MAbi ring system in <strong>1</strong> marks a new structural motif for cyclic peptides. The three-dimensional conformation of <strong>1</strong> was modelled to address the unique structural constraints imposed by the macrocyclization and the Aaa-7 (Δ<sup>Z</sup> βAla) residue as well as the stereochemistry of four as-yet to-be-determined chiral centres, Cys-11, α- carbon at aminoether 1,2 dithiol-27 (AED-27), and α- and β- carbons at Abu-22 (Fig. 2A). Computational modelling approaches have been successfully used to distinguish among possible diastereomers in the post-translationally modified peptides<sup>15,16</sup>. However, due to the high content of modified residues in these systems it is difficult to obtain accurate force fields and so less accurate potentials must be used. In the present work we have employed electronic structure calculations at the density functional tight-binding level which avoid the need for force field parameter sets suitable for the compound under study<sup>17</sup>. The self-consistent-charge extended tight binding method GFN2-xTB was selected as this provides accurate geometries, vibrational frequencies and non-bonded interactions whilst being rapid enough to permit molecular dynamics (MD) simulations of useful length to be performed on systems of the size of the peptide under study here. In the present work, atom pairs identified by NMR NOESY correlations (Supplementary Figures 34-36) were employed in conjunction with the GFN2-xTB-MD method. A NOESY spectrum acquired in CD<sub>3</sub>OH at 298K with the cross-referenced NOESY in CD<sub>3</sub>OD<sup>18,19</sup> was utilized for the calculations (Supplementary Figures 34-36). MD simulations were performed for multiple structures containing different combinations of either the *R* or *S* configurations at the chiral carbons in the vicinity of the *bis*-thioether linkage. The *S* configurations in both chiral centres at Abu-22, the *S* configuration at AED-27 and the *R* configuration of Cys-11 displayed better fits to the data, accounting for a greater number of NOESY-derived distance constraints, relative to other diastereomers (Supplementary Tables 4 and supplementary information of molecular dynamic animations). The calculated structure indicates a relatively rigid fused bicyclic structure held in place by the two crosslinks as well as important backbone and side chain hydrogen bonds (Fig. 3). Extensive hydrogen bonds among the backbone of amino acids residues (such as carbonyl at DAla-13 with NH at Val-15, carbonyl at Leu-14 with NHs of Ala-17 and DAla-18, respectively, carbonyl of Ala-17 with Dhb-20, carbonyl of DAla-18 with NH of MAbi residue) within the fused macrocycle could also be formed, giving the ring system a cage-like secondary structure (Fig. 3 A, molecular dynamics animation in supplementary document). Interestingly, the N-terminal tail is also structured through hydrogen bonds among the side chain of Glu-9 and the NH and carbonyl groups in the backbone, making the helix-like N-terminus coiled toward the fused macrocycle (Fig. 3). The *Z*-configuration at position 7 is likely to play an important role of the overall shape of N-terminal flexible chain, possessing an N-H••O=C intra-residue H-bond in an average distance of 2.14±0.20 Å (Fig. 3). Together, the studies above provide a conformational model for <strong>1</strong> and indicate that the crosslinks are generated with the *R*, *S*, *S,S* configurations, and the cage-like macrocycle C-terminus as well as the helix-like *N*-terminus are significantly stabilised via hydrogen bonding, a prediction that can be tested experimentally in the future. The extensive molecular interactions among the linear chain and the macrocyclic motif, based on the calculated model (Fig. 3), allowed us to re-examine the NOE data for previously unnoticed long-range correlations. Indeed, interpretation of complex NOE spectrum allow identification of two new long-range correlations between the Aaa-7 residue in the linear chain and the methyl group at MAbi motif in the macrocyclic ring, and between β-H at Leu-14 and α-H at aminoether 1,2 dithiol (S1, S2) of MAbi motif, respectively, further validating the calculated model (Fig. 2A).
+
+Overall, <strong>1</strong> displays unusual chemical features as well as a predicted sophisticated secondary structure, rendering <strong>1</strong> as a new RiPPs, which we named kintamdin, associated with the place, Kintampo water fall where *Streptomyces* sp. RK44 was originally isolated. When tested for various antibacterial and cytotoxic activities, <strong>1</strong> displays good anticancer activities against skin and breast cancer cell lines (2.4 ± 0.1 µM and 0.6 ± 0.1 µM, respectively) where possess weak or no inhibitory activities against bacterial strains available (Supplementary Table 5).
+
+## The minimal biosynthetic gene cluster of 1
+
+The structural novelty of <strong>1</strong> motivated us to probe its biosynthetic origin in the producing strain, RK44. Analysis of the surrounding genetic environment of *kin* A allowed identification of a candidate gene cluster (*kin*) (Fig. 4A, Supplementary Table 6). To validate the identity of the BGC, we carried out TAR cloning strategy for heterologous expression. To this end, we modified the construction method of pathway-specific capture vectors in order to improve the capture efficiency of the BGC of interest from the genomic DNA of RK44 as shown in Supplementary Fig. 37. One out of five clones after yeast transformation was identified to contain the correct length of the BGC in the construct. The construct pCAP03-*kin*2 was then transferred into various *streptomyces* hosts in our lab via *E. coli*-streptomyces conjugation. The production of <strong>1</strong> in *Streptomyces coelicolor* M1152<sup>20</sup> was confirmed through HRMS analysis as well as MS<sup><span class="CitationRef">2</span></sup> fragmentation by comparing with the ones generated from the authentic peptide <strong>1</strong> (Fig. 4C, i-ii and xi, Supplementary Figs. 38–39).
+
+To determine the boundary of the <strong>1</strong> BGC, a series of gene inactivation were carried out on pCAP03-*kin*2, followed by *E coli*-streptomyces conjugation and fermentation. HRMS analyses of the extracts of these variants demonstrated that gene inactivation of *orf*-1 and *orf*-2 as well *orf* 1 at the boundaries of the cloned DNA fragment showed no perturbation of the <strong>1</strong> production (Figure S40). Therefore, the minimal BGC directing the biosynthesis of <strong>1</strong> include fifteen *orf* s, encoding two metallopeptidases from the M16 family (KinE and KinF), one flavin-dependent decarboxylase (KinI), one SAM-dependent methyltransferase (KinO), one oxidoreductase (KinJ), two aminoglycoside phosphotransferases (KinD and H), four ABC transporters (KinK-N), two hypothetical proteins (KinB and C), and one transposase (KinG) (Fig. 4A and Supplementary Table 6). It is noteworthy that none of ORFs encoded in the BGC share any significant homologues (<30% aa sequence identities) identified in other classes of natural products. Genes involved in some of the posttranslational modifications of <strong>1</strong> were assigned (Supplementary Fig. 16 and Supplementary Table 6). Both KinE and KinF share low homologue with each other. They may function as proteases to generate mature <strong>1</strong>, a similar proposed function in the biosynthesis of ruminococcin C<sup><span class="CitationRef">21</span></sup>. KinI, a protein homologue to cypemycin *N*-methyltransferase<sup><span class="CitationRef">22</span></sup> may be responsible for the installation of *N*,*N*-dimethyl moiety at the *N*-terminus of <strong>1</strong>, while KinJ might be involved in the hydrogenation of Dha to D-Ala, as previously described for LanJ<sub>B</sub> enzymes such as CrnJ<sup><span class="CitationRef">13</span></sup> and BsjJ<sub>B</sub><sup><span class="CitationRef">14</span></sup>. Although sharing low homologue (<30% AA identity) to cypemycin decarboxylase (CypD)<sup><span class="CitationRef">23</span></sup>, KinO may account for the generation of the macrocyclic *bis* thioether ring (MAbi) system. The *kin* cluster does not encode any protein homologous to dehydratases presented in lanthipeptide (i.e. LanB) or linaridin (i.e. CypH) BGCs. The possible candidate gene products for dehydration are KinD and KinH, which share low structural homology (16% i.d.) to the ATP-dependent aminoglycoside phosphotransferase (APH, Pfam01636) superfamily, predicted in Phyre 2 server<sup><span class="CitationRef">24</span></sup>. It was proposed that KinD and Hmay catalyse the phosphorylation of Ser and Thr residues, followed by *anti*-elimination to yield Dha and Dhb residues respectively. The enzymes responsible for the formation of the unsaturated β-amino acid, Aaa-7, remain to be determined. Blast search using KinC as a sequence query in NCBI and structural modelling in Phyre2 server<sup><span class="CitationRef">24</span></sup> indicated that KinC share no significant homologue to any known proteins. However, part of the predicted structural model displays low degrees (83.8 confidence, 18% id) of similarity to the C-terminal domain of phosphoglucomutases, suggesting that KinC may be the candidate to catalyse the rearrangement of phosphorylated Ser-7 into Aaa-7 residue.
+
+To assess the *in vivo* role of these genes, we generated eight variants (Δ*kin* C-F, H-J and O) (Fig. 4C, traces iii-xi). Gene inactivation of Δ*kin* E resulted in significantly reduced production of <strong>1</strong> while knocking out *kin* F caused moderate reduction of the <strong>1</strong> production (Fig. 4C, traces v-vi), suggesting the presence of possible synergetic function between these two genes to ensure the efficiency of the proteolytic activities.
+
+The production of <strong>1</strong> was abolished in six other variants (Δ*kin* C, D and H-O), suggesting that these genes are essential for <strong>1</strong> biosynthesis (Figure 4C, traces iii, iv and viii-xi). Gene inactivation of *kin* O resulted in accumulation of a new metabolite <strong>2</strong> in the culture of the Δ*kin* O mutant (Figure 4C, trace xi). MS and MS<sup>n</sup> fragmentation analysis demonstrated that <strong>2</strong> is the non-methylated <strong>1</strong> (Supplementary Figure 41). Knocking out *kin* I, encoding a flavin-dependent decarboxylase resulted in accumulation of another new metabolite <strong>3</strong> in the culture of Δ*kin* I variant (Figure 4C, trace xi). <strong>3</strong> is the *N*,*N*-dimethylated 27-mer linear peptide containing the intact cysteine residue at the C-terminus and dehydrated amino acid residues as well as Ala residues derived from Dha, as evidenced in MS and tandem MS analysis (Supplementary Figure 42), suggesting that the dehydration on Thr and Ser and hydrogenation on the resulting Dha residues occur prior the decarboxylation.
+
+To assess the roles of the gene products of *kin* I and *kin* O in the <strong>1</strong> biosynthesis, biochemical analysis of recombinant KinI and KinO was carried out. Overexpression of *kin* I and *kin* O in *E*. *coli* allowed purification of both recombinant proteins to near homogeneity, as observed in SDS Page analysis (Supplementary Fig. 43), respectively. An *in vitro* assay of a 6-His-tagged recombinant KinO with <strong>2</strong> in the presence of *S*-adenosyl-L-methionine (SAM, 1 mM) was performed. The production of <strong>1</strong> was confirmed by HR-MS and MS<sup>n</sup> analysis (Supplementary Fig. 44), confirming that KinO is responsible for the *N*-terminal Ile dimethylation. Interestingly, <strong>3</strong> contains Dhb at position 20, suggesting that the Abu residue in <strong>1</strong> is derived from Michael addition of decarboxylated Cys with Dhb to likely yield AviMeCys, a biochemical precedent proposed in cypemycin<sup><span class="CitationRef">10</span></sup> and cacaoidin<sup><span class="CitationRef">11</span></sup>. Incubation of KinI with <strong>3</strong> and other necessary cofactors, however, failed to produce any new products, indicating that <strong>3</strong> is a shunt product in the variant.
+
+# Discussion
+
+Ribosomally synthesized and post-translationally modified peptides (RiPPs) are a group of structurally complex naturally occurring metabolites. Although exploring only 20 proteinogenic amino acids, their biosynthetic pathways recruit many post-translational modification (PTM) enzymes to install diversified structural features in RiPPs including lantionine/labionin/avionin cyclic systems that are unique in RiPPs. Kintamdin **1** contains an MAbi crosslink motif, which, to our knowledge, is unique in natural products. Furthermore, an unusual unsaturated β-amino acid, 3-amino acrylic acid (Aaa/(Δ<sup>Z</sup>βAla)), was identified at position 7 in the linear chain at the N-terminal of **1**. β-amino acid residues were largely thought to be a hallmark of non-ribosomal peptide (NPR) synthesis. However, recent studies indicated that β-amino acids are not impossible for peptide synthesized by the ribosome. Piel and co-workers<sup><span class="CitationRef">7</span></sup> discovered that a splicease enzyme, PlpX, together with its protein partner, PlpY, catalyses C-C bond cleavage of the peptide backbone to generate unsaturated β-amino acids, α-keto β-amides, in Nif11-like proteusin precursor peptides. The corresponding genes, <em>plp</em>X and <em>plp</em>Y, are located in the <em>plp</em> gene cluster in the genome of the cyanobacterium <em>Pleurocapsa</em> sp. PCC7319, which produces an as-yet unknown natural product<sup><span class="CitationRef">2</span></sup>. In this respect, kintamdin **1** is the first example in the structurally defined RiPPs that contain β-amino acid residues.
+
+Aaa (Δ<sup>Z</sup>βAla) residue is rare in natural product inventory. Only one cyclopeptide metabolite containing Aaa residue has been discovered from a marine gut fungus, <em>Aspergillus flavipes</em>, from <em>Ligia oceanica</em> thus far<sup><span class="CitationRef">25</span></sup>. Despite its rarity in nature, Aaa (Δ<sup>Z</sup>βAla) has been applied in man-made conformationally controlled, oligomeric molecules in the field of foldamers<sup><span class="CitationRef">26</span></sup>, due to its unique properties of conformational changes. This so-called dynamic foldamer system offers the possibility of triggering the switch of a foldamer between two distinct but structurally defined states. For example, a motion of a molecular spring could be generated if a foldamer undergoes elongation and contraction, resulting from the transition between two conformations<sup><span class="CitationRef">27</span></sup>. A conformational transition can also convey chemical information across a foldamer from a signalling unit at one end to a report at the other end at multinanometer distances<sup><span class="CitationRef">28</span></sup>. It has been found that Aaa (Δ<sup>Z</sup>βAla) residue in a peptide allows formation of an N-H••O=C intra-residue H-bond which closes a 6-membered <em>pseudo</em> ring system, allowing the peptide to form a well-defined 3D architecture of the flattened β sheet<sup><span class="CitationRef">29</span></sup>. Recent studies demonstrated that foldamers containing Δ<sup>Z</sup>βAla/Δ<sup>E</sup>βAla units display interesting conformational, electronic, and supramolecular aggregation properties that can be modulated by selective <em>E</em>-<em>Z</em> photoisomerization<sup><span class="CitationRef">30</span>–<span class="CitationRef">32</span></sup>. As a result, the Aaa (Δ<sup>Z</sup>βAla) system as a unique structural element may provide a new route of designing functional foldamers for biomimetic and nanotechnological applications. Aaa (Δ<sup>Z</sup>βAla) residue has a profound effect in the overall folding of **1**. In our calculated model, Aaa-7 residue in **1** is likely the key structural element of the helical N-terminal of the linear chain coiled toward the macrocyclic crosslink to yield a unique overall folding in the small molecule inventory. It raises an interesting question of whether the unique conjugation between Aaa(Δ<sup>Z</sup>βAla)-7 and Dha-8 could undergo a photoinduced isomerization to yield Δ<sup>E</sup>βAla motif, thus influencing the bioactivity of the resulting isomer. This will be our future investigation in our laboratory.
+
+The structural novelty of **1** motivated us to investigate its biosynthetic origin. Examination of the genetic surrounding of the gene encoding the precursor peptide KinA allowed identification of the <em>kin</em> BGC. Modified TAR cloning strategy and heterologous expression of the captured DNA fragment allowed the production of **1** in the resulting <em>Streptomyces coelicolor</em> M1152 variant. Subsequent gene inactivation determined the minimal cassette of the <em>kin</em> BGC, which encodes fifteen proteins. The genetic knockout experiments indicated that the genes encoding putative catalytic functions, <em>kin</em>C-F, H-J and <em>kin</em>O, are essential for the production of **1**. Attempts using comparative untargeted metabolomics<sup><span class="CitationRef">33</span></sup> failed to identify any intermediates in the Δ<em>kin</em>C, D, H and J variants, suggesting that these PTMs may occur prior to the cleavage of leader peptide, resulting in degradation of premature products<sup><span class="CitationRef">2</span></sup>. Intermediate **2**, the demethylated **1**, was produced in the variant where <em>kin</em>O was inactivated. Assay of **2** with recombinant KinO yielded **1** as observed in HR-MS analysis, indicating that KinO is the last step of the **1** biosynthesis (Supplementary Fig. 45). Knocking out <em>kin</em>I produced the shunt product, di-<em>N</em>-methylated linear peptide **3**. Biochemical analysis indicated that **3** is not the immediate substrate of recombinant KinI. It is likely that the substrate of KinI is a demethylated **3** attached with the leader peptide, similar to other RiPP systems in previous reports<sup><span class="CitationRef">2</span></sup> (Supplementary Fig. 45).
+
+The novelty of its structure and biosynthesis of **1** allowed us to further address the occurrence of its homologues in bacterial genomes. Searches with KinA as query in the National Centre for Biotechnology Information (NCBI) database, showed the prevalent presence of KinA homologues in actinobacteria, including <em>streptomyces</em>, rare actinomycetes (i.e. <em>actinopolyspora</em>, <em>kineococcus</em>, <em>arthrobacter</em>, <em>nocardioides</em>, thermophilic <em>microbispora</em>), and an opportunistic pathogen, <em>corynebacterium timonense</em><sup><span class="CitationRef">34</span></sup>. Comparisons among these KinA sequences reveal a high degree of similarity and the conserved signature motif of S/T-S/<strong>/</strong>T-X-x-<span class="Underline">C</span>-x<sub>n</sub>-<span class="Underline">T</span>-X-X-X-X-<span class="Underline">C</span> (S7-S8-X-X-X-C11-T22-C27 in the of **1**) (Supplementary Fig. 46). The underlined <span class="Underline">C</span>…<span class="Underline">T</span>…<span class="Underline">C</span> sequence, corresponding to the formation of the <em>bis</em>-thioether macrocyclic crosslink (MAbi) in the case of **1**, is entirely conserved in this subset of molecules. The first S/T residues, which corresponds to the unsaturated β-amino acid residue, Aaa(Δ<sup>Z</sup>βAla)-7 in the case of **1**, are also highly conserved.
+
+Analysis of the surrounding genetic environment of these identified <em>orf</em>s allowed the identification of candidate biosynthetic gene clusters (Fig. 5). The occurrence of these fifty precursor peptides always coincided with the presence, in close proximity, of genes with homology to eight genes in the <em>kin</em> BGC, namely <em>kin</em>A, C-F, H-J. Interestingly, in two cases of the BGCs, a single gene product contains KinE homologue in N-terminal and KinF homologues in the C-terminal, suggesting that these two gene products have evolved a hybrid construction of two historical activities. Given their invariant co-occurrence and structural uniqueness among RiPPs, we propose that Kintamdin **1** is the founding member of a previously undescribed family of RiPPs, which were proposed to be β-bithionin.
+
+# Conclusion
+
+We have discovered and characterized a new RiPP, kintamdin 1, containing unusual chemical features through a combination of chemical workup, tandem MS analysis and structural elucidation assisted by computational modelling studies. 1 contains a rare unsaturated β-amino acid, (Z)-3-amino-acrylic acid (Aaa or ΔZβAla), and an unprecedented bis-thioether macrocyclic crosslink (MAbi) motif which is the first discovered in natural products. Both motifs have profound effects on the overall shape of 1. Computational modelling combined with detailed NOE interpretation suggested that 1 should adopt an intriguing well-defined structure, a similar phenomenon observed in a synthetic peptidyl foldamer containing ΔZβAla residue. We have also established the minimal biosynthetic gene cluster of 1 which contain a unique set of biosynthetic enzymes, the majority of which is different to what have been established in the biosynthetic pathways of previously known RiPPs. We established that the SAM-dependent methyltransferase, KinI, catalyses the last step in 1 biosynthesis through in vivo and in vitro studies. This study also demonstrated that gene clusters generating similar bacterial RiPPs are widely spread, hinting a large untapped capacity of RiPP biosynthesis and the potential for discovering previously unknown β-bithionin.
+
+# Materials And Methods
+
+General information  
+H NMR spectra were obtained on Bruker AVANCE III HD 400 MHz (AscendTM 9.4 Tesla, UK) with Prodigy TCITM cryoprobe at 298 K in CD₃OD and DMSO-d₆ (Goss Scientific, Massachusetts, MA, USA). Chemical shifts are reported in parts per million (ppm), relative to the solvent signals. ¹³C NMR spectra were obtained with proton decoupling on the same NMR spectrometer and are reported in ppm with TMS for internal standard. Multiplicity is defined as: s = singlet; d = doublet; t = triplet; q = quartet; m = multiplet, br = broad, or combinations of the above. Coupling constants (J) are reported in Hertz. High resolution mass spectra (HRMS) for chemical workup and molecular networking were obtained on either a 12T SolariX 2XR FT-ICR MS (Bruker Daltonics) via direct infusion or an LTQ Orbitrap Thermo Scientific MS system coupled to a Thermo Instrument HPLC system (Accela PDA detector, Accela PDA autosampler, and Accela pump). The injected samples were chromatographically separated by a C18 (Sunfire 150 × 46 mm) column. The gradient elution for separation was CH₃CN/H₂O with 0.1% trifluoroacetic acid (TFA) (from 0–100% for 30 min, flow rate, 1.0 mL/min, UV detection max 340 nm).
+
+Structure elucidation of 1  
+Detail analysis of ¹H, ¹³C, COSY, gHSQC, HSQC-TOCSY, gHMBC and ROESY were used to elucidate the structure of kintamdin (Supplementary Fig. 1–7). ¹H spectra indicated the usual fingerprint of peptides exhibiting multiple exchangeable protons (δH 7.6–10.4 ppm), numerous overlapping α protons of amino acids (δH 3.4–4.6 ppm), various alkyls (δH 0.8–3.3 ppm), aromatic (δH 6.9–7.6 ppm) sidechains. Additionally, there were 9 olefinic protons (δH 5.3–7.2 ppm) and two chemically equivalent protons (δH 2.89 ppm) assigned to a N,N-dimethyl group found in the spectrum of 1. Analysis of COSY, HSQC and HSQC-TOCSY spectra enabled the assignment of various proteinogenic amino acids such as Trp (×1), Val (×3), Ile (×1), Glu (×1), Asp (×1), Gly (×1), Ser (×1), Ala (×5). Also observed were 7 dehydrated amino acids (DHAAs) such as Dha (×1), and Dhb (×5). It also accounted for one aminovinyl-3-methyl-cysteine (AviMeCys), and one cysteine linkage. Structural configurations of Dhb at positions 2, 3, 4, 6, and 20 have been deduced as Z based on observed NOE correlation between the Me(γ) group and its corresponding NH (Figure S2).
+
+Computational calculation of 1  
+Initial structures for the kintamdin peptide were constructed with standard amino acids using the Avogadro software (version 1.2.0). Amino acid modifications and cyclisations were then performed by hand and hydrogens were added to correspond to expected ionisation states at pH 7.4 and the resulting structures energy minimized using the MMFF94 molecular mechanics force field. Each of the stereoisomers corresponding to different combinations of chiral centres in the dithiol bridge were created at this stage and minimised in the same way to produce starting structures for further analysis. The chiral centre at the methyl group in the dithiol bridge (attached to β-carbon of the Abu(S₂)-22 residue) was kept fixed in the S-configuration based on biosynthetic precedent.  
+
+The resulting structures were then optimised at the density functional tight-binding level of theory with the xTB software (version 6.3.0) and the GFN2-xTB electronic structure method which incorporates the accurate D4 method for modelling dispersion effects. Whilst more computationally demanding than classical force-fields, this method was chosen because of the large number of non-standard features in the peptide being studied which meant that conventional force-field approaches were likely to be unreliable due to the lack of accurate force-field parameters. Solvent (methanol) effects were included through a GB/SA continuum treatment.  
+
+The GFN2-xTB optimised structures were then used as starting points for NVT ensemble molecular dynamics simulations which were performed at 298K, again including methanol solvent effects. All bonds were constrained using the SHAKE algorithm which permitted the use of a 4 fs time step in the simulations. A total of 300 ps of simulation was performed for each of the eight possible dithiol bridge stereoisomers. The initial structures in the MD simulations were seen to undergo large changes in the conformation of the acyclic peptide chain (residues 1–10) which folded rapidly within the first 50 to 100 ps and thereafter remained relatively fixed. The 100–200 ps period was treated as further equilibration and then in the final 100 ps of the simulations distance monitors corresponding to the NMR coupling data from the ROESY experiments were used to collect data to be used for the evaluation of the closeness of fit between the different structural models and the experimental observations. The full distance monitor data is provided in the Supporting Information.
+
+# References
+
+1. Arnison, P. G. et al. Ribosomally synthesized and post-translationally modified peptide natural products: Overview and recommendations for a universal nomenclature. *Nat. Prod. Rep.* **30**, 108–160 (2013).
+
+2. Montalbán-López, M. et al. New developments in RiPP discovery, enzymology and engineering. *Nat. Prod. Rep.* **38**, 130–239 (2021).
+
+3. Ortega, M. A. et al. Structure and mechanism of the tRNA-dependent lantibiotic dehydratase NisB. *Nature* **517**, 509–512 (2015).
+
+4. Xu, M. et al. Functional Genome Mining Reveals a Class V Lanthipeptide Containing a D-Amino Acid Introduced by an F420H2-Dependent Reductase. *Angew. Chemie - Int. Ed.* **59**, 18029–18035 (2020).
+
+5. Ren, H. et al. Discovery and Characterization of a Class IV Lanthipeptide with a Nonoverlapping Ring Pattern. *ACS Chem. Biol.* **15**, 1642–1649 (2020).
+
+6. Wiebach, V. et al. The anti-staphylococcal lipolanthines are ribosomally synthesized lipopeptides brief-communication. *Nat. Chem. Biol.* **14**, 652–654 (2018).
+
+7. Morinaka, B. I. et al. Natural noncanonical protein splicing yields products with diverse β-amino acid residues. *Science.* **359**, 779–782 (2018).
+
+8. Fang, Q. et al. Signalling and bioactive metabolites from Streptomyces sp. RK44. *Molecules* **25**, 460–473 (2020).
+
+9. Overbeek, R. et al. The SEED and the Rapid Annotation of microbial genomes using Subsystems Technology (RAST). *Nucleic Acids Res.* **42**, 206–214 (2014).
+
+10. Ma, S. & Zhang, Q. Linaridin natural products. *Nat. Prod. Rep.* **37**, 1152–1163 (2020).
+
+11. Ortiz-López, F. J. et al. Cacaoidin, First Member of the New Lanthidin RiPP Family. *Angew. Chemie - Int. Ed.* **59**, 12654–12658 (2020).
+
+12. Martin, N. I. et al. Structural Characterization of Lacticin 3147, A Two-Peptide Lantibiotic with Synergistic Activity. *Biochemistry* **43**, 3049–3056 (2004).
+
+13. Lohans, C. T., Li, J. L. & Vederas, J. C. Structure and biosynthesis of carnolysin, a homologue of enterococcal cytolysin with d-amino acids. *J. Am. Chem. Soc.* **136**, 13150–13153 (2014).
+
+14. Huo, L. & Van Der Donk, W. A. Discovery and Characterization of Bicereucin, an Unusual D-Amino Acid-Containing Mixed Two-Component Lantibiotic. *J. Am. Chem. Soc.* **138**, 5254–5257 (2016).
+
+15. Imai, Y. et al. A new antibiotic selectively kills Gram-negative pathogens. *Nature* **576**, 459–464 (2019).
+
+16. Schramma, K. R., Bushin, L. B. & Seyedsayamdost, M. R. Structure and biosynthesis of a macrocyclic peptide containing an unprecedented lysine-to-tryptophan crosslink. *Nat. Chem.* **7**, 431–437 (2015).
+
+17. Jornet-Somoza, J. et al. Insights into colour-tuning of chlorophyll optical response in green plants. *Phys. Chem. Chem. Phys.* **17**, 26599–26606 (2015).
+
+18. Tabudravu, J. N., Jaspars, M., Morris, L. A., Jantina Kettenes-van den Bosch, J. & Smith, N. Two distinct conformers of the cyclic heptapeptide phakellistatin 2 isolated from the Fijian marine sponge Stylotella aurantium. *J. Org. Chem.* **67**, 8593–8601 (2002).
+
+19. Tabudravu, J. N., Morris, L. A., Milne, B. F. & Jaspars, M. Conformational studies of free and Li+ complexed jasplakinolide, a cyclic depsipeptide from the Fijian marine sponge Jaspis splendens. *Org. Biomol. Chem.* **3**, 745–749 (2005).
+
+20. Gomez-Escribano, J. P. & Bibb, M. J. Engineering Streptomyces coelicolor for heterologous expression of secondary metabolite gene clusters. *Microb. Biotechnol.* **4**, 207–215 (2011).
+
+21. Chiumento, S. et al. Ruminococcin C, a promising antibiotic produced by a human gut symbiont. *Sci. Adv.* **5**, (2019).
+
+22. Claesen, J. & Bibb, M. Genome mining and genetic analysis of cypemycin biosynthesis reveal an unusual class of posttranslationally modified peptides. *Proc. Natl. Acad. Sci. U. S. A.* **107**, 16297–16302 (2010).
+
+23. Ding, W. et al. Cypemycin Decarboxylase CypD Is Not Responsible for Aminovinyl-Cysteine (AviCys) Ring Formation. *Org. Lett.* **20**, 7670–7673 (2018).
+
+24. Kelley, L. A., Mezulis, S., Yates, C. M., Wass, M. N. & Sternberg, M. J. E. The Phyre2 web portal for protein modeling, prediction and analysis. *Nat. Protoc.* **10**, 845–858 (2015).
+
+25. Xu, J., Zhao, S. & Yang, X. A new cyclopeptide metabolite of marine gut fungus from Ligia oceanica. *Natural Product Research* **28**, 994–997 (2014).
+
+26. Gellman, S. H. Foldamers : A Manifesto. *Acc. Chem. Res.* **31**, 173–180 (1998).
+
+27. Crisma, M. et al. Peptide α/310-helix dimorphism in the crystal state. *J. Am. Chem. Soc.* **129**, 15471–15473 (2007).
+
+28. Poli, M. De et al. Conformational photoswitching of a synthetic peptide foldamer bound within a phospholipid bilayer. *Science.* **352**, 575–580 (2016).
+
+29. Mazzier, D. et al. Helical foldamers incorporating photoswitchable residues for light-mediated modulation of conformational preference. *J. Am. Chem. Soc.* **138**, 8007–8018 (2016).
+
+30. Marafon, G., Crisma, M. & Moretto, A. Intrinsically Photoswitchable α/β Peptides toward Two-State Foldamers. *Angew. Chemie- Int. Ed.* **130**, 10374–10377 (2018).
+
+31. Marafon, G., Crisma, M. & Moretto, A. Tunable E- Z Photoisomerization in α,β-Peptide Foldamers Featuring Multiple (E/Z)-3-Aminoprop-2-enoic Acid Units. *Org. Lett.* **21**, 4182–4186 (2019).
+
+32. Marafon, G. et al. Photoresponsive Prion-Mimic Foldamer to Induce Controlled Protein Aggregation. *Angew. Chemie - Int. Ed.* **60**, 5173–5178 (2020).
+
+33. Crone, W. J. K. et al. Dissecting Bottromycin Biosynthesis Using Comparative Untargeted Metabolomics. *Angew. Chemie* **128**, 9791–9795 (2016).
+
+34. Merhej, V., Falsen, E., Raoult, D. & Roux, V. Corynebacterium timonense sp. nov. and Corynebacterium massiliense sp. nov., isolated from human blood and human articular hip fluid. *Int. J. Syst. Evol. Microbiol.* **59**, 1953–1959 (2009).
+
+35. Hanwell, M. D. et al. Avogadro: an advanced semantic chemical editor, visualization, and analysis platform. *J. Cheminform.* **4**, 1–17 (2012).
+
+36. Bannwarth, C., Ehlert, S. & Grimme, S. GFN2-xTB - An Accurate and Broadly Parametrized Self-Consistent Tight-Binding Quantum Chemical Method with Multipole Electrostatics and Density-Dependent Dispersion Contributions. *J. Chem. Theory Comput.* **15**, 1652–1671 (2019).
+
+37. Caldeweyher, E. et al. A generally applicable atomic-charge dependent London dispersion correction. *J. Chem. Phys.* **150**, 154122 (2019).
+
+38. Ryckaert, J.-P., Ciccotti, G. & Berendsen, H. J. C. Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. *J. Comput. Phys.* **23**, 327–341 (1977).
+
+39. Van Gunsteren, W. F. & Berendsen, H. J. C. Algorithms for macromolecular dynamics and constraint dynamics. *Mol. Phys.* **34**, 1311–1327 (1977).
+
+# Supplementary Files
+
+- [SuppIementaryinformationgeneralexperimentalinformation.docx](https://assets-eu.researchsquare.com/files/rs-352308/v1/c595a80279d789d75bb817e3.docx)  
+  Supplementary information - general experimental information
+
+- [SuppIementaryinformationtables.docx](https://assets-eu.researchsquare.com/files/rs-352308/v1/31716b3d76ac6ee556ade596.docx)  
+  Supplemental information - SI tables
+
+- [SuppIementaryinformationFigurespart1.docx](https://assets-eu.researchsquare.com/files/rs-352308/v1/2a6e9a8411d4a462bf0a4484.docx)  
+  Supplementary information - SI figures part 1
+
+- [SuppIementaryinformationFigurespart2.docx](https://assets-eu.researchsquare.com/files/rs-352308/v1/e7a32225e3ac4b98e75cd28a.docx)  
+  Supplementary information - SI figures part 2
+
+- [KintamdinMD.mp4](https://assets-eu.researchsquare.com/files/rs-352308/v1/d5723afaf7c2adec7c3f2219.mp4)  
+  molecular simulation with amino acid backbones and side chains
+
+- [KintamdinbackboneMD.mp4](https://assets-eu.researchsquare.com/files/rs-352308/v1/95d162923cca37c9be114c5d.mp4)  
+  molecular simulation with amino acid backbones only

0944c67707575528e5b0b90844d0835daa9bc0e7b8dcb4d47e6dd8c6cdb05e16/preprint/images_list.json

@@ -0,0 +1,34 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpeg",
+        "caption": "Schematic of synthetic COFs and structural representations of COFs. a Thesynthesis route of CoPc-O-COF and CoPc-S-COF. The simulated AA stacking of b, c CoPc-O-COF and d, e CoPc-S-COF (Co: orange; C: light gray; N: blue; O: red; S: yellow; F: green). PXRD of f CoPc-O-COF and g CoPc-S-COF: experimental PXRD profile (black), refined profile (red), the difference between the experimental and refined PXRD (gray), and simulation pattern based on the AA stacking manner (blue and green).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpeg",
+        "caption": "Morphology and characterization of COFs. TEM images of a CoPc-O-COF and d CoPc-S-COF. HR-TEM images of b CoPc-O-COF and e CoPc-S-COF. The EDX mapping analysis of c CoPc-O-COF and fCoPc-S-COF. High-resolution XPS spectra of g Co 2p for CoPc-O-COF and CoPc-S-COF, h O 1s for CoPc-O-COF and i S 2p for CoPc-S-COF.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpeg",
+        "caption": "H2O2 electroproduction and DFT calculation. a LSVs of CoPc-O-COF and CoPc-S-COF at 1600 rpm in O2-saturated 0.1 M KOH. b H2O2 selectivity and electron transfer number n of CoPc-O-COF and CoPc-S-COF. c Chronoamperometry measurement of CoPc-S-COF for 36000 s at 0.52 V versus RHE. d ORR polarization curves in O2-saturated 0.1 M KOH and H2O2RR polarization curves in Ar-saturated 0.1 M KOH containing 10 mM H2O2. e H2O2RR polarization curves and f current densities of\u00a0 CoPc-O-COF and CoPc-S-COF in Ar-saturated 0.1 M KOH containing different concentrations of H2O2. g Di\ufb00erential charge distribution on both simulated periodic fragment of both COFs (isosurface level=0.01). h Partial density of states (PDOS) of Co 3d-orbital in different models. i Reaction free energy change for 2e- ORR, 4e- ORR and H2O2RR process of both COFs.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpeg",
+        "caption": "H2O2 electroproduction in the flow cell. a Schematic diagram of the \ufb02ow cell. b LSVs of CoPc-O-COF and CoPc-S-COF in flow cell. c The chronoamperometry measurements at varied applied voltages of CoPc-S-COF. d FEH2O2 of CoPc-O-COF and CoPc-S-COF at varied applied voltages. e H2O2 yields of CoPc-S-COF. f Chronopotentiometry curve at a current density of 125 mA cm-2 and the corresponding FEH2O2 in the \ufb02ow cell for CoPc-S-COF.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

0944c67707575528e5b0b90844d0835daa9bc0e7b8dcb4d47e6dd8c6cdb05e16/preprint/preprint.md

@@ -0,0 +1,105 @@
+# Abstract
+
+Realization of stable and industrial-level H₂O₂ electroproduction still faces great challenge due largely to the easy decomposition of H₂O₂. Herein, a two-dimensional dithiine-linked phthalocyaninato cobalt (CoPc)-based covalent organic framework (COF), CoPc-S-COF, was afforded from the reaction of hexadecafluorophthalocyaninato cobalt (II) with 1,2,4,5-benzenetetrathiol. Introduction of the sulfur atoms with large atomic radius and two lone-pairs of electrons in the C-S-C linking unit leads to an undulated layered structure and an increased electron density of the Co center for CoPc-S-COF according to a series of experiments in combination with theoretical calculations. The former structural effect allows the exposition of more Co sites to enhance the COF catalytic performance, while the latter electronic effect activates the 2e⁻ oxygen reduction reaction (2e⁻ ORR) but deactivates the H₂O₂ decomposition capability of the same Co center, as a total result enabling CoPc-S-COF to display outstanding electrocatalytic H₂O₂ production performance with a remarkable H₂O₂ selectivity of 95% and a stable H₂O₂ production with a concentration of 0.48 wt% under a high current density of 125 mA cm⁻² at an overpotential of ca. 190 mV for 20 h in a flow cell, representing the thus far reported best H₂O₂ synthesis COFs electrocatalysts.
+
+Physical sciences/Chemistry/Catalysis/Electrocatalysis  
+Physical sciences/Chemistry/Electrochemistry  
+Phthalocyanine  
+Covalent organic framework  
+2e⁻ ORR  
+H₂O₂  
+Electrosynthesis
+
+# 1. Introduction
+
+Hydrogen peroxide (H₂O₂) is an important inorganic chemical and environmentally friendly oxidant with extensive applications in bleaching, disinfection, wastewater treatment, and organic synthesis.<sup>1–10</sup> In industry, anthraquinone method is employed to generate more than 90% of H₂O₂, which however is energy-intensive and produces a large amount of toxic by-products.<sup>11–13</sup> For a sustainable future, it is essential to develop an energy efficient and eco-friendly strategy for the synthesis of H₂O₂ that should operate onsite even on large or small scales. As a consequence, electrocatalytic 2e⁻ oxygen reduction reaction (2e⁻ ORR) has been considered as the most promising alternative approach since it can realize the green and distributed on-demand H₂O₂ generation under ambient conditions. However, large-scale electrocatalytic H₂O₂ production is still hard to be realized because of the limited solubility of oxygen in electrolyte solutions and easy decomposition of H₂O₂ especially in the presence of metal active centers, which usually result in small working currents (< 100 mA cm⁻²) and low H₂O₂ concentration (< 0.1 wt%).<sup>14</sup> In addition, the favorable thermodynamics to generate water molecules via the 4e⁻ pathway inevitably reduces the H₂O₂ generation capability during ORR.<sup>15–17</sup>
+
+Covalent organic frameworks (COFs) consist of organic building blocks linked by covalent bonds,<sup>18</sup> which have drawn great research attention for extensive applications in gas storage and separation,<sup>19,20</sup> optoelectronic devices,<sup>21</sup> catalysis,<sup>22</sup> and energy storage<sup>23</sup> owing to their superiority of high porosity, robust stability, and low density. As a consequence of the ordered pores that favor more exposed active sites to contact with substrate molecules, COFs have shown a great application potentials as promising electrocatalysts for various reactions including ORR,<sup>24</sup> oxygen evolution reaction,<sup>25,26</sup> hydrogen evolution reaction,<sup>27</sup> and CO₂ reduction reaction.<sup>28</sup> In particular, two-dimensional (2D) conjugated COFs with ultrastrong fused aromatic linkage have been revealed to exhibit intrinsically high conductivity and excellent thermal/chemical stability, promoting enhanced electrocatalytic performance.<sup>29,30</sup> However, the design of suitable linkers and optimization of reaction conditions for COFs construction remain a demanding task for synthetic chemists. The dioxin,<sup>31,32</sup> phenazine,<sup>33</sup> and piperazine<sup>34</sup> linkage formation in a 2D conjugated COF by nucleophilic aromatic substitution have been established as fused heterocyclic organic linkage to build up crystalline and stable COFs. Corresponding COFs upon these linkages have been applied for catalysis and energy storage devices. However, these linkages usually result in relatively close layers spacing associated with their typical planar interlayer π-stacking arrangement, leading to the inner active-sites being buried to some degree. Recently, Kaskel<sup>35</sup> and co-workers constructed a dithiine-linked COF with undulated layers due to the bending along the C−S−C bridge but the aromaticity and crystallinity of the overall COF structure still maintained, providing a heuristic for more efficient utilization of buried inner active-sites. In addition to the choice of linkage in 2D conjugated COFs, planar conjugated precursors including porphyrin,<sup>36,37</sup> phthalocyanine (Pc),<sup>38</sup> and hexabenzocoronene<sup>39</sup> have usually been selected as building blocks owing to their robust stability and intrinsic high electrical conductivity. Particularly, metallic phthalocyanine (MPc) building units with M-N₄ coordination configuration have been demonstrated to act as high-efficiency active sites for catalyzing a series of reactions as exemplified by the efficient 2e⁻ ORR activity of CoPc.<sup>40</sup>
+
+Herein, a new dithiine-linked 2D CoPc-based COF, CoPc-S-COF, was afforded from the reaction of hexadecafluorophthalocyaninato cobalt (II) (CoPcF₁₆) with 1,2,4,5-benzenetetrathiol (BTT). For the purpose of comparison, a conventional dioxin-linked 2D CoPc-based COF, CoPc-O-COF, was also prepared by reaction between CoPcF₁₆ and 1,2,4,5-tetrahydroxybenzene (THB). Powder X-ray diffraction (PXRD) and electron microscopy analysis results reveal the crystalline porous framework of CoPc-S-COF with an undulated layered structure due to the bending along the C-S-C bridge associated with the large atomic radius and two lone-pairs of electrons of the sulfur atoms in the linking unit, resulting in almost double exposed active Co sites for 2e⁻ ORR compared to CoPc-O-COF with an eclipsed π-stacking model according to the electrochemical analysis. This, in combination with the activated 2e⁻ ORR but deactivated H₂O₂ decomposition capability of the same Co center due to the electron-donating effect of S atoms, enables CoPc-S-COF to display a superior electrocatalytic 2e⁻ ORR performance with a remarkable H₂O₂ selectivity of 95% and a stable H₂O₂ production under a high current density of 125 mA cm⁻² at an overpotential of ca. 190 mV for 20 h in a flow cell, generating H₂O₂ solution with a concentration of 0.48 wt%.
+
+# 2. Materials synthesis and characterization
+
+The synthesis of CoPc-O-COF and CoPc-S-COF is illustrated in Fig. 1a and their simulated structural models are displayed in Figs. 1b–1e. Nucleophilic substitution reaction of CoPcF₁₆ with THB and BTT, respectively, in dimethylacetamide (DMAC) and p-xylene with trimethylamine (Et₃N) as catalyst affords CoPc-O-COF and CoPc-S-COF in the yield of 75 and 88%. Observation of the band at 1298 cm⁻¹ due to the C-O-C bonds¹³ in the Fourier-transform infrared (FT-IR) spectrum demonstrates the successful formation of dioxin bridge in CoPc-O-COF, Supplementary Fig. 1. The characteristic band of the C-S-C linkage³⁵ gets appeared at 717 cm⁻¹ in the FT-IR spectrum, Supplementary Fig. 2, verifying the successful formation of dithiine bridge in CoPc-S-COF. The solid-state ¹³C cross-polarization/magic-angle-spinning (CP/MAS) NMR spectroscopy reveals the characteristic aromatic carbon signals at 149 and 143 ppm for CoPc-O-COF and CoPc-S-COF, respectively, further supporting the generation of the dioxin/dithiine-linked COFs, Supplementary Figs. 3 and 4. Both CoPc-O-COF and CoPc-S-COF exhibit a signal at ca. -124 ppm in their solid-state ¹⁹F CP/MAS NMR spectra, indicating their same C-F group nature, Supplementary Fig. 5. The decomposition temperature was revealed to be above 300°C for both COFs according to thermogravimetric analysis, indicating their great thermal stability, Supplementary Fig. 6. Moreover, the PXRD patterns of both COFs recollected after soaking in different solutions including 1 M KOH, 1 M HCl, pure water, THF, DMF, acetone, and ethanol for three days remain unchanged, unveiling the good chemical stability of both COFs, Supplementary Figs. 7 and 8.
+
+The crystalline structures of these two COFs were estimated by PXRD measurement combined with computational simulation. As displayed in Fig. 1f, CoPc-O-COF shows two strong peaks at 4.13 and 27.48°, corresponding to (100) and (001) facets, respectively. Moreover, the experimental pattern of CoPc-O-COF agrees with the calculated one with AA layer stackings on the basis of Forcite geometrical simulation method, Figs. 1b and 1c. Furthermore, the Pawley-refined PXRD pattern of CoPc-O-COF using the P4/mmm space the observed experimental curve as proved by the good agreement factors of Rp = 2.73% and Rwp = 3.59%, Fig. 1f. The PXRD pattern of CoPc-S-COF exhibits one strong peak at 4.32° and four medium intensity reflections at 6.16, 8.66, 9.46, and 24.73°, corresponding to (110), (020), (220), (130), and (001) facets. Combination of the theoretical simulation and Pawley refinement indicates that CoPc-S-COF adopts undulated layer-stacked structure owing to the nonplanar configuration of the C-S-C units with a dihedral angle of ca. 101°, affording the lattice parameters of a = b = 28.65 Å, c = 3.61 Å, α = γ = 90.00°, and β = 89.15° in C2/m space group with Rp = 3.43% and Rwp = 4.40%, Figs. 1c, 1f, and 1g. The Raman spectra of both COFs were recorded to explore their vibrational splittings based on corresponding functional bonds, Supplementary Fig. 9. The obvious symmetric peaks due to the aromatic carbon and carbon-oxygen bonds indicate the in-plane vibration nature for CoPc-O-COF. Observation of the series of asymmetric stretching bands due to the aromatic carbon and carbon-sulfur vibrations for CoPc-S-COF indicates the different vibrational splittings of the energy states of the C-S bond associated with the bending or out-of-plane twisting of the bonds.¹⁵ Actually unlike the dioxin-linked structure for CoPc-O-COF, the C-S-C units in CoPc-S-COF are stabilized in a nonplanar configuration to minimize the lone pair electron repulsion of sulfur atoms with large atomic radius in neighboring layers,⁴¹ resulting in its undulated layer-stacked structure.
+
+The morphology of CoPc-O-COF and CoPc-S-COF was investigated by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images, Figs. 2a–2f and Supplementary Fig. 10. As can be found, CoPc-O-COF exhibits a micrometer scale irregular sheet morphology, different from the irregular cluster morphology of CoPc-S-COF due to the undulated layer-stacked mode. Both CoPc-O-COF and CoPc-S-COF exhibit distinct lattice fringes with a spacing of 1.93 ± 0.11 and 1.88 ± 0.12 nm in their high-resolution TEM (HR-TEM) images, which are attributed to the (100) plane of CoPc-O-COF and (110) plane of CoPc-S-COF, respectively, and in turn confirm their high crystallinity, Figs. 2b and 2e. In addition, clear lattice fringes belonging to the (001) plane of these two COFs get appeared at ca. 0.32 ± 0.02 nm for CoPc-O-COF and ca. 0.37 ± 0.02 nm for CoPc-S-COF. Nevertheless, the corresponding fast Fourier transform (FFT) analysis displays the crystalline spot, Supplementary Fig. 11, further demonstrating their good crystallinity. Energy dispersive X-ray (EDX) mapping analysis reveals the elemental composition of Co, C, N, and F in both COFs as well as O element in CoPc-O-COF and S element in CoPc-S-COF with corresponding atom ratios close to the theoretical values, Figs. 2c, 2f, Supplementary Fig. 12, and Supplementary Table 1. As displayed in Supplementary Fig. 13, N₂ adsorption-desorption measurements reveal their permanent porosity with a Brunauer-Emmett-Teller (BET) surface area of 183 m² g⁻¹ for CoPc-O-COF and 285 m² g⁻¹ for CoPc-S-COF.
+
+X-ray photoelectron spectroscopy (XPS) was also performed to explore the elemental composition and metal valence states in both COFs. The XPS spectra of both COFs disclose the peaks due to Co, C, N, and F elements, Supplementary Fig. 14 and Supplementary Table 2, in agreement with corresponding EDX mapping results. The Co 2p XPS spectrum of CoPc-O-COF displays two peaks at 781.5 and 796.7 eV, attributed to Co 2p₃/₂ and Co 2p₁/₂ of Co (II), Fig. 2g. Nevertheless, the Co 2p₃/₂ and Co 2p₁/₂ peaks of CoPc-S-COF shift to a lower energy of 780.8 and 796.1 eV compared to CoPc-O-COF, due to the significant electron-donating effect of S atoms in CoPc-S-COF. The characteristic peaks due to the C-O-C⁴² and chemisorbed H₂O/O₂ appear at 531.5 and 533.3 eV, respectively, in the high-resolution O 1s XPS spectrum of CoPc-O-COF, and two peaks centered at 164.8 and 163.5 eV due to C-S-C⁴³ are observed in the high-resolution S 2p XPS spectrum of CoPc-S-COF, Figs. 2h and 2i, further confirming the successful generation of dioxin/dithiine-connected CoPc-based COFs. Both CoPc-O-COF and CoPc-S-COF exhibit a F 1s peak at 687.2 eV in their F 1s XPS spectra, respectively, indicating their same C-F group nature, Supplementary Fig. 15. Additionally, the X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) spectroscopies were used to determine the chemical state and local coordination environment of the Co species. As displayed in Supplementary Fig. 16, the average oxidation state of Co centers in both COFs is close to +2 according to their similar absorption edge to that for CoPc in Co K-edge XANES spectra. Moreover, the Fourier transformation (FT) EXAFS spectra and corresponding fitting results of both COFs show a peak at 1.4 Å due to the Co-N scattering path with a coordination number of ca. 4, Supplementary Fig. 17 and Supplementary Table 3. In particular, the Co content of CoPc-O-COF and CoPc-S-COF amounts to 5.91 and 5.01 wt%, respectively, according to inductively coupled plasma-optical emission spectrometry (ICP-OES), very close to the theoretical values, Supplementary Table 4.
+
+### 3. Electrocatalytic ORR performance
+
+The ORR measurements were performed on the three-electrode system with the rotating ring-disk electrode (RRDE) used as the working electrode in the condition of alkaline media. As shown in Fig. 3a, the ORR polarization curves and H₂O₂ detection current of both COFs are collected on RRDE at 1600 rpm in O₂-saturated 0.1 M KOH. CoPc-O-COF and CoPc-S-COF electrodes offer high reactivity for oxygen reduction to render an onset potential of 0.82 V versus reversible hydrogen electrode (RHE). Moreover, the Tafel slopes of CoPc-O-COF and CoPc-S-COF are calculated to be 47 and 49 mV dec⁻¹, respectively, implying their same rapid reaction kinetics, Supplementary Fig. 18. Figure 3b presents the H₂O₂ selectivity and electron transfer number *n* during ORR for both COFs. The H₂O₂ selectivity of CoPc-S-COF amounts to larger than 90% in the voltage range of 0.20–0.70 V versus RHE with an *n* value of 2.0–2.2, suggesting its promising 2e⁻ ORR performance. Meanwhile, the H₂O₂ selectivity value of CoPc-O-COF is slightly lower than that of CoPc-S-COF in the same voltage range. According to the ORR polarization curves at different rotation rates and Koutecky–Levich (K–L) diffusion equation, the electron transfer number *n* of CoPc-O-COF and CoPc-S-COF is determined to be *ca*. 2.4 and 2.2, respectively, Supplementary Fig. 19, consistent with the RRDE result. The Co mass activity (MA) of both COFs was also calculated, Supplementary Fig. 20. CoPc-S-COF exhibits a MA of 280.1 A g<sub>Co</sub>⁻¹ at 0.3 V vs RHE, superior to that of CoPc-O-COF, 190.2 A g<sub>Co</sub>⁻¹ at 0.3 V vs RHE. In addition, the conductivity of both COF electrodes was analyzed by electrochemical impedance spectroscopy (EIS) measurements. As shown in Supplementary Fig. 21, CoPc-O-COF and CoPc-S-COF electrodes exhibit a small EIS semicircle diameter of 112 and 87 Ω, indicating the optimized charge transfer of these two COFs owing to their conjugated structure. Nevertheless, the double-layer capacitances (C<sub>dl</sub>) of both COFs, which is proportional to their electrochemical surface area, were derived from the CV curves at different sweep rates, Supplementary Fig. 22. The C<sub>dl</sub> of CoPc-S-COF is calculated as 241 µF cm⁻², much larger than that of CoPc-O-COF, 171 µF cm⁻², indicating the more available active sites within CoPc-S-COF originated from its twisted layered structure, in turn leading to its higher 2e⁻ ORR performance. In line with this point, the surface electrochemical active sites on the CoPc-S-COF electrode are calculated to be 62.3 nmol cm⁻² according to the peak current of CV curves as a function of scan rate, Supplementary Fig. 23, revealing 11.0% of the total cobalt-phthalocyanine units acting as active sites. This value is almost twice of that for the CoPc-O-COF electrode, 6.9%, confirming the more exposed active sites in CoPc-S-COF, Supplementary Fig. 24. This in turn becomes responsible for the much superior 2e⁻ ORR activity of CoPc-S-COF to CoPc-O-COF.
+
+The catalytic stability of both COFs on RRDE was assessed by the chronoamperometry measurements, Figs. 3c and Supplementary Fig. 25. The Co-S-COF electrode shows almost unchanged current signals and maintained H₂O₂ selectivity of >93% for 10 h at 0.52 V versus RHE, verifying its robust catalytic durability for 2e⁻ ORR in alkaline media. As displayed in Supplementary Fig. 25, Co-O-COF also shows good catalytic stability on RRDE. Moreover, both CoPc-O-COF and CoPc-S-COF exhibit similar PXRD patterns before and after the stability tests, proving their robust framework nature, Supplementary Fig. 26. The scanning transmission electron microscope and XANES measurements were also carried out for both COFs before and after the stability tests. As exhibited in Supplementary Fig. 27, no Co nanoparticles can be observed in the STEM images of the two COFs before and after the stability tests, excluding the Co leaching and aggregation during the electrochemical test. Particularly, the Co K-edge EXAFS spectra of both COFs after the stability tests also show only the peak at 1.4 Å due to the Co-N scattering path without giving the peak at 2.2 Å due to the Co-Co bond, further confirming the absence of Co nanoparticles formed from the Co leaching during the electrochemical test and in turn the excellent durability of the two COFs, Supplementary Fig. 28. The poison experiment was then carried out to identify the catalytic site by using SCN⁻ ions, which tend to bind the metal atoms of CoPc and block the adsorption of reaction intermediates, thus negatively affecting the catalytic performance. As shown in Supplementary Fig. 29, significant degradation of the catalytic activity occurs after adding SCN⁻ into the electrolyte, demonstrating the actual active site nature of the Co metal center for 2e⁻ ORR in the two COFs. Additionally, *in-situ* Co center poisoning experiments with nitrite has also been carried out and shown in Supplementary Fig. 30. As can be found, significant degradation of the catalytic activity for CoPc-S-COF occurs after adding nitrite into the electrolyte, further confirming the active site nature of the Co atoms towards 2e⁻ ORR in CoPc-S-COF. Particularly, the polarization curves of H₂O₂ reduction reaction (H₂O₂ RR) of both COFs were recorded in the Ar-saturated 0.1 M KOH solution containing 10 mM H₂O₂. As revealed in Fig. 3d, both samples display negligible H₂O₂ RR activity. To more clearly compare the inhibition of H₂O₂ decomposition for both COFs, the H₂O₂ RR polarization curves of CoPc-O-COF and CoPc-S-COF in Ar-saturated 0.1 M KOH solutions containing different concentrations of H₂O₂ were measured, Figs. 3e and 3f. CoPc-O-COF exhibits larger H₂O₂ RR currents than those of CoPc-S-COF especially in high H₂O₂ concentration (>10 mM). This result indicates the superior inhibition of CoPc-S-COF to H₂O₂ decomposition during the 2e⁻ ORR process due to the enriched electron nature of the Co center associated with the electron-donating sulfur atoms in the linking unit, favouring the production of high concentration H₂O₂ solution.
+
+To explore the impact of F atom on the catalytic activity and selectivity of the Co centers in CoPc-O-COF and CoPc-S-COF, the catalytic activity of the small molecule Co phthalocyanines including CoPc, CoPcF₁₆, and the metal-free phthalocyanine H₂PcF₁₆ electrodes have also been prepared and tested on RRDE at 1600 rpm in O₂-saturated 0.1 M KOH, Supplementary Fig. 31. It is worth noting that the CoPc and CoPcF₁₆ electrodes exhibit the same onset potential of 0.78 V versus RHE and similar H₂O₂ selectivity (>70%) in the voltage range of 0.20–0.60 V versus RHE with an *n* value of 2.5–2.6, demonstrating their obvious similar reactivity for 2e⁻ oxygen reduction with negligible effect of F atoms. However, the metal-free H₂PcF₁₆ electrode displays much inferior electrocatalytic performance with lower onset potential of 0.68 V versus RHE and H₂O₂ selectivity (<30%), confirming the nature of Co active centers. For the purpose of further clarifying the impact of O and S atoms on the catalytic activity and selectivity of the Co centers in CoPc-O-COF and CoPc-S-COF, two small molecule Co phthalocyanines including CoPc-S and CoPc-O containing C-S-C and C-O-C groups, respectively, were synthesized with their 2e⁻ ORR performance assessed, Supplementary Figs. 32 and 33. As can be seen, CoPc-S shows better 2e⁻ ORR activity and worse H₂O₂ RR activity in comparison with CoPc-O, proving the effect of S on enhancing the 2e⁻ ORR performance of CoPc-S-COF. In addition, the S-doped CoPc-O-COF (named CoPc-O-COF-S 20%) was prepared with its electrocatalytic performance tested. As expected, CoPc-O-COF-S 20% shows high ORR activity with an onset potential of 0.80 V versus RHE and an electron transfer number of 2.2–2.3, Supplementary Figs. 34–36. Nevertheless, CoPc-O-COF-S20% displays smaller H₂O₂ RR currents than those of CoPc-O-COF in the Ar-saturated 0.1 M KOH solution containing 10–50 mM H₂O₂, revealing the effect of S-doping on diminishing the catalytic activity towards H₂O₂ decomposition.
+
+To gain further insight into the 2e⁻ ORR activity of the COFs, the electronic structure of CoPc-O-COF and CoPc-S-COF was investigated by density functional theory (DFT) calculation. The adsorption energy (ΔG<sub>ads</sub>) of O₂ on various atoms of the two COFs including Co, N, O, S, and F has been firstly calculated to explore the active site of H₂O₂ production. As can be seen in Supplementary Fig. 37 and Supplementary Table 5, the significantly smaller value of ΔG<sub>ads</sub> for Co atom, *ca.* -0.3 eV, than those for other atoms (*ca.* -0.02 ~ -0.05 eV for N, *ca.* +0.4 ~ +0.6 eV for O/S/F) reveals the active site nature of Co atom in the two COFs towards ORR. In addition, the electron density on the CoPc moiety including around Co site in CoPc-S-COF is higher than that for CoPc-O-COF due to the electron-donating effect of S atoms, which facilitates the charge transfer between Co active sites and intermediates and in turn affords enhanced catalytic activity for 2e⁻ ORR, Fig. 3g and Supplementary Fig. 38. Moreover, the calculated projected density of states (pDOS) discloses a lower d band center position of -1.33 eV for CoPc-S-COF with higher intensity of peaks near the Fermi level (E<sub>f</sub>) compared to CoPc-O-COF with a d band center position of -1.17 eV, Fig. 3h. This indicates that larger density of active electrons around Co centers in CoPc-S-COF participates in the electrochemical ORR reaction, confirming the higher catalytic activity of CoPc-S-COF due to the electron-donating effect of S atoms. Nevertheless, Fig. 3i presents the calculated Gibbs free energy differences (ΔG) diagrams of the 2e⁻ ORR and 4e⁻ ORR processes on CoPc-S-COF and CoPc-O-COF. As can be found, the conversion of OOH* to H₂O₂ is the rate-determining step of 2e⁻ ORR on both CoPc-S-COF and CoPc-O-COF with an energy barrier of 0.63 and 1.11 eV, respectively. These values are smaller than the energy barrier of the OOH* to O* conversion process (the rate-determining step of 4e⁻ ORR) on both COFs, 1.44 and 1.48 eV, demonstrating the faster reaction kinetics of 2e⁻ ORR than 4e⁻ ORR on both COFs and in turn their outstanding selectivity towards 2e⁻ ORR. In particular, the lower OOH* to H₂O₂ conversion energy barrier of CoPc-S-COF in comparison with that for CoPc-O-COF indicates the enhanced 2e⁻ ORR activity of the former species over the latter one, while the higher H₂O₂ to OH⁻ conversion energy barrier of CoPc-S-COF than that for CoPc-O-COF illustrates the diminished activity towards H₂O₂ decomposition of CoPc-S-COF to CoPc-O-COF, rationalizing the more stable electrocatalytic H₂O₂ production of CoPc-S-COF.
+
+Gas diffusion electrode (GDE) devices were used to further explore the practical application potential of the as-prepared CoPc-based COFs towards 2e⁻ ORR. A three-phase flow cell, Fig. 4a, in which the catalyst is deposited on a gas diffusion layer (GDL) as the work electrode, is deemed to be able to afford higher reduction current densities by increasing oxygen concentration on GDL and improve the H₂O₂ production rate. Corresponding measurements of both COFs were carried out in 1 M KOH. As exhibited in Fig. 4b, both COFs show excellent catalytic activity with much higher current density in flow cell compared to RRDE measurements. The chronoamperometry measurements at varied applied voltages were conducted with the generated H₂O₂ determined by the Ce⁴⁺ titration method, Fig. 4c, and Supplementary Figs. 39 and 40. Figure 4d shows the determined Faradaic efficiency of H₂O₂ (FE<sub>H2O2</sub>) for both COFs. CoPc-O-COF and CoPc-S-COF could exhibit over 80% FE<sub>H2O2</sub> in the range of 0.73 to 0.33 V versus RHE, even higher than the RRDE measurements in the range from 0.73 to 0.53 V. Remarkably, the J<sub>H2O2</sub> of CoPc-S-COF reaches 152 mA cm⁻² at 0.63 V versus RHE (equivalent to an overpotential of 230 mV) with a FE<sub>H2O2</sub> of 98% in 1 M KOH at room temperature, which gets further increased to 415 mA cm⁻² at 0.33 V versus RHE (equivalent to an overpotential of 530 mV) with FE<sub>H2O2</sub> still higher than 80%, superior to CoPc-O-COF, Figs. 4c, 4d, and Supplementary Fig. 39. The H₂O₂ RR performance of both COFs was also explored in the flow cell. As displayed in Fig. 4b, CoPc-S-COF shows much smaller H₂O₂ reduction current densities compared to CoPc-O-COF, consistent with the result in the RRDE system. This actually suggests that H₂O₂ RR could be inhibited in CoPc-S-COF and therefore leads to higher FE<sub>H2O2</sub> and H₂O₂ concentration. Furthermore, the long standing and stable H₂O₂ production of CoPc-S-COF has been recorded at a fixed current density of 125 mA cm⁻², Figs. 4e and 4f. CoPc-S-COF maintains FE<sub>H2O2</sub> >95% in the continuous H₂O₂ electroproduction for 20 h. Nevertheless, the H₂O₂ amount produced gets linearly increased along with increasing the operating time with an almost constant production rate of *ca.* 9500 mmol g<sub>cat</sub>⁻¹ h⁻¹ and an almost unchanged operating voltage of *ca.* 0.67 V versus RHE, Fig. 4f, further confirming the excellent stability of CoPc-S-COF as well as its inactivation towards H₂O₂ decomposition. More importantly, the H₂O₂ yield reaches up to 377 mg after 20 h electrolysis corresponding to a H₂O₂ concentration of 0.48 wt%, Fig. 4e, representing the thus far reported best COFs electrocatalytic H₂O₂ production performance and demonstrating the great application potential of CoPc-S-COF in practical H₂O₂ production, Supplementary Table 6. Moreover, after the stability test, CoPc-S-COF shows constant structure according to the FTIR and XPS analysis, further proving its robust stability, Supplementary Figs. 41 and 42.
+
+# 4. Conclusion
+
+In summary, a porous dithiine-linked CoPc-based COF was fabricated. CoPc-S-COF possesses an undulated layer-stacked structure due to the bending along the C-S-C bridge to allow more exposed Co centers for 2e<sup>-</sup> ORR. This, in combination with the activated 2e<sup>-</sup> ORR but deactivated H<sub>2</sub>O<sub>2</sub> decomposition capability of the Co center because of the electron-donating effect of S atoms, enables CoPc-S-COF to display a high H<sub>2</sub>O<sub>2</sub> selectivity and realize the large-scale H<sub>2</sub>O<sub>2</sub> production in the flow cell. This work should be beneficial to the design and preparation of high-performance and low-cost electrocatalysts for H<sub>2</sub>O<sub>2</sub> electrosynthesis.
+
+# Methods
+
+**Synthesis of CoPc-O-COF.** CoPcF₁₆ (8.6 mg, 0.01 mmol) and THB (2.8 mg, 0.02 mmol) were added into the mixed solvent of 0.7 mL *p*-xylene and 0.5 mL DMAc in a 16 mL Pyrex tube. The mixture was sonicated for 5 min to form a homogeneous suspension. Then 100 µL triethylamine was added into the mixture. After three freeze-pump-thaw cycles, the Pyrex tube was sealed and heated in an oven at 100°C for 7 days. The black-green precipitate was collected by centrifugation and rinsed with acetone, dichloromethane, and THF in a Soxhlet extractor for one day. Finally, CoPc-O-COF was then obtained as black powder in a yield of 75%.
+
+**Synthesis of CoPc-S-COF.** CoPcF₁₆ (8.6 mg, 0.01 mmol) and BTT (4.2 mg, 0.02 mmol) were added into the mixed solvent of 0.7 mL *p*-xylene and 0.5 mL DMAc in a 16 mL Pyrex tube. The mixture was sonicated for 5 min to form a homogeneous suspension. Then 100 µL triethylamine was added into the mixture. After three freeze-pump-thaw cycles, the Pyrex tube was sealed and heated in an oven at 100°C for 7 days. A dark-green solid was formed inside the Pyrex tube during the reaction process. The product was collected by centrifugation and then solvent exchanged with dry ethanol for a week followed by supercritical drying. Finally, CoPc-S-COF was obtained as fluffy dark-green product in a yield of 88%.
+
+# References
+
+1. Yang, X., et al. Tuning Two-Electron Oxygen-Reduction Pathways for H₂O₂ Electrosynthesis via Engineering Atomically Dispersed Single Metal Site Catalysts. *Adv. Mater.* **34**, e2107954 (2022).
+2. Chen, J., et al. Kinetically restrained oxygen reduction to hydrogen peroxide with nearly 100% selectivity. *Nat. Commun.* **13**, 2808 (2022).
+3. Wang, N., et al. Recent Progress of Electrochemical Production of Hydrogen Peroxide by Two-Electron Oxygen Reduction Reaction. *Adv. Sci.* **8**, e2100076 (2021).
+4. Wang, Y., Waterhouse, G.I.N., Shang, L. & Zhang, T. Electrocatalytic Oxygen Reduction to Hydrogen Peroxide: From Homogeneous to Heterogeneous Electrocatalysis. *Adv. Energy Mater.* **11**, 2003323 (2020).
+5. Fei, H., et al. Atomic cobalt on nitrogen-doped graphene for hydrogen generation. *Nat. Commun.* **6**, 8668 (2015).
+6. Bu, Y., et al. Carbon-Based Electrocatalysts for Efficient Hydrogen Peroxide Production. *Adv. Mater.* **33**, e2103266 (2021).
+7. Jiang, K., et al. Catalyst Design for Electrochemical Oxygen Reduction toward Hydrogen Peroxide. *Adv. Funct. Mater.* **30**, 2003321 (2020).
+8. Kim, H.W., et al. Efficient hydrogen peroxide generation using reduced graphene oxide-based oxygen reduction electrocatalysts. *Nat. Catal.* **1**, 282–290 (2018).
+9. Tian, Y., et al. Edge-hosted Atomic Co-N₄ Sites on Hierarchical Porous Carbon for Highly Selective Two-electron Oxygen Reduction Reaction. *Angew. Chem. Int. Ed.* **61**, e202213296 (2022).
+10. Li, B.Q., et al. Electrosynthesis of Hydrogen Peroxide Synergistically Catalyzed by Atomic Co-Nx-C Sites and Oxygen Functional Groups in Noble-Metal-Free Electrocatalysts. *Adv. Mater.* **31**, e1808173 (2019).
+11. Sun, Y., et al. A comparative perspective of electrochemical and photochemical approaches for catalytic H₂O₂ production. *Chem. Soc. Rev.* **49**, 6605–6631 (2020).
+12. Li, D., et al. Metal-Free Thiophene-Sulfur Covalent Organic Frameworks: Precise and Controllable Synthesis of Catalytic Active Sites for Oxygen Reduction. *J. Am. Chem. Soc.* **142**, 8104–8108 (2020).
+13. Melchionna, M., et al. The Rise of Hydrogen Peroxide as the Main Product by Metal-Free Catalysis in Oxygen Reductions. *Adv. Mater.* **31**, e1802920 (2019).
+14. Lee, B.-H., et al. Supramolecular tuning of supported metal phthalocyanine catalysts for hydrogen peroxide electrosynthesis. *Nat. Catal.* **6**, 234–243 (2023).
+15. Lin, R., et al. Approaching theoretical performances of electrocatalytic hydrogen peroxide generation by cobalt-nitrogen moieties. *Angew. Chem. Int. Ed.* **62**, e202301433 (2023).
+16. Zhao, X. et al. Origin of Selective Production of Hydrogen Peroxide by Electrochemical Oxygen Reduction. *J. Am. Chem. Soc.* **143**, 9423–9428 (2021).
+17. Zhang, C., et al. Crystal Engineering Enables Cobalt-Based Metal-Organic Frameworks as High-Performance Electrocatalysts for H₂O₂ Production. *J. Am. Chem. Soc.* **145**, 7791–7799 (2023).
+18. Diercks, C. S., et al. The atom, the molecule, and the covalent organic framework. *Science* **355**, eaal1585 (2017).
+19. Pramudya, Y., et al. Design Principles for High H₂ Storage Using Chelation of Abundant Transition Metals in Covalent Organic Frameworks for 0–700 bar at 298 K. *J. Am. Chem. Soc.* **138**, 15204–15213 (2016).
+20. Wang, Z., et al. Covalent organic frameworks for separation applications. *Chem. Soc. Rev.* **49**, 708–735 (2020).
+21. Keller, N., et al. Optoelectronic processes in covalent organic frameworks. *Chem. Soc. Rev.* **50**, 1813–1845 (2021).
+22. Huang, S., et al. Porphyrin and phthalocyanine based covalent organic frameworks for electrocatalysis. *Coord. Chem. Rev.* **464**, 214563 (2022).
+23. Yang, X., et al. Ionothermal Synthesis of Fully Conjugated Covalent Organic Frameworks for High-Capacity and Ultrastable Potassium-Ion Batteries. *Adv. Mater.* **34**, e2207245 (2022).
+24. Bhunia, S., et al. [2,1,3]-Benzothiadiazole-Spaced Co-Porphyrin-Based Covalent Organic Frameworks for O₂ Reduction. *ACS Nano* **17**, 3492–3505 (2023).
+25. Liu, M., et al. Construction of Catalytic Covalent Organic Frameworks with Redox-Active Sites for the Oxygen Reduction and the Oxygen Evolution Reaction. *Angew. Chem. Int. Ed.* **61**, e202213522 (2022).
+26. Mondal, S., et al. A Thiadiazole-Based Covalent Organic Framework: A Metal-Free Electrocatalyst toward Oxygen Evolution Reaction. *ACS Catal.* **10**, 5623–5630 (2020).
+27. Lin, C.Y., et al. Covalent Organic Framework Electrocatalysts for Clean Energy Conversion. *Adv. Mater.* **30**, 1703646 (2018).
+28. Han, B., et al. Two-Dimensional Covalent Organic Frameworks with Cobalt(II)-Phthalocyanine Sites for Efficient Electrocatalytic Carbon Dioxide Reduction. *J. Am. Chem. Soc.* **143**, 7104–7113 (2021).
+29. Zhong, H., et al. Boosting the Electrocatalytic Conversion of Nitrogen to Ammonia on Metal-Phthalocyanine-Based Two-Dimensional Conjugated Covalent Organic Frameworks. *J. Am. Chem. Soc.* **143**, 19992–20000 (2021).
+30. Yang, C., et al. Theory-Driven Design and Targeting Synthesis of a Highly-Conjugated Basal-Plane 2D Covalent Organic Framework for Metal-Free Electrocatalytic OER. *ACS Energy Lett.* **4**, 2251–2258 (2019).
+31. Lu, M., et al. Stable Dioxin-Linked Metallophthalocyanine Covalent Organic Frameworks (COFs) as Photo-Coupled Electrocatalysts for CO₂ Reduction. *Angew. Chem. Int. Ed.* **60**, 4864–4871 (2021).
+32. Yue, Y., et al. Stable Bimetallic Polyphthalocyanine Covalent Organic Frameworks as Superior Electrocatalysts. *J. Am. Chem. Soc.* **143**, 18052–18060 (2021).
+33. Huang, N., et al. A Stable and Conductive Metallophthalocyanine Framework for Electrocatalytic Carbon Dioxide Reduction in Water. *Angew. Chem. Int. Ed.* **59**, 16587–16593 (2020).
+34. Zhi, Q., et al. Piperazine-Linked Metalphthalocyanine Frameworks for Highly Efficient Visible-Light-Driven H₂O₂ Photosynthesis. *J. Am. Chem. Soc.* **144**, 21328–21336 (2022).
+35. Haldar, S., et al. Porous Dithiine-Linked Covalent Organic Framework as a Dynamic Platform for Covalent Polysulfide Anchoring in Lithium-Sulfur Battery Cathodes. *J. Am. Chem. Soc.* **144**, 9101–9112 (2022).
+36. Wang, K., et al. Tetrapyrrole macrocycle based conjugated two-dimensional mesoporous polymers and covalent organic frameworks: From synthesis to material applications. *Coord. Chem. Rev.* **378**, 188–206 (2019).
+37. Lin, S., et al. Covalent organic frameworks comprising cobalt porphyrins for catalytic CO₂ reduction in water. *Science* **349**, 1208–1213 (2015).
+38. Qiu, X.F., et al. A Stable and Conductive Covalent Organic Framework with Isolated Active Sites for Highly Selective Electroreduction of Carbon Dioxide to Acetate. *Angew. Chem. Int. Ed.* **61**, e202206470 (2022).
+39. Yu, H.-Y., et al. A stack-guiding unit constructed 2D COF with improved charge carrier transport and versatile photocatalytic functions. *Chem. Eng. J.* **445**, 136713 (2022).
+40. Yuan, Y., et al. Deciphering the selectivity descriptors of heterogeneous metal phthalocyanine electrocatalysts for hydrogen peroxide production. *Chem. Sci.* **13**, 11260–11265 (2022).
+41. Tiekink, E.R.T. & Zukerman-Schpector, J. Stereochemical activity of lone pairs of electrons and supramolecular aggregation patterns based on secondary interactions involving tellurium in its 1,1-dithiolate structures. *Coord. Chem. Rev.* **254**, 46–76 (2010).
+42. Li, N., et al. Polyarylether-Based 2D Covalent-Organic Frameworks with In-Plane D-A Structures and Tunable Energy Levels for Energy Storage. *Adv. Sci.* **9**, e2104898 (2022).
+43. Shen, H., et al. Synergistic Effects between Atomically Dispersed Fe-N-C and C-S-C for the Oxygen Reduction Reaction in Acidic Media. *Angew. Chem. Int. Ed.* **56**, 13800–13804 (2017).
+44. Han, B., et al. Maximizing Electroactive Sites in a Three-Dimensional Covalent Organic Framework for Significantly Improved Carbon Dioxide Reduction Electrocatalysis. *Angew. Chem. Int. Ed.* **61**, e202114244 (2022).
+45. Chen, S., et al. Identification of the Highly Active Co-N₄ Coordination Motif for Selective Oxygen Reduction to Hydrogen Peroxide. *J. Am. Chem. Soc.* **144**, 14505–14516 (2022).
+46. Lin, L., et al. Atomic-Level Modulation-Induced Electron Redistribution in Co Coordination Polymers Elucidates the Oxygen Reduction Mechanism. *ACS Catal.* **12**, 7531–7540 (2022).
+47. Wu, C., et al. Polarization Engineering of Covalent Triazine Frameworks for Highly Efficient Photosynthesis of Hydrogen Peroxide from Molecular Oxygen and Water. *Adv. Mater.* **34**, 2110266 (2022).
+
+# Supplementary Files
+
+- [NCOMMS2334167SupplementaryInformation.docx](https://assets-eu.researchsquare.com/files/rs-3206425/v1/53eb9de438cbc1d0e9705bbd.docx)

096f68b6c3d806b969c69bed1e41f06bd80db5156ede8c7cd6984ab2b966a978/preprint/images_list.json

@@ -0,0 +1,18 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Elemental composition from LCROSS compared to the elemental composition of possible sources. LCROSS observations (black squares) are based on the assumption that the volatiles are either stored as \u201cClathrates\u201d or are condensed onto the regolith as \u201cCondensates.\u201d Note that no single source exactly matches all of the elemental ratios. Data are provided in (15).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Stability curves for clathrates stored in the PSRs. Comparison of the pressure-temperature profile, or P=f(T), in the upper lunar regolith (0.2 to 5 m) to stability curves for clathrates with SO2, H2S, and CO (16\u201318). The P=f(T) profile is calculated based on a temperature profile extrapolated from (13) and pressure based on a 1.66 g/cc lunar regolith (19). Mixed clathrate stability curves are based on clathrates formed from gas mixture of CO + SO2 or H2S (with a cometary C/S from 20); such clathrate is dominated by SO2 or H2S. Because the P=f(T) for the lunar regolith falls in the area above and to the left of all stability curves, the regolith is within the clathrate stability domain.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

096f68b6c3d806b969c69bed1e41f06bd80db5156ede8c7cd6984ab2b966a978/preprint/preprint.md

@@ -0,0 +1,519 @@
+# Abstract
+
+Returning humans to the Moon presents an unprecedented opportunity to determine the origin of volatiles stored in the permanently shaded regions (PSRs), which trace the history of lunar volcanic activity, solar wind surface chemistry, and volatile delivery to the Earth and Moon through impacts of comets, asteroids, and micrometeoroids. So far, the source of the volatiles sampled by the Lunar Crater Observation and Sensing Satellite (LCROSS) plume (1, 2) has remained undetermined. We show here that the source could not be volcanic outgassing and the composition is best explained by cometary impacts. Ruling out a volcanic source means that volatiles in the top 1–3 meters of the Cabeus PSR regolith may be younger than the latest volcanic outgassing event (~ 1 billion years ago; Gya) (3).
+
+Planetary Science  
+Astronomy  
+Geochemistry  
+volatiles  
+planetary science  
+LCROSS impact
+
+# 1. Introduction
+
+The LCROSS experiment impacted the upper stage of a spent Centaur rocket into the PSR of Cabeus crater, creating a plume that contained the first carbon-, nitrogen-, and sulfur-bearing volatiles detected in the lunar PSRs (1, 2, 4, 5). These ground-breaking observations not only provide ground truth for ongoing remote observations of water on the surface (e.g., 6, 7) and at depth (e.g., 8, 9), but provide vital clues to the origin of volatiles present on the Moon. The LCROSS plume was observed 30 seconds after impact by the Lunar Reconnaissance Orbiter (LRO) Lyman Alpha Mapping Project (LAMP), which detected H₂ and CO (2, 4). Meanwhile, the LCROSS shepherding spacecraft measured the abundance of several additional species relative to water for four minutes until it also impacted into Cabeus crater (1). The published abundances from LAMP (2, 4) were derived from the expanding shell of vapor traveling at 3–4 km/s that passed LRO >100 km away from the impact site. In contrast, the published abundances from the LCROSS shepherding spacecraft (1) were derived from vapor emanating from the impact site over time. Thus, the published LAMP observations were not made at the same time as the LCROSS measurements and require reanalysis for proper comparison (see supplemental materials).
+
+To determine the origin of the volatiles observed in the LCROSS plume we must consider how volatile composition changed between the source and storage in the PSR. Several processes occur between initial delivery by the source and detection in the plume that change the molecular composition. This means that species that were measured in the plume may not be the same as the molecular species found in the source. To simplify the analysis and eliminate as many influences as possible, we instead compare the elemental composition of the LCROSS volatiles with the elemental composition of the potential sources, evaluating abundances of four elements as they relate to carbon: hydrogen (C/H), nitrogen (N/C), oxygen (O/C), and sulfur (C/S).
+
+# 2. Elemental Composition
+
+The elemental composition of the volatiles in the regolith of the PSR indicated by LCROSS observations depends on the type of ice storing the volatiles. We consider two cases based on types of ice that would be stable in the PSR regolith: condensates and clathrates. Condensates are volatiles condensed onto regolith grains, while clathrates are volatiles trapped in water cages. If the volatiles are stored as condensates, then each species is released according to its volatility temperature (5, 10), as assumed in (4, 11). The long-term stability of each species depends on how the temperature varies diurnally with depth (12). Thermal modeling shows that temperatures are stable below ~ 0.2 m depth (13). The LCROSS impactor was estimated to have excavated material from 1–3 meters deep in the PSR (14), so the volatiles observed in the plume originated below the depth of thermal stability. Additionally, Cabeus is one of the coldest PSRs, with diurnal variation in surface temperature between 38.7 and 46.7 K and subsurface temperatures estimated to be 38 K (12). This means that condensed volatiles in this PSR should remain stable long-term on the surface and at depth. We use regolith volatile abundances estimated by (11) based on the LCROSS plume composition, adjusting CO and H₂ based on our reanalysis of LAMP observations (5). Elemental ratios for volatiles sampled by LCROSS, assuming they were condensed in the regolith, are identified in Fig. 1 as “Condensates” (15).
+
+If the volatiles are stored in clathrates, then the plume composition is a reasonable representation of the volatile abundance in the regolith. This is because all volatiles trapped in clathrates are released together when clathrates become destabilized. The LCROSS elemental ratios for volatiles stored as clathrates are identified in Fig. 1 as “Clathrates” (15). We show in Fig. 2 that clathrates are stable at the temperatures and pressures beneath the surface in the Cabeus PSR.
+
+### 3. Volatile Sources
+
+The potential source or combination of sources for volatiles sampled by LCROSS will depend on the timing for volatile delivery. Cabeus crater is estimated to be 3.5 billion years old (21), providing an upper limit for the age of these volatiles. A lower limit comes from modeling the influence of impact gardening on ice deposits. Based on the abundance of ice detected by LCROSS and the depth probed by the impactor, the volatiles sampled should be from more than 1 Gya (22). Although volcanic outgassing was most active more than 3 Gya, some activity continued until 1 Gya (3), so a volcanic source cannot be ruled out based on deposit age. Throughout its history the Moon has been subject to impacts, with the largest fluxes predating the formation of Cabeus, between 3.5 and 4.6 Gya (23). However, impacts by comets and meteorites have continued since that time at a lower rate. Comets and chondrites in the form of asteroid impactors and micrometeoroids (24) are also a reasonable volatile source. Finally, water molecules can form through surface chemistry initiated by solar wind protons and travel to the PSRs (22).
+
+In Fig. 1 we compare the elemental composition for the LCROSS observations with potential volatile sources (15). Comet composition is based on coma measurements of sublimated ices, and varies significantly. However, refractory material in comet nuclei is likely chondritic in composition, so comet impacts would provide a combination of material with what we designate as “cometary,” as well as “chondritic” composition. We provide the composition for the coma of 67P/Churyumov-Gerasimenko (67P/C-G), using the best effort to date at determining elemental composition with Rosetta observations (20). We also illustrate average or extreme values based on coma observations from several comets (20). The C/S measured in comets ranges between 2.2 and 8.0 when sulfur-bearing species have been detected. N/C in comets ranges between 0.06 and 0.37. Note that the nitrogen inventory for these comets do not include N₂, which is difficult to measure remotely. In 67P/C-G, N₂ contributed ~ 17% of the total nitrogen inventory in the coma. The volcanic composition is from (3) with N/C from (25).
+
+# 4. Source Mixtures
+
+As Fig. 1 shows, no source is a perfect fit for the LCROSS measurements. Volcanic sources and chondrites provide the right amount of sulfur, but do not provide sufficient hydrogen and nitrogen. Volcanic sources are also deficient in oxygen. Comets provide sufficient hydrogen, carbon, and nitrogen, but are depleted in sulfur – even when considering the most extreme value. Solar wind only contributes hydrogen and oxygen (see supplementary materials). We developed a model to determine if a mixture of sources can match the LCROSS observations, and found that no combination was able to match all four elemental ratios within the uncertainties of the LCROSS measurements – even when taking into account the uncertainties for the sources (see supplemental materials). The main limitation is fitting both the C/S and the N/C ratios observed by LCROSS. The two sources with sufficient sulfur to match C/S, volcanoes and chondrites, are too depleted in nitrogen and hydrogen for any cometary contribution to provide agreement with N/C and still match C/S. This is the case even using the maximum N/C and the minimum C/S for comets. The best fit is provided by 100% comets, which agrees with all ratios except for C/S.
+
+To improve our constraints on the source, or mixture of sources, we consider processes that could fractionate elemental ratios between delivery of the source volatiles to the lunar surface and observation in the LCROSS plume, including volcanic atmospheric processes, impact processes, clathrate formation, and cycles of sublimation and recondensation. Because these processes are complex and difficult to accurately quantify, we determine whether the LCROSS observations represent upper or lower limits for the elemental ratios and summarize the results in Table 1.
+
+## 4.1 Volcanic Atmosphere Fractionation
+
+Volcanic sulfur is thought to be released as S₂, which could rapidly be lost to the surface as solid elemental sulfur or aerosols before reaching a cold trap (26). This would result in a higher C/S ratio in the PSR compared to the source, so the observed C/S is an upper limit for volcanic C/S. This creates a challenge for explaining the LCROSS C/S as volcanic in origin, because volcanic C/S would need to be much lower than C/S in the LCROSS plume to provide sufficient sulfur to explain the observations.
+
+The relative abundances of elements in volcanic gas can also be changed by the escape of molecules from the top of the atmosphere. Unfortunately, loss rates depend on a wide range of complex parameters that are not well constrained (27), making it difficult to quantify how much elemental ratios can fractionate as a result of escape. However, we can estimate upper and lower limits for LCROSS measurements compared to the sources based on the relative masses of the dominant species for each element. Escape from a volcanic atmosphere would be dominated by H and H₂ (26, 27) that either originated in the volcanic gas as H₂, or was produced by dissociation of water molecules. This would increase the C/H of the volatiles in the PSR, making the observations an upper limit for the source ratio. Atomic oxygen and OH produced by water dissociation could also be lost, making O/C in the PSR a lower limit compared to the source. Any nitrogen present would be in the form of either N₂ or NH₃, which are either the same mass as or lighter than volcanic carbon-bearing molecules CO and CO₂. This means that the N/C in the PSR is a lower limit for N/C in a volcanic source when considering atmospheric escape. Because volcanic N/C is drastically lower than the LCROSS observations, escape does not provide a mechanism allowing for volcanic gas to be the source of nitrogen in the Cabeus PSR.
+
+Although escape of hydrogen and oxygen leads to limits that provide worse agreement between a volcanic source and the LCROSS observations, water produced by solar wind surface chemistry would decrease C/H and increase O/C over time by adding water to the PSR (28), cancelling out escape fractionation. These ratios would allow for a combination of volcanic and solar wind sources. However, measured N/C and C/S ratios disagree with volcanic source composition, even accounting for processes that change elemental ratios in a volcanically-produced atmosphere, conclusively demonstrating that the volatiles sampled by LCROSS are not from a volcanic source.
+
+## 4.2 Fractionation of Impact Material
+
+Next, we consider fractionation of volatiles delivered by impacts of comets, asteroids, and micrometeoroids. The elemental ratios can be fractionated by impact loss and by escape during transport to cold traps. The total percentage of volatiles retained after impact depends on the impact velocity and angle (29). Volatiles lost to space escape rapidly as part of the outward flow of the impact plume. Fractionation is similar to hydrodynamic escape, with preferential loss of lighter species. However, light species flow outward rapidly enough to drag heavier species with them (e.g., 30). Additional loss to space could occur by escape during subsequent transport to cold traps over several Earth days (31). Fractionation can be estimated in the same way as with the volcanic atmosphere, assuming that lighter species are removed at a faster rate than heavier species. Hydrogen would primarily be in light molecules like H, H₂, and water making the C/H in the PSR an upper limit compared to C/H of the source. Loss of oxygen and OH would make O/C in the PSR a lower limit compared to the source. According to simulations of impact chemistry of comets (32) and chondrites (33), nitrogen in an impact plume would primarily be in the form of N₂ with some NH₃ present, while carbon and sulfur are found in heavier molecules like CO, CO₂, H₂S, SO₂ and OCS. As with the volcanically produced atmosphere, N/C in the LCROSS observations is a lower limit compared to the source. We also note that LCROSS and LAMP did not have the ability to detect N₂, which is expected to be produced in impact plumes. The N/C in the LCROSS plume may have been higher than observed, arguing further for that the observation is a lower limit compared to the source. The masses for carbon-bearing species are generally lighter than sulfur-bearing species, suggesting that C/S in the LCROSS observations is a lower limit compared to the source. We applied our model again using these constraints (see Table 1) and found that only cometary ices, with some contribution from solar wind-produced water, can explain all four elemental ratios.
+
+## 4.3 Clathrate Formation
+
+During the cooling of an impact plume, clathrates can form with entrapped mixtures different from the coexisting gases. In this case, the entrapped mixture will be enriched in H₂S and SO₂, and depleted in CO compared to the initial mixture because H₂S and SO₂ have a higher propensity for trapping compared to CO at low pressure conditions (16). If insufficient water is available to trap all of the CO, H₂S and SO₂ present in the gas, C/S in the clathrates is lower than in the source. Ammonia is not trapped in clathrates, but would form ammonia hydrates at temperatures between 80 and 100 K, or condense as pure ammonia frost at temperatures below 80 K. If not all of the CO is trapped, but all of the NH₃ ends up in the PSR, the N/C observed by LCROSS is an upper limit compared to the source. In this case either comets or chondrites could agree with the C/S and N/C. However, based on the water to CO ratio in clathrates the O/C ratio for volatiles trapped in clathrates must be lower than 6.75 if not enough water was available for all of the CO to be trapped (15). The LCROSS O/C disagrees with this limit, so additional water must be supplied by the solar wind. We again modeled a combination of sources assuming that the LCROSS C/S and O/C are lower limits based on clathrate formation processes and escape, that C/H is an upper limit based on escape, and ignoring N/C because of the competing influences of clathrate formation and escape. We found that a combination of cometary and solar wind sources fits these constraints, but that the modeled O/C is too high to support clathrate formation even accounting for a solar wind water source. Therefore, it is unlikely that ices that formed as clathrates can explain the LCROSS observations.
+
+## 4.4 Sublimation and Recondensation
+
+Finally, we consider how a cycle of sublimation and recondensation of volatiles could fractionate the elemental ratios. As volatiles are transported to the PSR, they could condense to the surface at night and sublimate during the day. A similar cycle could also take place within a PSR if diurnal temperatures vary enough to cause sublimation of some species depending on their volatility. The temperatures in the Cabeus PSR are very low and not likely to cause diurnal variations, but volatiles in this PSR could have been influenced by these processes before being trapped. Additionally, recondensation could occur within the Cabeus PSR when volatiles are released through impact gardening. This cycle would increase the abundance of water relative to other species observed in the LCROSS plume that have lower volatility temperatures (5). It would also increase the abundance of NH₃, H₂S, and SO₂ relative to CO and N₂. This means that C/S, N/C, and C/H in the PSR are lower limits compared to the source, while O/C is an upper limit. We modeled the source contributions with these constraints and found that a combination of comets, chondrites, and solar wind was possible. To narrow the possibilities further we add four more constraints shown in Table 1. Including these constraints limits the possible combination of source volatiles to 30–45% cometary and 55–70% chondrites with no solar wind contributions. Finally, we consider the combination of loss to space and a cycle of sublimation and recondensation. In this case the only reliable constraints are C/S, N/C, N/S and O/H. These constraints allow for any combination of comets and chondrites with no water provided by solar wind.
+
+| Model constraints and results for determining the possible sources for the LCROSS plume based on understanding of fractionation processes. |
+|---|
+| <table border="1" float="Yes" id="Tab1"> |
+| <caption language="En"> |
+| <div class="CaptionNumber"> |
+| Table 1 |
+| </div> |
+| <div class="CaptionContent"> |
+| <p> |
+| Model constraints and results for determining the possible sources for the LCROSS plume based on understanding of fractionation processes. |
+| </p> |
+| </div> |
+| </caption> |
+| <colgroup cols="7"> |
+| <thead> |
+| <tr> |
+| <th align="left" colname="c1"> |
+| </th> |
+| <th align="left" colname="c2"> |
+| <p> |
+| No fractionation |
+| </p> |
+| </th> |
+| <th align="left" colname="c3"> |
+| <p> |
+| Volcanic atm. processes |
+| </p> |
+| </th> |
+| <th align="left" colname="c4"> |
+| <p> |
+| Impact and Escape |
+| </p> |
+| </th> |
+| <th align="left" colname="c5"> |
+| <p> |
+| Clathrate formation |
+| </p> |
+| </th> |
+| <th align="left" colname="c6"> |
+| <p> |
+| Sublimation and recondensation |
+| </p> |
+| </th> |
+| <th align="left" colname="c7"> |
+| <p> |
+| Escape, subl. &amp; recondensation |
+| </p> |
+| </th> |
+| </tr> |
+| </thead> |
+| <tbody> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| C/S |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c2" morerows="3" rowspan="4"> |
+| <p> |
+| Within errors of LCROSS obs. &amp; sources |
+| </p> |
+| </td> |
+| <td align="left" colname="c3"> |
+| <p> |
+| Upper limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c5"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| N/C |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c3"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c5"> |
+| <p> |
+| Un-constrained |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| O/C |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c3"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c5"> |
+| <p> |
+| &lt; 6.75 |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Upper limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Unconstrained |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| C/H |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c3"> |
+| <p> |
+| Upper limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Upper limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c5"> |
+| <p> |
+| &gt; 0.07 |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Unconstrained |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| N/S |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c2" morerows="3" rowspan="4"> |
+| <p> |
+| n/a |
+| </p> |
+| </td> |
+| <td align="left" colname="c3" morerows="3" rowspan="4"> |
+| <p> |
+| n/a |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c5" morerows="3" rowspan="4"> |
+| <p> |
+| n/a |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| S/O |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Upper limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Unconstrained |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| S/H |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Upper limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Unconstrained |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| O/H |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| Upper limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| Lower limit |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| Constrained by solar wind input |
+| </p> |
+| </td> |
+| </tr> |
+| <tr> |
+| <td align="left" colname="c1"> |
+| <p> |
+| <b> |
+| Results |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c2"> |
+| <p> |
+| <b> |
+| No good fit |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c3"> |
+| <p> |
+| <b> |
+| No good fit |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c4"> |
+| <p> |
+| <b> |
+| Comets &amp; Solar Wind |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c5"> |
+| <p> |
+| <b> |
+| No good fit |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c6"> |
+| <p> |
+| <b> |
+| 30–45% Comets |
+| </b> |
+| </p> |
+| <p> |
+| <b> |
+| 55–70% Chondrites |
+| </b> |
+| </p> |
+| </td> |
+| <td align="left" colname="c7"> |
+| <p> |
+| <b> |
+| Comets &amp; Chondrites |
+| </b> |
+| </p> |
+| </td> |
+| </tr> |
+| </tbody> |
+| </colgroup> |
+| </table> |
+
+# 5. Summary And Conclusions
+
+Because no combination of known sources is able to match the large abundances of both sulfur and nitrogen compared to carbon measured by LCROSS we had to consider fractionation of the elements between delivery of volatiles to the surface of the Moon and trapping in the PSRs. The large nitrogen abundance allows us to rule out a volcanic atmosphere as a source for any of the volatiles even accounting for fractionating process. The fractionation of the elemental ratios by loss of volatiles to space and a cycle of sublimation and recondensation allows for a combination of cometary and chondritic material for the volatiles observed by LCROSS. Recognizing that the refractory material in comets is likely chondritic in composition, comets alone are a reasonable source and are likely the primary source of these volatiles.
+
+Measuring the elemental composition and the isotope ratios of the five elements evaluated in this study as a function of depth within the Cabeus PSR would provide constraints on the relative contribution of the solar wind to the water sampled, as well as details about the impactors. Because the isotope ratios of each source differ enough to serve as a tracer of the source, mapping them with depth would allow us to map out the composition of impactors as a function of time. Additionally, noble gas abundances and their isotope ratios are extremely valuable for tracing the sources of volatiles delivered to the Moon (e.g., <em>34</em>). As humans prepare to return to the Moon (<em>35</em>), we have an unprecedented opportunity to make such measurements in Cabeus and other PSRs. It is essential that future lunar missions have a plan to characterize the elemental and isotopic composition of lunar volatiles as a function of depth as they are accessed prior to converting volatiles to resources needed for human exploration.
+
+# 6. Supplemental Material
+
+## 6.1 Reanalysis of LAMP Observations
+
+The LCROSS plume was observed 30–60 seconds after impact by the Lunar Reconnaissance Orbiter (LRO) Lyman Alpha Mapping Project (LAMP), which detected H<sub>2</sub> and CO (2, 4), and for the first 4 minutes after impact by the LCROSS spacecraft which measured the abundance of several additional species relative to water (1). The published abundances from LAMP did not include the observations that were made at the same time or covering the same region of space as the LCROSS measurements. To allow for direct comparison between the two sets of observations, we have reanalyzed the LAMP observations to find CO and H<sub>2</sub> abundances relative to water in the time period relevant to the LCROSS measurements.
+
+The previous analysis of the LAMP data primarily focused on the pulse of vapor that expanded out in a spherical shell from the impact. This shell crossed LAMP’s field of view in the period 30–60 s after impact, which was before LRO passed the impact site. LAMP’s field of view was still > 50 km away from the impact site. A second feature was detected as LAMP’s field of view passed above the impact site. This feature is more closely associated with the population of vapor detected by the LCROSS shepherding spacecraft. We used the model outlined in (4) to simulate that population of vapor for better comparison with abundances measured by the LCROSS Shepherding spacecraft. With this reanalysis, we find the sublimation rate of CO needed in the model to reproduce LAMP’s observed count rate when it encountered the impact site. The model assumes a sublimating source of CO with starting velocities consistent with a Maxwell-Boltzmann distribution at temperature of 200 K. We determine the model column density as LAMP would detect it looking across the impact site at an altitude spanning from 30–40 km above the surface. To reproduce the LAMP observations, a source rate of 0.1 kg/s was required.
+
+The particle dynamics are such that they disperse from the plume center quickly. Thus, we must make an assumption about the time profile of the source rate to relate LAMP measurements at 90 s after impact to the LCROSS observations continuing until 3 minutes after impact. We assume that the source rate is constant throughout the first 150 seconds, which likely leads to an upper limit. With these assumptions, then the LAMP observations are consistent with < 15 kg of CO released out of the ~ 5500 kg mass of the LCROSS ejecta, or < 5% relative to water observed by LCROSS.
+
+## 6.2 Results
+
+Table S1 provides the abundances of each species thought to be present in the plume or in the regolith based on the LCROSS observations using (1) for all species except for CO and H<sub>2</sub>, which were recalculated as described above in Materials and Methods. We provide two categories of abundances that depend on the type of ice present in the regolith. If we assume that the species measured in the plume were released all at once by the destabilization of clathrates in the regolith, then the plume composition would represent the regolith composition. This category is designated as “clathrates”. If we assume that all volatiles are condensed onto the surface individually, they their abundance in the plume is related to their volatility temperature (10). We determine this using the study conducted by (11) but correcting for the abundances of CO and H<sub>2</sub> that were calculated as part of this work. This category is designated as “condensates”.
+
+**Table S1.**
+
+Molecular abundance of the volatiles observed in the LCROSS impact plume and their predicted abundance in the regolith if the volatiles are stored as condensed material. The abundance in the LCROSS plume was measured by the LCROSS experiment (1) and LRO-LAMP (2). The LAMP observations have been reanalyzed to obtain abundances measured at the same time as the LCROSS abundances (This work). The regolith estimates are based on the analysis of each species volatility temperature conducted by (11) correcting CO and H<sub>2</sub> to agree with LAMP observations. Species in **bold** are used for the atmospheric escape analysis.
+
+|  | LCROSS plume (%) or “Clathrates” | Regolith (11) (%) or “Condensates” | Volatility temperature (K) (10) | Mass (amu) |
+|---|---|---|---|---|
+| **H<sub>2</sub>O** | 100±27.5 (1) | 5.60±1.54 | 100.8 | **18** |
+| **OH** | 0.030±0.001 (1) | n/a |  | **17** |
+| **H<sub>2</sub>** | 6.0±5.9 (This work) | 0.050±0.049 (This work) |  | **2** |
+| CH<sub>4</sub> | 0.65±0.41 (1) | 0.0030±0.0019 | 22.0 (36) | 16 |
+| C<sub>2</sub>H<sub>4</sub> | 3.12±2.46 (1) | 0.020±0.016 | 40.0 | 28 |
+| **CO** | 2.00±1.99 | 0.0800±0.0796 (This work) | 16.8 | **28** |
+| **CO<sub>2</sub>** | 2.17±1.38 (1) | 0.040±0.025 | 53.4 | **44** |
+| CH<sub>3</sub>OH | 1.55±7.92 (1) | 0.10±0.51 | 90.0 | 32 |
+| S | Not observed | Not observed | 181 (11) | 32 |
+| **H<sub>2</sub>S** | 16.75±2.82 (1) | 0.200±0.034 | 47.8 | **34** |
+| **SO<sub>2</sub>** | 3.19±0.08 (1) | 0.200±0.005 | 70.5 | **64** |
+| **OCS** | Not observed | Not observed | 46.8 (36) | **60** |
+| **NH<sub>3</sub>** | 6.03±1.26 (1) | 0.070±0.034 | 63 | **17** |
+| **N<sub>2</sub>** | Not observed | Not observed | 16.2 (36) | **28** |
+
+Table S2 applies the species composition from Table S1 to calculate the elemental abundances for the regolith ices if they are “clathrates” or “condensates”. Note that the greatest difference between the elemental abundances for the two types of ice are found in the nitrogen and oxygen relative to carbon. We also include the elemental abundances used for the sources shown in Fig. 1 based on the references indicated in the main article text and in the table references. Uncertainties are propagated from measured values reported in the literature to the elemental ratio uncertainties by standard means for error propagation. These elemental ratios and their uncertainties were used in the model described below for each of the cases outlined in Table 1 of the main text.
+
+**Table S2.**
+
+Elemental abundance of the volatiles in the top 1–3 meters of regolith in the Cabeus Crater PSR sampled by the LCROSS impact compared to possible sources for the volatiles. The LCROSS plume abundance assumes that the volatiles are stored as clathrates. The regolith abundance assumes that the volatiles are stored as condensates and that each species is released according to its volatility.
+
+|  | LCROSS plume (1) or “Clathrates” | Regolith (11) or “Condensates” | 67P/C-G (20) | Average comets (20) | CI Chondrites (38, 39) | CM Chondrites (38) | Volcanoes (3, 25) |
+|---|---|---|---|---|---|---|---|
+| C/S | 0.63±0.36 | 0.66±0.38 | 5.35±0.16 | ≥ 2.16 | 0.68±0.02 | 0.59±0.02 | 0.92±0.47 |
+| N/C | 0.48±0.24 | 0.27±0.14 | 0.10±0.05 | ≤ 0.37 | 0.085±0.003 | n/a | 0.006±0.001 |
+| O/C | 9.37±4.36 | 23.7±11.0 | 12.06±4.06 | 50.0±49.7 | 12.58±0.38 | 20.80±0.62 | 1.02±0.01 |
+| C/H | 0.04±0.03 | 0.02±0.01 | 0.05±0.02 | 0.01±0.0099 | 2.35±0.07 | 1.69±0.05 | 17.04±5.14 |
+
+## 6.2 Model Description
+
+We modeled the possible mixtures of sources by calculating C/S, N/C, O/C, and C/H for every combination of percent contribution of volcanic, chondrite, cometary, and solar wind contributions that add up to 100%. All elemental ratios used as input to the model are shown in Table S2 with the exception of solar wind contribution. Each time that we applied the model, we tested 167,002 possible combinations of these sources. In all cases, the solar wind was assumed to contribute only hydrogen and oxygen to the ratios by delivering water to the PSR. This is because the solar wind nitrogen and sulfur are orders of magnitude lower than solar wind hydrogen and any contribution would be negligible compared to other potential volatile sources as seen in the right panel of Fig. 1. The other sources would deliver all four elements according to their abundances in the source.
+
+The model was applied five times using the constraints outlined in Table 1. We did not apply the model for a volcanic atmosphere because we were only evaluating one source in this case. A volcanic source was allowed in all five cases and no models with volcanic contributions could fit in any of these cases because of the deficiency in nitrogen in volcanic gas. When fitting within the uncertainties, a modeled mixture of sources was determined to be an accurate fit if the simulated elemental ratios and uncertainties overlapped with the observations by LCROSS and the uncertainties in the measurements.
+
+# References
+
+1. A. Colaprete et al., Detection of water in the LCROSS ejecta plume. *Science* **330**, 463–468 (2010). doi:10.1126/science.1186986
+2. G. R. Gladstone et al., LRO-LAMP observations of the LCROSS impact plume. *Science* **330**, 472–476 (2010). doi:10.1126/science.1186474
+3. D. H. Needham, D. A. Kring, Lunar volcanism produced a transient atmosphere around the ancient Moon. *Earth and Planetary Science Letters* **478**, 175–178 (2017). doi:10.1016/j.epsl.2017.09.002
+4. D. M., Hurley et al., Modeling of the vapor release from the LCROSS impact: 2. Observations from LAMP. *Journal of Geophysical Research: Planets* **117**, (2012). doi:10.1029/2011JE003841
+5. See supplemental materials Table S1
+6. E. A. Fisher et al., Evidence for surface water ice in the lunar polar regions using reflectance measurements from the Lunar Orbiter Laser Altimeter and temperature measurements from the Diviner Lunar Radiometer Experiment. *Icarus* **292**, 74–85 (2017). doi:10.1016/j.icarus.2017.03.023
+7. S. Li et al., Direct evidence of surface exposed water ice in the lunar polar regions. *Proceedings of the National Academy of Sciences* **115**, 8907–8912 (2018). doi:10.1073/pnas.1802345115
+8. G. W. Patterson et al., Bistatic radar observations of the Moon using Mini-RF on LRO and the Arecibo Observatory. *Icarus* **283**, 2–19 (2017). doi:10.1016/j.icarus.2016.05.017
+9. A. B. Sanin et al., Testing lunar permanently shadowed regions for water ice: LEND results from LRO. *Journal of Geophysical Research: Planets* **117**, (2012). doi:10.1029/2011JE003971
+10. Volatility temperature is defined as the temperature at which pure solid evaporates from the surface to vacuum at a rate of 1 mm/billion years assuming a bulk density of 1 g/cm³ (36) as calculated by (11) using (37)
+11. A. A. Berezhnoy, E. A. Kozlova, M. P. Sinitsyn, A. A. Shangaraev, V. V. Shevchenko, Origin and stability of lunar polar volatiles. *Advances in space research* **50**, 1638–1646 (2012). doi:10.1016/j.asr.2012.03.019
+12. D. A. Paige et al., Diviner lunar radiometer observations of cold traps in the Moon’s south polar region. *Science* **330**, 479–482 (2010). doi:10.1126/science.1197135
+13. R. Woods‐Robinson, M. A. Siegler, D. A. Paige. A model for the thermophysical properties of lunar regolith at low temperatures. *Journal of Geophysical Research: Planets* **124**, 1989–2011 (2019). doi: 10.1029/2019JE005955
+14. P. O. Hayne, B. T. Greenhagen, M. C. Foote, M. A. Siegler, A. R. Vasavada, D. A. Paige, Diviner lunar radiometer observations of the LCROSS impact. *Science*, **330**, 477–479 (2010). doi:10.1126/science.1197135
+15. See supplemental materials Table S2
+16. J. I. Lunine, D. J. Stevenson, Thermodynamics of clathrate hydrate at low and high pressures with application to the outer solar system. *Astrophysical Journal Supplement Series*, **58** (3), 493–531 (1985). doi:10.1086/191050
+17. F. Hersant, D. Gautier, J. I. Lunine, Enrichment in volatiles in the giant planets of the Solar System. *Planetary and Space science*, **52** (7), 623–641 (2004). doi:10.1016/j.pss.2003.12.011
+18. A. H. Mohammadi, D. Richon, Phase equilibria of methane hydrates in the presence of methanol and/or ethylene glycol aqueous solutions. *Industrial & engineering chemistry research*, **49** (2), 925–928 (2010). doi:10.1021/ie901357m
+19. J. K. Mitchell, W. N. Houston, W. D. Carrier III, N. C. Costes, “Apollo soil mechanics experiment S-200.” (NASA Space Sciences Laboratory Series 15, Issue 7, 1974).
+20. M. Rubin et al., Elemental and molecular abundances in comet 67P/Churyumov-Gerasimenko. *Monthly Notices of the Royal Astronomical Society* **489**, 594–607 (2019). doi:10.1093/mnras/stz2086
+21. A. N. Deutsch, J. W. Head III, G. A. Neumann, Analyzing the ages of south polar craters on the Moon: Implications for the sources and evolution of surface water ice. *Icarus* **336**, 113455 (2020). doi:10.1029/2020GL087782
+22. D. M. Hurley et al., Two‐dimensional distribution of volatiles in the lunar regolith from space weathering simulations. *Geophysical Research Letters* **39** (2012). doi:10.1029/2012GL051105
+23. M. D. Hopkins, S. J. Mojzsis, A protracted timeline for lunar bombardment from mineral chemistry, Ti thermometry and U–Pb geochronology of Apollo 14 melt breccia zircons. *Contributions to Mineralogy and Petrology* **169**, 30 (2015). doi:10.1007/s00410-015-1123-x
+24. M. Benna, D. M. Hurley, T. J. Stubbs, P. R. Mahaffy, R. C. Elphic, Lunar soil hydration constrained by exospheric water liberated by meteoroid impacts. *Nature Geoscience* **12**, 333–338 (2019). doi:10.1038/s41561-019-0345-3
+25. B. Fegley, Thermodynamic models of the chemistry of lunar volcanic gases. *Geophysical Research Letters* **18**, 2073–2076 (1991). doi:10.1029/91GL02624
+26. I. Aleinov et al., Modeling a transient secondary paleolunar atmosphere: 3‐D simulations and analysis. *Geophysical Research Letters* **46**, 5107–5116 (2019). doi:10.1029/2019GL082494
+27. O. J. Tucker, R. M. Killen, R. E. Johnson, P. Saxena. Lifetime of a transient atmosphere produced by lunar volcanism. *Icarus* **359**, 114304 (2021). doi:10.1016/j.icarus.2021.114304
+28. D. Hurley Crider, R. R. Vondrak, Space weathering effects on lunar cold trap deposits. *Journal of Geophysical Research: Planets* **108** (2003). doi:10.1029/2002JE002030
+29. L. Ong, E. I. Asphaug, D. Korycansky, R. F. Coker. Volatile retention from cometary impacts on the Moon. *Icarus* **207**, 578–589 (2010). doi:10.1016/j.icarus.2009.12.012
+30. K. E. Mandt, O. Mousis, J. Lunine, D. Gautier. Protosolar ammonia as the unique source of Titan's nitrogen. *The Astrophysical Journal Letters* **788**, L24 (2014).
+31. P. Prem, N. A. Artemieva, D. B. Goldstein, P. L. Varghese, L. M. Trafton. Transport of water in a transient impact-generated lunar atmosphere. *Icarus*, **255**, 148–158 (2015). doi: 10.1016/j.icarus.2014.10.017
+32. R. Ishimaru, H. Senshu, S. Sugita, T. Matsui. A hydrocode calculation coupled with reaction kinetics of carbon compounds within an impact vapor plume and its implications for cometary impacts on Galilean satellites. *Icarus* **210**, 411–423 (2010). doi:10.1016/j.icarus.2010.06.016
+33. H. Kuwahara, S. Sugita. The molecular composition of impact-generated atmospheres on terrestrial planets during the post-accretion stage. *Icarus* **257**, 290–301 (2015). doi:10.1016/j.icarus.2015.05.007
+34. K. E. Mandt et al., Tracing the origins of the ice giants through noble gas isotopic composition. *Space Science Reviews* **216**, 1–37 (2020). doi:10.1007/s11214-020-00723-5
+35. Artemis III Science Definition Team Report (2020). https://www.nasa.gov/reports
+36. J. A. Zhang, D. A. Paige, Cold‐trapped organic compounds at the poles of the Moon and Mercury: Implications for origins. *Geophysical Research Letters* **36** (2009). doi:10.1029/2009GL038614
+37. Fray & Schmidt.
+38. C. M. D. Alexander, Quantitative models for the elemental and isotopic fractionations in chondrites: The carbonaceous chondrites. *Geochimica et Cosmochimica Acta*, **254**, 277–309 (2019). doi:10.1016/j.gca.2019.02.008
+39. Lodders, K., Solar system abundances of the elements. In *Principles and perspectives in cosmochemistry* (pp. 379–417). Springer, Berlin, Heidelberg (2010). doi: 10.1007/978-3-642-10352-0_8

098ddd9e6616a135d268798217d55aa269660b9e1b01881d2c24cabbd4ecf5b0/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "(a) Chemical structure of AOPIM-1 polymer and optical image of AOPIM-1 membrane; (b) The pH-tunable chargeability of AOPIM-1 under acidic or alkaline conditions, where the gray, white and red spheres represent C atoms, H atoms and O atoms, respectively; (c) Schematic diagram of microporous polymer membranes for adsorptive separation of organic molecules; (d) Three-dimensional view of an amorphous cell of the AOPIM-1 polymer. The brown surface indicates the van der Waals surface, and the orange surface is the Connolly surface with a probe radius of 1.6\u2009\u00c5; Cross-section SEM images of the AOPIM-1 membranes obtained by phase conversion of different coagulation bath compositions: (e) H2O:EtOH = 100:0 and (f) H2O:EtOH = 0:100.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "pH-tunable static adsorption feature of AOPIM-1 polymer. (a) Nitrogen absorption isotherms of AOPIM-1 as-prepared and after immersing in acid (pH=3) or base (pH=10) for at least 24 h; (b) Equilibrium adsorption capacity (qe) of AOPIM-1 for MO and MB dye molecules at different pH conditions; (c) Adsorption isotherm of MO and MB by AOPIM-1 at acidic and alkaline conditions, respectively (qe: equilibrium adsorption capacity, qm: maximum adsorption capacity, Ce: concentration of dye molecules at equilibrium); (d) Adsorption of MO and MB by AOPIM-1 as a function of contact time at acidic and alkaline conditions, respectively (qt: adsorption amount at time t). ",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Charge-selective dynamic adsorption of organic dye molecules by AOPIM-1 membranes. (a) Zeta potential of AOPIM-1 membrane at pH 3-10; (b) The dynamic separation performance of AOPIM-1 membrane at different pH condition for different dyes (concentration 20 ppm, applied pressure 0.2 MPa, pH = 3/10); (c, d) The separation (MO/MB) of two dye molecules with similar molecular weights but different charges under different pH condition (concentration 10 ppm MO + 10 ppm MB, applied pressure 0.2 MPa, pH = 3/10); (e) Comparison of AOPIM-1 membranes with reported adsorptive membranes with respect to their permeation flux and adsorptive capacity; (f) Comparison of AOPIM-1 membranes with reported membranes with respect to their permeation flux and dye rejection/removal efficiency.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.jpg",
+        "caption": "Adsorptive separation behavior of AOPIM-1 membrane in a multi-cycle dynamic adsorption-desorption process. (a) Flux and RHB removal rate of the membrane in multiple adsorption-desorption cycles (concentration: 20 ppm, applied pressure: 0.2 MPa, pH = 10); (b) Pore size distributions of pristine, adsorbed and desorbed AOPIM-1 membrane, respectively; (c) Nitrogen absorption-desorption isotherms of pristine, adsorbed and desorbed AOPIM-1, respectively; (d) optical pictures of the pristine, adsorbed and cleaned AOPIM-1 membrane coupon.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.jpg",
+        "caption": "Adsorptive separation of active pharmaceutical ingredients (APIs) by AOPIM-1 membranes. (a) Schematic illustration of the 2-step process for adsorptive separation of mixed API/polysaccharide/salt feed using AOPIM-1 membranes; (b) The adsorption/rejection ratio of API/polysaccharide/NaCl versus permeation volume of a synthetic water extract feed solution. ",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

098ddd9e6616a135d268798217d55aa269660b9e1b01881d2c24cabbd4ecf5b0/preprint/preprint.md

@@ -0,0 +1,188 @@
+# Abstract
+
+Trade-off between permeability and nanometer-level selectivity is an inherent shortcoming of membrane-based separation of molecules, while most highly porous materials with high adsorption capacity lack solution processability and stability for achieving adsorption-based molecule separation. We hereby report a hydrophilic amidoxime modified polymer of intrinsic microporosity (AOPIM-1) as a membrane adsorption material to selectively adsorb and separate small organic molecules from water with ultrahigh processing capacity. The membrane adsorption capacity for Rhodamine B reaches 26.114 g m⁻², 10~1000 times higher than previously reported adsorptive membranes. Meanwhile, the membrane achieves >99.9% removal of various nano-sized organic molecules with water flux 2 orders of magnitude higher than typical pressure-driven membranes of similar rejections. This work confirms the feasibility of microporous polymers for membrane adsorption with unprecedented capacity, and provides the possibility of adsorptive membranes for molecular separation.
+
+# Introduction
+
+Molecular separation is an essential component in many human daily activities and multiple industrial, medical, and environmental processes, such as water purification, oil and gas refining, energy generation and storage, pharmaceutical ingredient extraction and purification<sup>1–5</sup>. Among various materials and methods, synthetic membranes and membrane-based separation have been extensively studied due to its strong sustainability, scale-up feasibility, and no phase change during separation<sup>1,6</sup>. However, trade-off between membrane permeability and selectivity is an inherent shortcoming of membrane-based molecular separation<sup>7–8</sup>. Taking pressure-driven membrane processes for instance, microfiltration and ultrafiltration membranes exhibit high water permeation flux and reject large molecules with high efficiency, but their large pores (0.002–1 µm) are unable to separate nanometer-sized small organic molecules<sup>9</sup>. Processing of these small organic molecules requires membranes of nanofiltration level pore size and selectivity, yet the reduction in pore size inevitably results in much lower water permeation flux<sup>10</sup>. In contrast, adsorption can be a highly selective molecular separation process with specific physical or chemical interactions between adsorbents and target molecules<sup>11–14</sup>, but its application is often limited by its low processing rate, high internal diffusion resistance within absorbents, etc<sup>15–16</sup>. Membrane adsorption is a pressure-driven dynamic membrane-based adsorption process, which combines the merit of both adsorption and membrane separation<sup>17–19</sup>. Compared to traditional pressure-driven membranes with size sieving as the dominant separation mechanism, adsorptive membranes utilize more specific membrane-solute interactions like electrostatic interactions, π-π interactions, van der Waals forces, and hydrogen bonding to achieve highly selective and fast separation of small organic molecules<sup>20–21</sup>. Therefore, they are expected to break through the permeability-selectivity trade-off by simultaneously achieving the selectivity of dense membranes and permeability of porous membranes<sup>18,22–25</sup>.
+
+However, the application of membrane adsorption is hindered by the lack of adsorptive membranes with sufficient processing capacity. Adsorptive membranes are usually prepared by post-grafting affinity ligands or adsorptive filler mixing in traditional polymer materials<sup>15,20</sup>. Due to the limited specific surface area and adsorption sites incorporated, the overall processing capacity of these conventional adsorptive membranes is very low, mostly in the range of 0.01–1 g m<sup>−2</sup>. Such membranes could only handle solutions of very low concentration and require frequent cleaning and regeneration, which greatly affects their further development and practical application. Although porous materials such as metal-organic frameworks (MOFs) possess large porosities and rich affinity sites<sup>18–19,26–30</sup>, their poor solution processability hinders their engagement as absorptive membrane separation materials. Besides, most MOFs materials do not have sufficient stability in water, especially under acidic or alkaline conditions, resulting in membrane failure during long-term operation<sup>8</sup>.
+
+In this work, we report a microporous polymer based adsorptive separation membrane, which has extremely high adsorption capacity and can realize fast and selective molecular separation. The microporous polymer is amidoxime modified polymer of intrinsic microporosity (AOPIM-1) polymer, which owns a rigid and contorted three-dimensional structure in its backbone. The ineffective chain packing brings high specific surface area up to 550 m<sup>2</sup> g<sup>−1</sup> and produces interconnected free volumes with a size of less than 2 nm<sup>31–34</sup>. The amidoxime modification endows the polymer with good solution processability and provides abundant adsorption sites for selectively adsorbing charged molecules. Due to its unique chemical and physical structure, it achieves a high-efficiency removal of small organic molecules (>99.9%) with permeating flux 2 orders of magnitude higher than typical nanofiltration membranes of similar dye rejections, and such separation performance is maintained throughout multiple adsorption-elution cycles with >98% flux recovery rate (FRR). More importantly, the static adsorption capacity of the membrane material greatly surpasses traditional non-porous polymer adsorbents and comparable to MOF-based adsorbents, while the dynamic processing capacity reaches 26.114 g m<sup>−2</sup>, much higher than all the adsorptive membranes reported so far. The unprecedented processing capacity is expected to enable the membrane towards more realistic adsorption-separation application scenarios.
+
+# Results And Discussion
+
+## Material characterization of AOPIM-1 polymer
+
+Polymer of intrinsic microporosity (PIM-1) was synthesized via a polycondensation reaction between TTSBI and TFTPN (Figure S1) as reported previously<sup>35</sup>, which shows a high specific surface area (786 m<sup>2</sup> g<sup>−1</sup>). The prepared PIM-1 was further modified with hydroxylamine to obtain AOPIM-1 (Figure <span class="InternalRef">1</span> a, Figure S2)<sup>31,34</sup>. Spectroscopic characterizations of AOPIM-1 are provided in Figure S3-4, which signify the successful synthesis of the polymer. Figure <span class="InternalRef">1</span> d shows a three-dimensional view of a modeled amorphous cell of the AOPIM-1 polymer, which presents a highly microporous feature. Its specific surface area reaches 552.3 m<sup>2</sup> g<sup>−1</sup> as deduced from its N<sub>2</sub> adsorption isotherm (Figure <span class="InternalRef">2</span> a). Moreover, the amidoxime modified polymer changes its solubility parameter due to the introduction of polar groups, making it soluble in common casting solvents, such as DMF, NMP, DMSO, etc (Figure S5). Unlike PIM-1 that could only be dissolved in chloroform and tetrahydrofuran, the great solution processability of AOPIM-1 makes it feasible for the fabrication of asymmetric membranes via the industrially scalable phase inversion approach (Figure <span class="InternalRef">1</span> a). In addition, stability of AOPIM-1 under acidic and alkaline conditions are examined by immersing in acid (pH = 3) and base (pH = 10) for at least 24 h, and the chemical and physical structure of the polymer is found to remain unchanged (Figure <span class="InternalRef">2</span> a and S4). Interestingly, the AOPIM-1 polymer shows a pH-tunable chargeability (Figure <span class="InternalRef">1</span> b). Such tunable chargeability is attributed to the protonation/deprotonation of amidoxime groups, and they could thus be utilized as effective affinity sites with pH-tunability for selective adsorption and separation of oppositely charged molecules (Figure <span class="InternalRef">1</span> c). The processability, hydrophilicity, high specific surface area and pH-tunable affinity sites make AOPIM-1 a promising membrane adsorption material for aqueous-based molecular separations, which will be demonstrated in the following sections.
+
+## pH-tunable static adsorption feature of AOPIM-1 polymer
+
+Two organic dye molecules with similar molecular weight but different charge properties, Methyl Orange (MO, MW = 327, negatively charged) and Methylene Blue (MB, MW = 320, positively charged) are used to evaluate the static adsorption behavior of AOPIM-1. As shown in Figure <span class="InternalRef">2</span> b, the amidoxime modified polymer favors the capture of negatively charged MO molecules under acidic conditions while capturing the oppositely charged MB under alkaline conditions, exhibiting a pH-tunable adsorption feature. The adsorption capacity of the polymer is found to be as high as 445.02 mg g<sup>−1</sup> (MO, pH = 3.3) and 735.75 mg g<sup>−1</sup> (MB, pH = 10.9), respectively, owing to the combination of high specific surface area and pH-tunable chargeability of amidoxime groups within the polymer. And lower equilibrium adsorption capacity is observed under neutral pH conditions where the amidoxime possess minimal chargeability, which also illustrates that the pH-tunable affinity sites make indispensable contribution to the adsorption capacity. As shown in Figure S7, the mixture of MB and MO is green with some precipitation due to the opposite charges of MB and MO molecules<sup>14</sup>. After AOPIM-1 treatment under different pH conditions, one of the colors is completely removed, and the precipitation disappears. The UV-Vis spectra clearly show the disappearance of one of the absorption peaks. The result proves that AOPIM-1 achieves the selective adsorption of MB and MO mixture by adjusting pH environment. In Figure <span class="InternalRef">2</span> c, d, Langmuir adsorption isotherms and pseudo-second-order equation (R<sup>2</sup> > 0.99) are used to fit the adsorption of AOPIM-1 on target molecules in different pH environments. Under acidic and alkaline conditions, the maximum adsorption capacity for MO and MB reaches 491.63 mg g<sup>−1</sup> and 765.09 mg g<sup>−1</sup>, respectively. AOPIM-1 shows a faster adsorption rate, exceeding 50% in 40 min, and reaching equilibrium in 300 min. The obtained adsorption capacity of AOPIM-1 is much higher than traditional non-porous polymer adsorbents and comparable to that of newly developed MOF adsorbents (Table S1)<sup>18,30,36−38</sup>.
+
+## Adsorption separation of dye molecules by AOPIM-1 membranes
+
+The adsorptive membranes based on AOPIM-1 are constructed via a wet-phase inversion method. The membrane structure is altered through tuning phase inversion conditions including various coagulation bath composition, casting solution concentration, and casting solution composition as summarized in Table S2. With increasing ethanol content in coagulation bath, it could be observed that the structure of the membrane changes from finger-like pores (M1) to sponge-like pores (M3) as demonstrated in the cross-sectional SEM images in Figure <span class="InternalRef">1</span> e and <span class="InternalRef">1</span> f. Meanwhile, the pure water flux of resulted membrane decreases from 1505.7 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup> (M1) to 249.8 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup> (M3). However, the processibility capacity of M3 increases from 3.7 to 11.4 g m<sup>−2</sup> which is 3 times more than that of M1 (Table S2, Figure S8-10). It should be mentioned that the effective processing capacity of membranes is calibrated under the criterion of 99% rejection ratio of Rhodamine B (RHB) (20 ppm). With increase of the concentration of the casting solution, it could be observed that the membrane thickness (M3-M5 in Table S2) is gradually increased. The increasing in thickness and processing capacity has a linear relationship from M3 11.4 g m<sup>−2</sup> (64 µm) to M5 26.1 g m<sup>−2</sup> (119 µm). Figure <span class="InternalRef">3</span> e and Table S3 present the comparison of AOPIM-1 membranes with reported adsorptive membranes with respect to their permeation flux and adsorptive capacity, and the dynamic adsorption capacity of the membrane appears to be 10~1000 times higher than previously reported adsorptive membranes. The above experiment results prove that by adjusting the phase inversion process, the membrane structure can be reasonably designed to achieve the purpose of rapidly removing small molecular organic pollutants in aqueous system using dynamic adsorption processes.
+
+The dynamic adsorption separation process was conducted on a dead-end setup using a sponge-like AOPIM-1 membrane (M3). According to zeta potential measurement results, AOPIM-1 membranes exhibit positive charge at acidic pH < 4 while turning into negative charge at alkaline aqueous environment (Figure <span class="InternalRef">3</span> a). Thus, six types of organic dyes with different sizes and charges were used as target molecules for separation. Negatively charged dyes including Methyl Orange (MO), Congo Red (CR) and (BB) were tested in acidic condition and positive charged dyes including Methyl Blue (MB), RHB and Crystal Violet (CV) (Figure S11) was tested in alkaline condition. It can be seen from Figure <span class="InternalRef">3</span> b and Table S4 that under acidic conditions (pH=3), MO (99.9% rejection, 203.8 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup>), CR (99.9% rejection, 180.9 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup>) and BB (99.9% rejection, 192.5 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup>) can be selectively retained. Under alkaline conditions (pH = 10), MB (99.9% rejection, 177.9 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup>), RHB (99.9% rejection, 191.3 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup>) and CV (99.9% rejection, 170.0 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup>) can be retained, indicating highly efficient separation of dye molecules with different charges. The dynamic separation of two dye molecules mixture (MO/MB) is further demonstrated (Figure <span class="InternalRef">3</span> c-d). Under different pH conditions, one of the colors of the mixed solution disappears while the other color remains after filtering. The UV-Vis spectra clearly show the disappearance of one of the absorption peaks. The results prove that the AOPIM-1 membrane has the characteristic of charge selective dynamic adsorption of dye molecules. We summarized the state-of-the-art nanofiltration membranes and other membrane adsorption materials reported in the literature which are utilized for separating small organic molecules in aqueous systems (Figure <span class="InternalRef">3</span> f), and the current AOPIM-1 membranes possess 2 orders of magnitude higher flux than typical nanofiltration membranes of similar dye rejections, and surpass other reported adsorptive separation membranes in terms of water flux and dye rejection.
+
+The adsorptive separation behavior of the AOPIM-1 membrane is further demonstrated in a multi-cycle dynamic membrane adsorption process. It can be seen from Figure <span class="InternalRef">4</span> a that the flux of the membrane decreases with the increasing permeation volume in each cycle, which can be attributed to the accumulation of dyes on the surface and inside of the adsorptive membrane. In the first seven cycles, 20-30 mL of methanol was adopted as a feed solution to desorb the dye molecules from the membrane for a short time (within 5 minutes). It can be seen in digital photo in Figure <span class="InternalRef">4</span> d that the color of the membrane due to the adsorption of dye disappears after cleaning. At the same time, the flux recovery rate (FRR) in each cycle reaches 98%. After 7 cycles, a long-term desorption treatment (about 2 hours by methanol) on the membrane was performed, and a higher recovery rate is got. The Brunauer-Emmet-Teller (BET) surface area of the AOPIM-1 adsorptive membrane is reduced from 552.3 m<sup>2</sup> g<sup>−1</sup> to 415.4 m<sup>2</sup> g<sup>−1</sup> after the adsorption test, and it can be easily regenerated to the original level after desorption (Figure <span class="InternalRef">4</span> b). Meanwhile, it can be seen in Figure <span class="InternalRef">4</span> c, the pore size distribution of AOPIM-1 is also restored to the original level after cleaning. As a control group, a negatively charged polyethersulfone ultrafiltration membrane prepare by phase inversion was selected to test positively charged organic small molecules. It is found that almost all dye molecules pass through the membrane without noticeable retention (Figure S12), which is common for conventional ultrafiltration membranes that rejects large molecules and particulate matter (such as proteins, suspended solids, bacteria, viruses, and colloids) but cannot accurately separate small organic molecules<sup>10</sup>. Without necessary specific surface area and affinity sites for effective adsorption, membranes of conventional polymer materials require further reduced the pore size for separation of small molecules, but it will inevitably lead to a significant reduction in water flux and huge energy consumption. The above results reveal that the electrostatic attraction, micropore and membrane pore structure co-govern the dynamic adsorption performance of the membrane.
+
+## Adsorptive separation of active pharmaceutical ingredients (APIs)
+
+In the pharmaceutical industry, precise separation of organic molecules such as drugs, proteins, and polysaccharides are indispensable for the production of active pharmaceutical ingredients (APIs)<sup>5,7,56</sup>. For instance, the raw material water extracts of phytochemical drugs, an important category of APIs extracted from natural plants<sup>45</sup>, usually possess a complex composition including polysaccharides (molecular weights usually range from 10,000 to 100, 000 Da), APIs (molecular weight < 2000 Da), and inorganic salts such as sodium chloride. The feasibility of membrane adsorption for the separation of such 3-component systems was evaluated in this section. As schematically illustrated in Figure <span class="InternalRef">5</span> a and Figure S13, a 2-step process is proposed for adsorptive separation of a phytochemical drug water extract using AOPIM-1 membranes. In the first step, the water extract of natural plants passes through the surface of the membrane, the plant polysaccharides with large molecular weight are trapped on the feed side of the membrane, the active pharmaceutical molecules with small molecular weight are adsorbed within the membrane matrix, and inorganic salts are permeated through the membrane. After the membrane is saturated with API adsorption, the second step uses methanol as the eluent to flow in from the feed side, and the API eluate is obtained on the permeate side.
+
+As a proof of concept, an AOPIM-1 membrane with altered casting conditions (3:1 DMF/1,4-dioxane co-solvent<sup>47–48</sup>) and tightened membrane pores (water flux of ~121.3 L m<sup>−2</sup> h<sup>−1</sup> bar<sup>−1</sup>, MWCO of ~20 kDa) are prepared, and the MWCO curve of M6 and the pore size distribution are shown in Figure S14. A synthetic water extract (Table S5) is prepared with Dextran T-200 (MW = 20 kDa) used as the target macromolecular polysaccharide, berberine (MW = 336.4 Da) as the target small API molecule, and sodium chloride as the inorganic salt. As can be seen in Figure <span class="InternalRef">5</span> b, during a cross-flow filtration cycle, berberine is continuously adsorbed in the membrane, and the rejection rate of Dextran T-200 is maintained above 90%. At the same time, sodium chloride permeates through the membrane without rejection. At the end of one filtration cycle, the feed solution is changed to methanol to elute the berberine enriched in the membrane. Figure S15 clearly shows the UV-Vis absorption peaks in the stock solution and filtrate before and after processing 50 mL of the synthetic feed. After the filtration, the characteristic absorption peak of berberine disappeared, which proves that the berberine in the feed solution are completely adsorbed in the membrane. At the same time, it can be seen in the illustration that 50 mL of the feed solution is enriched in the membrane and finally eluted by about 5 mL of methanol, achieving a 10-fold enrichment. Compared with other separation methods, the traditional distillation method is energy-intensive and time-consuming, whereas the membrane filtration method is difficult to achieve sufficient accuracy and efficiency, and the obtained permeate usually requires further purification operations afterwards<sup>57</sup>. The membrane adsorption separation method shows the merit of smallest energy consumption and highest product purity. More importantly, this experiment demonstrates that the ultrahigh processing capacity of AOPIM-1 adsorptive membranes omits the need of frequent cleaning and regeneration, effectively broaden the application feasibility of membrane adsorption from treating trace organic compounds towards mass chemical productions.
+
+# Conclusions
+
+This work has developed a new type of pH-tunable high-capacity adsorptive membranes based on AOPIM-1, in which the high specific surface area (high adsorption capacity), abundant adsorption sites (adsorption selectivity), reversible adsorbate-adsorbent interaction (fast adsorption/desorption rates), good solubility processability (scale-up feasibility) and hydrophilicity (easy-wet micropores) of AOPIM-1 are fully utilized. The processing capacity and permeability of the membrane is adjusted by manipulating the phase inversion process. While achieving the retention of small molecules, the membrane permeation flux is in line with the level of common ultrafiltration, which is 2 orders of magnitude higher than typical nanofiltration membranes of similar rejections. The best processing capacity reaches 26.114 g m⁻², which is 10~1000 times higher than the value reported in the literature for existing adsorptive membranes. This newly developed membrane material realizes a dynamic operation process of membrane adsorption-desorption, cleaning, and regeneration with high efficiency, which can be a good supplement to conventional pressure-driven membrane separation processes. Owing to the unprecedented adsorption capacity, the AOPIM-1 membrane exhibits high separation efficiency of actual complex systems and highly guarantees the purity of the product obtained. The development of AOPIM-1 adsorptive membranes thereby broadens the prospects of membrane adsorption for practical applications.
+
+# Methods
+
+**Fabrication of PIM-1 and AOPIM-1 polymers.** Polymer of intrinsic microporosity (PIM-1) was obtained following a previously reported method. Under a nitrogen atmosphere, 3.001 g (15 mmol) tetra-fluoroterephthalonitrile (TFTPN), 5.106 g (15 mmol) 5,5’,6,6’-tetrahydroxy-3,3,3’,3’- tetramethylspirobisindane (TTSBI) and 30 mL anhydrous DMAc were added into a 100 mL three-necked flask. After the chemicals were completely dissolved, 6.21 g (45 mmol) anhydrous milled K₂CO₃ was added and the flask was placed into a 160°C oil bath under mechanical stirring. After approximately 3 min, a viscous yellow solution formed, and 20 mL of toluene was added. Several minutes later, a further 20 mL of toluene was added to dilute the solution. Then, the mixture was poured into 300 mL methanol, and an elastic, threadlike, light yellow polymer was observed. The polymer product was dissolved in chloroform and reprecipitated in methanol, and then refluxed in Milli-Q water for 4–5 h and dried at 80°C under vacuum for 48 h.
+
+Hydrophilic amidoxime modified PIM-1 (AOPIM-1) was synthesized by dissolving 0.5 g PIM-1 in 30 mL THF and heating to reflux under N₂. Then, 5.0 mL hydroxyl amine was added dropwise, and the solution was further refluxed for 20 hours. The resulting polymer was precipitated by the addition of ethanol, filtered, washed thoroughly with ethanol and water, and then dried at 110°C for 24 h.
+
+**Fabrication of AOPIM-1 membranes.** The polymer dope was obtained by dissolving AOPIM-1 in DMF or DMF/1,4-dioxane mixed solvent, and stirred continuously at room temperature overnight to ensure that the polymer dissolves evenly in the solvent. Then the mixture was left at room temperature for 24 h to remove air bubbles. After that, the polymer solution was used to cast films on a clean glass plate at 25 ºC and 40% relative humidity. For the casting solution with co-solvent, the solvent was allowed to evaporate from the surface of the film in 20 s to produce a denser selective skin. Next, the glass plate was immersed into a coagulation bath. After 1 h, membranes were transferred to a fresh water bath and kept for 24 h to finish phase separation. Finally, the membranes were immersed in methanol for future use.
+
+**Membrane material characterization.** The morphology of the as-prepared membranes was observed using a field emission scanning electron microscope (Hitachi S4800, Japan). Before capturing SEM image, a thin Au layer was sputtered onto the membrane under 10 mA for 2 min (Emitech K550X sputtering).¹ ¹H nuclear magnetic resonance (NMR) spectra were recorded on a Bruker 400 MHz spectrometer using dimethyl sulfoxide-d₆ as a solvent. FTIR spectra of membranes were obtained using a Nicolet 6700 FTIR spectrometer (USA). Nitrogen absorption/desorption measurements were performed on a Quantachrome Autosorb IQ-MP-MP at 77 K. All samples were degassed at 120°C for 12 h before nitrogen absorption measurements were performed. The surface charge of the membrane was determined by streaming potential measurement using a SurPASS 3 electrokinetic analyzer with a flat-plate measuring cell (10 mm × 20 mm).
+
+**Static adsorption behavior of AOPIM-1 polymer.** The static adsorption behavior of AOPIM-1 was investigated using dyes with different chargeability (negative, MO; positive, MB) as model solutes. A dry AOPIM-1 membrane coupon (~10 mg) was placed in dye solution with certain concentration (10–500 mg L⁻¹) and pH value. The mixture was stirred continuously at room temperature at least 24 h or a certain time. The concentration of dyes was analyzed by a UV-Vis spectrophotometer (PerkinElmer, Inc.). The amount of dye adsorbed by the AOPIM-1, q (mg g⁻¹), was determined from the Eq. (1)⁴⁹–⁵⁰,
+
+$$q=\frac{(C_0-C_*)V}{w}$$ (1)
+
+where q represents the adsorption amount at equilibrium (qₑ) or the adsorption amount at time t (qₜ). C₀ (mg L⁻¹), C∗ (mg L⁻¹), V (L), and W (g) represent initial concentration of dye, concentration of dye at equilibrium or time t, volume of solution and amount of AOPIM-1, respectively.
+
+The effect of pH on the adsorption was studied by adjusting pH of the dye solutions to 3–10 with the help of 0.1 M NaOH and 0.1 M HCl.
+
+The adsorption kinetics of dyes by AOPIM-1 was studied using a second-order equation in nonlinear form by Eq. (2)⁵⁴–⁵⁵,
+
+$$q_t=\frac{k_2 q_e^2 t}{1+k_2 q_e t}$$ (2)
+
+where k₂ (g mg⁻¹ min⁻¹) is the second-order rate constant. The adsorption capacity of dyes by AOPIM-1 was studied using a Langmuir isotherm model by Eq. (3)⁵⁴−⁵⁵,
+
+$$q_e=\frac{K_L q_m C_e}{1+K_L C_e}$$ (3)
+
+where Kₗ (L mg⁻¹) represents the affinity constant and qₘ (mg g⁻¹) is the maximum adsorption capacity of the adsorbate.
+
+**Dynamic adsorptive separation properties of AOPIM-1 membranes** The dynamic adsorption properties of AOPIM-1 membranes were investigated using six types of dyes with different chargeability (negative: MO, CR, BB; positive: MB, RHB, CV). Filtration experiments were performed in a dead-end filtration cell with an effective membrane area of 3.14 cm² (Figure S16). Each tested membrane was compacted by the filtration of deionized water under 0.2 MPa for 2 h in order to achieve a steady flux. Then, the permeate flux (J (L m⁻² h⁻¹ bar⁻¹)) used different dyes as feed solution (20 mg L⁻¹, pH=3/10) was measured under 0.2 MPa at room temperature and calculated using the following Eq. (4),
+
+$$J=\frac{V}{\Delta PAt}$$ (4)
+
+where ΔP (bar), V (L), A (m²), and t (h) represent permeate flux, transmembrane pressure, permeate volume, membrane area and filtration time, respectively. The rejection rate (R (%)) of dyes was calculated from Eq. (5),
+
+$$R=1-\frac{C_P}{C_F}$$ (5)
+
+where Cₚ and C_F correspond to the dye concentrations in the permeate and feed solutions, respectively. The dye concentrations in the permeate and feed solutions were determined using a UV/Vis spectrometer (Biochrom Libra S32). The processing capacity used RHB as feed solution (20 mg L⁻¹, pH=10) was measured under 0.2 MPa at room temperature and determined by permeation volume when rejection is higher than 99%.
+
+The multi-cycle adsorptive separation performance stability of the AOPIM-1 adsorptive membrane was performed in a dead-end filtration cell with an effective membrane area of 3.14 cm² for 8 cycles. For a typical cycle, the RHB solution (20 mg L⁻¹, pH = 10) was filtrated through the membrane for 500 L m⁻² at 0.2 MPa, and the permeate flux (J) was calculated by Eq. (4). Afterwards, the membrane is subjected to a short time (five minutes) desorption process by switching the feed solution of the membrane to about 20 ml methanol for cleaning, and the eluent can be obtained at the outlet. After that, fresh feed dye solution was applied for the next cycle.
+
+**Adsorptive separation of APIs by AOPIM-1 membranes** Separation of a synthetic 3-component mixture feed representing inorganic salt, polysaccharides and active pharmaceutical ingredient (API) chemicals in the water extract of natural plants was performed in a lab-scale cross-flow cell (Figure S13) at room temperature. The feed solution included NaCl, dextran T-200 and berberine with detailed composition listed in Table S5. The concentrations of each substance in the feed and permeate solution were measured by conductometer (FE30K, Mettler Toledo), total organic carbon (TOC) analyzer (Aurora 1030W) and UV-Vis spectrophotometer (PerkinElmer, Inc.), respectively.
+
+# References
+
+1. Sholl, D. S. & Lively, R. P. Seven chemical separations to change the world. *Nature* **532**, 435–437 (2016).
+
+2. Shannon, M. A., Bohn, P. W., Elimelech, M., Georgiadis, J. G., Marinas, B. J. & Mayes, A. M. Science and technology for water purification in the coming decades. *Nature* **452**, 301–310 (2008).
+
+3. Pintor, A. M. A., Vilar, V. J. P., Botelho, C. M. S. & Boaventura, R. A. R. Oil and grease removal from wastewaters: Sorption treatment as an alternative to state-of-the-art technologies. A critical review. *Chem. Eng. J.* **297**, 229–255 (2016).
+
+4. Buonomenna, M. G. & Bae, J. Organic Solvent Nanofiltration in Pharmaceutical Industry. *Sep. Purif. Rev.* **44**, 157–182 (2015).
+
+5. Abels, C., Carstensen, F. & Wessling, M. Membrane processes in biorefinery applications. *J. Membrane. Sci.* **444**, 285–317 (2013).
+
+6. Marchetti, P., Jimenez Solomon, M. F., Szekely, G. & Livingston, A. G. Molecular separation with organic solvent nanofiltration: a critical review. *Chem. Rev.* **114**, 10735–10806 (2014).
+
+7. Park, H. B., Kamcev, J., Robeson, L. M., Elimelech, M. & Freeman, B. D. Maximizing the right stuff: The trade-off between membrane permeability and selectivity. *Science* **356**, 1138–1148 (2017).
+
+8. Wang, Z., Wang, Z., Lin, S., Jin, H., Gao, S., Zhu, Y. & Jin, J. Nanoparticle-templated nanofiltration membranes for ultrahigh performance desalination. *Nat. Commun.* **9**, 2004 (2018).
+
+9. Fane, A. G., Wang, R. & Hu, M. X. Synthetic membranes for water purification: status and future. *Angew. Chem. Int. Ed.* **54**, 3368–3386 (2015).
+
+10. Bellona, C., Drewes, J. E., Xu, P. & Amy, G. Factors affecting the rejection of organic solutes during NF/RO treatment--a literature review. *Water. Res.* **38**, 2795–2809 (2004).
+
+11. Alsbaiee, A., Smith, B. J., Xiao, L., Ling, Y., Helbling, D. E. & Dichtel, W. R. Rapid removal of organic micropollutants from water by a porous beta-cyclodextrin polymer. *Nature* **529**, 190–194 (2016).
+
+12. Satilmis, B., Isık, T., Demir, M. M. & Uyar, T. Amidoxime functionalized Polymers of Intrinsic Microporosity (PIM-1) electrospun ultrafine fibers for rapid removal of uranyl ions from water. *Appl. Surf. Sci.* **467-468**, 648–657 (2019).
+
+13. Wang, Z., Cui, F., Pan, Y., Hou, L., Zhang, B., Li, Y. & Zhu, L. Hierarchically micro-mesoporous beta-cyclodextrin polymers used for ultrafast removal of micropollutants from water. *Carbohyd. Polym.* **213**, 352–360 (2019).
+
+14. Dong, Z., Wang, D., Liu, X., Pei, X., Chen, L. & Jin, J. Bio-inspired surface-functionalization of graphene oxide for the adsorption of organic dyes and heavy metal ions with a superhigh capacity. *J. Mater. Chem. A.* **2**, 5034–5040 (2014).
+
+15. Boi, C., Dimartino, S. & Sarti, G. C. Modelling and simulation of affinity membrane adsorption. *J. Chromatogr. A.* **1162**, 24–33 (2007).
+
+16. Ghosh, R. Protein separation using membrane chromatography: opportunities and challenges. *J. Chromatogr. A.* **952**, 13–27 (2002).
+
+17. Uliana, A. A., Bui, N. T., Kamcev, J., Taylor, M. K., Urban, J. J. & Long, J. R. Ion-capture electrodialysis using multifunctional adsorptive membranes. *Science* **372**, 296–299 (2021).
+
+18. Wang, H., Zhao, S., Liu, Y., Yao, R., Wang, X., Cao, Y., Ma, D., Zou, M., Cao, A., Feng, X. & Wang, B. Membrane adsorbers with ultrahigh metal-organic framework loading for high flux separations. *Nat. Commun.* **10**, 4204 (2019).
+
+19. Fenton, J. L., Burke, D. W., Qian, D., Cruz, M. O. & Dichtel, W. R. Polycrystalline Covalent Organic Framework Films Act as Adsorbents, Not Membranes. *J. Am. Chem. Soc.* **143**, 1466–1473 (2021).
+
+20. Hao, S., Jia, Z., Wen, J., Li, S., Peng, W., Huang, R. & Xu, X. Progress in adsorptive membranes for separation-A review. *Sep. Purif. Technol.* **255**, 117772 (2021).
+
+21. Ting, H., Chi, H.-Y., Lam, C. H., Chan, K.-Y. & Kang, D.-Y. High-permeance metal–organic framework-based membrane adsorber for the removal of dye molecules in aqueous phase. *Environ. Sci-Nano.* **4**, 2205–2214 (2017).
+
+22. Zong, L., Yang, Y., Yang, H. & Wu, X. Shapeable Aerogels of Metal-Organic-Frameworks Supported by Aramid Nanofibrils for Efficient Adsorption and Interception. *ACS Appl. Mater. Inter.* **12**, 7295–7301 (2020).
+
+23. Wang, Z., Zhang, B., Fang, C., Liu, Z., Fang, J. & Zhu, L. Macroporous membranes doped with micro-mesoporous beta-cyclodextrin polymers for ultrafast removal of organic micropollutants from water. *Carbohyd. Polym.* **222**, 114970 (2019).
+
+24. Wang, Z., Guo, S., Zhang, B., Fang, J. & Zhu, L. Interfacially crosslinked beta-cyclodextrin polymer composite porous membranes for fast removal of organic micropollutants from water by flow-through adsorption. *J. Hazard. Mater.* **384**, 121187 (2020).
+
+25. Liang, H. W., Cao, X., Zhang, W. J., Lin, H. T., Zhou, F., Chen, L. F. & Yu, S. H. Robust and Highly Efficient Free-Standing Carbonaceous Nanofiber Membranes for Water Purification. *Adv. Funct. Mater.* **21**, 3851–3858 (2011).
+
+26. Furukawa, H., Cordova, K. E., O’Keeffe, M. & Yaghi, O. M. The chemistry and applications of metal-organic frameworks. *Science* **341**, 1230444 (2013).
+
+27. McKeown, N. B. & Budd, P. M. Polymers of intrinsic microporosity (PIMs): organic materials for membrane separations, heterogeneous catalysis and hydrogen storage. *Chem. Soc. Rev.* **35**, 675–683 (2006).
+
+28. Wang, Z., Wang, D., Zhang, S., Hu, L. & Jin, J. Interfacial Design of Mixed Matrix Membranes for Improved Gas Separation Performance. *Adv. Mater.* **28**, 3399–3405 (2016).
+
+29. Wang, Z., Wang, D. & Jin, J. Microporous Polyimides with Rationally Designed Chain Structure Achieving High Performance for Gas Separation. *Macromolecules* **47**, 7477–7483 (2014).
+
+30. Zhang, C., Pei, L., Huang, W. & Bing, C. Selective adsorption and separation of organic dyes in aqueous solutions by hydrolyzed PIM-1 microfibers. *Chem. Eng. Res. Des.* **109**, 76–85 (2016).
+
+31. Patel, H. A. & Yavuz, C. T. Noninvasive functionalization of polymers of intrinsic microporosity for enhanced CO₂ capture. *Chem. Commun.* **48**, 9989–9991 (2012).
+
+32. Tan, R., Wang, A., Malpass-Evans, R., Williams, R., Zhao, E. W., Liu, T., Ye, C., Zhou, X., Darwich, B. P., Fan, Z., Turcani, L., Jackson, E., Chen, L., Chong, S. Y., Li, T., Jelfs, K. E., Cooper, A. I., Brandon, N. P., Grey, C. P., McKeown, N. B. & Song, Q. Hydrophilic microporous membranes for selective ion separation and flow-battery energy storage. *Nat. Mater.* **19**, 195–202 (2020).
+
+33. Luo, X., Wang, Z., Wu, S., Fang, W. & Jin, J. Metal ion cross-linked nanoporous polymeric membranes with improved organic solvent resistance for molecular separation. *J. Membrane. Sci.* **621**, 119002 (2021).
+
+34. Wang, Z., Ren, H., Zhang, S., Zhang, F. & Jin, J. Polymers of intrinsic microporosity/metal-organic framework hybrid membranes with improved interfacial interaction for high-performance CO₂ separation. *J. Mater. Chem. A.* **5**, 10968–10977 (2017).
+
+35. Wang, Z. G., Liu, X., Wang, D. & Jin, J. Tröger’s base-based copolymers with intrinsic microporosity for CO₂ separation and effect of Tröger’s base on separation performance. *Polym. Chem.* **5**, 2793–2800 (2014).
+
+36. Liu, L., Ma, Y., Yang, W., Chen, C., Li, M., Lin, D. & Pan, Q. Reusable ZIF-8@chitosan sponge for the efficient and selective removal of congo red. *New. J. Chem.* **44**, 15459–15466 (2020).
+
+37. Satilmis, B. & Budd, P. M. Selective dye adsorption by chemically-modified and thermally-treated polymers of intrinsic microporosity. *J. Colloid. Interf. Sci.* **492**, 81–91 (2017).
+
+38. Jing, R., Yan, C., Liu, Q., Yang, Q. & Li, Y. Preparation of amidoxime-modified polyacrylonitrile nanofibrous adsorbents for the extraction of copper(II) and lead(II) ions and dye from aqueous media. *J. Appl. Polym. Sci.* **135**, 45697 (2017).
+
+39. Zhao, S., & Wang, Z. A loose nano-filtration membrane prepared by coating HPAN UF membrane with modified PEI for dye reuse and desalination. *J. Membrane. Sci.* **524**, 214–224 (2017).
+
+40. Lin, J., Ye, W., Zeng, H., Yang, H., Shen, J., Darvishmanesh, S., & Van der Bruggen, B. Fractionation of direct dyes and salts in aqueous solution using loose nanofiltration membranes. *J. Membrane. Sci.* **477**, 183–193 (2015).
+
+41. Yang, Z., Zhou, Z. W., Guo, H., Yao, Z., Ma, X. H., Song, X., & Tang, C. Y. Tannic acid/Fe³⁺ nanoscaffold for interfacial polymerization: toward enhanced nanofiltration performance. *Environ. Sci. Technol.* **52**, 9341–9349 (2018).
+
+42. Yu, L., Zhang, Y., Zhang, H., & Liu, J. Development of a molecular separation membrane for efficient separation of low-molecular-weight organics and salts. *Desalination* **359**, 176–185 (2015).
+
+43. You, F., Xu, Y., Yang, X., Zhang, Y. & Shao, L. Bio-inspired Ni (2+)-polyphenol hydrophilic network to achieve unconventional high-flux nanofiltration membranes for environmental remediation. *Chem. Commun.* **53**, 6128–6131 (2017).
+
+44. Hu, M., & Mi, B. Enabling graphene oxide nanosheets as water separation membranes. *Environ. Sci. Technol.* **47**, 3715–3723 (2013).
+
+45. Zhao, R., Ma, T., Cui, F., Tian, Y. & Zhu, G. Porous Aromatic Framework with Tailored Binding Sites and Pore Sizes as a High-Performance Hemoperfusion Adsorbent for Bilirubin Removal. *Adv. Sci.* **7**, 2001899 (2020).
+
+46. Tan, L., Gong, L., Wang, S., Zhu, Y., Zhang, F., Zhang, Y. & Jin, J. Superhydrophilic Sub-1-nm Porous Membrane with Electroneutral Surface for Nonselective Transport of Small Organic Molecules. *ACS Appl. Mater. Inter.* **12**, 38778–38787 (2020).
+
+47. Soroko, I., Lopes, M. P. & Livingston, A. The effect of membrane formation parameters on performance of polyimide membranes for organic solvent nanofiltration (OSN): Part A. Effect of polymer/solvent/non-solvent system choice. *J. Membrane. Sci.* **381**, 152–162 (2011).
+
+48. See-Toh, Y. H., Silva, M. & Livingston, A. Controlling molecular weight cut-off curves for highly solvent stable organic solvent nanofiltration (OSN) membranes. *J. Membrane. Sci.* **324**, 220–232 (2008).
+
+49. Xing, T., Kai, H. & Chen, G. Study of adsorption and desorption performance of acid dyes on anion exchange membrane. *Color. Technol.* **128**, 295–299 (2012).
+
+50. Yuan, L. Y., Gao, G., Feng, C. Q., Chai, Z. F. & Shi, W. Q. A new family of actinide sorbents with more open porous structure: Fibrous functionalized silica microspheres. *Chem. Eng. J.* **385**, 123892 (2020).
+
+# Supplementary Files
+
+- [Supportinginformation.docx](https://assets-eu.researchsquare.com/files/rs-1132837/v1/dd4e177137522e358efb3dac.docx)  
+  supporting information

0ae095e4601335f806966eda85cd44ddaab029846fb71b47faa2768ba5776ad3/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Mechanoluminescent behaviors under rapid compression at given rates. a Microscopic optical images of self-recoverable ML light emission in SrZnOS: Mn2+ under compression at rates of 0.6 GPa/s (\u2160) and 6.5 GPa/s (\u2161) at room temperature. b The ML spectra obtained under rapid compression at 4.7 GPa/s. c Comparison of the ML and PL spectra (\u03bbex=375 nm) at ~1.0 GPa. d Peak position of ML as a function of pressure.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Oscillatory ML behavior of SrZnOS: Mn2+ at different loading rates. a and (b) ML spectra of Mn-doped SrZnOS as a function of pressure (or time) at ~0.6 GPa/s and 1.5 GPa/s at room temperature, presenting temporal characteristics. The PL peaks around 695 nm (marked by \u201c*\u201d) come from ruby. c Instantaneous ML intensity of SrZnOS: Mn2+ as a function of time under rapid compression at various compression rates. The data was collected using PTM at 0.2-3 GP/s and time-resolved fluorescent spectra above 3 GPa/s.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Experimental summary and temperature-dependent average \u03c4c and \u03c4ML at different rates. a Summary of the ML emission under compression at different temperatures and compression rates. The compression rate was estimated from the pressure difference and loading time. b, c Experimental summary of the average \u03c4c and \u03c4ML as a function of rate at different temperatures. The thick yellow lines in Fig. 3b and 3c are guide to the eye. The inset figure in Fig. 3c is the total intensity as a function of compression rate at room temperature.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Pressure-induced evolution of crystal structures and photoluminescence. a In-situ high-pressure synchrotron XRD patterns of Mn-doped SrZnOS. b d-space and (c) unit cell volume and lattice parameters as a function of the pressure. d High-pressure PL spectra of SrZnOS: Mn2+ excited by the laser of 375nm. e PL intensity as a function of pressure. The inset figure in Fig. 4e is the lifetime of SrZnOS: Mn2+ as a function of pressure.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Rate- and temperature-dependent kinetics of the recoverable ML emission. a Schematic diagram of the self-recoverable ML emission in SrZnOS: Mn2+ through piezoelectrically-induced excitation (PIE) and photon emission. b and c Interpretation of the time-dependent instantaneous ML intensity using Chandra\u2019s model at different loading rates below and at the critical rates. d and e \u03c4ML and the difference between minimum \u03c4ML and \u03c4c at the critical rates as a function of temperature.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

0ae095e4601335f806966eda85cd44ddaab029846fb71b47faa2768ba5776ad3/preprint/preprint.md

@@ -0,0 +1,152 @@
+# Abstract
+
+Photon emission may be continuously produced from mechanical work through the piezoelectrically-induced excitation (PIE) and self-recovery mechanoluminescence (ML) process. Significant progress has been made in high-performance ML materials in last decades, but the rate-dependent ML kinetics remains poorly understood. Here, we have conducted systematic studies on the self-recoverable ML of SrZnOS: Mn²⁺ under rapid compression up to ~10 GPa. A rate-dependent distinct kinetics is revealed: a diffuse-like ML behavior below ~1.2 GPa/s, oscillatory emission with a series of ML peaks at critical rate of ~1.2–1.5 GPa/s, and suppression above 1.5 GPa/s. Analysis from the rate-independent structural evolution and photoluminescence under high pressures show that the oscillatory ML emission at the critical rate corresponds to multi-cyclic PIE and self-recoverable processes. Both characteristic time (τ) for the PIE and self-recovery processes are minimized at the critical rate, indicating the time limit of ML in the dynamic response to rapid compression. The temperature is slightly favorable for PIE, but is unfavorable for the self-recovery process. The present work uncovers the temporal characteristics of self-recoverable ML, which provides a new insight into understanding the rate-dependent ML kinetics in the mechanical-photon energy conversion, conducive to the design of optoelectronic devices.
+
+[Physical sciences/Optics and photonics/Optical materials and structures/Photonic crystals](/browse?subjectArea=Physical%20sciences%2FOptics%20and%20photonics%2FOptical%20materials%20and%20structures%2FPhotonic%20crystals)
+
+[Physical sciences/Materials science/Materials for optics/Lasers, LEDs and light sources](/browse?subjectArea=Physical%20sciences%2FMaterials%20science%2FMaterials%20for%20optics%2FLasers%2C%20LEDs%20and%20light%20sources)
+
+# Introduction
+
+Mechanoluminescence is an intriguing luminescent behavior that converts mechanical energy into light emission under mechanical stimulation, such as compression, stretching, friction, impact, fracture, grinding, crushing, and so on<sup>1,2</sup>. It has been broadly investigated and can be categorized into fractoluminescence, triboluminescence, plasticoluminescence, and elasticoluminescence (EML) based on applied mechanical stimuli<sup>3,4</sup>. Among them, self-recoverable EML, that possesses potential applications in pressure sensing, dynamic stress visualization, bionic robotics, and optoelectronics, has attracted considerable attention due to its reproducibility, high efficiency, and self-recovery in dynamic response to the repetitive elastic deformation, since first discovery in ZnS: Mn<sup>2+</sup> by Xu et al. in 1999<sup>5-11</sup>. To date, increasing numbers of EML materials with high emission performance have been synthesized and optimized by modulating concentration of luminescent ions doped, composites, categories of matrix and dopants in chemical and materials science<sup>12,13</sup>. Great progress has been achieved in realizing tunable emission colors and enhanced ML intensity<sup>14</sup>. On the other hand, pressure and loading rate have significant influence on the emission intensity and wavelength of ML, which are the basis for the designs and applications of ML devices. The effect of pressure on the ML performance has been mainly considered in previous designs, yet ignoring the time and rate dependence of ML kinetics. It leads to the limitations in the development of time-dependent functions in the ML devices. This is because the mechanism that underlies the complex ML emission process in the mechanical-photon energy conversion remains unclear, and the detailed information on the rate-dependent ML kinetics is still in lack.
+
+Different mechanisms have been proposed to interpret kinetic process of the self-recoverable ML emission, e.g., piezoelectrically-induced de-trapping and triboelectrically-controlled models<sup>1,3,15-18</sup>. As a majority of the ML materials have been found in the piezoelectric host doped with lanthanide or transition metal ions, it is generally believed that the ML light emission is derived from piezoelectrically-induced excitation (PIE) of the luminescent ions through multiple interactions between the piezoelectrical potential, energy bandgap, defect states, carriers, and dopant activators. Several intermediate steps are involved, including deformation-induced polarization, piezoelectrically-induced de-trapping of carriers, recombination of the de-trapped electrons and holes with release of radiation energy, excitation of the luminescent ions, and photon emission, as previously indicated in ZnS: Mn<sup>2+</sup> and other materials<sup>17,19-21</sup>. It is considered briefly by Wang et al. as a two-way coupling between the piezoelectricity and photoexcitation processes<sup>15-16,22</sup>. The self-recovery process with the re-trapping (or re-charging) of the de-trapped electron carriers is required for the multi-cyclic PIE and photon emission of the luminescent activators<sup>15,17,23-24</sup>. Since the self-recoverable EML emission may undergo complex multiple interactions and intermediate steps, the characteristic time for the PIE and self-recovery processes in the response to the external mechanical stimulation (e.g., compression in the GPa pressure range) should be considerably longer than that of the well-known photoluminescence (PL) that is directly photon-excited<sup>25-27</sup>. Previous studies have shown that the ML process is affected by loading rate, pressure, piezoelectrical field, etc<sup>1,28-30</sup>. The ML intensity was observed to increase linearly with compression rate under slow compression, but exhibit frequency-dependent nonlinear saturation at high rate<sup>29</sup>. This implies significant influence of the competition between the compression rate and intrinsic self-recoverable ML time. However, how the compression rate affects the ML kinetics and what is the characteristic time for the PIE and self-recovery processes remain unclear, which provides key performance in designing optoelectronic devices using the coupled effect of pressure and rate on the emission wavelength and intensity.
+
+The models describing the time (or pressure)-dependent instantaneous ML intensity play a crucial role in understanding the kinetic process of the self-recoverable ML emission<sup>31-34</sup>. Based on the Boltzmann statistical distribution assumption on trap-depth<sup>17,19</sup>, Chandra et al. proposed a general EML emission process in which the luminescent activators involve one cycle of the PIE and photon emission<sup>18</sup>. Under continuous compression at a given rate, the instantaneous EML intensity is predicted to increase first with time (or pressure) and decrease after approaching a peak value due to small amount of the trapped carriers remaining<sup>19-20,34</sup>. The peak width corresponds to the time it takes for all the luminescent ions involved to be piezoelectrically excited to emit photons, and decreases at higher compression rates. This has been confirmed by the experimental observation of the ML peak in the time (or pressure)-dependent instantaneous ML intensity in the MPa pressure range<sup>1,5,19,29</sup>. The self-recovery process was assumed previously to occur during release of pressure, and therefore the EML kinetics for the multi-cyclic processes of the PIE and self-recovery was studied mostly during the cyclic compression-decompression processes at MPa level. However, recent studies revealed the ML behavior with an optimal emission efficiency at ~3.5 GPa in the GPa pressure range<sup>35,36</sup>, different from the observation at MPa level. This means the self-recovery for EML could occur under compression at the GPa level. It raises questions regarding how to describe the ML kinetics with the multi-cyclic PIE and self-recovery processes during continuous compression at a given rate, whether it is rate-dependent, and what the feature are if the time scale during compression process is comparable to the PIE and self-recovery.
+
+In the present work, piezoelectric SrZnOS: Mn<sup>2+</sup> was chosen to investigate the kinetic process of the EML emission under rapid compression from ambient pressure to ~10 GPa at different rates and various temperatures in dynamic diamond anvil cells using time-resolved fluorescent spectroscopy, optical imaging, high-pressure X-ray diffraction (XRD), and thermoluminescence spectroscopy. The experimental results reveal compression-rate-dependent kinetics of the self-recoverable ML emission. An oscillatory mechanoluminescent phenomenon was observed with the periodic light emission occurring under rapid compression at certain compression rates, corresponding to the multi-cyclic processes of the PIE and self-recovery. Temperature-dependent characteristic time show different effect of temperature on the PIE and self-recovery processes. The present work offers new insights into the kinetics of the self-recoverable EML emission under compression in the GPa pressure range.
+
+# Results
+
+## Rate-dependent mechanoluminescence under rapid compression
+
+The piezoelectric SrZnOS doped with luminescent ions of 1% Mn²⁺ has been synthesized by solid state reaction and confirmed to have a non-centrosymmetric hexagonal structure with lattice parameters *a* = *b* = 3.909(4) Å and *c* = 11.615(1) Å by X-ray diffraction, analogous to the wurtzite ZnS³⁷. The scanning electron microscopy and energy-dispersive spectroscopy show micron-sized particle grains with the homogeneous distribution of elemental Mn in the SrZnOS matrix (Supplementary Fig. 1). The PL measurements show that the Mn²⁺ ions are successfully doped in the SrZnOS sample with the emission peak at 608 nm and occupy the Zn²⁺ sites (Supplementary Fig. 2), as the similar ionic radii between Zn²⁺ and Mn²⁺⁶.
+
+Millisecond time-resolved optical imaging demonstrate strong dependence of the ML emission behavior on loading rates under rapid compression from ambient pressure to ~ 10 GPa (Fig. 1a), showcasing excellent conversion of mechanical energy to light emission in Mn²⁺-doped SrZnOS. Before the ML measurement, the sample was kept in the dark environment for several hours without the UV-light or irradiation to make sure the ML emission was self-recoverable ML. Generally, SrZnOS: Mn²⁺ exhibits the ML emission process that first brightens and then darkens under continuous compression at a given compression rate. At ~ 6.5 GPa/s, it shows brighter ML emission with higher intensity than that at ~ 0.6 GPa/s. There is no ML light emission occurring when keeping pressure constant. The fluorescence spectroscopy confirms the changes of the ML emission in the intensity and wavelength under high pressures (Fig. 1b). The ML spectrum centering at ~ 615 nm is consistent with PL of the Mn²⁺ at ~ 1 GPa (Fig. 1c), implying that the ML emission of SrZnOS: Mn²⁺ is derived from the ⁴T₁ to ⁶A₁ transition of Mn²⁺. The ML peak shifts from ~ 608 nm at ambient pressure to ~ 670 nm at ~ 11 GPa with the slope of ~ 5.0 nm/GPa (Fig. 1d). The red-shift of the ML emission wavelength could be attributed to the enhancement of the strength of the crystal field with increasing pressure³⁵–³⁶,38–40.
+
+Time-resolved ML fluorescent spectra of SrZnOS: Mn²⁺ further demonstrate rate-dependent ML light emitting under rapid compression (Fig. 2a, 2b, and Supplementary Fig. 3). Interestingly, SrZnOS: Mn²⁺ exhibits the feature of alternating brightening and darkening with the oscillatory emission peaks when compressing the samples at certain loading rates, rather than the monotonous enhancement or weakening of the emission intensity. For instance, the ML intensity increases first with pressure from 0 to ~ 0.9 GPa and then decreases from ~ 0.9 GPa to ~ 3 GPa under ramp compression at 0.6 GPa/s, forming a diffusive-like broad peak in the pressure-dependent ML intensity (Fig. 2a). Under further compression, a second broad peak of the ML emission is observed at ~ 3–5.6 GPa. In contrast, SrZnOS: Mn²⁺ exhibits a series of oscillatory ML emission peaks that brightens and weakens alternately under rapid compression at a critical rate of ~ 1.5 GPa/s (Fig. 2b). This oscillatory ML emission at the critical compression rate is analogous to the resonant phenomenon. Above 1.5 GPa/s, the number of the emission peaks decreases with increasing rate.
+
+In order to investigate the effect of the compression rate on the ML emission in detail, microsecond time-resolved photomultiplier tube (PMT) was employed to record the instantaneous ML emission intensity of SrZnOS: Mn²⁺ varying with time under ramp compression at rates of 0.2–3 GPa/s (Fig. 2c). Distinct kinetic processes of the self-recoverable ML emission in SrZnOS: Mn²⁺ are confirmed in the dynamic response to the rapid compression at different rates, namely, the diffuse-like emission with the broad emission peaks in the time-dependent ML intensities at 0.28–1.04 GPa/s and oscillatory ML behaviors with a series of emission peaks at the critical rates of ~ 1.2–1.5 GPa/s. The oscillatory ML emission is gradually suppressed with the decrease in the number of the emission peaks and total intensity above the critical rates (Supplementary Fig. 4). At higher rate above 5 GPa/s, single ML emission peak is observed. The oscillatory ML emission that brightens and darkens alternately is also confirmed in the time-resolved optical imaging (Supplementary Movie 1), consistent with the time-resolved fluorescent spectra and PMT. Meanwhile, we observed the feature of the oscillatory ML emission in SrZnOS doped with different concentrations of Mn²⁺ and Mn-doped different matrix materials (Supplementary Movie 2–7). Therefore, the oscillatory behaviors occurring at the critical compression rates should represent the inherent temporal characteristics of the mechanically-excited ML photon emission in the dynamic response to rapid compression.
+
+To study the effect of temperature on the ML kinetics, the ML behaviors were investigated under compression at different rates and various temperatures ranging from 298 K to 423 K, as summarized in Fig. 3a (Supplementary Fig. 5). The oscillatory feature at the critical rates is further confirmed at 323 K, 348 K, 373K, and 398 K, respectively. Generally, distinct regions are observed in the temperature-rate diagram, namely, diffusive-like ML behavior with broad emission peaks below ~ 1.2 GPa/s and the oscillatory ML emission at ~ 1.2–1.5 GPa/s. At higher temperature, the ML intensity becomes weaker with the ML peaks overlapped and the ML behavior is suppressed gradually with increasing temperature, leading to the disruption of the oscillatory ML behavior at 423 K.
+
+The characteristic time τₘₗ and τ꜀ are used to describe the temporal features of the mechanically-excited ML emission. τₘₗ is defined as the full width at half maximum (FWHM) of the ML emission peak. τ꜀ is interval time between adjacent ML peaks. Below 1.2 GPa/s, the ML curve may consist of several broad ML peaks that are overlapped and distinguished hardly. For comparison, τₘₗ is obtained by assuming the ML curves consist of two broad ML peaks. It should be noted that τₘₗ and τ꜀ vary slightly with pressure at a given rate. To understand the rate-dependent ML emission kinetics, τₘₗ and τ꜀ are averaged and plotted as a function of compression rate (Fig. 3b and 3c). Both average τₘₗ and τ꜀ exhibit a similar trend over compression rate (Fig. 3a and 3b). They decrease rapidly with increasing rates from ~ 0.28 GPa/s to 1.2 GPa/s, approach a minimum value (~ 140 ms for τₘₗ and ~ 270 ms for τ꜀) at the critical rate of ~ 1.2–1.5 GPa/s, and then change slightly above 1.5 GPa/s. The existence of minimum τₘₗ and τ꜀ at the critical rate implies that the dynamic response of the mechanical excitation to rapid compression reaches the time limit. It is also indicated by the rate-dependent changes of the total ML intensity with compression rate that has different slopes above and below the critical rate (inset of Fig. 3c).
+
+## Structural evolution and photoluminescence under pressures
+
+To gain a deeper understanding of the rate-dependent ML kinetics, the structural evolution and effect of the lattice contraction on the energy level of Mn²⁺ were investigated under high pressures by in-situ synchrotron X-ray diffraction, Raman, and photoluminescence measurements. Figure 4 shows the diffraction patterns of SrZnOS: Mn²⁺ under slow compression up to ~ 40 GPa. The diffraction peaks shift smoothly to higher angle with the decrease of *d*-spacing as pressure is increased. New diffraction peaks appear at ~ 18 GPa, indicating a phase transition occurring at ~ 18 GPa. Below 18 GPa, the initial hexagonal structure shrinks smoothly in the unit cell volume and lattice parameters (Fig. 4c). The structural evolution was also investigated at the rates where oscillatory ML emission was observed. We did not observe oscillatory changes in the lattice parameters and the intensities of the diffraction peaks. The Raman experiment further confirms smooth change of the lattice vibration with pressure (Supplementary Fig. 6). It is clear from these results that the experimental compression rate is far from being able to affect lattice change below 10 GPa in which the oscillatory ML emission occurs. In fact, it can be understood as the strain rate (< 0.1 s⁻¹) is well below the lattice response (i.e., speed of sound).
+
+In principle, the ⁴T₁ to ⁶A₁ transition for isolated Mn²⁺ is spin-forbidden with extremely low photon absorption. In SrZnOS: Mn²⁺, the defect traps and asymmetric crystal field exist in the localized tetrahedral environment due to the interaction between the host lattice and dopant results, leading to the non-radiative ⁴T₁ to ⁶A₁ transition of Mn²⁺. Therefore, the strong PL emission of Mn²⁺ could be observed in SrZnOS: Mn²⁺, and is affected by the enhanced interaction between the host lattice and Mn²⁺ dopant under high pressures. The PL spectra show gradual changes in peak shape, intensity and wavelength without rate-dependent oscillatory behavior under compression (Fig. 4d, 4e and Supplementary Fig. 7). The PL shape becomes asymmetric when pressure is increased, as the tetrahedron becomes more distorted due to the anisotropic compressibility of the Zn/Mn-S and Zn/Mn-O bonds (Supplementary Fig. 8–9 and Supplementary Table 1). The emission intensity deceases slightly under compression up to ~ 5 GPa, then is enhanced at ~ 6–10 GPa, and finally diminishes rapidly with pressure above 10 GPa (Fig. 4e). The PL peak exhibits a continuous red-shift with pressure. The PL decay curve of Mn²⁺ shows the lifetime is around 1.5–2.5 ms at 0–10 GPa (The inset of Fig. 4e and Supplementary Fig. 10), much smaller than the characteristic time (τₘₗ and τ꜀) of the ML emission and the experimental time scale under rapid compression. Hence, these results show that the photon-excited PL is mainly affected by pressure-enhanced interaction between the lattice and dopant.
+
+# Discussion
+
+The time-resolved optical measurements reveal rate-dependent ML emission behaviors under rapid compression at 0–10 GPa, namely, diffuse-like ML behavior with broad peaks at the rate below 1.2 GPa/s and oscillatory ML emission with a series of the peaks at the critical rate of ~1.2–1.5 GPa/s. It means the compression rates has significant influence on the mechanically-excited process of ML. Before discussing rate-dependent emission kinetics, let us recall the self-recoverable ML mechanism previously reported<sup>18,29,32–34</sup>. The piezoelectric SrZnOS: Mn<sup>2+</sup> ion has a non-centrosymmetric hexagonal structure (<em>P63mc</em>) with excellent piezoelectric and EML properties<sup>6,37,41</sup>, similar to the wurtzite zinc ZnS: Mn<sup>2+</sup>. Based on previously reported de-trapping mechanism in ZnS: Mn<sup>2+</sup>/Cu<sup>2+1,15–17,20,35</sup>, here, the ML emission process is proposed for SrZnOS: Mn<sup>2+</sup> through PIE of the luminescent Mn<sup>2+</sup> in the <sup>4</sup>T<sub>1</sub> to <sup>6</sup>A<sub>1</sub> transition (Fig. <span class="InternalRef" refid="Fig5">5</span>a). Upon rapid compression with deviatoric stress at a given rate, a piezoelectric field is generated via deformation-induced polarization, accompanied by the de-trapping process with the separation of the trapped electrons and holes around the activator of Mn<sup>2+</sup>. The escaped de-trapped electron may be thermally activated into the conduction band. The thermoluminescence (TL) results for SrZnOS: Mn<sup>2+</sup> at ambient pressure confirm the existence of the defect traps with the trap-depth at 0.73–0.85 eV, following Gaussian-shaped trap-depth distribution<sup>42–43</sup> (Supplementary Fig. 11). Under piezoelectrical perturbation, the conduction band is tilted due to the piezoelectric effect, leading to a recombination of the de-trapped electrons and holes with the release of radiative energy. The luminescent Mn<sup>2+</sup> is excited from the ground state (<sup>6</sup>A<sub>1</sub>) to the excited state (<sup>4</sup>T<sub>1</sub>) by the energy transfer, and then the photon emission occurs in the <sup>4</sup>T<sub>1</sub> to <sup>6</sup>A<sub>1</sub> transition of Mn<sup>2+</sup>. The above ML kinetic process, given by the piezoelectrically-induced de-trapping mode, describes a cycle of the PIE and photon emission process of the luminescent ions, in which an emission peak will appear in the ML curve. This has been confirmed by the ML experiments<sup>17,19–20,31-34</sup>. Self-recovery that may involve re-trapping (or energy re-charging) of the carriers is required for multi-cyclic PIE processes. It can be speculated that a series of oscillatory ML emission peaks with adjacent ones distinguished will be observed during the multi-cyclic PIE and self-recovery processes if the time scale required for the PIE and self-recovery is comparable to or longer than that of the width of the emission peaks.
+
+Combined with previous studies in the MPa pressure range<sup>1,15–20,28-34</sup>, the following conditions may lead to the occurrence of the oscillatory ML emission peaks in the time-dependent instantaneous intensity: Multi-cyclic PIE and self-recovery processes that the carriers and luminescent ions involve (i) under ramp compression at a given rate, (ii) during multi-cyclic compression-decompression processes at a given frequency, and (iii) under continuous stepwise compression; (iv) ML peaks will appear if the trap-depth of the carriers have an oscillatory distribution in the GPa pressure range. In our experiments, we compressed sample continuously at a given rate, and the derivative stress should increase with pressure, ruling out the possibilities of (ii) and (iii) (Supplementary Fig. 12)<sup>18,29</sup>. For the case of (iv), we checked the ML emission under compression from different initial pressure to ~10 GPa at a given rate, and found the ML intensity did not follow the oscillatory change with initial pressure, excluding the oscillatory distribution of the trap-depth (Supplementary Fig. 13). Therefore, we believe that the oscillatory ML emission under rapid compression at the critical rates should stem from the multi-cyclic processes of the PIE and self-recovery,<em> viz</em>, the first possibility.
+
+Here, the ML emission peaks in the time-dependent instantaneous ML intensity of SrZnOS: Mn<sup>2+</sup> are interpreted tentatively using the equation proposed by Chandra <em>et al</em> (Fig. <span class="InternalRef" refid="Fig5">5</span>b, c)<sup>17,19</sup>. The Chandra’s model predicted a ML peak in the time-dependent instantaneous ML intensity,<em> viz</em>, the cumulative ML intensity has a sigmoidal profile over time<sup>19</sup>. The ML peak corresponds to a process in which the luminescent ions involved undergo a cyclic PIE and photon emission process. The characteristic time (τ<sub>ML</sub>), determined by the peak width, is predicted to decrease linearly with increasing rate. The fitting result at ~1.25 GPa/s shows that the model can describe the time-dependent ML intensity very well, implying that the oscillatory emission peaks correspond to multi-cyclic processes of the PIE and self-recovery that the luminescent ions involve. The characteristic time (τ<sub>c</sub>) corresponds to the time required for the PIE and self-recovery process. Below the critical rate, the ML curve may consist of several broad emission peaks which are hardly distinguished due to the large τ<sub>ML</sub> (Fig. <span class="InternalRef" refid="Fig5">5</span>c). It is difficult to determine how many cycles the luminescent ions have been piezoelectrically excited.
+
+The above analysis shows that the oscillatory ML behavior is the result of the competition between τ<sub>c</sub> and τ<sub>ML</sub>. The measurement of the PL lifetime shows the time scale of ~2 ms for the photon emission, much shorter than τ<sub>ML</sub> and τ<sub>c</sub> (Fig. S9). This means the PIE and self-recoverable processes is much longer than the photon-excitation in PL. The shorter the τ<sub>ML</sub> and τ<sub>c</sub>, the faster the PIE and self-recovery processes. The minimum value of τ<sub>ML</sub> and τ<sub>c</sub> observed at the critical rates of ~1.2–1.5 GPa/s implies the shortest time required for the PIE and self-recovery process, a limit in the dynamic response to the rapid compression. To study the temperature effect, τ<sub>ML</sub> and τ<sub>c</sub> at the critical rate are plotted as a function of temperature (Fig. <span class="InternalRef" refid="Fig5">5</span>d, e). When Fig. <span class="InternalRef" refid="Fig5">5</span> (c) is plotted with the natural logarithm of the critical τ<sub>ML</sub> and inverse temperature, the result shows an approximately linear relationship between τ<sub>ML</sub> and 1/T,<em> i.e.</em>, there is an Arrhenius relationship between the critical τ<sub>ML</sub> and temperature with 1/τ = C* exp(-Q/k<sub>B</sub>T)<sup>45–47</sup>, where C is a constant and Q is the thermal activation energy required to be overcome for the PIE processes. Fitting the experimental data yields Q of 0.042(1) eV. Q is close to the thermal energy (k<sub>B</sub>T), but much smaller than the trap-depth (0.73–0.85 eV) of the carriers. This indicates that the thermal effect exerts a subtle, favorable influence on the PIE process.
+
+On contrast, τ<sub>c</sub> almost changes negligibly with temperature. However, the difference between τ<sub>c</sub> and τ<sub>ML</sub>, that corresponds to the self-recovery process, increases slightly over temperature. This suggests unfavorable effect of temperature on the self-recovery process (Fig. <span class="InternalRef" refid="Fig5">5</span>e). Previous studies indicate the self-recovery process is related to the re-trapping of the carriers or recharging of the piezoelectric field<sup>29,31–34</sup>. Regardless of them, the mechanical work (W = P*ΔV) per unit cell is calculated to be ~0.04 eV if assuming that the sample is compressed from ambient pressure to 1 GPa, significantly smaller than the photon energy of ~2.0 eV emitted in ML. This implies the mechanical work contributed by the atoms around the luminescent ions will be essential for photon emission. Meanwhile, the ML light emission only occurs when pressure changes continuously at a given rate, indicating the mechanical work needs to be accumulated over time. Therefore, the self-recovery may be a process of the energy accumulation in time and space for the mechanical-photon energy conversion. The temperature dependence of the difference between the τ<sub>c</sub> and τ<sub>ML</sub> shows the thermal effect may disturb the energy accumulation in the self-recovery process (Fig. <span class="InternalRef" refid="Fig5">5</span>e). It should be noted that the above interpretation is based on the de-trapping model, but more theoretical and experimental work is required to confirm it.
+
+In summary, we reveal a compression rate-dependent distinct ML kinetics in the Mn<sup>2+</sup> dopped SrZnOS at different rates and various temperatures. The optical fluorescent, PMT and imaging measurements present a diffuse-like emission with broad peaks in the ML intensity as a function of time (or pressure) at compression rate below ~1.2 GPa/s, and resonant ML emission character with the distinguishable oscillatory peaks at critical compression rate between 1.2 GPa/s and 1.5 GPa/s. The oscillatory ML emission is gradually suppressed with increasing compression rates above ~5 GPa/s, where the number of the emission peaks and total ML intensities decrease. It could be counted for the multi-cycles of the PIE and self-recoverable processes that the luminescent ions undergo during the compression process when the compression time is comparable to the intrinsic time scales of the piezoelectrically-induced excitation and light emission. The rate dependence of the characteristic time in the ML kinetics indicates the limit of the ML light emission in the dynamic response to the rapid compression. The present work offers new technical methods and notions for the study of the ML in the GPa pressure range, and uncovers rate-dependent distinctive kinetics of the ML emission, presenting new perspectives for understanding the mechanoluminescent mechanism that facilitates the design and development of high-performance ML materials in materials science, and opening new research direction for high-pressure science.
+
+# Methods
+
+## Synthesis of the SrZnOS: Mn²⁺ and ambient characterizations
+
+The SrZnOS: Mn²⁺ sample was synthesized through high-temperature solid-state reactions. The pristine powders of ZnS (99.99%), SrCO₃ (99.99%), and MnCO₃ (99.99%) were first weighed, mixed, and ground in an agate mortar to create stoichiometric mixtures. Subsequently, these mixtures were sintered in a 1323 K alumina crucible under an argon atmosphere for 8 hours in a horizontal tubular furnace. To make a uniform phase, the heating process was repeated twice with an intermediate grinding step. After the completion of the heating process, the samples in the horizontal tubular furnace were allowed to cool down to room temperature. The phase purity was confirmed by X-ray diffraction (XRD) at room temperature and ambient pressure using a PANalytical Empyrean diffraction meter with Cu Kα (40 kV, 40 mA) radiation (λ = 1.5418 Å). The morphology of samples was measured using a Quanta 250 FEG FEI scanning electron microscope (SEM).
+
+## Preparation of High-Pressure Samples
+
+A symmetric DAC with a pair of 400 µm-diameter culet-sized diamond anvils were used for all *in-situ* measurements under high pressures⁴⁸–⁴⁹. A steel gasket (T301) was pre-indented to 30–40 µm in thickness and a 200µm-diameter hole was laser-drilled to serve as the sample chamber. The SrZnOS: Mn²⁺ powder sample were loaded into the chamber together with a small ruby ball for the pressure determination with the ruby fluorescence method⁵⁰. Silicone oil was used as pressure-transmitting medium for high-pressure X-ray diffraction (XRD), Raman and PL measurements. ML experiments were conducted without pressure medium.
+
+## *In situ* High-Pressure structural characterizations
+
+The *in-situ* high pressure X-ray diffractions experiments were conducted at the beamline BL15U1 of Shanghai Synchrotron Radiation Facility (SSRF) with the wavelength of 0.6199 Å and beam spot size of 15×9 µm². Two-dimensional diffraction images were collected continuously during the compression process with a PILATUS 2M detector, and CeO₂ was used for standard calibration. The diffraction patterns were integrated into a one-dimensional profile by using the Dioptas software⁵¹. Structure refinements were carried out using the Rietveld method to obtain the lattice parameters using General Structure Analysis software (GSAS)⁵². High-pressure Raman spectra were recorded on a Renishaw Raman microscope using a 532 nm laser. The system was calibrated by the Raman signal of Si, and spectra were collected in the range of 100–1000 cm⁻¹.
+
+## *In situ* high-pressure optical measurements
+
+The *in-situ* high-pressure PL spectra and lifetime measurements were recorded using a home-designed spectroscopy system (Ideaoptics, Shanghai, China). The optical photographs of ML were captured by a high-speed camera (G516-C, Revealer). Here, a home-made time-resolved optical system, combined with dynamic diamond anvil cells (dDAC) and heat system⁵³, has been developed to collect the ML signal on different loading rates. It mainly consists of the fast fluorescence spectroscopy (Zyla 4.2 Plus, Andor) and photomultiplier tube (C8855, Hamamatsu Photonics) on microsecond time scale. Thermoluminescence experiments were performed using a TL spectrometer (LTTL-3DS, RongFan, Guangzhou, China) in the temperature range from 298 to 773 K at the heating rate of 5 K/s.
+
+# References
+
+1. Zhang JC et al (2019) Trap-controlled mechanoluminescent materials. Prog Mater Sci 103:678–742
+2. Chandra BP, Shrivastava KK (1978) Dependence of mechanoluminescence in rochelle-salt crystals on the charge-produced during their fracture. J Phys Chem Solids 39:939–940
+3. Walton AJ, Triboluminescence (2006) Adv Phys 26:887–948
+4. Chandra BP, Rathore AS (1995) Classif Mechanoluminescence Cryst Res Technol 30:885–896
+5. Wong MC et al (2017) Temporal and Remote Tuning of Piezophotonic-Effect-Induced Luminescence and Color Gamut via Modulating Magnetic Field. Adv Mater 29:1701945
+6. Zhao Y et al (2021) Stimuli responsive lanthanide ions doped layered piezophotonic microcrystals for optical multifunctional sensing applications. Nano Energy 87:106177
+7. Zhang Y et al (2020) Remote Regulation of Optogenetic Proteins by a Magneto-Luminescence Microdevice. Adv Funct Mater 31:2006357
+8. Li C et al (2022) Highly robust and soft biohybrid mechanoluminescence for optical signaling and illumination. Nat Commun 13:3914
+9. Jeong SM et al (2014) Bright, wind-driven white mechanoluminescence from zinc sulphide microparticles embedded in a polydimethylsiloxane elastomer. Energy Environ Sci 7:3338–3346
+10. Xu CN et al (1999) Artificial skin to sense mechanical stress by visible light emission. Appl Phys Lett 74:1236–1238
+11. Tang Y et al (2024) Dynamic multicolor emissions of multimodal phosphors by Mn²⁺ trace doping in self-activated CaGa₄O₇. Nat Commun 15:3209
+12. Zhang J-C et al (2015) Color Manipulation of Intense Multiluminescence from CaZnOS:Mn²⁺ by Mn²⁺ Concentration Effect. Chem Mater 27:7481–7489
+13. Liang L et al (2023) Controlling persistent luminescence in nanocrystalline phosphors. Nat Mater 22:289–304
+14. Zhuang Y, Xie R-J (2021) Mechanoluminescence rebrightening the prospects of stress sensing: a review. Adv Mater 33:2005925
+15. Wang X et al (2015) Dynamic pressure mapping of personalized handwriting by a flexible sensor matrix based on the mechanoluminescence process. Adv Mater 27:2324–2331
+16. Wang X et al (2019) Piezophotonic effect based on mechanoluminescent materials for advanced flexible optoelectronic applications. Nano Energy 55:389–400
+17. Chandra VK, Chandra BP, Jha P (2013) Self-recovery of mechanoluminescence in ZnS:Cu and ZnS:Mn phosphors by trapping of drifting charge carriers. Appl Phys Lett 103:161113
+18. Li W et al (2023) Unraveling Mechanoluminescent Mechanisms in Doped CaZnOS Materials: Co-Mediation of Trap‐Controlled and Non‐Trap‐Controlled Processes. Adv Funct Mater 33:2305482
+19. Chandra VK, Chandra BP, Jha P (2014) Microscopic theory of elastico-mechanoluminescent smart materials. Appl Phys Lett 104:031102
+20. Chandra BP et al (2010) Luminescence induced by elastic deformation of ZnS:Mn nanoparticles. J Lumin 130:442–450
+21. Joos JJ et al (2020) Identification of Dy³⁺/Dy²⁺ as Electron Trap in Persistent Phosphors. Phys Rev Lett 125:033001
+22. Wang X et al (2015) Recent Progress in Electronic Skin. Adv Sci 2:1500169
+23. Wang W et al (2024) Contact-separation-induced self-recoverable mechanoluminescence of CaF₂:Tb³⁺/PDMS elastomer. Nat Commun 15:2014
+24. Huang BL, Sun MZ, Peng DF (2018) Intrinsic energy conversions for photon-generation in piezo-phototronic materials: A case study on alkaline niobates. Nano Energy 47:150–171
+25. Hernández I, Rodríguez F, Hochheimer HD (2007) Pressure-Induced Two-Color Photoluminescence in MnF₂ at Room Temperature. Phys Rev Lett 99:027403
+26. Zhao Y et al (2020) Tricolor Ho³⁺ Photoluminescence Enhancement from Site Symmetry Breakdown in Pyrochlore Ho₂Sn₂O₇ after Pressure Treatment. Phys Rev Lett 125:245701
+27. Chen M et al (2021) Pressure-induced robust emission in a zero-dimensional hybrid metal halide (C₉NH₂₀)₆Pb₃Br₁₂. Matter Radiat Extremes 6:058401
+28. Yang Y et al (2021) Time-Resolved Bright Red to Cyan Color Tunable Mechanoluminescence from CaZnOS: Bi³⁺, Mn²⁺ for Anti-Counterfeiting Device and Stress Sensor. Adv Opt Mater 9:2100668
+29. Zhou T et al (2023) Unrevealing Temporal Mechanoluminescence Behaviors at High Frequency via Piezoelectric Actuation. Small 19:2207089
+30. Zhou S et al (2022) Design of Ratiometric Dual-Emitting Mechanoluminescence: Lanthanide/Transition-Metal Combination Strategy. Laser Photonics Rev 16:2100666
+31. Chandra BP, Zink JI (1980) Triboluminescence and the dynamics of crystal fracture. Phys Rev B 21:816–826
+32. Chandra VK, Chandra BP, Jha P (2013) Models for intrinsic and extrinsic elastico and plastico-mechanoluminescence of solids. J Lumin 138:267–280
+33. Chandra VK, Chandra BP, Jha P (2015) Modelling of fracto-mechanoluminescence damage sensor for structures. Sens Actuator A: Phys 230:83–93
+34. Chandra VK, Chandra BP, Jha P (2015) Piezoelectrically-induced trap-depth reduction model of elastico-mechanoluminescent materials. Phys B 461:38–48
+35. Wang H et al (2023) Pressure- and Rate-Dependent Mechanoluminescence with Maximized Efficiency and Tunable Wavelength in ZnS: Mn²⁺, Eu³⁺. ACS Appl Mater Interfaces 15:28204–28214
+36. Zhang L et al (2021) Unraveling the anomalous mechanoluminescence intensity change and pressure-induced red-shift for manganese-doped zinc sulfide. Nano Energy 85:106005
+37. Liu W et al (2017) Synthesis and characterization of sulfide oxide SrZnSO with strongly polar crystal structure. J Solid State Chem 246:225–229
+38. Wiśniewski K et al (2010) High pressure spectroscopy study of SCF Tb₃Al₅O₁₂:Mn. J Phys Conf Ser 249:012015
+39. Rodríguez-Lazcano Y, Nataf L, Rodríguez F (2009) Electronic structure and luminescence of [(CH₃)₄N]₂MnX₄ (X = Cl,Br) crystals at high pressures by time-resolved spectroscopy: Pressure effects on the Mn-Mn exchange coupling. Phys Rev B 80:085115
+40. Kaminska A et al (2012) Pressure-induced luminescence of cerium-doped gadolinium gallium garnet crystal. Phys Rev B 85:155111
+41. Zhao Y et al (2021) Multiresponsive Emissions in Luminescent Ions Doped Quaternary Piezophotonic Materials for Mechanical-to-Optical Energy Conversion and Sensing Applications. Adv Funct Mater 31:2010265
+42. Medlin WL (1961) Decay of Phosphorescence from a Distribution of Trapping Levels. Phys Rev 123:502–509
+43. Van den Eeckhout K et al (2013) Revealing trap depth distributions in persistent phosphors. Phys Rev B 87:045126
+44. Chen C et al (2023) Revealing the Intrinsic Decay of Mechanoluminescence for Achieving Ultrafast-Response Stress Sensing. Adv Funct Mater 33:2304917
+45. Lin C et al (2020) Temperature- and Rate-Dependent Pathways in Formation of Metastable Silicon Phases under Rapid Decompression. Phys Rev Lett 125:155702
+46. Zhu T et al (2008) Temperature and Strain-Rate Dependence of Surface Dislocation Nucleation. Phys Rev Lett 100:025502
+47. Bai Z, Fan Y (2018) Abnormal Strain Rate Sensitivity Driven by a Unit Dislocation-Obstacle Interaction in bcc Fe. Phys Rev Lett 120:125504
+48. Xu J-A, Mao H-K, Bell P-M (1986) High-Pressure Ruby and Diamond Fluorescence: Observations at 0.21 to 0.55 Terapascal. Science 232:1404–1406
+49. Dubrovinsky L et al (2012) Implementation of micro-ball nanodiamond anvils for high-pressure studies above 6 Mbar. Nat Commun 3:1163
+50. Mao H-K, Xu J, Bell P-M (1986) Calibration of the ruby pressure gauge to 800 kbar under quasi-hydrostatic conditions. J Geophys Res 91:4673–4676
+51. Prescher C, Prakapenka VB (2015) DIOPTAS: a program for reduction of two-dimensional X-ray diffraction data and data exploration. High Press Res 35:223–230
+52. Toby BH (2001) EXPGUI, a graphical user interface for GSAS. *J. Appl. Crystallogr.* 34, 210 – 213
+53. Sinogeikin SV (2015) Online remote control systems for static and dynamic compression and decompression using diamond anvil cells. Rev Sci Instrum 86:072209
+
+# Supplementary Files
+
+- [MovieS1SrZnOS.mp4](https://assets-eu.researchsquare.com/files/rs-4729676/v1/d1863d8d98634c80a61bb05c.mp4)  
+  Movie S1
+
+- [MovieS2SZOS0.2X.mp4](https://assets-eu.researchsquare.com/files/rs-4729676/v1/9cd1be7aec109b175a163bbf.mp4)  
+  Movie S2
+
+- [MovieS3SZOS0.5X.mp4](https://assets-eu.researchsquare.com/files/rs-4729676/v1/5415172c780f251bcb89e647.mp4)  
+  Movie S3
+
+- [MovieS4SZOS0.8X.mp4](https://assets-eu.researchsquare.com/files/rs-4729676/v1/578add2f7e7687bd8c681e33.mp4)  
+  Movie S4
+
+- [MovieS5SZOS2X.mp4](https://assets-eu.researchsquare.com/files/rs-4729676/v1/b12266f8ae8272649723fb5f.mp4)  
+  Movie S5
+
+- [MovieS6ZnS.mp4](https://assets-eu.researchsquare.com/files/rs-4729676/v1/bc92892ad1b6eb77ea589c77.mp4)  
+  Movie S6
+
+- [MovieS7CaZnOS.mp4](https://assets-eu.researchsquare.com/files/rs-4729676/v1/2da15e530c182a839ba609f8.mp4)  
+  Movie S7
+
+- [SupplementaryInformationofSrZnOSNC20240709.docx](https://assets-eu.researchsquare.com/files/rs-4729676/v1/1ede9a5d95d75c42dc3f4a8c.docx)

0afc5fd8bc20909da9586f52f7ba74888c5d3809eb3782695b4832adf305cb35/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Reaction scheme of PsG3Ox, phylogenetic analysis of P2Oxs/G3Oxs and PsG3Ox C-deglycosilation gene cluster. (a) The reaction catalyzed by PsG3Ox to convert D-Glc to 2-keto-D-Glc 20 and the conversion of Mang to 3-keto-Mang. (b) Maximum likelihood phylogenetic relationship between characterized P2Oxs; cholesterol oxidase sequence, ChOx, from Streptomyces sp. was used as outgroup. The blue cluster represent the bacterial enzymes that shows higher specificity for glycosides and the green cluster group the dimeric bacterial enzyme (KaP2Ox) and the tetrameric fungal enzymes that display higher specificity towards D-Glc. (c) The gene coding for PsG3Ox is part of a putative C-deglycosylation catabolic pathway. Four proteins with similarity to known C-deglycosylates (CGDs) are encoded in close vicinity to the psg3ox gene in P. siccitolerans 4J27 genome: C1 and C2 display 37 %, and 33 % identity compared to CarC and B1 and B2 which displayed 49 % and 38 % identity compared to CarB present from Microbacterium sp. 5-2b 9.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "The structural fold of bacterial PsG3Ox. (a) The overall structure of PsG3Ox displays the flavin- and substrate-binding domains highlighted in dark red and purple, respectively. The monomer of TmP2Ox (PDB 1TT0, light grey) is superimposed. The solvent-accessible surface of PsG3Ox (b) and of one subunit of TmP2Ox (c) are represented according to the a.d.p values, blue (6 \u00c52) to red (107 \u00c52). (d) Structural elements that surround the FAD cavity in PsG3Ox and delimiting amino acid residues (defined using a 1.4-\u00c5 rolling probe). (e) Comparison of PsG3Ox (residues in green) and TmP2Ox (PDB 2IGK, residues in purple) flavinylation site. The H bonds are shown as black dashed lines. In all structures, the FAD is shown as sticks in yellow color. (f) Amino acid sequence alignment of bacterial and fungal POxs based on 3D superpositions of the crystal structure. The \u03b1-helices or \u03b2-chains are numbered and colored as in (a). Catalytic residues are highlighted with *. The flavinylation motif and the substrate loop in fungal P2Oxs are marked with light green and orange boxes, respectively. The insertions and deletions regions are highlighted with dashed boxes. The cyan-marked residues correspond to the non-visible regions. Strictly conserved amino acids are represented on black background, whereas dark grey represents the most conserved residues among the selected sequences.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Conformational changes of PsG3Ox upon substrate binding. X-ray structure of (a) substrate-free PsG3Ox, (b) PsG3Ox-Glc, and (c) PsG3Ox-Mang complexes with thickness proportional to a.d.p. values, color-coded from blue (6 \u00c52) to red (107 \u00c52). Regions without electron density are highlighted near the structures. Structural models of (d) substrate-free PsG3Ox, (e) PsG3Ox-Glc, and (f) PsG3Ox-Mang with the non-visible regions in the crystal structure modeled by Rosetta. The insertion-1 and substrate loop (including the modeled segments) are colored dark blue and orange, respectively. Cartoon representation of the active site highlighting the insertion-1 and the conformation of the substrate loop (residues 346-354) in (g) substrate-free PsG3Ox, (h) PsG3Ox-Glc, and (i) PsG3Ox-Mang complexes. The catalytic residues are shown as sticks colored in dark red or blue for the crystal structures or models. The residues in the substrate loop are shown as sticks in grey color and orange for crystal structures and models. The purple triangles represent the interatomic distances between the catalytic pair and the residue P348 of the substrate loop. In all structures, the FAD and the substrates D-Glc and Mang are shown as sticks and colored yellow, green, and cyan, respectively.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Conformational transitions of loops close to PsG3Ox active site. Histograms of the substrate-loop distance to FAD (A352-FADN5) in GaMD simulations run for (a) Model I, which has no substrate and starts with a closed loops conformation, (b) Model II, which contains D-Glc and starts at semi-open conformation; and for (c) Model III*, that starts at open conformation and has no substrate. (d) Projection of the Model III* first trajectory onto the two first principal components obtained by PCA analysis, colored accordingly to A352-FADN5 distance, two main areas of structures are populated along PC1, which roughly correspond to the closed (green) and open (yellow and red) conformations of substrate loop. Histograms of the insertion-1 distance to FAD (G84-FADN5) in GaMD simulations run for (e) Model I, (f) Model II, and (g) Model III*. (h) Projection of the Model III* first trajectory onto the two first principal components obtained by PCA analysis, colored accordingly to G84-FADN5 distance, two main areas of structures are populated along PC1, which roughly correspond to the closed (green) and open (yellow and red) conformations of the insertion 1. In all histograms of GaMD simulations, the black and grey colors represent two distinct simulations of 600 ns. (i) Cartoon representation of the first and last (600 ns) frames of the GaMD simulation of Model III*. The transition from open to closed state of substrate loop and insertion-1 are represented by a black arrow. FAD is colored yellow. (j) root-mean-square fluctuation (RMSF) of Model III* GaMD. Histograms of the (k) substrate-loop and (i) insertion-1 distances to FADN5 in cMD simulation for Model IV that contains Mang and starts with open loops conformations. (m) Visual representation of the final frame (400 ns) of the corresponding simulation, where residues are colored according to hydrophobicity (white to red).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Interactions of PsG3Ox with substrates. \u00a0Binding of D-Glc in (a) PsG3Ox-Glc crystal structure and (b) docking model. Binding of Mang in (c) PsG3Ox-Mang crystal structure and in (d) docking model. The catalytic residues (H440 and N484) are shown as sticks colored in dark-red and the non-catalytic interacting residues are colored in orange. The FAD and the substrates D-Glc and Mang are shown as sticks and colored in yellow, green, and cyan, respectively. The hydrogen bonds are shown as black dashed lines. (e)Fold-change of the catalytic parameters compared with the wild-type PsG3Ox for the alanine mutants at the non-catalytic interacting residues. The catalytic parameters for D-Glc were estimated using the HRP-AAP/DCHBS coupled assay, whereas reactions with Mang were monitored by oxygen consumption in an Oxygraph. All reactions were performed in 100 mM sodium phosphate buffer at pH 7.5 at 37 \u00baC. The kinetic parameters were determined by fitting the data directly on the Michaelis-Menten equation using OriginLab. The catalytic parameters are displayed in Table S6.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

0afc5fd8bc20909da9586f52f7ba74888c5d3809eb3782695b4832adf305cb35/preprint/preprint.md

@@ -0,0 +1,153 @@
+# Abstract
+
+C-glycosides are natural products with important biological activities but are recalcitrant to degradation. Glycoside 3-oxidases (G3Oxs) are newly identified bacterial flavo-oxidases from the glucose-methanol-coline (GMC) superfamily that catalyze the oxidation of C-glycosides with the concomitant reduction of O₂ to H₂O₂. This oxidation is followed by C-C acid/base-assisted bond cleavage in two-step C-deglycosylation pathways. Soil and gut microorganisms have different oxidative enzymes, but the details of their catalytic mechanisms are largely unknown. Here, we report that Ps GO3x oxidizes at 50,000-fold higher specificity (kcat/Km) the glucose moiety of mangiferin to 3-keto-mangiferin than free D-glucose to 2-keto-glucose. Analysis of Ps G3Ox X-ray crystal structures and Ps GO3x in complex with glucose and mangiferin, combined with mutagenesis and molecular dynamics simulations, revealed distinctive features in the topology surrounding the active site that favors catalytically competent conformational states suitable for recognition, stabilization, and oxidation of the glucose moiety of mangiferin. Furthermore, their distinction to pyranose 2-oxidases (P2Oxs) involved in wood decay and recycling is discussed from an evolutionary, structural, and functional viewpoint.
+
+**Structural Biology**  **flavoenzymes**  **carbohydrate oxidases**  **CAZy database**  **C-glycosides**  **enzyme mechanisms**  **substrate specificity**  **structural biology**
+
+# Introduction
+
+C-glycosides represent a large group of natural products in which the anomeric carbon of glucose is directly connected via carbon-carbon bonding to an aglycone moiety (anthraquinone, flavone, terpenoids, phenols, among others). These compounds are secondary metabolites produced by plants and microorganisms and exhibit great structural diversity, wide natural distribution, and significant biological activities, including antioxidant, anti-inflammatory, antibacterial, antiviral, and antitumor<sup>1,2</sup>. C-glycosides have shown increasing importance in the pharmaceutical, agricultural, and food industries, and a great effort has been focused on their synthesis<sup>3,4</sup>. Several C-glycosides like puerarin (daidzein 8-C-β-D-glucoside) have been the precursor of clinical drugs, and biotechnological strategies have been optimized for large-scale production of plant C-glycosides via heterologous expression systems<sup>2,5</sup>.
+
+Gut microbiota that performs bioconversion of C-glycosides to aglycones with beneficial health effects has been identified over the past decade<sup>6–8</sup>. However, very recently, microbial screenings and biochemical studies suggest a ubiquitous process of C-C bond cleavage reactions in nature<sup>9,10</sup>. Compared with other glycosides (O-, N- and S-glycosides), C-glycosides are considerably more stable against chemical and enzymatic treatments. Because of this, C-glycosides are not deglycosylated by glycoside hydrolases, the so-called glycosidases. Instead, their microbial catabolic pathway includes enzymes that catalyze an oxidative step followed by the C-C bond cleavage by a complex of deglycosylation enzymes<sup>9</sup>. Soil and intestinal microorganisms have similar C–C bond-cleaving enzymes<sup>9</sup>. In contrast, the initial oxidation step by C-glycoside-3-oxidases that oxidize the C3 of the sugar moiety is catalyzed by NAD(H) anaerobic oxidoreductase in intestinal microorganisms<sup>6,7,11</sup> and by FAD-dependent oxidoreductases in soil microorganisms<sup>8,10</sup>. These latter enzymes display ~ 60% similarity to pyranose oxidases (POx, pyranose: oxygen 2-oxidoreductase; EC 1.1.3.10) from the glucose-methanol-choline (GMC) superfamily of enzymes<sup>12</sup>.
+
+G3Oxs from soil bacteria such as Microbacterium sp. 5-2b CarA, Arthrobacter globiformis NBRC12137 Ag CarA, and Microbacterium trichothecenolyticum NBRC15077 Mt CarA do not exhibit detectable activity for glucose and instead, oxidize C-glycosides such as carminic acid, mangiferin, and C6-glycosylated flavonoids, as well as O-glycosides but at a significantly lower rate, at C3 to form the corresponding 3-keto C-glycosides<sup>10</sup>. An analogous substrate preference was observed in bacterial Sc P2Ox from Streptomyces canus that displayed 100 to 1000-fold higher enzymatic activity towards the oxidation of C-glycoside puerarin compared to monosaccharide<sup>13</sup>. It was suggested that FAD-dependent G3Oxs from the POx family are ancestors of enzymes that oxidize glucose at the C2 position, the pyranose 2-oxidases (P2Oxs)<sup>10,13</sup>. The fungal P2Oxs, have most likely been acquired by horizontal gene transfer from bacteria and could have evolved and specialized over time to oxidize lignocellulose-derived sugars such as D-glucose, D-xylose, or D-galactose. Fungal P2Oxs are secreted to the extracellular space and are involved in wood decay and recycling. They are the most extensively studied POx, particularly from Trametes multicolor, Peniophora sp, and Phanerochaete chrysosporium<sup>14–16</sup>. They comprise a highly conserved flavin-binding domain with a Rossman-like-fold where FAD covalently binds and a substrate-binding domain. P2Oxs are homotetrameric, and the access to the active site is restricted by four channels that route the substrate from the enzyme surface to the active site cavity<sup>17–19</sup>. The structural characterization of bacterial GO3x, Mt CarA, and Sc P2Ox revealed a few structural and functional aspects of these enzymes<sup>10,13</sup>. However, many fundamental questions remain, particularly the mechanisms behind the diverse substrate specificities.
+
+Here, we investigated the bacterial enzyme As P2Ox from Arthrobacter siccitolerans (now Pseudoarthrobacter siccitolerans<sup>20,21</sup>), renamed for Ps P2Ox, using a combination of experimental and computational approaches that reveal functional and structural details that explain these enzymes ability to bind and oxidize larger glycoside substrates. This work contributed to unveiling the catalytic mechanism of a critical catabolic enzyme involved in the degradation of recalcitrant C-glycosides in nature that remain to be fully disclosed and advancing our understanding of the structure-function relationships among members of the POxs family of enzymes.
+
+# Results And Discussion
+
+**Biochemical and kinetic characterization.** The pseudo-second-order constants measured by transient state kinetics for the reductive and oxidative-half reactions (0.15 ± 0.02 M⁻¹ s⁻¹ and (0.76 ± 0.06) × 10⁶ M⁻¹ s⁻¹) indicate that the rate-limiting step of *Ps*P2Ox is the reductive-half reaction, *i.e.*, the oxidation of D-Glc to 2-keto-D-Glc, similarly to other studied POx (Supplementary Figures 1 and 2)²²,²³. The compound used to measure *Ps*P2Ox enzymatic activity in the coupled assay with horseradish peroxidase (HRP)²⁰, 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), is after oxidation to a green cation radical, reduced by the enzyme to the dicationic form, leading to an underestimation of the enzymatic activity (Supplementary Table 2). Therefore we have optimized the assay using substrates 4-aminophenazone (AAP) and 3,5-dichloro-2-hydroxybenzene-sulfonic acid (DCHBS), which originate after oxidation of a pink chromogen, N-(4-antipyryl)-3-chloro-5-sulfonate-p-benzoquinone-monoimine²⁴, which is not reduced by *Ps*P2Ox (see Supplementary Information). This assay allowed re-estimate the catalytic parameters for monosaccharides D-Glc, D-Xyl, D-Gal, D-Ara, and D-Rib (Table 1 and Supplementary Figure 3). Still, the obtained results confirmed that the enzyme is a poor biocatalyst for the oxidation of monosaccharides. In contrast, and notably, the Kₘ of *Ps*P2Ox for molecular oxygen is 1 to 3 orders of magnitude lower than the described for other POxs (Supplementary Table 4). Following the substrate preference of recently characterized CarA enzymes¹⁰ and *Sc*P2Ox¹³ towards the oxidation of *C*- and *O*-glycosides, the activity of *Ps*P2Ox was tested for mangiferin (Mang), carminic acid, and rutin (Supplementary Figure 4); *Ps*P2Ox is inactive towards carminic acid and rutin; however, Mang is oxidized at a catalytic efficiency (*k*cat/Kₘ) four orders of magnitude higher than D-Glc, displaying a *k*cat around 40-fold higher and a Kₘ 1000-fold lower (Table 1), revealing a striking selectivity for the oxidation of the sugar moiety of the C-glycoside. To identify the product(s) of the reaction, enzymatic reactions with Mang were monitored by TLC (Supplementary Figure 5) and characterized using a combination of 1D and 2D-NMR (Supplementary Figures 6, 7, and 8 and Supplementary Table 5). The NMR data showed that oxidation must occur at the 3-OH, affording the 3-keto mangiferin (Fig 1a). Similarly to Mang¹ H NMR, the anomeric proton of the reaction product was easily identified at 5.17 ppm (Fig. S6). Using this peak as a starting point, COSY NMR (Supplementary Figure 8a) was used to assign neighbor protons: the anomeric H1 correlates to the H2 (4.69 ppm), and the H5 (3.27 ppm) is correlated to H4 (4.14 ppm) and H6 (3.55 ppm). The peaks of H2 and H4 are doublets (Supplementary Table 5 and Supplementary Figure 8a), indicating the absence of the coupling constants with a vicinal H3. The chemical shift values of H2 and H4 also appear to have shifted downfield, as well as their carbons (correlated through HMQC in Supplementary Figure 8b), indicating the existence of a more vicinal electronegative group due to the oxidation at C3. Due to the significantly higher specificity of the enzyme towards the C3 position of Mang compared to the oxidation of the C2 position of D-Glc, we renamed this enzyme from *Ps*P2Ox to *Ps*G3Ox. The phylogenetic analysis of bacterial and fungal POxs (Fig 1b; ¹³) corroborates a clear division between those with a higher specificity towards the oxidation of D-Glc and other monosaccharides at the C2 position, represented by well-characterized fungal P2Oxs (Fig. 1b, green clade), and those with lower specificity for D-Glc and higher specificity towards the oxidation of *C*-glycosides at the C3-position, the bacterial G3Ox (Fig. 1b, blue clade). Interestingly, the bacterial *Ka*P2Ox, the closest bacterial member to fungal counterparts, shows a high specificity for D-Glc²⁵; (Supplementary Table 4). To investigate the presumable role of *Ps*G3Ox in *C*-glycosides catabolism, a Basic Local Alignment Search Tool (BLAST) was performed in the genome of the drought-tolerant *P. siccitolerans* to find putative *C*-deglycosylases in the neighborhood of *Ps*G3Ox coding gene. The *Ps*G3Ox belongs to a gene cluster that has a similar organization to other soil bacteria⁹. Downstream of the *psg3ox* gene, two homologs, *Ps*G3OxC1, and *Ps*G3OxC2, were found showing 37% and 33% identity when compared to CarC enzyme and two other ORFs, *Ps*G3OxB1 and *Ps*G3OxB2, shows 49% and 38% identity to CarB, which is *C*-glycoside deglycosylases (CGDs) that catalyze the cleaving of the C-C bond between the sugar and the aglycone moiety (⁹; Fig 1c). These results hint at the role of *Ps*G3Ox as part of the *C*-deglycosylation catabolic pathway with a putative biological function in, e.g., sugar uptake as a carbon source from natural glycosylated compounds.
+
+**PsG3Ox overall structure.** The crystal structure of *Ps*G3Ox was refined to 2.01 Å resolution, and the crystal structures of *Ps*G3Ox-Glc and *Ps*G3Ox-Mang complexes were refined at 2.35 and 2.60 Å resolutions (Supplementary Table 2). All crystals belong to space group C2221, showed similar cell dimensions, and contained one molecule in the asymmetric unit (a.u.). The overall structure of *Ps*G3Ox is composed of a flavin- and a substrate-binding domain, comparable to other POx structures (Fig. 2a). The flavin domain has a Rossmann fold-like structure of class α/β and contains one non-covalently-bond-FAD molecule. The substrate-binding domain comprises six-stranded central β-sheets (β8-β13) and six α-helices (α2-α6 and α9). *Ps*G3Ox shows an r.m.s.d. value of 0.84 Å compared with homologous Cα positions of the recently characterized bacterial *Mt*CarA (PDB 7DVE)¹⁰. This is in contrast to the significantly higher r.m.s.d. values (1.70-1.81 Å) of structures of P2Oxs from the fungal origin, *T. multicolor* (PDB 1TT0), *Peniophora sp*. (PDB 1TZL) and *P. chrysosporium* (PDB 4MIF). *Ps*G3Ox shows, when compared to the fungal enzymes, i) a significantly more solvent-exposed FAD cavity (Fig. 2b,c), ii) smaller monomers’ size (~500 residues instead of the ~600 residues) (Fig. 2f), iii) a monomeric, instead of the tetrameric fungal state (Fig. 2b,c), and, iv) significantly higher <a.d.p.>’s values (52 Ų) than fungal enzymes (12-30 Ų). The structural model of *Sc*P2Ox predicted using RoseTTaFold¹³ shows a monomeric enzyme and a similar overall fold to *Mt*CarA and *Ps*GO3x.
+
+The structural alignment revealed that *Ps*G3Ox (as well as *Mt*CarA) display nine deletions and four insertions when compared to fungal enzymes (Fig. 2f). Two of the deletions (boxes 1 and 3) correspond to the oligomerization loop and arm regions involved in the fungal P2Ox inter-subunit interactions (Fig. 2a)¹⁷. An additional deletion (boxes 9, 10, and 11) corresponds to a region in fungal enzymes known as the head domain. A shorter insertion 319-321 (box 12 in Fig. 2f) is located near the flavin cofactor, while the insertions 186-190 (box 5) and 506-509 (box 13) are more distant from the active site (Supplementary Figure 9). The major insertion in *Ps*G3Ox is the insertion-1 segment containing 33-residues (box 2), forming two α-helices, α2 (60-71) and α3 (83-88), together with two loops (72-82 and 89-93), which are close to the substrate-binding domain and in the neighboring of the FAD access (Fig. 2d). Interestingly, insertion-1 is located in an equivalent structural position to the oligomerization domains of fungal enzymes (Supplementary Figure 10a), i.e., on the interaction interfaces; mainly, the region 60-70 of *Ps*G3Ox is close to the fungal oligomerization loop, while region 71-93 (the most flexible part of insertion-1, from now on referred as the insertion-1 loop) is located near the fungal oligomerization arm domain. Insertion-1 is also present in bacterial homologs (Supplementary Figure 10b-f), except for the dimeric bacterial *Ka*P2Ox, which contains two oligomerization domains comparable to those in fungal enzymes (Supplementary Figure 10g). The segments that differentiate monomers from oligomers tend to be located on the interaction interfaces, where they mediate or disrupt oligomerization and are usually loops²⁶. It has been claimed that homooligomerization in glycosyltransferases and other proteins might be crucial for their function²⁷. We speculate that the evolutionary mechanism of POxs homooligomerization can hypothetically occur through the deletion and insertion of segments in the region where insertion-1 locates, promoting the stabilization of dimers and tetramers²⁸.
+
+**Catalytic center in PsG3Ox.** The FAD cofactor in *Ps*G3Ox is non-covalently bonded, similar to other characterized bacterial enzymes, except *Ka*P2Ox, and in contrast to fungal enzymes where FAD is covalently linked through its 8α-methyl group, *e.g.*, H167 NE2 in *Tm*P2Ox (PDB 2IGK). *Ps*G3Ox’s equivalent residue, H127, is approximately 6.5 Å away from the FAD C8M atom (Fig. 2e). *In vitro* deflavinylation (Supplementary Figure 11) followed by FAD incorporation, revealed that Apo-*Ps*G3Ox binds exogenous FAD with an estimated dissociation constant Kₐ = (2.0 ± 0.7) × 10⁻⁷ M restoring the enzymes’ total activity (Supplementary Figure 12). The estimated Kₐ value indicates a high affinity for FAD, comparable to other flavoenzymes (for more details, see Supplementary Information)²⁹⁻³¹. The access to the FAD cavity is made through a cavity that contains (i) residues 125 AAHW 128 that is at the same structural position of fungal flavinylation motif, 165 STHW 168 in *Tm*P2Ox, that covalently binds FAD, (ii) the substrate loop (346 ASPVPLADD 354), which in fungal enzymes, is reportedly a dynamic gating segment that fine-tunes the enzymes’ reactivity¹⁵,³², and (iii) the insertion-1 segment (60-93 residues) (Fig. 2d).
+
+We have attempted to trigger a covalent attachment of FAD to H127 in *Ps*G3Ox by replacing residues A125 and A126 with serine and threonine to mimic the fungal flavinylation motif (Fig. 2f). The single variants A125S and A126T still show non-covalently bonded FAD (Supplementary Figure 13) and 2- to 3-fold decreased catalytic efficiency than wild-type (Supplementary Table 6). The double variant S125A-T126A was produced in the apo-form. Considering that the N120 side-chain interacts through a hydrogen bond with H127, which may preclude the covalent binding to FAD (Fig. 2e), site-directed mutagenesis was used to construct N120V, since a valine is present in the fungal *Tm*P2Ox at that position. Still, this replacement also impaired FAD incorporation leading to the production of an apo-enzyme. The site-saturation mutagenesis of N120 resulted in two active variants, N120C and N120Y, displaying a FAD non-covalently bond (Supplementary Figure 13). The replacement of conserved H440 and N484 residues by alanine resulted in inactive enzymes, confirming their key catalytic role (Supplementary Table 6); H440 is expected to act as a proton acceptor for the C2-OH group of sugar substrates, with the support of N484, that stabilizes the protonated intermediate through a hydrogen bond, after the hydride transfer of C2 hydrogen to the FAD N5 atom³²,³³. *Ps*G3Ox’s substrate loop, 346 ASPVPLADD 354, has three polar and six hydrophobic residues and shows comparable structural flexibility to the fungal *Tm*P2Ox substrate loop (452 DAFSYGAVQ 460) (Supplementary Table 7). In the primary sequence alignment, this region is very well conserved among bacterial enzymes (Supplementary Figure 10h), except in the case of *Ka*P2Ox, which exhibits an amino acid composition (369 DAFHYGDVP 377) comparable to the fungal enzymes. To assess the role of the loop in the catalytic properties of *Ps*G3Ox, a substrate loop truncated variant (∆loop 345-359) was constructed and characterized. This variant shows a comparable catalytic efficiency (*k*cat/Kₘ) for D-Glc and Mang to the full-length protein, even though the productive binding is negatively affected: whereas the affinity for both substrates increases, the turnover number (*k*cat) is 3 to 5-fold lower (Supplementary Table 6). The insertion-1 segment in *Ps*G3Ox lowers the substrate loop and active site exposure to solvent (Supplementary Table 8). Deleting the insertion-1 loop between residues 73-93 (its most flexible region) resulted in an inactive, FAD-depleted variant. After *in vitro* flavinylation, a fully loaded FAD enzyme was obtained, but enzymatic activity was not recovered (Table 1). These results suggest that insertion-1 is involved in FAD incorporation and proper stabilization at the active site and plays a crucial role in catalysis.
+
+**Conformational loop changes upon substrate binding and catalysis.** The structures of complexes *Ps*G3Ox-Glc and *Ps*G3Ox-Mang are similar to the native structure with r.m.s.d. values of 0.40 Å and 0.42 Å. (Fig. 3a-c; Supplementary Figure 14a) but display significantly (almost 2-fold) higher FAD cavity volumes (Supplementary Table 9). Furthermore, these structures show higher a.d.p.s values and lack visible electron density maps of residues close to the insertion-1 and substrate loop regions (Supplementary Table 10; Fig. 3b,c; Supplementary Figure 14c,d). The conformation of these non-visible regions was modeled using Rosetta, and the loop candidates were scored based on the lowest possible Rosetta energies (Fig. 3d-f). In the *Ps*G3Ox substrate-free, similarly to *Mt*CarA, the substrate loop displays a closed conformation, and the active site has a smaller dimension and higher content of hydrophobic residues (Fig. 3d,g). In *Ps*G3Ox-Glc and *Ps*G3Ox-Mang structures, the substrate loop adopts a semi-open and open conformation, respectively. The insertion-1 loop, accordingly to the Rosetta best models, also displays a closed and an open conformation in the substrate-free structure and substrate-complexes, respectively (Fig. 3h-i). In fungal P2Oxs, the substrate loop follows open-to-closed conformational transitions that discriminate between electron-donor and electron-acceptor substrates and probe the regioselective oxidation at the C2 or C3 position of monosaccharides (Supplementary Figure 15)³⁴⁻³⁶. Therefore, to provide additional insights into the conformational landscape of the substrate loop and insertion-1 loop along the catalytic cycle of *Ps*G3Ox conventional MDs (Fig. 4k-m; Supplementary Figure 16-18) and GaMDs (Fig. 4a-j; Supplementary Figure 19-21) simulations were performed. The A352-FAD N5 and G84-FAD N5 distances were used to discriminate between closed (≤ 20 Å) and open (≥ 25 Å) conformations of the loops. For the substrate-free Model I, the substrate loop and the insertion-1 loop remain closed during simulations (Supplementary Figure 18), even when an enhanced sampling is pursued with the GaMD simulations (Fig. 4a,e; Supplementary Figure 21). This observation suggests a clear preference for a closed active site in the absence of substrates, in agreement with crystallographic data. The hydrophobic composition of the substrate loop hypothetically favors this conformation. Contrary, when D-Glc is bound in the active site, the simulations suggest that the semi-open substrate loop conformation (with closed insertion 1, Model II) is the only one capable of retaining D-Glc properly bound, with stable interactions with catalytic and non-catalytic residues (Fig. 4b,f, and Supplementary Figures 16-22); in a semi-open substrate loop conformation there is more available space in the A346-P348 zone to accommodate D-Glc (purple triangle in Fig. 3h). In the first Model II simulation, the A352-FAD N5 and G84-FAD N5 distances remain ~15-20 Å (Supplementary Figures 18 and 21). However, in the second GaMD simulation, the loops adopt a more open form (distances A352-FAD N5 > 20 Å and G84-FAD N5 ~ 20 Å, which may relate to D-Glc movement inside the binding cavity, losing some of the required catalytic interactions and finally leaving the active site (Supplementary Figures 21 and 22). Supporting the idea that the open loop conformation cannot properly retain D-Glc, in Model III simulations (open loops, with initial A352-FAD N5 and G84-FAD N5 distances ~ 30 Å), D-Glc goes away from the active site after 300 ns in cMD and before 200 ns in GaMD (Supplementary Figure 22). Interestingly, when long GaMD simulations are run starting from open loops conformations in Model III*, without D-Glc in the active site, substrate loop, and insertion-1 loop follow open to closed transition (Fig. 4c,g; Supplementary Figures 19 and 21). This agrees with the preference mentioned above for closed states in substrate-free *Ps*G3Ox. The open-to-closed transitions (Fig 4i) are also clear by the main movements (PC1) of the protein in this trajectory, as seen in the Principal Component Analysis (PCA) (Fig. 4d,h). Significant higher r.m.s.f. values were found for insertion-1 and substrate loop that started with open conformations (Fig. 4j). Other simulations, e.g., starting from the closed models, showed much lower fluctuations (Supplementary Figures 17 and 20), corroborating the higher stability of the enzyme in a closed state. Overall, the simulations performed with D-Glc suggest that the substrate loop adopts a narrow range of conformations that allow effective oxidation of this substrate, affecting productive binding and explaining the poor catalytic parameters obtained for D-Glc. *In silico* molecular dynamics using Mang as substrate (Model IV) indicates that the substrate and insertion-1 loops adopt an open conformation (Fig. 4k,l) which is required to enlarge the active site pocket and allow bulkier substrates such as glycosides to bind (Fig. 3f). During the 400 ns of MD simulations of Model IV (open loops), both loops remain open (distance A352-FAD N5 > 30 Å and G84-FAD N5 > 20-25 Å) (Fig. 4k,l; Supplementary Figure 18), and Mang remains correctly oriented for ~280 ns, suggesting a stable complex. However, at the end of the simulation, Mang leaves its oxidation position in the active site. Interestingly, it moves towards the substrate loop, where the mangiferin’s aglycon motif is accommodated between the loop’s hydrophobic residues P348 and V349 and the residue F431 (Fig. 4m). This suggests that the aglycon motif of substrates such as Mang can interact with the substrate loop during substrate recruitment and/or product release. The dynamic transitions of the substrate loop and insertion-1 loop hint at a joint function of these two regions, not only in the access to the active site but also in the proper accommodation and oxidation of substrates.
+
+**Enzyme-substrate interactions.** In the crystal structure of the *Ps*G3Ox-Glc complex, D-Glc O2 is located at 4.2 and 4.7 Å of the catalytic residues H440 NE2 and N484 ND2, and at 4.2 Å from FAD N5 (Fig. 5a). The hydrogen atom at the C2 atom of D-Glc points to the flavin N5 atom, which might facilitate a hydride transfer and support D-Glc oxidation at the C2 position (Mendes et al., 2016). In fungal P2Oxs, D-Glc establishes shorter distances (2.5-3.0 Å) to the FAD N5 and catalytic residues³⁴,³⁷. Molecular docking of D-Glc into *Ps*G3Ox revealed a wider network of interactions than those observed in the crystal structure (Fig. 5a,b; Supplementary Figures 23, 24, and 25). The binding of D-Glc positioned for C2, and C3 oxidation (based on the corresponding catalytic distances) was abundant (Supplementary Figure 23). Remarkably, in the semi-open system, a higher C2/C3 ratio was obtained, suggesting a preference for oxidation of D-Glc at the C2 oxidation (Supplementary Figure 23). This observation supports that a semi-open substrate loop conformation is essential to bring and retain D-Glc close to the FAD and catalytic residues and properly orient D-Glc for C2 oxidation. In the *Ps*G3Ox-Mang crystal complex (and dockings), more hydrogen bonds were observed, which may explain the higher affinity and catalytic efficiency of *Ps*G3Ox for this substrate compared with D-Glc (Fig 5c,d; Supplementary Figures 26 and 27). Notably, in both experimental and computational dockings, a clear preference for the positioning of Mang oriented towards the C3 oxidation was observed in contrast to the C2 positioning followed for D-Glc (Fig. 5c,d; Supplementary Figure 26). Computational dockings were performed to explore the structural reasons behind the lack of activity toward carminic acid and rutin (Table 1; Supplementary Figure 28a). Carminic acid binds in a non-catalytically competent manner: H2/C2 is relatively well oriented towards FAD N5, but it is the OH3 that interacts with catalytic H440; no interaction with the catalytic residue N484 was observed. In the case of rutin, the dockings predict binding in an orientation compatible with catalysis (Supplementary Figure 28b). Inhibition assays show a 5-fold lower inhibition constant (Kᵢ) for rutin (0.1 mM) as compared to carminic acid (Supplementary Figure 29), suggesting that rutin can bind more strongly to the active site, supporting the dockings’ data; FAD reduction was not observed in enzymatic assays performed in anaerobic conditions in the presence of rutin (data not shown). Therefore we hypothesize that the absence of activity with rutin might be associated with its intrinsic properties, e.g., a hypothetic redox potential that impairs electron transfer. However, this possibility was not investigated here.
+
+Alanine mutagenesis was used to investigate the importance of residues K55, R94, T129, Q297, and Q340 in substrate binding and catalysis (Fig. 5e; Supplementary Table 6). In general, the replacement to Ala resulted, with a few exceptions, in decreased *k*cat and comparable Kₘ values to the wild type. The absence of Q297 and, particularly, of Q340 has a significant detrimental effect in catalysis for both substrates, even if slightly higher for D-Glc. On the other hand, the oxidation of Mang is particularly affected in R94A and T129A variants, whereas the oxidation of D-Glc is primarily unaffected. These two variants, along with K55A and Q340A, display higher Kₘ values for Mang and are thus expected to be involved in Mang's binding and productive pose in the active site. K55A shows a similar *k*cat for Mang following its furthest location from the FAD and catalytic residues in the docking simulation models (Fig. 5d), contrary to the complex crystal structure where K55 is located nearby the aglycon motif of the mangiferin (Fig. 5b). The results obtained with R94 and Q297 are by the MDs that suggest a role in anchoring bulky substrates. In *Ps*G3Ox, the residue T129 establishes hydrogen bonds with P348 N and FAD N5 (Supplementary Figure 14), similar to residue T169 in *Tm*P2Ox (Supplementary Figure 15); however, whereas T169 was suggested to trigger the transition between the different conformations of the substrate loop³⁴⁻³⁷, T129 shows the same configuration independently of the substrate loop conformation (Supplementary Figure 14).
+
+# Concluding Remarks
+
+This work establishes the distinct structural, functional, and mechanistic features between two groups within the POx family of enzymes. These enzymes are known to have a clear preference for the oxidation of D-Glc, at the C2 or C3 positions, yielding the respective ketoaldoses as products. However, it is clear the existence of two distinct phylogenetic groups with different activity profiles in this family: those with a higher specificity towards the oxidation of the monosaccharide D-Glc, the P2Oxs, and those with a higher specificity towards the oxidation of the D-Glc moiety of C-glycosides, the G3Ox group. These latter are primarily inactive towards D-Glc oxidation. At the same time, the first group is most likely inactive to the bulkier C-glycosides, considering their overall tetrameric fold with buried active sites, in contrast with monomeric G3Ox with a more accessible active site. This work revealed that insertion-1, a striking structural segment of G3Ox’s active site, plays a vital role in catalysis. Insertion-1 locates in bacterial GO3xs at the same region where the oligomerization loop and domain locates in dimeric and tetrameric P2Ox. Therefore we speculate that structural changes in these (non-conserved) regions may contribute to the formation of different functional oligomeric states that might be important in the regulation of enzyme activity; i.e., we propose that the catalytic mechanisms and binding specificities within the POx family of enzymes are modulated through different homooligomerization states. All POx members show, at their active site, a substrate loop that plays a prominent role in regioselectivity control by modulating its conformation depending on the type/size of the substrate. In *Ps* G3Ox and *Mt* CarA, the substrate loop is projected into the active site cavity, sterically occluding the binding of substrates, contrary to P2Oxs fungal enzymes. This propensity to adopt a closed conformation probably relates to its hydrophobic composition and articulated interaction with the insertion-1 segment, which contributes to lowering the substrate loop and active site exposure to solvent. In these enzymes, the dynamic transitions of substrate loop and insertion-1 hint at a joint function in the access to the active site and in allowing proper accommodation of the substrates. In the presence of small substrates such as D-Glc, a semi-open conformation must be adopted to create enough space for substrate binding and orientation towards C2 oxidation while simultaneously avoiding an excessive enlargement of the active-site cavity that would result in the loss of substrate, impairing the catalysis. Most likely, the difficulties in establishing this equilibrium, either by a lack of trigger(s) for the formation of the semi-open state or its maintenance, are reflected in the low specificity of G3Ox to D-Glc. In contrast, an open conformation is mandatory for the binding of Mang. In *Ps* G3Ox, hydrogen-bond interactions between the substrate loop and the substrate were not observed, in opposition to fungal P2Oxs; however, the *Ps* G3Ox substrate loop can create a “hydrophobic clamp” that interacts with the aglycone moiety of mangiferin, suggesting a role of the substrate loop to assist the substrate recruitment or product release. Active-site residues also play a role in anchoring the substrate/product to the enzyme. Overall, a combined experimental and computational investigation of *Ps* G3Ox allowed mapping the relationships between the enzyme sequence-structure-function and elucidating functional transitions that accompany substrate binding and release. Such changes highlight the fine control of access to the catalytic site required by the enzyme mechanism and, in turn, the specificity offered by the enzyme towards different substrates. Work is ongoing to explore the molecular determinants for substrate specificity among bacterial POxs. This undertaking is essential for advancing fundamental biochemical insights in protein science and shedding light on the diversity of microbial catabolic pathways of natural compounds.
+
+# References
+
+1. Hultin, P.G. Bioactive C-glycosides from bacterial secondary metabolism. *Curr Top Med Chem* **5**, 1299-331 (2005).
+2. Zhang, Y.Q., Zhang, M., Wang, Z.L., Qiao, X. & Ye, M. Advances in plant-derived C-glycosides: Phytochemistry, bioactivities, and biotechnological production. *Biotechnol Adv* **60**, 108030 (2022).
+3. Yang, Y. & Yu, B. Recent Advances in the Chemical Synthesis of C-Glycosides. *Chem Rev* **117**, 12281-12356 (2017).
+4. Bililign, T., Griffith, B.R. & Thorson, J.S. Structure, activity, synthesis and biosynthesis of aryl-C-glycosides. *Nat Prod Rep* **22**, 742-760 (2005).
+5. Kytidou, K., Artola, M., Overkleeft, H.S. & Aerts, J.M.F.G. Plant Glycosides and Glycosidases: A Treasure-Trove for Therapeutics. *Front Plant Sci* **11** (2020).
+6. Braune, A., Engst, W. & Blaut, M. Identification and functional expression of genes encoding flavonoid O- and C-glycosidases in intestinal bacteria. *Environ Microbiol* **18**, 2117-2129 (2016).
+7. Nakamura, K., Zhu, S., Komatsu, K., Hattori, M. & Iwashima, M. Deglycosylation of the Isoflavone C-Glucoside Puerarin by a Combination of Two Recombinant Bacterial Enzymes and 3Oxo-Glucose. *Appl Environ Microbiol* **86** (2020).
+8. Kim, E.M., Seo, J.H., Baek, K. & Kim, B.G. Characterization of two-step deglycosylation via oxidation by glycoside oxidoreductase and defining their subfamily. *Sci Rep* **5** (2015).
+9. Mori, T. et al. C-Glycoside metabolism in the gut and in nature: Identification, characterization, structural analyses and distribution of C-C bond-cleaving enzymes. *Nat Commun* **12** (2021).
+10. Kumano, T. et al. FAD-dependent C-glycoside-metabolizing enzymes in microorganisms: Screening, characterization, and crystal structure analysis. *Proc Nat Acad Sci USA* **118** (2021).
+11. Sanugul, K., Akao, T., Nakamura, N. & Hattori, M. Two proteins, Mn2+, and low molecular cofactor are required for C-glucosyl-cleavage of mangiferin. *Biol Pharm Bull* **28**, 2035-2039 (2005).
+12. Sutzl, L., Foley, G., Gillam, E.M.J., Boden, M. & Haltrich, D. The GMC superfamily of oxidoreductases revisited: analysis and evolution of fungal GMC oxidoreductases. *Biotechnol Biofuels* **12** (2019).
+13. Kostelac, A. et al. Biochemical Characterization of Pyranose Oxidase from *Streptomyces canus* -Towards a Better Understanding of Pyranose Oxidase Homologues in Bacteria. *Int J Mol Sci* **23** (2022).
+14. Abrera, A.T., Sutzl, L. & Haltrich, D. Pyranose oxidase: A versatile sugar oxidoreductase for bioelectrochemical applications. *Bioelectrochem* **132** (2020).
+15. Wongnate, T. & Chaiyen, P. The substrate oxidation mechanism of pyranose 2-oxidase and other related enzymes in the glucose-methanol-choline superfamily. *FEBS J* **280**, 3009-27 (2013).
+16. Sutzl, L. et al. Multiplicity of enzymatic functions in the CAZy AA3 family. *Appl Microbiol Biotechnol* **102**, 2477-2492 (2018).
+17. Hallberg, B.M., Leitner, C., Haltrich, D. & Divne, C. Crystal structure of the 270 kDa homotetrameric lignin-degrading enzyme pyranose 2-oxidase. *J Mol Biol* **341**, 781-96 (2004).
+18. Bannwarth, M., Bastian, S., Heckmann-Pohl, D., Giffhorn, F. & Schulz, G.E. Crystal structure of pyranose 2-oxidase from the white-rot fungus *Peniophora* sp. *Biochemistry* **43**, 11683-90 (2004).
+19. Hassan, N. et al. Crystal structures of *Phanerochaete chrysosporium* pyranose 2-oxidase suggest that the N-terminus acts as a propeptide that assists in homotetramer assembly. *FEBS Open Bio* **3**, 496-504 (2013).
+20. Mendes, S. et al. Characterization of a bacterial pyranose 2-oxidase from *Arthrobacter siccitolerans*. *J. Mol. Cat. B: Enzymatic* **133**, 34-43 (2016).
+21. Busse, H.J. Review of the taxonomy of the genus *Arthrobacter*, emendation of the genus *Arthrobacter sensu lato*, proposal to reclassify selected species of the genus *Arthrobacter* in the novel genera *Glutamicibacter* gen. nov., *Paeniglutamicibacter* gen. nov., *Pseudoglutamicibacter* gen. nov., *Paenarthrobacter* gen. nov and *Pseudarthrobacter* gen. nov., and emended description of *Arthrobacter roseus*. *Int J Syst Evol Microbiol* **66**, 9-37 (2016).
+22. Prongjit, M., Sucharitakul, J., Wongnate, T., Haltrich, D. & Chaiyen, P. Kinetic mechanism of pyranose 2-oxidase from *Trametes multicolor*. *Biochemistry* **48**, 4170-80 (2009).
+23. Sucharitakul, J., Wongnate, T. & Chaiyen, P. Kinetic isotope effects on the noncovalent flavin mutant protein of pyranose 2-oxidase reveal insights into the flavin reduction mechanism. *Biochemistry* **49**, 3753-65 (2010).
+24. Fossati, P., Prencipe, L. & Berti, G. Use of 3,5-dichloro-2-hydroxybenzenesulfonic acid/4-aminophenazone chromogenic system in direct enzymic assay of uric acid in serum and urine. *Clin Chem* **26**, 227-31 (1980).
+25. Herzog, P.L. et al. Versatile Oxidase and Dehydrogenase Activities of Bacterial Pyranose 2-Oxidase Facilitate Redox Cycling with Manganese Peroxidase In Vitro. *Appl Environ Microbiol* **85** (2019).
+26. Hashimoto, K., Madej, T., Bryant, S.H. & Panchenko, A.R. Functional states of homooligomers: insights from the evolution of glycosyltransferases. *J Mol Biol* **399**, 196-206 (2010).
+27. Hashimoto, K. & Panchenko, A.R. Mechanisms of protein oligomerization, the critical role of insertions and deletions in maintaining different oligomeric states. *Proc Natl Acad Sci U S A* **107**, 20352-7 (2010).
+28. Kumari, N. & Yadav, S. Modulation of protein oligomerization: An overview. *Prog Biophys Mol Biol* **149**, 99-113 (2019).
+29. Caldinelli, L., Pedotti, M., Motteran, L., Molla, G. & Pollegioni, L. FAD binding in glycine oxidase from Bacillus subtilis. *Biochimie* **91**, 1499-508 (2009).
+30. Fraaije, M.W., van Den Heuvel, R.H., van Berkel, W.J. & Mattevi, A. Structural analysis of flavinylation in vanillyl-alcohol oxidase. *J Biol Chem* **275**, 38654-8 (2000).
+31. Villegas, J.M. et al. FAD binding properties of a cytosolic version of Escherichia coli NADH dehydrogenase-2. *Biochim Biophys Acta* **1844**, 576-84 (2014).
+32. Wongnate, T., Surawatanawong, P., Chuaboon, L., Lawan, N. & Chaiyen, P. The Mechanism of Sugar C-H Bond Oxidation by a Flavoprotein Oxidase Occurs by a Hydride Transfer Before Proton Abstraction. *Chem Eur J* **25**, 4460-4471 (2019).
+33. Wongnate, T., Sucharitakul, J. & Chaiyen, P. Identification of a catalytic base for sugar oxidation in the pyranose 2-oxidase reaction. *Chembiochem* **12**, 2577-86 (2011).
+34. Kujawa, M. et al. Structural basis for substrate binding and regioselective oxidation of monosaccharides at C3 by pyranose 2-oxidase. *J Biol Chem* **281**, 35104-15 (2006).
+35. Tan, T.C., Haltrich, D. & Divne, C. Regioselective control of beta-d-glucose oxidation by pyranose 2-oxidase is intimately coupled to conformational degeneracy. *J Mol Biol* **409**, 588-600 (2011).
+36. Spadiut, O., Tan, T.C., Pisanelli, I., Haltrich, D. & Divne, C. Importance of the gating segment in the substrate-recognition loop of pyranose 2-oxidase. *FEBS J* **277**, 2892-909 (2010).
+37. Tan, T.C. et al. H-bonding and positive charge at the N5/O4 locus are critical for covalent flavin attachment in trametes pyranose 2-oxidase. *J Mol Biol* **402**, 578-94 (2010).
+38. Kabsch, W. Xds. *Acta Crystallogr. D* **66**, 125-132 (2010).
+39. Evans, P.R. An introduction to data reduction: space-group determination, scaling and intensity statistics. *Acta Crystallogr D* **67**, 282-92 (2011).
+40. Evans, P.R. & Murshudov, G.N. How good are my data and what is the resolution? *Acta Crystallogr D* **69**, 1204-14 (2013).
+41. Vonrhein, C. et al. Data processing and analysis with the autoPROC toolbox. *Acta Crystallogr D* **67**, 293-302 (2011).
+42. Kantardjieff, K.A. & Rupp, B. Matthews coefficient probabilities: Improved estimates for unit cell contents of proteins, DNA, and protein-nucleic acid complex crystals. *Prot Sci* **12**, 1865-1871 (2003).
+43. Matthews, B.W. Solvent Content of Protein Crystals. *J. Mol. Biol.* **33**, 491-497 (1968).
+44. Oeffner, R.D., Bunkoczi, G., McCoy, A.J. & Read, R.J. Improved estimates of coordinate error for molecular replacement. *Acta Crystallogr D* **69**, 2209-15 (2013).
+45. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. *Acta Crystallogr D* **60**, 2126-32 (2004).
+46. Blanc, E. et al. Refinement of severely incomplete structures with maximum likelihood in BUSTER-TNT. *Acta Crystallogr D* **60**, 2210-21 (2004).
+47. Roversi, P., Blanc, E., Vonrhein, C., Evans, G. & Bricogne, G. Modelling prior distributions of atoms for macromolecular refinement and completion. *Acta Crystallogr D* **56**, 1316-23 (2000).
+48. Borges, P.T. et al. Unveiling molecular details behind improved activity at neutral to alkaline pH of an engineered DyP-type peroxidase. *Comput Struct Biotechnol J* **20**, 3899-3910 (2022).
+49. Stein, A. & Kortemme, T. Improvements to Robotics-Inspired Conformational Sampling in Rosetta. *Plos One* **8**, e63090 (2013).
+50. Kocourek, J., Ticha, M. & Kostir, J. The use of diphenylamine-alanine-phosphoric acid reagent in the detection and differentiation of monosaccharides and their derivatives on paper chromatograms. *J Chromatogr* **24**, 117-24 (1966).
+51. Krieger, E., Nielsen, J.E., Spronk, C.A.E.M. & Vriend, G. Fast empirical pK(a) prediction by Ewald summation. *J Mol Graph Model* **25**, 481-486 (2006).
+52. Krieger, E. & Vriend, G. New ways to boost molecular dynamics simulations. *J Comput Chem* **36**, 996-1007 (2015).
+53. Hornak, V. et al. Comparison of multiple amber force fields and development of improved protein backbone parameters. *Proteins* **65**, 712-725 (2006).
+54. Mark, P. & Nilsson, L. Structure and dynamics of the TIP3P, SPC, and SPC/E water models at 298 K. *J. Phys. Chem. B* **105**, 9954-9960 (2001).
+55. Miao, Y., Feher, V.A. & McCammon, J.A. Gaussian Accelerated Molecular Dynamics: Unconstrained Enhanced Sampling and Free Energy Calculation. *J Chem Theory Comput* **11**, 3584-3595 (2015).
+56. D.A. Case, H.M.A., K. Belfon, I.Y. Ben-Shalom, J.T. Berryman, S.R. Brozell, D.S. Cerutti, T.E. Cheatham, III, G.A. Cisneros, V.W.D. Cruzeiro, T.A. Darden, R.E. Duke, G. Giambasu, M.K. Gilson, H. Gohlke, A.W. Goetz, and P.A. Kollman AMBER 2020. (University of California, San Francisco., 2020).
+57. Wang, J.N. et al. Gaussian accelerated molecular dynamics: Principles and applications. *Wiley Interdisciplinary Reviews-Computational Molecular Science* **11** (2021).
+58. Canutescu, A.A. & Dunbrack, R.L. Cyclic coordinate descent: A robotics algorithm for protein loop closure. *Prot Sci* **12**, 963-972 (2003).
+59. Trott, O. & Olson, A.J. Software News and Update AutoDock Vina: Improving the Speed and Accuracy of Docking with a New Scoring Function, Efficient Optimization, and Multithreading. *J Comput Chem* **31**, 455-461 (2010).
+60. Land, H. & Humble, M.S. YASARA: A Tool to Obtain Structural Guidance in Biocatalytic Investigations. *Methods Mol Biol* **1685**, 43-67 (2018).
+61. Krieger, E., Darden, T., Nabuurs, S.B., Finkelstein, A. & Vriend, G. Making optimal use of empirical energy functions: Force-field parameterization in crystal space. *Proteins-Structure Function and Bioinformatics* **57**, 678-683 (2004).
+
+# Methods
+
+**Bacterial strains, plasmids, and cultivation media.** *Escherichia coli* strains, plasmids, and primers used in this work are summarized in **Table S1**. *E. coli* strain DH5α (Novagen, Darmstadt, Germany) was used to propagate and amplify plasmid constructs. *E. coli* Rosetta pLysS (DE3, Novagen, Darmstadt, Germany) was used to express the wild-type *PsG3Ox* and its variants cloned in the pET-15b plasmid (Novagen, Darmstadt, Germany). KRX (Promega, Wiscosin, USA) was used as expression strain in the SSM library screenings. Luria Bertani medium (LB) was used for cell cultivation, supplemented with 100 μg ml⁻¹ of ampicillin (NZYTech, Lisbon, Portugal) and, in the case of Rosetta pLysS, also with 20 μg ml⁻¹ of chloramphenicol (NZYTech, Lisbon, Portugal).
+
+**Construction of variants by site-directed mutagenesis.** The Quick-change mutagenesis protocol (Stratagene, California, USA) was used in the construction of single variants, as well as to delete two regions of the enzyme, the insertion-1 loop, between residues 73 and 93, and the substrate-binding loop between residues 345 and 359, in variant Δloop (345-359) (**Table S1**). The plasmid pSM-1, carrying the wild-type *PsG3Ox* gene, was used as a DNA template with appropriate primers. PCRs were performed in a thermal cycler (MyCycler™ thermocycler, Biorad) in 50 μL reaction volume containing 100 ng of DNA template, 1 μM of primers (forward and reverse), and 200 μM of dNTPs (NZYTech, Lisbon, Portugal). 1 U of NZYProof polymerase (NZYTech, Lisbon, Portugal) was used to amplify the DNA, except for pAT27 and pAT28 where 1 U of Q5 High-Fidelity DNA polymerase (New England BioLabs, Massachusetts, USA) was used. For the single and double mutants, after an initial denaturation of 4 min at 95 ºC, 25 cycles of 1 min at 95 ºC, 1.5 min at 72 ºC, and 10 min at 72 ºC were performed, followed by a final elongation of 10 min at 72 ºC. Amplification of truncated variants was performed after an initial denaturation of 30 sec at 98 ºC, 35 cycles of 10 sec at 98 ºC, 30 sec at 72 ºC and 4 min at 72 ºC, followed by a final elongation of 2 min at 72 ºC. For all the PCR products, the DNA template was digested with 10 U of *Dpn* I (ThermoFisher, Massachusetts, USA) at 37 ºC for 6 h, followed by purification using Illustra GFX PCR DNA kit (GE Healthcare, Illinois, USA). In the truncated variants, an overnight ligation with T4 ligase (ThermoFisher, Massachusetts, USA) was performed at room temperature, followed by purification with the abovementioned kit. The PCR products were transformed in *E. coli* strains using electroporation, and the presence of the desired mutation(s) or deletions was confirmed by DNA sequencing.
+
+**Production and purification of *PsG3Ox*.** The recombinant strains *E. coli* Rosetta pLysS carrying the genes coding for wild-type *PsG3Ox* and variants were grown in 2.5 L LB media supplemented with 100 μg ml⁻¹ of ampicillin and 20 μg ml⁻¹ of chloramphenicol in Corning® 5L Baffled PETG Erlenmeyer flasks. The cultures were incubated at 37 °C, 100 rpm (Innova 44 incubator shaker, New Brunswick Scientific). Cultures were induced with 100 μM of isopropyl β-D-1-thiogalactopyranoside (IPTG) at an OD₆₀₀ nm = 0.8, the temperature was lowered to 25 °C, and cells were collected by centrifugation (4420 × g, 10 min, 4°C) after 16 h of cultivation. The purification of wild-type *PsG3Ox* and variants was performed using a Histrap HP column (Cytiva, Massachusetts, USA) as previously described (Mendes et al., 2016). Enzyme preparations for crystallographic trials were purified using a 1 ml-Resource-Q column (Cytiva, Massachusetts, USA) using 20 mM Tris-HCl pH 7.6 as running buffer and a gradient of 0-500 mM of NaCl for elution. Before the enzyme crystallization, the His(6x)-tag was cut from the enzyme using the thrombin cleavage kit (Abcam, Cambridge, UK) following the manufacturer protocol at 20ºC for 16 h. Afterward, the preparation was loaded on the Histrap HP column, and the flowthrough (containing the untagged enzyme) was collected. The total protein concentration was determined by Bradford assay using bovine serum albumin as standard. Abs₄₅₀nm (M⁻¹ cm⁻¹) of purified preparations was measured to assess the functional fraction of enzyme preparations (e.g., for kinetic measurements).
+
+**Crystallization and cryoprotection.** Crystallization conditions were screened with a nanodrop crystallization robot (Cartesian, Genomic Solutions) using the sitting drop vapour diffusion method with round-bottom Greiner 96-well CrystalQuick™ plates (Greiner Bio-One, Kremsmünster, Austria). The Structure Screen I and II (Molecular Dimensions) led to the formation of *PsG3Ox* crystals within seven days at 20 ºC using 2 M ammonium sulfate and 0.1 M Tris-HCl, pH 8.5 in drops of 0.1 μl protein solution (18 mg ml⁻¹) plus 0.1 μl reservoir solution. Following this crystallization hit, microliter-scale crystals optimization proceeded using the hanging drop vapour diffusion method in XRL 24-well crystallization plates with 500 μL of the reservoir solution (Molecular Dimensions, Newmarket, UK). Several conditions were tested: ammonium sulfate concentration ranging from 0.5 to 2M, Tris-HCl, pH 7.0-8.5, and different ratios (1:2, 1:1, and 2:1) of protein: reservoir solution volumes. Yellow and round crystals appeared after 7-10 days, reaching dimensions of 100 µm in their three dimensions when using 2 M ammonium sulfate and 0.1 M Tris-HCl, pH 8.5, and 1 μL of protein (18 mg ml⁻¹) and 2 μL of reservoir solution at 20 °C. Crystals of the *PsG3Ox*-substrate complex were obtained by soaking *PsG3Ox* crystals in 1) the reservoir solution containing 2M D-Glc for 30 min and 2) the reservoir solution containing 1 mM Mang for 1 min. Crystals were cryo-protected by plunging in a reservoir solution supplemented with 20% (v/v) glycerol before flash-cooling in liquid nitrogen.
+
+**Data collection and processing.** Diffraction data were measured in ID23-2 and ID30A-1 beamlines at the European Synchrotron Radiation Facility (ESRF, Grenoble, France) and in XALOC beamline at ALBA (Barcelona, Spain) for *PsG3Ox*, *PsG3Ox*-Glc and *PsG3Ox*-Mang complex crystals, respectively. Diffraction images of *PsG3Ox* were obtained with a DECTRIS PILATUS3 X 2M detector, using 0.8731 Å radiation wavelength, crystal-to-detector distance 232 mm, and oscillations width 0.20º in a total of 360º rotation. The diffraction data of the *PsG3Ox*-Glc complex were obtained with a PILATUS3 2M detector, using 0.9654 Å radiation wavelength, crystal-to-detector distance of 238 mm, and oscillations width of 0.20º in a total of 180º rotation. The diffraction data of the *PsG3Ox*-Mang complex were obtained with a DECTRIS PILATUS 6M detector, radiation wavelength 0.97926 Å, crystal-to-detector distance of 553.95 mm, and oscillations width of 0.20º in a total of 180º rotation. Data were indexed and integrated with XDS³⁸ in a space group determined with POINTLESS³⁹, and the data scaled with AIMLESS³⁹,⁴⁰. These programs were used within the autoPROC data processing pipeline⁴¹. Data collection details and processing statistics are listed in **Table S2**.
+
+**Structure determination, refinement, and analysis.** The three crystals belong to the same space group and show similar cell dimensions. The distribution of their Matthews coefficient⁴²,⁴³ indicated a high probability of a single molecule in their asymmetric units. The phase problem of *PsG3Ox* was solved by molecular replacement using MORDA that selected the coordinates of *P. chrysosporium* POx (*Pc* P2Ox, PDB 4MIF) and *Alkalihalobacillus halodurans* ATP phosphoribosyltransferase regulatory subunit structure (PDB 3OD1) as search models, which led to a 99% probability solution. PHASER, within the PHENIX suite, was used to localize the two *PsG3Ox*-substrates structures using native *PsG3Ox* structure as a search model, which led to TFZ values of 35 and 42, indicating successful structures solutions⁴⁴. Automated model rebuilding and completion were performed with PHENIX.AUTOBUILD followed by manual model building, performed with COOT⁴⁵, and iterative refinement cycles, using PHENIX.REFINE. Structure refinement included atomic coordinates, isotropic atomic displacement parameters (a.d.p.s), and domains of translation, libration, and screw refinement of anisotropic a.d.p.s (TLS), previously defined with TLSMD server (http://skuld.bmsc.washington.edu/~tlsmd). Approximately 1.5% of reflections were randomly excluded from monitoring the refinement strategy. Solvent water molecules were automatically assigned from σA difference maps peaks neighboring hydrogen bonding acceptors/donors within 2.45-3.40 Å distances. Other solvent molecules were identified through a comparison of their shapes against electron density blobs, as well as by comparing their refined a.d.p.s with those of neighboring atoms. Some atoms were modeled with partial occupancies when hinted by the difference in Fourier maps and neighboring a.d.p.s values. As some regions of the *Ps* G3Ox and *Ps* G3Ox-substrate complex structures were not visible in the electron density maps, a BUSTER protocol was applied to search for missing atoms⁴⁶,⁴⁷. The stereochemistry of the refined structures was analyzed with MOLPROBITY. Three-dimensional superposition of polypeptide chains was performed with MODELLER. Figures of structural models were prepared with PyMOL. Refinement statistics are presented in **Table 1**. Structure factors and associated structure coordinates of *Ps* G3Ox, *Ps* G3Ox-Glc, and *Ps* G3Ox-Mang complex were deposited in the Protein Data Bank (www.rcsb.org) with PDB codes 7QF8, 7QFD, and 7QVA, respectively.
+
+**Loop modeling.** Rosetta loop modeling was used to build the non-visible loops in the *Ps* G3Ox (202-204 and 309-318), *Ps* G3Ox-Glc (77-89, 201-206, 308-320, and 345-359), and *Ps* G3Ox-Mang (74-90, 201-206, 309-324 and 351-359) complex structures using a methodology previously described⁴⁸; a full-atom refinement step was performed with the next-generation kinematic (NGK) closure robotics-inspired conformational sampling protocol⁴⁹. The crystal structures of *Ps* G3Ox, *Ps* G3Ox-Glc, and *Ps* G3Ox-Mang were kept intact except for the loop regions that were created, with repacking of the side chains within 10Å of the remodeled region. A total of 500 loops were built, and to find the best loop candidate; each model was scored by its Rosetta energy score and contacts with the substrate molecules.
+
+**Apparent steady-state kinetics.** Apparent steady-state kinetics measurements were performed at 37 ºC in 100 mM sodium phosphate buffer, pH 7.5, and reactions were started with the addition of enzyme. The kinetic parameters for D-glucose (D-Glc, PanReac Applichem, Darmstadt, Germany), D-galactose (PanReac Applichem, Darmstadt, Germany), D-ribose (VWR, Pennsylvania, USA), D-xylose (Sigma Aldrich, Missouri, USA), and L-arabinose (Sigma Aldrich, Missouri, USA) were measured using coupling assay containing 0.1 mM 4-Aminoantipyrine (AAP, Acros organics, Geel, Belgium), 1 mM 3,5-dichloro-2-hydroxybenzenesulfonic acid sodium salt (DCHBS, Alfa Aesar, Massachusetts, USA), 8 U ml⁻¹ Horseradish peroxidase (HRP, PanReac Applichem, Darmstadt, Germany) and different concentrations of substrate. Enzymatic activity was monitored using a Synergy2 microplate reader (BioTek, Vermont, USA) following the formation of N-(4-antipyryl)-3-chloro-5-sulfonate-p-benzoquinone-monoimine (a pink chromogen) at 515 nm (ε₅₁₅ = 26,000 M⁻¹ cm⁻¹). The kinetic parameters for molecular oxygen were measured in an Oxygraph system (Hansatech instruments, Pentney, UK) to follow oxygen consumption in reactions containing 1 M D-Glc as an electron donor and different oxygen concentrations pre-set by bubbling O₂ or N₂ gas. The oxidation of the glycosides mangiferin (Sigma Aldrich, Missouri, USA), rutin (Acros Organics, Geel, Belgium), and carminic acid (Sigma Aldrich, Missouri, USA) were followed by oxygen consumption in the Oxygraph apparatus in reactions containing 0 - 2 mM of Mang, 0 - 0.5 mM of rutin or 0 – 1 mM of carminic acid. Specific activity was calculated considering the preparation's functional (FAD-loaded) enzyme ratio. Apparent steady-state kinetic parameters (*k*cat and *K*m) were determined by fitting data directly into the Michaelis-Menten equation using Origin-Lab software. For inhibition assays, the steady-state kinetics for D-Glc were performed as described below in the presence of 0 - 0.5 mM rutin or 0 - 0.2 mM carminic acid. The data was represented using a Lineweaver Burk plot and the inhibition constants were estimated based on a secondary plot of the slopes against inhibitor concentration.
+
+**Identification of mangiferin oxidation product.** Oxidation of Mang was performed under aerobic conditions at 25ºC, pH 7.5 in 30 mL of Milli-Q water containing 20 mg of Mang, 1 U ml⁻¹ of Catalase (Sigma Aldrich, Missouri, USA), and 1 U ml⁻¹ of *Ps* G3Ox. To estimate the time needed to have a high yield of oxidized Mang, a time-course of the reaction was performed in a thin layer chromatography (TLC) on silica gel 60 F254 sheet (Merck, Darmstadt, Germany) using a mixture of butanol, acetic acid, and water in the proportions 4:1:2.2 (v/v) as mobile phase. The TLC revealed a diphenylamine-aniline-phosphoric acid reagent⁵⁰, a system used to distinguish sugars. For the NMR characterization, the reaction occurred for 30 min, and then the enzymes were removed by ultrafiltration using a vivaspin20 of 30 kDa cutoff (Cytiva, Massachusetts, USA). The water in the mixture was evaporated under low pressure on a rotary vacuum evaporator, and the resulting sediment was resuspended in ~ 600 of dimethyl sulfoxide-d6 (Merck, Darmstadt, Germany). The Mang and the reaction product, both in DMSO-d6, were analyzed through ¹H, ¹³C APT, COSY, and HMQC NMR in a Bruker Avance II+400. ¹H NMR spectra were obtained at 400 MHz and ¹³C at 100.61 MHz.
+
+**Molecular dynamics simulations.** Molecular dynamics simulations (MD) of *Ps* G3Ox were carried out to further explore the conformational preferences of the enzyme at different stages of the catalytic cycle. First, MD simulations of 400 ns production runs were performed of four systems built as follows: i) Model I: taken from the *Ps* G3Ox crystal structure (without substrates), which shows closed substrate and insertion-1 loops; ii) Model II: obtained from the same crystal structure but with the substrate loop sampled with the Yasara Sample Loop function and D-Glc docked (see below) in a binding mode compatible with C2 oxidation; this model presents a semi-open substrate loop and a closed insertion-1 loop; iii) Model III: prepared from the *Ps* G3Ox-Mang crystal structure, with Mang removed, the missing parts of the loops were built with the Yasara Build Loop function, and D-Glc docked in a C2 binding mode; this model shows open substrate and insertion-1 loops; iv) Model IV: derived from the *Ps* G3Ox-Mang crystal structure (in a binding mode compatible with C3 oxidation) by building the missing parts of the open substrate and insertion-1 loops. Missing hydrogen atoms were added, and the protonation state of the titratable residues was assigned with the Yasara hydrogen bond networks optimization and pKa prediction tools at pH 7⁵¹. The systems were solvated with a solvation box and neutralized with NaCl. The conventional MDs, cMDs, were set up and run using Yasara⁵²; the AMBER14 force field⁵³ and TIP3P water model⁵⁴ were used. A two-step equilibration was carried out for 400 ps: first, the system’s temperature was increased in 10 steps from 30K up to 300K, followed by a second step with constant temperature and box dimensions. The bonds and angles involving a hydrogen atom were fixed. A restrain on all non-hydrogen atoms of the complexed protein-ligand was applied so that the equilibration would mainly affect the system’s solvent. Bonded and non-bonded forces were updated every 2 and 5 fs, respectively. The protein-ligand complex was then released, and a production step of 400 ns was carried out. Gaussian accelerated MDs (GaMDs)⁵⁵ were then run for Model I - III to access effective longer simulation times and wider conformational explorations, especially of the substrate loop. The starting structures for these GaMD simulations were the 100 ns structure from the corresponding classical MDs. An extra model (Model III*) was built from Model III by removing the bound D-Glc substrate to see if loop transitions could be observed. The GaMD simulations were run with the AMBER 20 program⁵⁶. The GaMD protocol consisted of an initial equilibration stage where the potential boost was applied, boost parameters were updated, and production runs were updated with fixed boost parameters. A dual boost on dihedral and total potential energy was applied (igamd = 3). Two simulations of 600 ns were run for each Model I, II, and III*, whereas the simulation for Model III was stopped after 200 ns as D-Glc was observed to left the active-site. GaMD inputs were generated following the recommendations from the developers⁵⁷. The convergence of all simulations was assessed by calculating the RMSD values of the protein Cα atoms. The distances between FAD N5 and A352, as well as G84, were used as an indicator of the conformational state of the substrate and insertion-1 loops, respectively.
+
+**Protein-ligand docking.** Protein-ligand docking calculations were carried out on different protein structures from the cMD trajectories using AutoDock VINA⁵⁸–⁶⁰. D-Glc was docked in *Ps* G3Ox’s active site among 100 frames from semi-open (Model II) and open (Model III) substrate’s loop MDs, respectively. On the other hand, Mang was docked among 100 frames from the open substrate loop MD (Model IV). The rest of the substrate’s loop conformations, particularly those with the closed loop, did not allow proper positioning of the substrate in the active site due to steric clashes, and they were discarded. D-Glc was docked using the YAMBER force field⁶¹ on a 20 × 20 × 20 Å cuboid docking cell centered on FAD N5, while Mang was docked on a 34 × 34 × 34 Å cuboid docking cell centered on the same atom. A total of 16 ligand conformations were generated per frame, potentially yielding 1600 ligand-enzyme combinations per ligand. Distances between the glycoside’s D-Glc group and FAD N5 (N5-HC2; N5-HC3), as well as between H440 (NE2-HO2; NE2-HO3), were measured to filter structures in potential catalytically relevant C2 (NE2-HO2 and N5-HC2 < 4 Å) and C3 (NE2-HO2 and N5-HC2 < 4 Å) binding modes that could go into an optimization protocol, with a minimization step of the whole complex. The initial (pre-optimization) and final distances between FAD N5/H440 and the D-Glc group of each glycoside, as well as the total energies of the entire system and the ligand binding energies, were obtained for further analysis. The minimum distances between the ligand and residues Q297, Q340, R94, T129, K55, were measured. The number of events (frequency) for each measured distance and energy, depending on the ligand-binding mode and substrate loop conformation, were plotted in different histograms. Rutin and carminic acid were docked following the same protocol as for Mang.
+
+# Table
+
+**Table 1. Apparent steady-state kinetic parameters of wild-type and variants** *Ps* **G3Ox for different substrates.** The catalytic parameters for D-Glc and other monosaccharides were estimated using the HRP-AAP/DCHBS coupled assay; for molecular oxygen, the reactions were followed in an Oxygraph in the presence of 1 M D-Glc. The reactions with Mang were monitored by oxygen consumption in an Oxygraph. All reactions were performed in 100 mM sodium phosphate buffer at pH 7.5 and 37 ºC. The kinetic parameters were determined by fitting the data directly on the Michaelis-Menten equation using OriginLab.
+
+| Enzyme | Substrate | $k_{cat}$ (s$^{-1}$) | $K_m$ (M) | $k_{cat}/K_m$ (M$^{-1}$ s$^{-1}$) |
+|--------|-----------|------------------------|-----------|-----------------------------------|
+| Wild-type | D-Glucose | 0.19 ± 0.03 | 0.46 ± 0.13 | 0.45 ± 0.09 |
+| Wild-type | D-Xylose | 0.13 ± 0.02 | 1.00 ± 0.20 | 0.13 ± 0.01 |
+| Wild-type | D-Galactose | 0.02 ± 0.00 | 0.58 ± 0.16 | 0.03 ± 0.01 |
+| Wild-type | L-Arabinose | 0.02 ± 0.00 | 0.57 ± 0.04 | 0.03 ± 0.00 |
+| Wild-type | D-Ribose | 0.01 ± 0.00 | 0.19 ± 0.08 | 0.06 ± 0.01 |
+| Wild-type | Dioxygen<sup>a</sup> | 0.14 ± 0.04 | (5.83 ± 2.10) ×10<sup>-6</sup> | (19.87 ± 4.82) ×10<sup>3</sup> |
+| Wild-type | Mangiferin | 8.13 ± 1.67 | (0.49 ± 0.10) ×10<sup>-3</sup> | (19.22 ± 2.71) ×10<sup>3</sup> |
+| Wild-type | Rutin | nd | - | - |
+| Wild-type | Carminic acid | nd | - | - |
+| ∆loop (345-359) | D-Glucose | 0.06 ± 0.01 | 0.36 ± 0.13 | 0.19 ± 0.06 |
+| ∆loop (345-359) | Mangiferin | 1.4 ± 0.4 | (0.08 ± 0.02) ×10<sup>-3</sup> | (23.6 ± 0.3) ×10<sup>3</sup> |
+| ∆insert1 (73-93)<sup>b</sup> | D-Glucose | - | - | (0.19 ± 0.01) ×10<sup>-2</sup> |
+| ∆insert1 (73-93)<sup>b</sup> | Mangiferin | <sup>c</sup> | | |
+
+<sup>a</sup> D-Glc was used as an electron donor; nd – not detected  
+<sup>b</sup> Assays performed after *in vitro* flavinylation of the purified preparation; $k_{cat}/K_m$ was obtained from the first-order approximation of the Michaelis-Menten equation ([S] « $K_m$)  
+<sup>c</sup> Residual activity was detected using 2.5 mM of Mang ($V_{max}$ = 13.4 ± 1.7 nmol min$^{-1}$ mg$^{-1}$)
+
+# Supplementary Files
+
+- [SIPsG3OxFeb13.docx](https://assets-eu.researchsquare.com/files/rs-2662172/v1/6444979a686c55c8d7103236.docx)

0b5b3a4de9e1a58732fa866fb016e8496c69453bd9df0d262d283e6453d11c4b/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Metal difluorocarbene involved catalytic coupling. a, Coupling free difluorocarbene. \u00a0b, Metal difluorocarbene involved catalytic coupling. c, Previous nucleophilic or electrophilic addition of metal difluorocarbene. d, New mode of catalytic difluorocarbene transfer via 1,2-migration. e, Our report on the copper difluorocarbene involved catalytic gem-difluoropropargylation. f, Unique properties of the CF2 group and representative CF2-containing bioactive molecules.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Mechanistic studies of gem-difluoropropargylation via 1,2-migration of copper difluorocarbene. a, Stoichiometric reaction of alkynylcopper complex A1 with difluorocarbene. b, Stoichiometric reaction of A1 with difluorocarbene and allyl chloride 2a. c, Preparation of alkynylcopper species and their reactions with 2a. The number in the parenthesis is the 19F NMR chemical shift. d, Reaction of alkynyl nucleophiles with 2a and copper difluorocarbene G. e, Possible pathway for the copper difluorocarbene involved catalytic gem-difluoropropargylation.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Copper difluorocarbene involved catalytic gem-difluoropropargylation. a, Substrate scope of potassium propiolates and allyl chlorides. b, Substrate scope of propargyl sulfonates. a9 (0.5 mmol, 1.0 equiv), 1 (1.5 equiv), 2 (1.2 equiv), MeCN (5 mL). b11 (0.5 mmol, 1.0 equiv), 1 (1.5 equiv), 9 (1.2 equiv). All reported yields are isolated yields.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Diverse transformations of the gem-difluoropropargylated products. a, Gram-scale synthesis of 7 and its transformation. b, Synthesis of difluoroalkylated indole 46. c, Synthesis of difluoroalkylated pyrrole 49. d, Synthesis of the key intermediate 51 for pheromone derivative 52. e, Synthesis of the key intermediate 53 for PGF2a analogue 54.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/[IMAGE_FULL_TEXT_1].png",
+        "caption": "",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

0b5b3a4de9e1a58732fa866fb016e8496c69453bd9df0d262d283e6453d11c4b/preprint/preprint.md

@@ -0,0 +1,156 @@
+# Abstract
+
+The use of metal for catalytic difluorocarbene transfer reactions has long been hindered by the lack of understanding of metal difluorocarbene chemistry, despite the prospect of a new dimension to create novel fluorine structures for medicinal chemistry and advanced materials science. Here, we report a new mode of copper catalyzed difluorocarbene transfer reaction via 1,2-migration of copper difluorocarbene, in sharp contrast to the previous nucleophilic addition of copper difluorocarbene pathway. This innovative reaction enables the development of a modular catalytic *gem*-difluoropropargylation reaction using a variety of simple and widely available potassium propiolates, terminal alkynes, and allyl/propargyl electrophiles to couple difluorocarbene, opening a new avenue to precise synthesis of organofluorine compounds without tedious synthetic procedure. The impact of this protocol has been demonstrated by the efficient synthesis of complex fluorinated skeletons and the rapid synthesis of the key intermediates of pheromone derivative and PGF<sub>2a</sub> agonist. Mechanistic studies reveal that the migratory insertion of difluorocarbene into the C-Cu bond of the alkynylcopper species is involved in the reaction.
+
+[Physical sciences/Chemistry/Organic chemistry/Synthetic chemistry methodology](/browse?subjectArea=Physical%20sciences%2FChemistry%2FOrganic%20chemistry%2FSynthetic%20chemistry%20methodology)
+
+[Physical sciences/Chemistry/Chemical synthesis/Synthetic chemistry methodology](/browse?subjectArea=Physical%20sciences%2FChemistry%2FChemical%20synthesis%2FSynthetic%20chemistry%20methodology)
+
+# Full Text
+
+The precise fluorine editing of organic molecules has emerged as a powerful tool in modern drug discovery due to the beneficial effect of fluorine atom(s) that can significantly improve the metabolic stability, lipophilicity, and binding affinity of bioactive compounds.  
+<sup>1-4</sup> Consequently, impressive achievements have been made in the fluoroalkylation reactions over the past decades.  
+<sup>5-9</sup> However, most developed methods focus on the transformations of fluorinated carbanions, carbocations, and carbon-centered radicals.  
+<sup>5-9</sup> Compared to these three active intermediates, difluorocarbene, the smallest fluorocarbon unit, has the advantage of forming two chemical bonds,  
+<sup>10-12</sup> providing a new dimension to expand the chemical space and create new fluorine structures for medicinal chemistry. Ideally, coupling difluorocarbene with two simple and readily available feedstocks would enable more efficient access to organofluorine compounds without the tedious synthesis of fluoroalkylating reagents (Fig. 1a). Nevertheless, this straightforward synthetic route is regulated by the high reactivity of difluorocarbene. As a result, only limited reaction types of difluorocarbene transfer reactions have been reported so far.  
+<sup>13-16</sup> To overcome this limitation, the complexation of difluorocarbene with metal would be an attractive strategy, as the reactivity of difluorocarbene can be modulated by metal (Fig. 1b). However, due to the lack of catalytic activity in those isolated metal difluorocarbene complexes, the metal-catalyzed difluorocarbene transfer reaction remains a substantial challenge,  
+<sup>17</sup> in sharp contrast to the classic metal catalyzed carbene transfer reactions that have been proven to be a powerful transformation in organic synthesis.  
+<sup>18-20</sup> This problem is further underscored by the lack of understanding of metal difluorocarbene chemistry, though investigating metal difluorocarbene complexes has been around for over 40 years.  
+<sup>21</sup>
+
+We recently isolated palladium(0) (ref 22) and copper(I) (ref 23) difluorocarbene complexes ([Pd<sup>0</sup>]=CF<sub>2</sub> and [Cu<sup>I</sup>]=CF<sub>2</sub>) and found they possess opposite reactivities ([Pd<sup>0</sup>]=CF<sub>2</sub>, nucleophilic; [Cu<sup>I</sup>]=CF<sub>2</sub>, electrophilic), though Pd<sup>0</sup> and Cu<sup>I</sup> have the same d electron count. These findings have been applied in catalytic organic synthesis.  
+<sup>24-27</sup> However, the initial step of these catalytic difluorocarbene transfer reactions requires the formation of the low valent metal difluorocarbene ([M]=CF<sub>2</sub>, M = Pd<sup>0</sup>, Cu<sup>I</sup>) intermediates, followed by attacking the carbene carbon center with an electrophile or a nucleophile to generate a difluoroalkyl metal species ([M]-CF<sub>2</sub>R) (Fig. 1c).  
+<sup>22, 23</sup> We envision that the formation of the M-CF<sub>2</sub>R species by migratory insertion of difluorocarbene into the C-M bond would open a new dimension to harness metal difluorocarbene chemistry for catalytic synthesis of organofluorine compounds, as the C-M bond can be easily constructed by transmetalation or oxidative addition,  
+<sup>28</sup> which would provide a more general pathway for catalytic difluorocarbene transfer reactions (Fig. 1d). To realize this hypothesis, one critical factor is the rapid formation of a metal difluorocarbene complex C-[M]=CF<sub>2</sub>, followed by a facile migratory insertion pathway without the influence of coupling [M]-C with an electrophile or a nucleophile. Since copper is low-cost, earth-abundant, and easy to form a [Cu<sup>I</sup>]-C species via transmetalation between [Cu<sup>I</sup>] and a nucleophile,  
+<sup>29</sup> we assume that using copper as a catalyst under suitable conditions may address the above crucial issue and provide a cost-efficient route for modular construction of fluorinated structures (Fig. 1e), thus expanding copper difluorocarbene chemistry and opening a new avenue to efficient, precise synthesis of organofluorine compounds.
+
+Here, we disclose a copper catalyzed *gem*-difluoropropargylation reaction via 1,2-migration of copper difluorocarbene (Fig. 1e). This innovative reaction uses inexpensive and industrial feedstock potassium bromodifluoroacetate (BrCF<sub>2</sub>CO<sub>2</sub>K) as the difluorocarbene precursor,  
+<sup>30, 31</sup> allowing various widely available potassium propiolates, terminal alkynes, and allyl/propargyl electrophiles to couple difluorocarbene, providing a facile route to accessing synthetically valuable *gem*-difluoropropargylated compounds. The distinct feature of this approach is the synthetic simplicity without the tedious synthesis of fluoroalklyating reagents or moisture-sensitive organometallic reagents. The diverse transformations of the resulting products and the applications of the current protocol in the rapid synthesis of the key intermediates of bioactive molecules demonstrate the synthetic utility of this new mode of catalytic difluorocarbene transfer reaction, showing the prospective in modern drug discovery. Mechanistic studies reveal that the fast migratory insertion of difluorocarbene into the C-Cu bond of alkynylcopper(I) species is the key step for the catalytic difluorocarbene transfer.
+
+To test our hypothesis, we chose terminal alkynes as the nucleophiles, as the resulting *gem*-difluoropropargyl structure is a synthetically versatile synthon for diverse transformations. Notably, it has been widely used in copper-free click chemistry  
+<sup>32, 33</sup> because of the unique properties of the difluoromethylene (CF<sub>2</sub>) group that can lower the lowest unoccupied molecular orbital (LUMO) of the alkynes.  
+<sup>32, 34</sup> Furthermore, the CF<sub>2</sub> group is a bioisostere of the oxygen atom and the carbonyl group (Fig. 1f).  
+<sup>2, 35</sup> Incorporating the CF<sub>2</sub> group at the metabolic liable position can increase the metabolic stability of bioactive molecules.  
+<sup>1-4</sup> It has been one of the valuable strategies for discovering new bioactive molecules by tactically site-selective difluoromethylenation (Fig. 1f).  
+<sup>1-4, 35, 36</sup> However, efficient methods for such a *gem*-difluoropropargyl structure are limited. The developed methods either rely on the deoxyfluorination of alkynyl ketones with sulfur fluorides  
+<sup>37</sup> or coupling *gem*-difluoropropargyl bromides with organometallic reagents,  
+<sup>38, 39</sup> aldehydes,  
+<sup>40</sup> or imines  
+<sup>41</sup>. However, the requirement of multiple steps to prepare the substrates, such as alkynyl ketones and organometallic reagents, as well as the poor functional group of sulfur fluorides and of using strong base *n*-butyllithium to prepare *gem*-difluoropropargyl bromides  
+<sup>42</sup> regulate the widespread applications of these methods. Yet, the current catalytic modular synthesis harnessing copper difluorocarbene chemistry would overcome these limitations and provide straightforward access to the *gem*-difluoropropargyl structure.
+
+Initially, to ascertain the feasibility of the migratory insertion of difluorocarbene into the C-Cu bond, we prepared the 1,10-phenanthroline-supported alkynyl complex **A1**.  
+<sup>43</sup> Unexpectedly, treatment of **A1** with inexpensive and widely available difluorocarbene precursor BrCF<sub>2</sub>CO<sub>2</sub>K **1** in CH<sub>3</sub>CN at 50<sup>o</sup>C afforded a trifluoroalkene **3** instead of *gem*-difluoropropargyl copper complex **C1** (Fig. 2a). A similar result was also observed in DMF. Although low yields of **3** were obtained due to the decomposition of **A1**, these results demonstrate the feasibility of the migratory insertion pathway. Once **C1** was formed via the copper difluorocarbene complex **B1**, it underwent another difluorocarbene insertion to generate a tetrafluoroalkylcopper **E1** (ref 23). Finally, b-fluoride (b-F) elimination of **E1** produced **3**. This possible pathway indicates that the difluorocarbene elongation in the alkynyl copper complex **C1** is thermodynamically favorable, and the tetrafluoroalkyl copper **E1** is prone to b-F elimination due to its instability. Complex **A1** could also be used as a nucleophile to react with **1** and allyl chloride **2a** in CH<sub>3</sub>CN, providing the three-component coupling product **4** in 38% yield along with a side product **5** (13%) generated between **A1** and **2a** (Fig. 2b). Replacing CH<sub>3</sub>CN with DMF led to a lower yield of **4**. No **4** was observed using DMSO. These results suggest that the formation of **C1** via an alkynylcopper difluorocarbene complex **B1** through 1,2-migration is reasonable, which should be faster than the cross-coupling of **A1** with **2a** in a suitable reaction media, such as CH<sub>3</sub>CN and DMF, thereby facilitating the formation of *gem*-difluoropropargyl structure in the catalytic reaction. Given the difficulty in obtaining **C1** through the current difluorocarbene pathway, we prepared *gem*-difluoropropargyl cadmium species **F1** and **F2** by reaction of *gem*-difluoropropargyl bromide **6** with cadmium in DMF.  
+<sup>44</sup> These two organocadmium reagents were assigned according to the literature.  
+<sup>45</sup> Transmetalation of the mixture of **F1** and **F2** with CuI at -40<sup>o</sup>C afforded the *gem*-difluoropropargyl copper **C2** and bis(*gem*-difluoropropargyl)copper species **C3** in 41% yield and 8% yield, respectively. Since it is hard to isolate these two species, they were directly used to react with allyl chloride **2a**, providing **7** in 95% yield, thus demonstrating the feasibility of coupling *gem*-difluoroproparyl copper with an electrophile (Fig. 2c). To investigate the possibility of nucleophilic addition of alkynyl species to the carbene carbon center, we prepared copper difluorocarbene complex **G**.  
+<sup>23</sup> However, no desired product **4** was obtained when **G** was treated with **2a** and alkynyl nucleophiles, including alkynyl lithium/zinc reagents (**8a**, **8b**) and potassium propiolate **9a** (Fig. 2d). Thus, the pathway beginning with the formation of [Cu<sup>I</sup>]=CF<sub>2</sub>, followed by a reaction with an alkynyl nucleophile, is less likely.
+
+Based on the above results, a new mode of copper catalyzed difluorocarbene transfer reaction should be feasible for the catalytic modular synthesis of *gem*-difluoropropargylated compounds. In this copper catalyzed process, the reaction is initiated by the formation of an alkynylcopper species **A**, which subsequently undergoes complexation with a difluorocarbene to generate an alkynylcopper difluorocarbene intermediate **B**. This key intermediate undergoes 1,2-difluorocarbene migratory insertion to produce the *gem*-difluoropropargyl copper species **C**. Finally, **C** reacts with an electrophile to produce the *gem*-difluoropropargylated compound and releases copper catalyst simultaneously (Fig. 2e).
+
+Inspired by the above observations and the possible pathway illustrated in Fig. 1e, we explored catalytic coupling reaction between terminal alkyne **8c** and allyl chloride **2a** to couple with difluorocarbene (Table 1). When **8c** (1.0 equiv) was treated with **2a** (1.5 equiv) and difluorocarbene precursor **1** (2.0 equiv) in the presence of CuCl (10 mol%) and 1,10-phenanthroline **L1** (10 mol%) in CH<sub>3</sub>CN at 50<sup>o</sup>C using K<sub>2</sub>CO<sub>3</sub> as the base, 10% yield of the desired product **4** was obtained along with 5% yield of side product **5** (entry 1). A survey of the ligands showed that ligand **L4** could suppress the generation of **5** and increase the yield of **4** to 34% (entries 2-4, supplementary Table 2). Replacing K<sub>2</sub>CO<sub>3</sub> with Na<sub>2</sub>CO<sub>3</sub> slightly improved the reaction efficiency (entry 5, supplementary Table 3). However, the undesired defluorination of **8c** and the sensitivity of [Cu<sup>I</sup>]=CF<sub>2</sub> to the base make it difficult to increase the yield further. To circumvent these limitations, we chose readily available potassium propiolate **9b** as the alternative substrate. We envisioned that the relatively faster release of alkynyl nucleophile through decarboxylation of **9b** without needing a base would benefit the reaction efficiency. Similar to terminal alkyne **8c**, the ligand is critical for the reaction (entries 6-8, supplementary Table 5), and **L4** remained the optimal ligand, providing **7** in 73% yield at 80<sup>o</sup>C (entry 6). Decreasing the reaction temperature to 70<sup>o</sup>C increased the yield to 85% (entry 9). To optimize the reaction conditions further, we examined a series of reaction parameters, including copper catalysts, solvents, the loading amount of the catalyst, the ratio of the reactants, and the reaction time (entries 10, 11, supplementary Tables 6-11). Finally, the optimized reaction conditions were identified by shortening the reaction time to 30 min with 7.5% mol CuCl/**L4** as the catalyst (entry 12). Under these conditions, 1.5 equiv of **1** and 1.2 equiv of **2a** could provide **7** in 80% isolated yield. Notably, this reaction proceeded smoothly, even shortening the reaction time to 10 min (entry 13). This distinct feature is in sharp contrast to the conventional copper-catalyzed fluoroalkylation reactions that usually take a long time, thus underscoring the advance of the current copper catalyzed difluorocarbene transfer reaction and providing a potential opportunity to prepare <sup>18</sup>F-labeling difluoromethylenation tracers for positron emission tomography (PET).  
+<sup>46</sup> No product was observed without copper salt or ligand (entries 14, 15), demonstrating the essential role of Cu/**L** in promoting the reaction.
+
+**Table 1. Representative Results for the Optimization of the Reaction Conditions** <em><sup>a</sup></em>
+
+[IMAGE_FULL_TEXT_1]  
+<br/>
+
+| entry | [Cu] | <strong>L</strong> | base | temp (<sup>o</sup>C) | <strong>4</strong> or <strong>7</strong>, yield (%) <em><sup>b</sup></em> | <strong>5</strong> or <strong>10</strong>, yield (%) <em><sup>b</sup></em> |
+|-------|------|-------------------|------|------------------------|--------------------------------------------------|--------------------------------------------------|
+| 1 <em><sup>c</sup></em> | CuCl | <strong>L1</strong> | K<sub>2</sub>CO<sub>3</sub> | 50 | <strong>4</strong>, 10 | <strong>5</strong>, 5 |
+| 2 <em><sup>c</sup></em> | CuCl | <strong>L2</strong> | K<sub>2</sub>CO<sub>3</sub> | 50 | <strong>4</strong>, 25 | <strong>5</strong>, 30 |
+| 3 <em><sup>c</sup></em> | CuCl | <strong>L3</strong> | K<sub>2</sub>CO<sub>3</sub> | 50 | <strong>4</strong>, 2 | <strong>5</strong>, 37 |
+| 4 <em><sup>c</sup></em> | CuCl | <strong>L4</strong> | K<sub>2</sub>CO<sub>3</sub> | 50 | <strong>4</strong>, 34 | <strong>5</strong>, nd |
+| 5 <em><sup>c</sup></em> | CuCl | <strong>L4</strong> | Na<sub>2</sub>CO<sub>3</sub> | 50 | <strong>4</strong>, 38 | <strong>5</strong>, nd |
+| 6 <em><sup>d</sup></em> | CuCl | <strong>L4</strong> | -- | 80 | <strong>7</strong>, 73 | <strong>10</strong>, -- |
+| 7 <em><sup>d</sup></em> | CuCl | <strong>L5</strong> | -- | 80 | <strong>7</strong>, 48 | <strong>10</strong>, -- |
+| 8 <em><sup>d</sup></em> | CuCl | <strong>L6</strong> | -- | 80 | <strong>7</strong>, 27 | <strong>10</strong>, -- |
+| 9 <em><sup>d</sup></em> | CuCl | <strong>L4</strong> | -- | 70 | <strong>7</strong>, 85 | <strong>10</strong>, -- |
+| 10 <em><sup>d</sup></em> | CuBr | <strong>L4</strong> | -- | 70 | <strong>7</strong>, 80 | <strong>10</strong>, -- |
+| 11 <em><sup>d</sup></em> | CuI | <strong>L4</strong> | -- | 70 | <strong>7</strong>, 55 | <strong>10</strong>, -- |
+| 12 <em><sup>d,e</sup></em> | CuCl | <strong>L4</strong> | -- | 70 | <strong>7</strong>, 82 (80) | <strong>10</strong>, -- |
+| 13 <em><sup>d,e,f</sup></em> | CuCl | <strong>L4</strong> | -- | 70 | <strong>7</strong>, 83 | <strong>10</strong>, -- |
+| 14 <em><sup>d</sup></em> | none | <strong>L4</strong> | -- | 70 | <strong>7</strong>, nd | <strong>10</strong>, -- |
+| 15 <em><sup>d</sup></em> | CuCl | none | -- | 80 | <strong>7</strong>, 3 | <strong>10</strong>, -- |
+
+<em><sup>a</sup></em> Reaction conditions (unless otherwise specified): **8c** or **9a** (0.2 mmol, 1.0 equiv), **1** (2.0 equiv), **2a** (1.5 equiv), MeCN (2 mL), 6 h.  
+<em><sup>b</sup></em> Determined by <sup>19</sup>F-NMR using fluorobenzene as an internal standard; nd, not detected.  
+<em><sup>c</sup></em> Using **8c** as the substrate.  
+<em><sup>d</sup></em> Using **9b** as the substrate.  
+<em><sup>e</sup></em> **9b** (0.5 mmol, 1.0 equiv), **1** (1.5 equiv), **2a** (1.2 equiv), CuCl/**L4** (7.5 mol%), 30 min. The data in the parenthesis is the isolated yield.  
+<em><sup>f</sup></em> The reaction was run in 10 min.
+
+With the viable reaction conditions in hand, we examined the scope of this copper difluorocarbene involved catalytic *gem*-difluoroproparylation reaction (Fig. 3). Various potassium arylpropiolates were applied to this transformation (Fig. 3a), providing the corresponding *gem*-difluoropropargylated products efficiently (**4**, **7**, **12**-**32**). Generally, aromatic propiolates bearing an electron-donating substituent provided higher yields than electron-deficient substrates. The reaction exhibited high functional group tolerance. Base and nucleophile sensitive functional groups, such as ketone (**13**), ester (**14**), and nitrile (**16**), were compatible with the reaction; aryl fluoride (**4**), chloride (**21**), bromide (**17**, **19**-**21**), and iodide (**18**) moieties underwent the current copper-catalyzed process smoothly. Additionally, the position of bromide in the aromatic ring did not affect the reaction efficiency. Para-, meta-, and ortho-aryl bromides efficiently delivered the corresponding *gem*-difluoropropargylated products (**17**, **19**, **20**). The high compatibility of chlorobromoaryl moiety (**21**) offers a good opportunity for diversified transformations by sequential aryl bromide and chloride functionalization. Ferrocene- and thiophene-containing substrates were also applied to the reaction, with moderate to good yields obtained (**22**-**24**). The reaction was not restricted to allyl chloride **2a**, as substituted allyl chlorides, including linear, branched, and cyclic allyl chlorides (**25**-**32**), underwent smooth coupling. Even highly reactive allyl chlorides bearing vinyl chloride (**30**) or unsaturated ester (**31**) were still amenable to the reaction. In the case of linear allyl chloride (**25**), no branched product was observed. In addition to arylpropiolates, alkyl- and silyl-substituted propiolates were competent coupling partners (**33**-**38**), and the aliphatic side chain bearing benzyloxy (**34**), chloride (**35**), sulfamide (**36**), or cyclopropyl (**37**) did not interfere with the reaction efficiency. This approach could also be extended to propargyl electrophiles (Fig. 3b). One problem with this type of substrate is forming a copper-allenylidene complex between the copper catalyst and propargyl electrophile.  
+<sup>47</sup> This competitive side reaction significantly influences the current copper catalyzed difluorocarbene transfer process. After extensive efforts (supplementary Table 12), we found that using propargyl sulfonates as the limiting substrates could suppress this undesired side reaction, producing various *gem*-difluoropropargylated allenes with high efficiency. Although the synthesis of allenes has been well-established,  
+<sup>48-51</sup> efficient methods for such fluoroalkylated allenes have yet to be reported. Given the synthetic versatility of allene and alkyne, the resulting *gem*-difluoroporpargylated allenes should be a valuable structure for diverse transformations. As depicted in Fig. 3b, arylpropiolates underwent smooth coupling with good functional group tolerance (**39**-**43**). Versatile synthetic handles, such as nitrile (**40**), thiophene (**41**), aryl bromide (**43**), and alkyl chloride (**42**) moieties, tolerate the reaction well. In contrast to the allyl electrophiles, arylpropiolate bearing an electron-withdrawing group provided a higher yield (**40**). However, alkylpropiolates led to low yields (20-30%). Of note, for all the coupling reactions described above, no [2+1] cycloaddition side products generated between difluorocarbene and the unsaturated carbon-carbon bond were observed,  
+<sup>15</sup> thus demonstrating the advance of this copper catalytic system further.
+
+The reaction is readily scalable, as demonstrated by the gram-scale synthesis of **7** with a high yield (Fig. 4a). The resulting *gem*-difluoropropargyl products can be elaborated through a myriad of transformations to create a diversity of new organofluorine compounds. Selective oxidative cleavage of the carbon-carbon double bond of **7** with ozone, followed by reduction with NaBH<sub>4</sub>, afforded alcohol **44** efficiently. Cyclization of **37** with phenidone **45** via rhodium catalysis produced difluoroalkylated indole **46** with high efficiency (Fig. 4b).  
+<sup>52</sup> The *gem*-difluoropropargyl structure could also be used to construct the difluoroalkylated pyrrole **49** through deprotection of **38**, followed by silver-catalyzed [3+2] reaction with ethyl isocyanoacetate **49** (Fig. 4c).  
+<sup>53</sup> Given the unique properties of the CF<sub>2</sub> group and critical applications of indole and pyrrole in medicinal chemistry, the rapid access to these complex fluorinated molecules that otherwise require tedious steps to prepare through conventional methods provides a good opportunity to discover new interesting bioactive molecules. Remarkably, this copper catalyzed difluorocarbene transfer reaction could be used as a key step to introduce the CF<sub>2</sub> group at the metabolic liable allylic position of bioactive molecules. As shown in Fig. 4d, pheromone derivative **52**, a probe used to study hydrophobic interaction in pheromone reception, was rapidly accessed from **50** via two steps, followed by a reported procedure.  
+<sup>54</sup> Since the *Z*-difluoroalkylated alkenes have been found in a series of pheromone analogs,  
+<sup>54</sup> this copper difluorocarbene involved catalytic *gem*-difluoropropargylation should have applications in such a kind of compounds. Furthermore, using (-)-corey lactone diol derived terminal alkyne **8d** as the substrate could directly afford **53** by harnessing the current copper difluorocarbene chemistry (Fig. 4e). Although a 42% yield of **53** was obtained, 46% of **8d** could be recovered. Notably, compound **53** could be used as a potential key intermediate for synthesizing tafluprost **54**, a PGF<sub>2a</sub> agonist for treating glaucoma,  
+<sup>36</sup> thus underscoring the synthetic utility of this transformation.
+
+In summary, a new mode of copper difluorocarbene involved catalytic coupling reaction has been developed. The stoichiometric reactions reveal that a difluorocarbene migratory insertion into the C-Cu bond is involved in the catalytic cycle. This innovative approach allows a wide range of readily available simple components, including potassium propiolates, terminal alkynes, and allyl/propargyl electrophiles, for the rapid modular synthesis of valuable *gem*-difluoropropargyl structure, opening a new avenue to the efficient, precise synthesis of organofluorine compounds. We anticipate that this copper difluorocarbene chemistry will be attractive to creating novel fluorinated structures of interest in medicinal chemistry and <sup>18</sup>F PET-CT for diagnosis. Most importantly, this work should also prompt the development of new metal difluorocarbene involved catalytic coupling reactions in methodology development.
+
+# References and Notes
+
+1. Muller, K., Faeh, C. & Diederich, F. Fluorine in pharmaceuticals: looking beyond intuition. *Science* **317**, 1881–1886 (2007).
+2. Hagmann, W. K. The many roles for fluorine in medicinal chemistry. *Med. Chem.* **51**, 4359–4369 (2008).
+3. Meanwell, N. A. Fluorine and fluorinated motifs in the design and application of bioisosteres for drug design. *Med. Chem.* **61**, 5822–5880 (2018).
+4. Inoue, M., Sumii, Y. & Shibata, N. Contribution of organofluorine compounds to pharmaceuticals. *ACS Omega* **5**, 10633–10640 (2020).
+5. Umemoto, T. Electrophilic perfluoroalkylating agents. *Rev.* **96**, 1757–1778 (1996).
+6. Prakash, G. K. S. & Mandal, M. Nucleophilic trifluoromethylation tamed. *Fluorine Chem.* **112**, 123–131 (2001).
+7. Furuya, T., Kamlet, A. S. & Ritter, T. Catalysis for fluorination and *Nature* **473**, 470–477 (2011).
+8. Tomashenko, O. A. & Grushin, V. V. Aromatic trifluoromethylation with metal complexes. *Rev.* **111**, 4475–4521 (2011).
+9. Zhang, F., Xiao, Y.-L. & Zhang, X. Transition-metal (Cu, Pd, Ni)-catalyzed difluoroalkylation via cross-coupling with difluoroalkyl halides. *Chem. Res.* **51**, 2264–2278 (2018).
+10. Brahms, D. L. S. & Dailey, W. P. Fluorinated Carbenes. *Rev.* **96**, 1585–1632 (1996).
+11. Ni, C.-F. & Hu, J.-B. Recent Advances in the Synthetic Application of Difluorocarbene. *Synthesis* **46**, 842–863 (2014).
+12. Dilman, A. D. & Levin, V. V. Difluorocarbene as a Building Block for Consecutive Bond-Forming Reactions. *Chem. Res.* **51**, 1272–1280 (2018).
+13. Miller, T. G. & Thanassi, J. W. The Preparation of Aryl Difluoromethyl Ethers. *Org. Chem.* **25**, 2009–2012 (1960).
+14. Fuqua, S. A. Duncan, W. G. & Silverstein, R. M. A one-step synthesis of 1,1-difluoroolefins from aldehydes by a modified Wittig synthesis. *Tetrahedron Lett.* **5**, 1461–1463 (1964).
+15. Dolbier, W. R. & Battiste, M. A. Structure, Synthesis, and Chemical Reactions of Fluorinated Cyclopropanes and Cyclopropenes. *Rev.* **103**, 1071–1098 (2003).
+16. Yuan, W.-J., Tong, C.-L., Xu, X.-H. & Qing, F.-L. Copper-Mediated Oxidative Chloro- and Bromodifluoromethylation of Phenols. *Am. Chem. Soc.* **145**, 23899–23904 (2023).
+17. Brothers, P. J. & Roper, W. R. Transition-metal dihalocarbene *Chem. Rev.* **88**, 1293–1326 (1988).
+18. Dotz, K. H. *Topics in Organometallic Chemistry: Metal Carbenes in Organic Synthesis* 13 (Springer, 2004).
+19. Wang, J.-B., Che, C.-M. & Doyle, M. P. *Transition Metal-Catalyzed Carbene Transformations* (Wiley, 2022).
+20. Zhang, L., DeMuynck, B. M., Paneque, A. N., Rutherford, J. E. & Nagib, D. A. Carbene reactivity from alkyl and aldehydes. *Science* **377**, 649–654 (2022).
+21. Zhou, W., Pan, W.-J. & Xiao, J.-C. Transition-metal difluorocarbene complexes. *Commun.* **57**, 9316–9329 (2021).
+22. Fu, X.-P., Xue, X.-S., Houk, K. N. & Zhang, X. Controllable catalytic difluorocarbene transfer enables access to diversified fluoroalkylated arenes. *Chem.* **11**, 948–956 (2019).
+23. Zeng, X., Li, Y., Min, Q.-Q., Xue, X.-S. & Zhang, X. Copper-catalysed difluorocarbene transfer enables modular synthesis. *Chem.* **15**, 1064–1073 (2023).
+24. Feng, Z., Min, Q.-Q. & Zhang, X. Access to difluoromethylated arenes by Pd-catalyzed reaction of arylboronic acids with bromodifluoroacetate. *Lett.* **18**, 44–47 (2016).
+25. Feng, Z., Min, Q.-Q., Fu, X.-P., An, L. & Zhang, X. Chlorodifluoromethane-triggered formation of difluoromethylated arenes catalysed by palladium. *Chem.* **9**, 918–923 (2017).
+26. Zhang, X.-Y., Fu, X.-P., Zhang, S. & Zhang, X. Palladium Difluorocarbene Involved Catalytic Coupling with Terminal Alkynes. *CCS Chem.* **2**, 293–304 (2020).
+27. Zeng, X., Xue, X.-S. & Zhang, X. Copper Difluorocarbene Enables Catalytic Difluoromethylation. *Am. Chem. Soc.* **146**, 16902–16911 (2024).
+28. Crabtree, R. H. *The Organometallic Chemistry of the Transition Metals* (Wiley, 2014).
+29. Krause, N. *Modern Organocopper Chemistry* (Wiley, 2002).
+30. Tissot, P. & Waefler, J. P. Thermal decomposition of the alkaline salts of difluorochloro- and difluorobromoacetic a *Thermochim. Acta* **66**, 315–321 (1983).
+31. Su, D.-B., Duan, J.-X. & Chen, Q.-Y. A simple, novel method for the preparation of trifluoromethyl d *J. Chem. Soc., Chem. Commun.* 807–808 (1992).
+32. Baskin, J. M. & Bertozzi, C. R. Copper-free click chemistry for dynamic in vivo imaging. *Natl. Acad. Sci. U.S.A.* **104**, 16793–16797 (2007).
+33. Laughlin, S. T., Baskin, J. M., Amacher, S. L. & Bertozzi, C. R. In vivo imaging of membrane-Associated glycans in developing zebrafish. *Science* **320**, 664–667 (2008).
+34. Ess, D. H., Jones, G. O. & Houk, K. N. Transition states of strain-promoted metal-free click chemistry: 1,3-dipolar cycloadditions of phenyl azide and cyclooctynes. *Lett.* **10**, 1633–1636 (2008).
+35. Shurtleff, V. W. et al. Invention of MK-7845, a SARS-CoV‑2 3CL Protease inhibitor employing a novel difluorinated glutamine mimic. *Med. Chem.* **67**, 3935–3958 (2024).
+36. Nakajima, T. & Hara, H. New fluoroprostaglandin F<sub>2α</sub> derivatives with prostanoid FP-receptor agonistic activity as potent ocular-hypotensive agents. *Pharm. Bull.* **26**, 1691–1695 (2003).
+37. Hirao, K.-I., Yamashita, A. & Yonemitsu, O. Fluoroalkylnorbornadienes and their corresponding valence isomer quadricyclanes — a light energy storage system. *Fluorine Chem.* **36**, 293–305 (1987).
+38. Yu, Y.-B., He, G.-Z. & Zhang, X. Synthesis of α,α-difluoromethylene alkynes by palladium-catalyzed *gem*-difluoropropargylation of aryl and alkenyl boron reagents. *Chem. Int. Ed.* **53**, 10457–10461 (2014).
+39. An, L., Xu, C. & Zhang, X. Highly selective nickel-catalyzed *gem*-difluoropropargylation of unactivated alkylzinc reagents. *Commun.* **8**, 1460 (2017).
+40. Mae, M. & Hammond, G. B. Mg(0)-promoted debromometalation of *gem*-difluoropropargyl bromides. *Tetrahedron Lett.* **46**, 1787–1789 (2005).
+41. Surmont, R., Verniest, G. & Kimpe, N. D. Gold-Catalyzed Synthesis of 2-aryl-3-fluoropyrroles. *Lett.* **11**, 13, 2920–2923 (2009).
+42. Xu, B. & Hammond, G. B. An efficient synthesis of difluoropropargyl bromides. *Synthesis* **5**, 803–806 (2006).
+43. Wu, W. & Weng, Z. Trifluoroacetic anhydride promoted copper(I)-catalyzed interrupted click reaction: From 1,2,3-triazoles to 3-trifluoromethyl-substituted 1,2,4-triazinones. *Chem. Int. Ed.* **56**, 10476–10480 (2017).
+44. Platonov, V., Yang, Z.-Y. & Burton, D. J. The preparation, stability and reactivity of perfluorobenzyl-cadmium and -copper reagents *Fluorine Chem.* **66**, 23–24 (1994).
+45. Hartgraves, G. A. & Burton, D. J. The preparation and allylation of difluoromethylcadmium. *Fluorine Chem.* **39**, 425–430 (1988).
+46. J. B. I. & Gouverneur V. [<sup>18</sup>F]Difluorocarbene for positron emission tomography. *Nature* **606**, 102–108 (2022).
+47. Nakajima, K., Shibata, M. & Nishibayashi, Y. Copper-catalyzed enantioselective propargylic etherification of propargylic esters with alcohols. *Am. Chem. Soc.* **137**, 2472–2475 (2015).
+48. Ma, S. Transition metal-catalyzed/mediated reaction of allenes with a nucleophilic functionality connected to the α-carbon atom. *Chem. Res.* **36**, 701–712 (2003).
+49. Hoffmann-Röder, A. & Krause, N. Synthesis and properties of allenic natural products and pharmaceuticals. *Chem. Int. Ed.* **43**, 1196–1216 (2004).
+50. Ye, J. & Ma, S. Palladium-catalyzed cyclization reactions of allenes in the presence of unsaturated carbon-carbon bonds. *Chem. Res.* **47**, 989–1000 (2014).
+51. Huang, X. & Ma, S. Allenation of terminal alkynes with aldehydes and ketones. *Chem. Res.* **52**, 1301–1312 (2019).
+52. Lin, S. & Yi, W. Rh(III)-catalysed switchable and chemoselective synthesis of difluorinated pyrazolo[1,2-a]indazolone and indole frameworks. *Asian Org. Chem.* **11**, e202200019 (2022).
+53. Liu, J. & Bi, X. Silver-catalyzed isocyanide-alkyne cycloaddition: a general and practical method to oligosubstituted pyrroles. *Chem. Int. Ed.* **52**, 6953–6957 (2013).
+54. Sun, W.-C., Ng, C.-S. & Prestwich, G. D. Synthesis of partially fluorinated analogs of (Z)-5-decenyl acetate: probes for hydrophobic interaction in pheromone reception. *Org. Chem.* **57**, 132–137 (1992).
+
+# Supplementary Files
+
+- [ZXNCatalSI.pdf](https://assets-eu.researchsquare.com/files/rs-5324940/v1/0db366e26bef6cf177bf87ed.pdf)

0b8d3b368b3c5edca8508af2150ac69abfc0e67a6a884ddab975ce75e19e86e1/preprint/images_list.json

@@ -0,0 +1,42 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.jpg",
+        "caption": "Addition of 4,4\u2032\u2212bpy to two equivalents of Cp\u20323M (M = Sm, Gd, Cm) yields (Cp\u20323M)2(\u03bc\u22124,4\u2032\u2212bpy).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.jpg",
+        "caption": "Top. Structure of 1\u2212Cm with thermal ellipsoids at 50% probability. Green = Curium, Blue = Nitrogen, Orange = Silicon, Gray = Carbon, and hydrogen has been omitted for clarity purposes. Bottom Left. A concentrated solution of 1\u2212Cm in toluene. Bottom right. Crystals of 1\u2212Cm.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.jpg",
+        "caption": "Photoluminescence of putative Cp\u20323Cm microcrystals (blue) compared to 1\u2212Cm (red), excited at 420 nm at \u2013180 \u2103. Top Right. Cp\u20323Cm glowing red upon irradiation at 420 nm.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "Solid-state absorption spectra of 1\u2212Sm (orange), 1\u2212Gd(green), and 1\u2212Cm (red).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Natural Localized Molecular Orbitals (NLMOs) involved in the metal-ligand interactions occurring in 1\u2212Sm, 1\u2212Gd, 1\u2212Cm, and Cp\u20323Cm.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]

0b8d3b368b3c5edca8508af2150ac69abfc0e67a6a884ddab975ce75e19e86e1/preprint/preprint.md

@@ -0,0 +1,167 @@
+# Abstract
+
+Structural and electronic characterization of (Cp′₃Cm)₂(μ−4,4′−bpy) (Cp′ = trimethylsilylcyclopentadienyl, 4,4′−bpy = 4,4′−bipyridine) is reported and provides a rare example of curium−carbon bonding. Cp′₃Cm displays unexpectedly low energy emission that is quenched upon coordination by 4,4′−bipyridine. Electronic structure calculations on Cp′₃Cm and (Cp′₃Cm)₂(μ−4,4′−bpy) rule out significant differences in the emissive state, rendering 4,4′−bipyridine as the primary quenching agent. Comparisons of (Cp′₃Cm)₂(μ−4,4′−bpy) with its samarium and gadolinium analogues reveal atypical bonding patterns and electronic features that offer insights into bonding between carbon with f-block metal ions.
+
+Physical sciences/Chemistry/Inorganic chemistry/Organometallic chemistry  
+Physical sciences/Chemistry/Nuclear chemistry  
+actinides  
+curium  
+cyclopentadienyl ligands  
+organoactinide  
+organometallic
+
+# Introduction
+
+Discovered by Seaborg, James, and Ghiorso in 1944 via α-particle bombardment of ²³⁹Pu, curium (Z = 96) is one of the heaviest elements available in quantities suitable for traditional synthetic chemistry.¹ It is most stable in the +3 oxidation state, and possesses a [Rn]5f⁷ electron configuration.¹ Its half-filled shell creates increased stability with respect to other 5fⁿ configurations, and is often associated with an expected decrease in f-electron contributions to bonding, as well as high resistance to changes in oxidation state.²–⁴ Additionally, similarities in ionic radii lead to challenges in separating curium from americium and the lanthanides. This separation is an essential component of recycling used nuclear fuel owing to curium’s significant contribution to the radiotoxicity of nuclear waste.⁵,⁶
+
+Despite the distinct electronic properties of the wide variety Cm³⁺ compounds that have been prepared to date, no single-crystal structural characterization of a complex containing a Cm−C bond has been reported.⁶–¹⁶ Examination of this interaction could provide insights into methods for engaging the frontier orbitals of curium and other actinides in forming partially covalent bonds with selected ligands. This could allow us to gain some control over the electronic structure of these complex elements.⁷–²⁵ For example, it has been shown that the engagement of 5f orbitals in Cm-ligand bonds can be increased by using soft-donor ligands that can be further enhanced by applying mechanical pressure.³
+
+Quantum mechanical evaluation of a recently reported Am(III) cyclopentadienyl (Cp) complex showed that a variety of metal frontier orbitals were mixing with Cp′ ligand orbitals to create partially covalent bonds, but also revealed a surprising degree of ionicity in the Am−N interactions with 4,4′-bipyridine in the same complex.²⁶,²⁷ However, complexes of this type remain rare largely due to low available quantities of these isotopes (reactions are completed with < 5 mg of metal content), the need for specialized research facilities, and their exceptional air and moisture sensitivity.²⁰,²⁶−²⁹ Owing to these difficulties, syntheses allowing the structural characterization of An−C (An = Pu, Am, Cf) have only recently become accessible and can be applied to curium.²⁰,²⁶−²⁹ These synthetic challenges also necessitate the use of lanthanide analogues that possess similar ionic radii and/or electron configurations for optimizing the chemistry and for providing benchmarks for comparisons with the 5f series.
+
+The synthesis and first single-crystal structural characterization of Cm−C bonding is reported, in addition to its lanthanide analogues with Sm(III) and Gd(III) (Fig. 1). This multinuclear organometallic curium complex, (Cp′₃Cm)₂(μ−4,4′−bpy) (Cp′ = trimethylsilylcyclopentadienyl, 4,4′−bpy = 4,4′−bipyridine) (1−Cm), serves as gateway into the field of organometallic curium chemistry and provides a further understanding of soft donor coordination with curium. Additionally, 1−Cm and its putative Cp′₃Cm precursor present unexpected spectroscopic properties atypical of curium systems. Comparison to lanthanide analogues, (Cp′₃Sm)₂(μ−4,4′−bpy) (1−Sm) and (Cp′₃Gd)₂(μ−4,4′−bpy) (1−Gd), are discussed based on similarities in ionic radii and valence, respectively. This study further enforces the influence of a crowded coordination environment on the degree of covalency in soft donor transuranic systems and introduces bonding variances between Cm−C bonding from its organometallic lanthanide analogues.
+
+# Results And Discussion
+
+Addition of 4,4′−bpy to two equivalents of a putative Cp′₃M (M = Sm, Gd, Cm) in toluene yields (Cp′₃M)₂(µ−4,4′−bpy) (Fig. 2). All three molecules crystallized in the P$\bar{1}$ space group and were isomorphous, demonstrating a pseudo-tetrahedral geometry from the 4,4′−bpy and centroids of three Cp′ rings coordinated to a metal center. In each case, two metal centers are bridged by 4,4′−bpy, creating a dinuclear barbell shape with an inversion center in the center of the 4,4′−bpy. Additional synthetic and crystallographic details can be found in the Supplementary Information.
+
+1−Sm, 1−Gd, and 1−Cm possess M−N distances of 2.626(3) Å, 2.592(3) Å, and 2.5962(16) Å, respectively. Surprisingly, the M−N distance observed in 1−Cm is within error of the 4f congener, 1−Gd, but is substantially shorter than its ionic radii-based analogue, 1−Sm.¹⁰ The M−N distance presented in 1−Sm is slightly shorter than that of a similar molecule, Cp₃Sm(py) (py = pyridine), 2.656(3) Å.¹¹ A longer distance would be anticipated in 1−Sm based on the provided steric bulk from the trimethylsilyl groups of Cp′ competing with 4,4′−bpy. However, 1−Gd exhibits a notable longer Gd−N distance in comparison to that reported in Cp₃Gd(NH₃), 2.501(6) Å, due to the significantly smaller size of the coordinated NH₃.¹² As there has been no single-crystal structural characterization of organometallic curium to date, comparison to similar coordination environments proves challenging. The Cm−N distance of 1−Cm is greater than those reported in Cm(HDPA)₃·H₂O (H₂DPA = 2,6−pyridinedicarboxylic acid) but less than Cm(S₂CNEt₂)₃ (N₂C₁₂H₈).¹³,¹⁴ The average Cm−N distance of Cm(HDPA)₃ is 2.541(4) Å and 2.550(4) Å for the Λ and Δ enantiomers, respectively, and the reported Cm−N distance Cm(S₂CNEt₂)₃ (N₂C₁₂H₈) is 2.601(6) Å.¹³,¹⁴ Curium sets a trend for these bridged organometallic actinide systems in the trivalent state. Previously reported values of isostructural systems containing uranium¹⁵ and americium¹⁶ possess notably longer M−N bonds than 1−Cm: 2.626(7) Å, 2.618(2) Å, and 2.5962(16) Å respectively, introducing a trend of decreasing M−N distances across the actinide series. This trend excludes thorium owing to its low stability in the trivalent state, resulting in the reduction of 4,4′−bpy and oxidation to Th⁴⁺.¹⁷
+
+The M−Cent (Cent = Centroid) bond distances observed in 1−Sm, 1−Gd, and 1−Cm are 2.516(3) Å, 2.498(4) Å, and 2.517(3) Å, respectively. Distorted symmetry and steric competition resulting from the increased coordination number causes a deviation in M−Cent distances compared to their Cp′₃M counterparts (Table S5-S6).¹⁸,¹⁹ A range of M−Cent distances from 2.498(2) Å to 2.538(2) Å is seen in 1−Cm, 2.496(3) Å to 2.537(3) Å in 1−Sm, and 2.477(4) Å to 2.523(4) Å in 1−Gd, presenting a statistically significant variation of about 0.04 Å in all three molecules. A decrease in M−Cent avg is observed in comparison to isostructural uranium and americium systems.²⁰,²¹ The average M−Cent length and range observed in 1−Gd are longer and spread over a broader range compared to the reported M−Cent distance range seen in Cp′₃Gd, 2.434 Å – 2.441 Å owing to increased coordination.²² Similarly, the Cent−M−Cent angles of 1−Sm, 1−Gd, and 1−Cm are lower than the reported Cent−M−Cent angles of Cp′₃M (M = lanthanide), due to the effect of 4,4′−bpy on the coordination environment.²³,²⁴ The observed average Cent−M−Cent angle in 1−Cm is 117.3°. The Cent−M−N angles of 1−Cm range from 93.761° to 103.628°. This considerable difference can be attributed to the bulkiness of the trimethylsilyl groups of Cp′. Additionally, 4,4′−bpy is angled downward at just over 9°, that is likely due to an interaction between hydrogen from the trimethylsilyl groups and nitrogen on the bridge.
+
+Average M−C bond distances of 1−Sm, 1−Gd, and 1−Cm are 2.788(3) Å, 2.771(4) Å, and 2.789(2) Å, respectively. 1−Sm and 1−Cm share similar average M−C distances, within error, owing to nearly identical ionic radii.¹⁰ A significant variation in M−C distances is observed due to ring shifting upon the coordination of 4,4′−bpy. A broad range of M−C distances, between 2.714(4) Å and 2.889(3) Å, is noted in 1−Gd, differing greatly from those reported in Cp′₃Gd, 2.690(2) Å – 2.7427(19) Å.²² This comparison remains consistent between 1−Sm and Cp′₃Sm.²³ 1−Cm experiences the same phenomenon, exhibiting M−C distances between 2.728(2) Å and 2.9029(2) Å within the same Cp′⁻ ring. This substantial range indicates that coordination of 4,4′−bpy not only directly impacts M−C distances, but also the angle at which Cp′ coordinates. The average M−C distance in 1−Cm is shorter than those reported in (Cp′₃U)₂(µ−4,4′−bpy)²⁴ and (Cp′₃Am)₂(µ−4,4′−bpy),²⁵ resulting from a decrease in ionic radii across the actinide series and is consistent with the observed M−N bond distance trend.¹⁰
+
+Coordination of Cp′ to curium leads to unique emission properties not before reported in a curium complex. Photoluminescence of putative Cp′₃Cm microcrystals was collected at room temperature upon irradiation with an excitation wavelength of 420 nm (Fig. 3). A bathochromic shift and considerable splitting were observed, yielding a band centered at ~670 nm (ca. 14,925 cm⁻¹) with a full-width half-maximum value of about 52 nm (ca. 1,225 cm⁻¹), noticeably larger than most previously reported.²⁶,²⁷,²⁸,²⁹,³⁰ While Cm³⁺ typically phosphoresces red-orange from 590–620 nm (ca. 19,949–16,129 cm⁻¹),³¹,³²,³³,³⁴,³⁵ red luminescence has been reported previously in CmCp₃, albeit not redshifted to this degree.³⁶,³⁷ After 24 hours of air exposure, no photoluminescence was observed. Surprisingly, 1−Cm exhibits no photoluminescence from known excitation wavelengths of 365 nm and 420 nm used for Cm³⁺, suggesting the change in coordination environment by the coordination of 4,4′−bpy to Cp′₃Cm quenches emission. This is likely due to a non-radiative deactivation mechanism attributable to the resonance between C−H vibrational modes of the 4,4′-bpy with the electronic emissive state of the curium ion (vide infra).
+
+Solid-state absorption spectroscopy of 1−Sm, 1−Gd, and 1−Cm were measured from 350–1,700 nm (ca. 28,571–5,882 cm⁻¹) (Fig. 4). Spectra of each were collected at room temperature; additionally, 1−Cm was measured at −180 °C. Assignment of the Laporte forbidden f−f transitions of these molecules has been completed in terms of total angular momentum, J. Since actinides exhibit a complex interplay between electron repulsion, relativistic, and ligand-field effects, the analysis was performed under the intermediate coupling scheme.³⁸,³⁹ A charge transfer (CT) band in 1−Sm is observed beginning at 575 nm (ca. 17,391 cm⁻¹), masking the high energy fingerprint f−f transitions indicative of Sm³⁺; however, lower energy f−f transitions in the range of 900–1,700 nm (ca. 11,111–5,882 cm⁻¹) are still seen.⁴⁰,⁴¹ Similar to other organometallic f-block molecules,⁴²,⁴³,⁴⁴ notable splitting of these transitions is detected due to the unique coordination environment resulting from the coordination of Cp′. The CT band of 1−Gd, beginning at 600 nm (ca. 16,667 cm⁻¹), masks the f−f transitions indicative of Gd³⁺ that are characteristically detected between 270–320 nm (ca. 37,037–31,250 cm⁻¹) (Fig. 4).⁴⁵ Following the trend of 1−Sm and 1−Gd, the absorption spectrum of 1−Cm contains a CT band beginning at 615 nm (ca. 16,260 cm⁻¹) that masks high energy fingerprint transitions (Fig. 4).⁸,¹⁴,¹⁶,³⁴,⁴⁰ Transitions at 587 nm and 597 nm (ca. 17,036 cm⁻¹ and 16,750 cm⁻¹) (J = 7/2 and 5/2, respectively) typically detected at lower energy in Cm³⁺ spectra and two transitions from 630–650 nm (ca. 15,873–15,385 cm⁻¹) (J = 7/2) are observed, further displaying the unique photophysical properties of 1−Cm with respect to only a small handful of solid-state Cm³⁺ compounds reported to date. These transitions are split and redshifted about 30 nm from those previously reported.³⁸
+
+Solution absorption spectra of 1−Sm, 1−Gd, 1−Cm, and putative Cp′₃Cm were collected from 250–1,700 nm (ca. 40,000–5,882 cm⁻¹) at room temperature (See Supplementary Information). Characteristic f−f transitions of Cm³⁺ are generally very weak, possessing only single or double digit molar absorptivities (ε), and are often difficult to observe in the solution phase.⁸,¹⁴,³⁴ As such, fingerprint transitions of Cp′₃Cm were not seen due to the low concentration. The previously stated transitions, 587 nm (J = 7/2), 596 nm (J = 5/2) and 630–650 nm (J = 7/2), of 1−Cm are very weak in the solution phase. However, a transition at 411 nm (ca. 24,331 cm⁻¹) is observed, and has undergone a bathochromic shift of about 15 nm (ca. 919 cm⁻¹) compared to previously reported values.⁸,¹⁴
+
+Spin-orbit CASSCF along with MC-pDFT methods (hereinafter referred to as SO-pDFT) were employed to calculate the electronic states of 1−Sm, 1−Gd, 1−Cm and Cp′₃Cm. Since the size of the systems represents a limitation from the theoretical perspective, a model consisting of one Cp′₃M unit (M = Sm, Gd, Cm) coordinated to pyridine was used instead. The assignment of the different electronic states was done indicating the total angular momentum quantum number J along with the predominant ²S+1L term in parenthesis. The assignment of the spectroscopic transitions was performed by considering vertical excitations from the ground state (GS) to different excited states (ES).
+
+In 1−Sm, the spin-orbit ground state corresponds to a J = 5/2 (⁶H). The region between ~1,200 cm⁻¹ (~8,333 nm) to ~11,150 cm⁻¹ (~897 nm) exhibit a continuum of excited states belonging to J = 7/2−15/2 (⁶H) and J = 1/2−11/2 (⁶F) manifolds. If the barycenter of the manifolds is considered, the energy gap between them is no greater than ~1,500 cm⁻¹. Moving to higher energies, the first excited state being predominantly quartet appears at ~18,270 cm⁻¹ (~547 nm) and corresponds to a J = 5/2 (⁴G). The assignment of the experimental absorption features and the detail about the calculated SO-states is shown in Table S7-S8. As observed, theoretical predictions agree with the experimental data, particularly in the range of 11,111 cm⁻¹ (900 nm) to 5,882 cm⁻¹ (1700 nm), where most of the spectroscopic features were observed. The errors associated to the calculation are in the range of 105 and 760 cm⁻¹. This has also been observed in other Sm(III) complexes, where the 4f-4f transitions were observed in the same energy region with identical assignments.⁴⁶–⁴⁹
+
+In 1−Gd, the scenario is simpler as it corresponds to a half-filled f-shell. Given its 4f⁷ configuration, the main transitions are expected to be in the high energy region involving the J = 7/2–3/2 (⁶P) and J = 9/2−1/2 (⁶D) manifolds. In this case, the SO-GS corresponds to a J = 7/2 (⁸S) with a splitting of ~15 cm⁻¹. The first excited state appears at ~29,344 cm⁻¹ (341 nm) and was ascribed to the J = 7/2 (⁶D) manifold. The energetic stabilization of the first excited multiplet has been previously reported for other f⁷ systems at different levels of theory.³,⁵⁰,⁵¹ A detailed assignment of each transition is shown in Table S9-S10. Since the CT band in 1−Gd masks the fingerprint f−f transitions associated to the Gd³⁺ ion, no experimental counterpart exists to assess the accuracy of the predicted transitions. Reports on other Gd(III) compounds have found the position of the first excited state to be between 32,467 cm⁻¹ (308 nm) and 31,645 cm⁻¹ (316 nm).⁵²–⁵⁵ The calculated value for 1−Gd exhibits a bathochromic shift of ~2,300 cm⁻¹ (25 nm) with respect to the range of energy found in previous reports.
+
+Since Cm³⁺ is the isoelectronic analog of Gd³⁺, similar electronic states at different energy ranges are expected. As observed in Table S11-S12, the SO-GS of 1−Cm corresponds to a J = 7/2 (⁸S) with a splitting of ~385 cm⁻¹. The first two excited manifolds J = 7/2 (⁶D) and J = 5/2 (⁶D), usually considered the emissive states, start to appear around 15,216 cm⁻¹ (657 nm) and 16,650 cm⁻¹ (601 nm), respectively. Therefore, the experimental absorption peaks observed between 15,873 cm⁻¹ (630 nm) and 15,384 cm⁻¹ (650 nm) can be ascribed to transitions towards these couple of excited manifolds. Regarding the Cp′₃Cm system, the assignment and position of the electronic states is almost identical to the ones found in 1−Cm. The SO-GS corresponds to a J = 7/2 (⁸S) exhibiting a splitting of ~351 cm⁻¹, whereas the two first excited manifolds are located at ~15,121 cm⁻¹ (661 nm) and 16,365 cm⁻¹ (611 nm). As shown in the experimental section (Fig. 3), the photoluminescence spectrum of this complex shows a band between ~16,260 cm⁻¹ (615 nm) and ~14,285 cm⁻¹ (700 nm) with three prominent peaks observed at ~15,504 cm⁻¹ (645 nm), 15,152 cm⁻¹ (660 nm) and 14,925 cm⁻¹ (670 nm). This band can be assigned to transitions from the J = 7/2 (⁶D) manifold to the SO-GS, where the three observed peaks can be attributed to the splitting of this multiplet (Table S12). In general, curium emits in the range of 16,949 cm⁻¹ (590 nm) to 16,129 cm⁻¹ (620 nm).³,⁷,³³,³⁴,³⁷,³⁸ In this case, a non-negligible bathochromic shift that was accurately reproduced by the calculations, is observed. This phenomenon has usually been attributed to the nephelauxetic effect⁵⁶ where the ligand-field of Cp′ causes a unique reduction in the interelectronic repulsion between the f-electrons of curium(III).
+
+According to SO-pDFT calculations, 1−Cm and Cp′₃Cm systems are similar in terms of assignment and positions of the SO states. Nonetheless, while 1−Cm photoluminescence is quenched, Cp′₃Cm display a prominent emission band in the vis-NIR region. It is known that C−H vibrations can lead to non-radiative routes when the low quanta superior harmonics and vibrational modes of a coordinated ligand strongly resonate with the emissive state of the metal center.⁵⁷–⁵⁹ For instance, it has been reported that the fundamental vibrational modes of the C−H stretching in 4,4′−bipyridine are found at around 3,000 cm⁻¹.⁶⁰,⁶¹ Since C−H vibrational modes has proven to have a cumulative quenching effect with respect to the number of C−H bonds,⁵⁷ a resonance between the electronic emissive state of curium(III) and the fifth vibrational harmonic of 4,4′−bipyridine seems to be a reasonable explanation for the absence of emission in 1−Cm.
+
+Given the differences observed in the spectroscopy of curium in two similar environments, it is important to analyze the subtle differences that may arise in curium-ligand bonds when dimerizing two Cp′₃Cm units bridged by 4,4′−bpy to form 1−Cm. To do so, the natural bond orbital (NBO) approximation and the Quantum Theory of Atoms in Molecules (QTAIM) were relied upon. The analysis of the ground-state wavefunction in a localized formalism such as natural localized molecular orbitals (NLMOs) within the NBO analysis provide a unique and simple way to rationalize the chemical bond in the context of Lewis’ chemical intuition. Conversely, the topological analyses of the electron density through QTAIM metrics lay out a more holistic picture of the interactions and energies associated with the electron density in the interatomic region.
+
+The nature of the metal−Cp′ interaction is something that is not well-understood because of the delocalization on the Cp′ ring. Under the NBO localization formalism, these interactions are shown as one π-type bond with the metal-center per ligand (Fig. 5). It is interesting to note that among the three analog complexes, 1−Sm displays the major engagement of f-orbitals to bond formation (Table S15). This is unexpected as we generally see a greater participation of their 5f-electrons in bonding than the lanthanide 4f-orbitals. This serves as additional evidence that the cyclopentadienyl-derived ligands have unique electronic properties that result in a differential interaction between lanthanides and actinides. This scenario differs significantly when the metal−N bpy bonds are analyzed, where the classic trend is recovered with 1−Cm showing the stronger orbital mixing (almost twice as much the lanthanide mixing) and greater participation of the f-orbitals in bonding compared to 1−Sm and 1−Gd. Thus, for the metal-Cp′ bonds, 1−Sm shows an increased degree of orbital mixing with respect to 1−Cm and 1−Gd, whose metal-carbon bonds have similar hybrid contributions to their NLMOs. Whereas for metal-N bpy bonds 1−Sm and 1−Gd resemble each other with 1−Cm showing higher metal hybrid contributions to the bond than the lanthanide analogs (Table S15). On the other hand, the role of the coordination of 4,4′−bpy in orbital mixing on 1−Cm is not significant as Cm−Cp′ bonds in Cp′₃Cm are almost identical to those in 1−Cm.
+
+As a good complement to the view of the chemical bond given from localized molecular orbitals, analyzing the topology of the electron density provides a good opportunity to evaluate the accumulation of electron density, ρ(r), and the balance of kinetic and potential energies at the point where two basins (atomic fragments) contact each other along the bond path (bond critical point, BCP). Despite 1−Sm showed a more pronounced Sm−Cp′ orbital mixing, the accumulation of electron density at the BCP is very similar among the three congener systems (Table 1). Conversely, the balance of potential and kinetic energy densities, V(r) and G(r), respectively, show that 1−Sm presents a similar stability of electron density at the BCP to that of 1−Cm. However, when the total energy density, H(r), is normalized against ρ(r) the difference in energy can be interpreted as effective differences in covalency. According to this metric, frequently referred to as covalency degree, it is suggested that Sm-C bonds in 1−Sm have a greater energetic stabilization caused by covalent interactions compared to those in 1−Cm and 1−Gd by ~4 kJ mol⁻¹ and ~17 kJ mol⁻¹, respectively (Table 1). This supports the idea that the higher degree of mixing shown in Sm−Cp′ bonds correlate with the degree of covalency from an energetic perspective. On the other hand, metal−N bpy bonds show slightly larger ρ(r) values with Cm−N being the highest. The most striking difference is that kinetic energies associated are significantly increased, reducing significantly H(r) values and even switching to positive values as in Sm−N and Gd−N bonds. This indicates that these bonds have no covalent character and only Cm−N bpy bonds have a (low) covalent character. Moreover, these bonds can be compared to the previously reported values for 1−Nd and 1−Am, where both show positive values and set an unusual precedent for differences in Am(III)−N and Cm(III)−N bonds. A final comparison can be made between 1−Cm and Cp′₃Cm bonds, where from the NLMO analysis no major differences were found, but from a topological perspective we see significant differences. The coordination of the 4, 4′−bpy seems to reduce the accumulation of electron density at the BCP while decreasing the magnitude of the potential energy with respect to kinetic energy resulting in less than half of the total energy density in 1−Cm (−21.9 kJ mol⁻¹ Å⁻³) compared to Cp′₃Cm (−51.0 kJ mol⁻¹ Å⁻³) (Table 1). This suggests that direct correlations between orbital mixing and the degree of covalency cannot be assumed and must be contrasted carefully with alternative methods, where energies can be derived.
+
+# Conclusions
+
+Characterization of (Cp′₃Cm)₂(µ−4,4′−bpy) and its lanthanide analogues has been accomplished and provides new insights into Cm−C interaction as well as laying the groundwork for expanding organometallic curium chemistry. The Cm−C bonds in the putative Cp′₃Cm complex show lower-energy emission compared traditional curium complexes. Additionally, coordination of 4,4′−bipyridine to Cp′₃Cm leads to the complete quenching of curium’s fingerprint emissive states owing to the resonance between the C−H vibrational states in 4,4′−bipyridine and the emission energy of Cm(III). The bonding patterns calculated from these three complexes shows atypical features, especially the unusual orbital mixing observed in Sm−Cp′ bonds when compared to the curium and gadolinium complexes. Overall, the coordination of 4,4′−bpy does not significantly affect the orbital mixing in Cm−Cp′, but it reduces the electron density accumulated in the interatomic region that ultimately translates to lower degrees of covalency.
+
+# Methods
+
+Caution! $^{248}\text{Cm}$ ($t_{1/2} = 348,000$ y) presents serious health hazards due to α-emission (5.078 MeV, 75%) and spontaneous fission (~8.4%), resulting in considerable neutron emission (80 mrem/h). All reactions and handling of $^{248}\text{Cm}$ were completed in a Category II radiological facility with HEPA equipped fume hoods and gloveboxes utilizing strict safety controls.
+
+Anhydrous SmCl$_3$ (Sigma, 99.9%), anhydrous GdCl$_3$ (Sigma, 99.99%), 4,4′−bpy (Sigma, 98%), Bromotrimethylsilane (Sigma, 97%), Hydrobromic Acid (Sigma, 8.77 M), and distilled water were used as received. KCp′ and Cp′$_3$Sm were synthesized by their respective literature procedures. $^{23}$, $^{62}$ Toluene (Sigma), hexane (Fisher Scientific), diethyl ether (Fisher Scientific), and dimethoxyethane (Sigma) were distilled using sodium benzophenone ketyl and stored on activated 3 Å molecular sieves (Sigma). Toluene, hexane, and diethyl ether were further dried over NaK for 24 hours and filtered through activated alumina neutral alumina immediately before use. Dimethoxyethane was stored on activated neutral alumina for 24 hours and filtered through more activated neutral alumina immediately before use.
+
+All reactions were completed using Schlenk line and glovebox techniques in an argon atmosphere with exclusion of air and water unless noted otherwise. All handling of $^{248}\text{Cm}$ was completed in a HEPA filter equipped fume hood and negative pressure glovebox attached to the HEPA line.
+
+**(Cp′$_3$Sm)$_2$($\mu$−4,4′−bpy), 1−Sm.** Using a scintillation vial (20 mL), the addition of 4,4′−bpy (2.6 mg, 0.017 mmol) to Cp′$_3$Sm (18.7 mg, 0.033 mmol) in toluene (2 mL) resulted in the immediate formation of an orange-yellow precipitate. The slurry was stirred vigorously overnight at room temperature, dried under reduced pressure, rinsed with hexane to remove any unreacted Cp′$_3$Sm, and once again dried under the reduced pressure. The powder was dissolved in toluene and heated to 120°C (with the cap removed) while gentle stirring, resulting in an orange solution. Upon reaching a gentle boil, stirring was ceased and the sample was manually cooled down 70°C in 10°C increments over a period of 20 minutes. At 70°C the cap was reapplied to the vial carefully so as not to agitate the sample. The heat was then turned off and the vial was slowly cooled to room temperature, resulting in the growth of yellow crystals suitable for single-crystal x-ray diffraction studies after 3 hours. UV-vis-NIR (toluene): $\lambda_{\text{max}}$ nm (cm$^{-1}$) = 944 (10,593), 1,059 (9,443), 1,079 (9,268), 1,105 (9,050), 1,212 (8,251), 1,235 (8,097), 1,248 (8,013), 1,274 (7,849), 1,335 (7,491), 1,377 (7,262), 1,430 (6,993), 1,460 (6,849), 1,485 (6,734), 1,515 (6,601), 1,569 (6,373), 1,631 (6,131).
+
+**(Cp′$_3$Gd)$_2$($\mu$−4,4′−bpy), 1−Gd.** In a slight modification to the literature procedure, $^{24}$ Cp′$_3$Gd was synthesized by the addition KCp′ (115 mg, 0.65 mmol) to anhydrous GdCl$_3$ (50 mg, 0.19 mmol) in toluene (2 mL) and stirred at 70°C overnight. The green slurry was centrifuged and KCl was filtered off, followed by rinsing the pellet with toluene (3 × 1 mL). Toluene was evaporated under reduced pressure and the resulting powder was taken up in hexane to extract any unreacted KCp′. The slurry was again centrifuged, rinsed with hexane (3 × 1 mL), and dried under reduced pressure, resulting in a light green powder (89.4 mg, 0.157 mmol, yield 82.7%). (Cp′$_3$Gd)$_2$($\mu$−4,4′−bpy) was synthesized by the addition of 4,4′−bpy (2.5 mg, 0.016 mmol) to Cp′$_3$Gd (18.4 mg, 0.032 mmol) in toluene (2 mL). The bright yellow precipitate was dried under reduced pressure, rinsed with hexane, and taken up once again in toluene. Crystallization was completed analogous to **1−Sm**, resulting in bright orange-yellow crystals suitable for single-crystal X-ray diffraction studies.
+
+**(Cp′$_3$Cm)$_2$($\mu$−4,4′−bpy), 1−Cm.** A stock solution of Cm$^{3+}$ (3.3 mg Cm content, 0.013 mmol) in HCl (2 M) was dried under a nitrogen stream, dissolved in water, and transferred to a falcon tube (15 mL). Excess NH$_4$OH (~2.5 mL) was added dropwise, resulting in the immediate precipitation of Cm(OH)$_3$ $\bullet$ nH$_2$O. The slurry was centrifuged and the pellet was rinsed with water (3 × 1 mL). Cm(OH)$_3$ $\bullet$ nH$_2$O was dissolved in minimal HBr (1 mL, 8.77 M), pipetted to a scintillation vial (20 mL), and dried under a stream of nitrogen. The colorless CmBr$_3$ $\bullet$ nH$_2$O was rinsed with diethyl ether (2 × 1 mL) and transported into a glovebox overnight.
+
+Following actinide drying procedures reported previously, $^{27}$, $^{29}$, $^{63}$ DME (1.5 mL) was added dropwise to CmBr$_3$ $\bullet$ nH$_2$O. The slurry was stirred for 10 minutes, followed by the dropwise addition of bromotrimethylsilane (TMS−Br, 1.5 mL). The slurry was stirred at 50°C for 2 hours and the powder began to dissolve slowly. After cooling to room temperature, hexane (4 mL) was added, resulting in a white precipitate. The sample was allowed to settle for 15 minutes before the supernatant was pipetted away, and then was rinsed with hexane (3 × 2 mL), stirring for 5 minutes and resting for 15 minutes between each rinse, followed by drying under reduced pressure 30 minutes. The resulting white powder was taken up in diethyl ether (1.5 mL), stirred for 20 minutes, and further precipitated out by the addition of hexane (4 mL). The slurry was stirred vigorously for 5 minutes and allowed to settle for 15 minutes before pipetting away the supernatant and drying under reduced pressure for 2 hours, resulting in a white powder of putative CmBr$_3$(DME)$_n$.
+
+KCp′ (8.2 mg, 0.046 mmol) was dissolved in toluene (1.5 mL), added dropwise to CmBr$_3$(DME)$_n$, and stirred vigorously at 70°C for 2 hours. A color change from colorless to tan was observed. The sample was centrifuged to remove the KBr byproduct and filtered, followed by rinsing the pellets with toluene (3 × 0.5 mL each pellet). Toluene was evaporated under reduced pressure and hexane (1.5 mL) was added to precipitate and extract excess KCp′. The slurry was centrifuged, supernatant filtered, and the pellets were washed with hexane (3 × 0.5 mL each pellet). About 40 µL of the resulting champagne-colored solution was isolated and diluted to 1.5 mL for solution phase absorption studies. Putative CmCp′$_3$ isolated and was dried under reduced pressure, resulting in a tan oil (7.1 mg, 0.011 mmol, yield 81%). Two crystal clusters were extracted for photoluminescence studies. The crystal quality was not suitable for single-crystal X-ray diffraction but were suitable for study emission studies.
+
+An immediate color change to bright yellow was observed upon the addition of 4,4′−bpy (1.0 mg, 0.006 mmol) to CmCp′$_3$ in toluene (2 mL). The solution was transferred from a 20 mL scintillation vial to a 6 mL scintillation vial and concentrated under reduced pressure until an orange-yellow powder was observed (~0.3 mL toluene). The slurry was slowly stirred and heated to 120°C resulting in the orange powder dissolving. Similar to **1−Sm** and **1−Gd**, the solution was slowly cooled to 70°C in 10°C increments. At 70°C the vial was capped and the hotplate was turned off, allowing the sample to slowly cool to room temperature. Around 60°C a color change from orange to an emerald green was observed, followed by a slow color change to a sapphire blue over a period of about an hour. The blue solution rested overnight at room temperature. The following morning the sample was again yellow and gold crystals suitable for single-crystal X-ray diffraction were retrieved. UV-vis-NIR (toluene): $\lambda_{\text{max}}$ nm (cm$^{-1}$) = 587 (17,036), 597 (16,750), 631 (15,848), 641 (15,601).
+
+# References
+
+1. Seaborg, G. T. & Segrè, E. The trans-uranium elements. Nature **159**, 863–865 (1947).
+2. Mikheev, N. B. & Kamenskaya, A. N. Complex formation of the lanthanides and actinides in lower oxidation states. Chem. Rev. **109**, 1–59 (1991).
+3. Sperling, J. M. et al. Compression of curium pyrrolidine-dithiocarbamate enhances covalency. Nature **583**, 396–399 (2020).
+4. Wybourne, B. G. Spectroscopic Properties of Rare Earths. (1965).
+5. Abergel, R. J. & Ansoborlo, E. Curious Curium. Nat. Chem. **8**, 516 (2016).
+6. Matveev, P. I. et al. A first phosphine oxide-based extractant with high Am/Cm selectivity. Dalton Trans. **48**, 2554–2559 (2019).
+7. Kim, J. I., Klenze, R. & Wimmer, H. Fluorescence spectroscopy of curium (III) and application. Eur. J. Solid State Inorg. Chem. **28**, 347–356 (1991).
+8. Carnall, W. T. & Rajnak, K. Electronic energy level and intensity correlations in the spectra of the trivalent actinide aquo ions. II. Cm³⁺. J. Chem. Phys. **63**, 3510–3514 (1975).
+9. Carnall, W. T. A systematic analysis of the spectra of trivalent actinide chlorides in D₃h site symmetry. J. Chem. Phys. **96**, 8713–8726 (1992).
+10. Yusov, A. B. Luminscence of transplutonium elements and its application. Radiokhimiya **35**, 3–25 (1993).
+11. Carnall, W. T. & Wybourne, B. G. Electronic Energy Levels of the Lighter Actinides: U³⁺, Np³⁺, Pu³⁺, Am³⁺, and Cm³⁺. J. Chem. Phys. **40**, 3428–3433 (1964).
+12. Carnall, W. T. Energies and Intensities of Some Low-Lying Levels of Trivalent Curium. J. Chem. Phys. **47**, 3081–3082 (1967).
+13. Wybourne, B. G. Low-Lying Levels of Trivalent Curium. J. Chem. Phys. **40**, 1456–1457 (1964).
+14. Edelstein, N. M., Klenze, R., Fanghänel, T. & Hubert, S. Optical properties of Cm(III) in crystals and solutions and their application to Cm(III) speciation. Coord. Chem. Rev. **250**, 948–973 (2006).
+15. Carnall, W. T. & Fields, P. R. A Study of the Complexes of Curium(III) by Absorption Spectrometry. J. Am. Chem. Soc. **81**, 4445–4449 (1959).
+16. Carnall, W. T., Fields, P. R. & Stewart, D. C. The absorption spectrum of aqueous curium (III). J. Inorg. Nucl. Chem. **6**, 213–216 (1958).
+17. Streitwieser, A. & Müller-Westerhoff, U. Bis(cyclooctatetraenyl)uranium (Uranocene). A New Class of Sandwich Complexes That Utilize Atomic f Orbitals. J. Am. Chem. Soc. **90**, 7364 (1968).
+18. Streitwieser, A. et al. Preparation and Properties of Uranocene, Di-π-cyclooctatetraeneuranium(IV). J. Am. Chem. Soc. **95**, 8644–8649 (1972).
+19. Evans, W. J. Tutorial on the Role of Cyclopentadienyl Ligands in the Discovery of Molecular Complexes of the Rare-Earth and Actinide Metals. Organometal. **35**, 3088–3100 (2016).
+20. Windorff, C. J. et al. Identification of the Formal + 2 Oxidation State of Plutonium: Synthesis and Characterization of {Pu²⁺[C₅H₃(SiMe₃)₂]₃}⁻. J. Am. Chem. Soc. **139**, 3970–3973 (2017).
+21. Su, J. et al. Identification of the Formal + 2 Oxidation State of Neptunium: Synthesis and Structural Characterization of {Np²⁺[C₅H₃(SiMe₃)₂]₃}¹⁻. J. Am. Chem. Soc. **140**, 7425–7428 (2018).
+22. MacDonald, M. R. et al. Identification of the + 2 Oxidation State for Uranium in a Crystalline Molecular Complex, [K(2.2.2-Cryptand)][(C₅H₄SiMe₃)₃U]. J. Am. Chem. Soc. **135**, 13310–13313 (2013).
+23. Fieser, M. E. et al. Structural, Spectroscopic, and Theoretical Comparison of Traditional vs Recently Discovered Ln²⁺ Ions in the [K(2.2.2-cryptand)][(C₅H₄SiMe₃)₃Ln] Complexes: The Variable Nature of Dy²⁺ and Nd²⁺. J. Am. Chem. Soc. **137**, 369–382 (2015).
+24. MacDonald, M. R., Bates, J. E., Ziller, J. W., Furche, F. & Evans, W. J. Completing the Series of + 2 Ions for the Lanthanide Elements: Synthesis of Molecular Complexes of Pr²⁺, Gd²⁺, Tb²⁺, and Lu²⁺. J. Am. Chem. Soc. **135**, 9857–9868 (2013).
+25. MacDonald, M. R. et al. Expanding Rare-Earth Oxidation State Chemistry to Molecular Complexes of Holmium(II) and Erbium(II). J. Am. Chem. Soc. **134**, 8420–8423 (2012).
+26. Long, B. N. et al. Cyclopentadienyl Coordination Induces Unexpected Ionic Am – N Bonding in an Americium Bipyridyl Complex. Nat. Commun. **13**, 1–7 (2021).
+27. Goodwin, C. A. P. et al. [Am(C₅Me₄H)₃]: An Organometallic Americium Complex. Angew. Chemie **131**, 11821–11825 (2019).
+28. Goodwin, C. A. P. et al. Isolation and characterization of a californium metallocene. Nature **599**, 421–424 (2021).
+29. Windorff, C. J. et al. A Single Small-Scale Plutonium Redox Reaction System Yields Three Crystallographically-Characterizable Organoplutonium Complexes. Inorg. Chem. **59**, 13301–13314 (2020).
+30. Shannon, R. D. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta. Cryst. **A32**, 751–767 (1976).
+31. Deacon, G. B., Gatehouse, B. M., Platts, S. N. & Wilkinson, D. L. Organolanthanoids. XI. The Crystal and Molecular Structures of Two Tris(η⁵-cyclopentadienyl)(pyridine)-lanthanoid(III) Compounds. Aust. J. Chem. **40**, 907–914 (1987).
+32. Baisch, U. et al. Nanocrystalline lanthanide nitride materials synthesised by thermal treatment of amido and ammine metallocenes: X-ray studies and DFT calculations. Chem. Eur. J. **12**, 4785–4798 (2006).
+33. Cary, S. K. et al. Emergence of californium as the second transitional element in the actinide series. Nat. Commun., 1–8 (2015).
+34. Morss, L. R., Edelstein, N. & Fuger, J. The Chemistry of the Actinide and Transactinide Elements Third Edition. 1364–1365 (Springer, 2006).
+35. Mehdoui, T., Berthet, J. C., Thuery, P. & Ephritikhine, M. CCDC 958634: Experimental Crystal Structure Determination. (2013).
+36. Formanuik, A. et al. Double Reduction of 4,4′–Bipyridine and Reductive Coupling of Pyridine by Two Thorium(III) Single-Electron Transfers. Chem. Eur. J. **23**, 2290–2293 (2017).
+37. Cary, S. K. et al. Spontaneous Partitioning of Californium from Curium: Curious Cases from the Crystallization of Curium Coordination Complexes. Inorg. Chem. **54**, 11399–11404 (2015).
+38. Cary, S. K. et al. A series of dithiocarbamates for americium, curium, and californium. Dalton Trans. **47**, 14452–14461 (2018).
+39. Nugent, L. J., Laubereau, P. G., Werner, G. K. & Vander Sluis, K. L. Noncovalent Character in the Chemical Bonds of the Lanthanide(III) and the Actinide(III) Tricyclopentadienides. Organometal. Chem. **27**, 365–372 (1971).
+40. Laubereau, P. G. & Burns, J. H. Tricyclopentadienyl-curium. Inorg. Nucl. Chem. Lett. **6**, 59–63 (1970).
+41. Bagus, P. S., Ilton, E. S., Martin, R. L., Jensen, H. A. A. & Knecht, S. Spin-orbit coupling in actinide cations. Chem. Phys. Lett. **546**, 58–62 (2012).
+42. Bagus, P. S., Freund, H. J., Kuhlenbeck, H. & Ilton, E. S. A new analysis of X-ray adsorption branching ratios: Use of Russell-Saunders coupling. Chem. Phys. Lett. **455**, 331–334 (2008).
+43. Sperling, J. M. et al. Pronounced Pressure Dependence of Electronic Transitions for Americium Compared to Isomorphous Neodymium and Samarium Mellitates. Inorg. Chem. **60**, 476–483 (2021).
+44. Carnall, W. T., Fields, P. R. & Rajnak, K. Electronic Energy Levels in the Trivalent Lanthanide Aquo Ions. I. Pr³⁺, Nd³⁺, Pm³⁺, Sm³⁺, Dy³⁺, Ho³⁺, Er³⁺, and Tm³⁺. J. Chem. Phys. **49**, 4424–4442 (1968).
+45. Binnemans, K. & Görller-Walrand, C. On the color of the trivalent lanthanide ions. Chem. Phys. Lett. **235**, 163–174 (1995).
+46. Carnall, W. T., Fields, P. R. & Rajnak, K. Electronic Energy Levels in the Trivalent Lanthanide Aquo Ions. I. Pr³⁺, Nd³⁺, Pm³⁺, Sm³⁺, Dy³⁺, Ho³⁺, Er³⁺, and Tm³⁺. J. Chem. Phys. **49**, 4424 (1968).
+47. Rast, H. E., Fry, J. L. & Caspers, H. H. Energy Levels of Sm³⁺ in LaF₃. J. Chem. Phys. **46**, 1460 (1967).
+48. Sailaja, S. et al. Optical properties of Sm³⁺-doped cadmium bismuth borate glasses. J. Mol. Struct. **1038**, 29–34 (2013).
+49. Matthes, P. R. et al. The Series of d Rare Earth Complexes [Ln₂Cl₆(µ-4,4′-bipy)(py)₆], Ln = Y, Pr, Nd, Sm-Yb: A Molecular Model System for Luminescence Properties in MOFs Based on LnCl₃ and 4,4′-Bipyridine. **19**, 17369–17378 (2013).
+50. Murdoch, K. M., Nguyen, A. D., Edelstein, N. M., Hubert, S. & Ga⁁con, J. C. Two-photon absorption spectroscopy of Cm³⁺ in LuPO₄. Phys. Rev. B **56**, 3038 (1997).
+51. Illemassene, M., Murdoch, K. M., Edelstein, N. M. & Krupa, J. Optical spectroscopy and crystal field analysis of Cm³⁺ in LaCl₃. J. Lumin. **75**, 77–87 (1997).
+52. Strasser, A. & Vogler, A. Phosphorescence of gadolinium(III) chelates under ambient conditions. Inorganica Chim. Acta **357**, 2345–2348 (2004).
+53. Lanje, M. M. & al., e. J. Phys.: Conf. Ser. **1913** (2021).
+54. Carnall, W. T., Fields, P. R. & Rajnak, K. Spectral Intensities of the Trivalent Lanthanides and Actinides in Solution. II. Pm³⁺, Sm³⁺, Eu³⁺, Gd³⁺, Tb³⁺, Dy³⁺, and Ho³⁺. J. Chem. Phys. **49**, 4412 (1968).
+55. Carnall, W. T., Fields, P. R. & Rajnak, K. Electronic Energy Levels of the Trivalent Lanthanide Aquo Ions. II. Gd³⁺. J. Chem. Phys. **49**, 4443 (1968).
+56. Goodwin, C. A. P. et al. Structural and Spectroscopic Comparison of Soft-Se vs. Hard-O Donor Bonding in Trivalent Americium/Neodymium Molecules. Angew. Chem. Int. Ed. **60**, 9459–9466 (2021).
+57. HemmilÄ, I., Mukkala, V.-M. & Takalo, H. Effect of C-H bonds on the quenching of luminescent lanthanide chelates. J. Fluoresc. **5**, 159–163 (1995).
+58. Rabouw, F. T. et al. Quenching Pathways in NaYF₄:Er³⁺,Yb³⁺ Upconversion Nanocrystals. ACS Nano **12**, 4812–4823 (2018).
+59. Dimitrije, M. et al. Vibrational Quenching in Near-Infrared Emitting Lanthanide Complexes: A Quantitative Experimental Study and Novel Insights. Chem. Eur. J. **25**, 15944–15956 (2019).
+60. Topacli, A. & Akyüz, S. 4,4′-Bipyridyl: vibrational assignments and force field. Spectrochim. Acta A Mol. Biomol. Spectrosc. **51**, 633–641 (1995).
+61. Neto, N., Muniz-Miranda, M., Angeloni, L. & Castellucci, E. Normal mode analysis of 2,2′-bipyridine—I. Internal modes. Spectrochim. Acta A Mol. Spectrosc. **39**, 97–106 (1983).
+62. Peterson, J. K., MacDonald, M. R., Ziller, J. W. & Evans, W. J. Synthetic Aspects of (C₅H₄SiMe₃)₃Ln Rare-Earth Chemistry: Formation of (C₅H₄SiMe₃)₃Lu via [(C₅H₄SiMe₃)₂Ln]⁺ Metallocene Precursors. Organometallics **32**, 2625–2631 (2013).
+63. Wilkerson, M. P., Burns, C. J., Paine, R. T. & Scott, B. L. Synthesis of Crystal Structure of UO₂Cl₂(THF)₃: A Simple Preparation of an Anhydrous Uranyl Reagent. Inorg. Chem. **38**, 4156–4158 (1999).
+
+# tables
+
+**Table 1**. Topological metrics of the electron density of **1−Sm**, **1−Gd**, **1−Cm**, and Cp′₃Cm. The electron densities were derived from ground state-specific CASSCF calculations.
+
+| | **1−Sm** | | **1−Gd** | | **1−Cm** | | Cp′₃Cm |
+|--- | --- | --- | --- | --- | --- | --- | ---|
+| | Sm−C<sub>avg</sub> | Sm−N | Gd−C<sub>avg</sub> | Gd−N | Cm−C<sub>avg</sub> | Cm−N | Cm−C<sub>avg</sub> |
+| ρ(r) | 0.214 | 0.238 | 0.213 | 0.244 | 0.224 | 0.275 | 0.244 |
+| ∇²ρ(r) | 2.435 | 3.516 | 2.503 | 3.747 | 2.817 | 3.940 | 2.796 |
+| V(r) | −486.8 | −584.6 | −491.7 | −620.4 | −556.7 | −751.6 | −611.0 |
+| G(r) | 465.0 | 612.3 | 473.6 | 651.2 | 534.7 | 734.4 | 559.9 |
+| H(r) | −21.8 | 27.8 | −18.1 | 30.8 | −21.9 | −17.2 | −51.0 |
+| |V(r)/G(r)| | 1.047 | 0.955 | 1.038 | 0.953 | 1.041 | 1.023 | 1.091 |
+| H(r)/ρ(r) | −101.5 | 116.4 | −85.0 | 126.3 | −98.0 | −62.6 | −209.0 |
+| ε(r) | 3.797 | 0.066 | 3.713 | 0.063 | 3.766 | 0.052 | 3.349 |
+| δ(r) | 0.135 | 0.180 | 0.132 | 0.183 | 0.147 | 0.227 | 0.163 |
+| WBI | 0.129 | 0.161 | 0.123 | 0.168 | 0.172 | 0.243 | 0.159 |
+
+*Metrics tabulated correspond to values measured at the bond critical point except for the delocalization, δ(r), and Wiberg bond indices (WBI). The electron density, ρ(r) is expressed in e Å⁻³, whereas the Laplacian in e Å⁻⁵. Potential V(r), kinetic G(r), and total H(r) energy densities are expressed in kJ mol⁻¹ Å⁻³. Metal−Cp′ bonds are shown as averages of all metal−C bonds.*
+
+# Supplementary Files
+
+- [SupplementaryInformation.docx](https://assets-eu.researchsquare.com/files/rs-2558183/v1/c421e39a8d206ef6fc5880c2.docx)

0baf27384321bec138bb51b4ad842f61b51161ccb0c9ed323e1fc6c143bbf9d7/preprint/images_list.json

@@ -0,0 +1,50 @@
+[
+    {
+        "type": "image",
+        "img_path": "images/Figure_1.png",
+        "caption": "Schematic illustrations depicting the particle size dependent phase evolutions of olivine FePO4 particles during lithiation or sodiation. Different color codes denote different phases during lithiation or sodiation. The dashed boxes in the diagram indicate phase evolutions that have not been observed experimentally.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_2.png",
+        "caption": "Particle morphology features. a, Schematic illustration of the ion insertion process within a carbon-coated FePO4 crystal. Ion enters from the (010) facet and migrates along the [010] direction. b, Constructed Wulff shape of LiFePO4 using the calculated surface energies of specific orientations. c-g, Scanning transmission electron micrographs (STEM) and selected area electron diffraction (SAED) patterns (top-right) of Platelet-20 nm (c), Platelet-45 nm (d), Cuboid-87 nm (e), Platelet-600 nm (f) and Platelet-1200 nm (g) particles. The SAED patterns were taken along the axis labeled in the red across. The red arrows denote some specific orientations of the particles. Scale bars in c-e, 100 nm. Scale bars in f-g, 2 \u03bcm. h, Scanning electron micrograph (SEM) of Cuboid-6000 nm particles. Scale bars, 2 \u03bcm. i, Schematic diagram illustrating some characteristic morphology features.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_3.png",
+        "caption": "Electrochemical response during lithiation or sodiation. a, 1st electrochemical cycle under 17 mA/g (equivalent to 0.1C based on theoretical capacity of LiFePO4) in 60 ml 1M LiCl aqueous solution (paired with Ag/AgCl/KCl (4.0 M) reference and LixFePO4 counter electrodes). b, 1st electrochemical cycle under 15.4 mA/g (equivalent to 0.1C based on theoretical capacity of NaFePO4) in 60 ml 1M NaCl aqueous solution (paired with Ag/AgCl/KCl (4.0 M) reference and NayFePO4 counter electrodes). c, Bar chart comparisons of end potential collected right after different depth-of-discharge (DOD) in 60 ml 1M LiCl \u00a0(left panel) or NaCl (right panel) aqueous solution, which corresponds to the value at the bottom of the bar, and open-circuit potential after 20 hours of relaxation without currents, corresponding to the value at the top of the bar. See Methods for electrode preparation and DOD calculation. The dashed lines denotes the calculated thermodynamic voltage for specific reactions (See Supplementary Note 1 for computation details).",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_4.png",
+        "caption": "In situsynchrotron X-ray diffraction (XRD) tracking of phase evolutions during lithiation or sodiation. a, Lithiation of Platelet-20 nm particles at 0.43C. b, Lithiation of Cuboid-87 nm particles at 0.88C. c, Sodiation of Platelet-20 nm particles at 0.42C. d, Sodiation of Cuboid-87 nm particles at 0.44C. e, Lithiation of Platelet-600 nm particles at 0.092C. f, Lithiation of Platelet-1200 nm particles at 0.077C. g, Sodiation of Platelet-600 nm particles at 0.090C. h, Sodiation of Platelet-1200 nm particles at 0.086C. i, Snapshots (initial scan and last scan) of in situ synchrotron XRD during sodiation as well as the ex situsynchrotron XRD of the electrodes after ~ 10 hours relaxation in the open air for Platelet-1200 nm particles. j, Schematic showing the lattice distortion and relaxation processes with the corresponding XRD peak features.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_5.png",
+        "caption": "Li extraction performance and non-faradic ion-exchange. a, Li/(Li+Na) ratios after recovery of different electrodes from 1:1000 Li to Na solution using 70% accessible capacity and 0.1C' extraction rate. b-c, Li/(Li+Na) ratios after recovery of electrodes used under different extraction rates from 1:1000 Li to Na solution (b) and 1:100 Li to Na solution (c) using 70% accessible capacity. d, Intercalation curves of Platelet-20 nm and Platelet-600 nm particles under different extraction rates. e, Measured Na/(Na+Li) ratios of the electrodes after soaking in 1 mM LiCl and 1 M NaCl mixed solution for 7 hours, using Li pre-intercalated particles. DOD_Li0.1'/0.35'/0.5' denotes the seeding percentage based on the accessible capacity. f-g, Measured Na/(Na+Li) ratios at different times of Platelet-20 nm and Cuboid-6000 nm particles after soaking in 1 mM LiCl and 1 M NaCl mixed solution, using DOD_Li0.1'/0.35'/0.5' pre-intercalated particles. h, In situ synchrotron XRD tracking of Platelet-20 nm pre-intercalated particles (DOD_Li0.50') during ion-exchange in 1 mM LiCl and 1 M NaCl mixed solution. Mass loading for electrodes used in a-g is ~ 2.5 mg/cm2. Error bars represent the standard deviation of three replicate measurements.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    },
+    {
+        "type": "image",
+        "img_path": "images/Figure_6.png",
+        "caption": "Quantitative correlation maps for each group illustrating Li competitiveness, particle morphology, and electrochemical characteristics features. a, Coefficient of correlation (R) map for particles in Group 1. b, Coefficient of correlation (R) map for particles in Group 2. Each value calculated in this map represents the degree of relationship between two variables. Specifically, \u201c0.1C'_Li%\u201d represents the recovered Li/(Li+Na) atomic ratios from 1 mM:1 M LiCl: NaCl(aq) mixed solutions at 0.1C'. Features labeled in orange correspond to particle morphology features, while those labeled in green pertain to electrochemical characteristics features. See Supplementary Tables 8, 9, and Supplementary Notes 3 for the definition of each variable or summary of the values for each particle.",
+        "footnote": [],
+        "bbox": [],
+        "page_idx": -1
+    }
+]