Dynamic Behavior and Interaction Mechanism of Soil Organic Matter in Water Systems: A Coarse-Grained Molecular Dynamics Study.

Investigating soil organic matter's (SOM) microscale assembly and functionality is challenging due to its complexity. This study constructs comparatively realistic SOM models, including diverse components such as Leonardite humic acid (LHA), lipids, peptides, carbohydrates, and lignin, to unveil their spontaneous self-assembly behavior at the mesoscopic scale through microsecond coarse-grained molecular dynamics simulations. We discovered an ordered SOM aggregation creating a layered phase from its hydrophobic core to the aqueous phase, resulting in an increasing O/C ratio and declining structural amphiphilicity. Notably, the amphiphilic lipids formed a bilayer membrane, partnering with lignin to constitute SOM's hydrophobic core. LHA, despite forming a layer, was embedded within this structure. The formation of such complex architectures was driven by nonbonded interactions between components. Our analysis revealed component-dependent diffusion effects within the SOM system. Lipids, peptides, and lignin showed inhibitory effects on self-diffusion, while carbohydrates facilitated diffusion. This study offers novel insights into the dynamic behavior and assembly of SOM components, introducing an effective approach for studying dynamic SOM mechanisms in aquatic environments.

Bioaccumulation and Trophodynamics of Novel Brominated Flame Retardants (NBFRs) in Marine Food Webs from the Arctic and Antarctic Regions.

The pervasive contamination of novel brominated flame retardants (NBFRs) in remote polar ecosystems has attracted great attention in recent research. However, understanding regarding the trophic transfer behavior of NBFRs in the Arctic and Antarctic marine food webs is limited. In this study, we examined the occurrence and trophodynamics of NBFRs in polar benthic marine sediment and food webs collected from areas around the Chinese Arctic Yellow River Station (n = 57) and Antarctic Great Wall Station (n = 94). ∑7NBFR concentrations were in the range of 1.27-7.47 ng/g lipid weight (lw) and 0.09-1.56 ng/g lw in the Arctic and Antarctic marine biota, respectively, among which decabromodiphenyl ethane (DBDPE) was the predominant compound in all sample types. The biota-sediment bioaccumulation factors (g total organic carbon/g lipid) of NBFRs in the Arctic (0.85-3.40) were 4-fold higher than those in the Antarctica (0.13-0.61). Trophic magnification factors (TMFs) and their 95% confidence interval (95% CI) of individual NBFRs ranged from 0.43 (95% CI: 0.32, 0.60) to 1.32 (0.92, 1.89) and from 0.34 (0.24, 0.49) to 0.92 (0.56, 1.51) in the Arctic and Antarctic marine food webs, respectively. The TMFs of most congeners were significantly lower than 1, indicating a trophic dilution potential. This is one of the very few investigations on the trophic transfer of NBFRs in remote Arctic and Antarctic marine ecosystems, which provides a basis for exploring the ecological risks of NBFRs in polar regions.

New Mechanism for the Apoptosis of Human Neuroblastoma Cells by the Interaction between Fluorene-9-Bisphenol and the G Protein-Coupled Estrogen Receptor 1.

Fluorene-9-bisphenol (BHPF) is an emerging contaminant. Presently, there is no report on its interaction with G protein-coupled estrogen receptor 1 (GPER). By using an integrated toxicity research scenario that combined theoretical study with experimental methods, BHPF was found to inhibit the GPER-mediated effect via direct receptor binding. Molecular dynamics simulations found that Trp2726.48 and Glu2756.51 be the key amino acids of BHPF binding with GPER. Moreover, the calculation indicated that BHPF was a suspected GPER inhibitor, which neither can activate GPER nor is able to form water channels of GPER. The role of two residues was successfully verified by following gene knockout and site-directed mutagenesis assays. Further in vitro assays showed that BHPF could attenuate the increase in intracellular concentration of free Ca2+ induced by G1-activated GPER. Besides, BHPF showed an enhanced cytotoxicity compared with G15, indicating that BHPF might be a more potent GPER inhibitor than G15. In addition, a statistically significant effect on the mRNA level of GPER was observed for BHPF. In brief, the present study proposes that BHPF be a GPER inhibitor, and its GPER molecular recognition mechanism has been revealed, which is of great significance for the health risk and assessment of BHPF.

Lithium Pollution and Its Associated Health Risks in the Largest Lithium Extraction Industrial Area in China.

Lithium (Li) is an important resource that drives sustainable mobility and renewable energy. Its demand is projected to continue to increase in the coming decades. However, the risk of Li pollution has also emerged as a global concern. Here, we investigated the pollution characteristics, sources, exposure levels, and associated health risks of Li in the Jinjiang River basin, the largest area for Li2CO3 production in China. Our results revealed the dominant role of Li extraction activities in the pollution of the river, with over 95% of dissolved Li in downstream river water being emitted from this source. Moreover, the Li concentration in aquatic plants (i.e., water hyacinth) and animals (i.e., fish) significantly increased from upstream to downstream areas, indicating a significant risk to local aquatic ecosystems. More importantly, our study found that local residents were suffering potential chronic noncarcinogenic health risks primarily from consuming contaminated water and vegetables. We also investigated the pollution characteristics of associated elements present in Li ores (e.g., Rb, Cs, Ni, and F-). By uncovering the remarkable impact of Li extraction activities on the Li content in ecosystems for the first time, our study emphasizes the importance of evaluating Li pollution from Li-related industrial activities, including mining, extraction, and recovery.

Remarkable Contamination of Short- and Medium-Chain Chlorinated Paraffins in Free-Range Chicken Eggs from Rural Tibetan Plateau.

Rapid social-economic development introduces modern lifestyles into rural areas, not only bringing numerous modern products but also new pollutants, such as chlorinated paraffins (CPs). The rural Tibetan Plateau has limited industrial activities and is a unique place to investigate this issue. Herein we collected 90 free-range chicken egg pool samples across the rural Tibetan Plateau to evaluate the pollution status of CPs. Meanwhile, CPs in related soils, free-range chicken eggs from Jiangxi, and farmed eggs from markets were also analyzed. The median concentrations of SCCPs (159 ng g-1 wet weight (ww)) and MCCPs (1390 ng g-1 ww) in Tibetan free-range chicken eggs were comparable to those from Jiangxi (259 and 938 ng g-1 ww) and significantly higher than those in farmed eggs (22.0 and 81.7 ng g-1 ww). In the rural Tibetan Plateau, the median EDI of CPs via egg consumption by adults and children were estimated to be 81.6 and 220.2 ng kg-1 bw day-1 for SCCPs and 483.4 and 1291 ng kg-1 bw day-1 for MCCPs, respectively. MCCPs might pose potential health risks for both adults and children in the worst scenario. Our study demonstrates that new pollutants should not be ignored and need further attention in remote rural areas.

High-Efficiency Effect-Directed Analysis Leveraging Five High Level Advancements: A Critical Review.

Organic contaminants are ubiquitous in the environment, with mounting evidence unequivocally connecting them to aquatic toxicity, illness, and increased mortality, underscoring their substantial impacts on ecological security and environmental health. The intricate composition of sample mixtures and uncertain physicochemical features of potential toxic substances pose challenges to identify key toxicants in environmental samples. Effect-directed analysis (EDA), establishing a connection between key toxicants found in environmental samples and associated hazards, enables the identification of toxicants that can streamline research efforts and inform management action. Nevertheless, the advancement of EDA is constrained by the following factors: inadequate extraction and fractionation of environmental samples, limited bioassay endpoints and unknown linkage to higher order impacts, limited coverage of chemical analysis (i.e., high-resolution mass spectrometry, HRMS), and lacking effective linkage between bioassays and chemical analysis. This review proposes five key advancements to enhance the efficiency of EDA in addressing these challenges: (1) multiple adsorbents for comprehensive coverage of chemical extraction, (2) high-resolution microfractionation and multidimensional fractionation for refined fractionation, (3) robust in vivo/vitro bioassays and omics, (4) high-performance configurations for HRMS analysis, and (5) chemical-, data-, and knowledge-driven approaches for streamlined toxicant identification and validation. We envision that future EDA will integrate big data and artificial intelligence based on the development of quantitative omics, cutting-edge multidimensional microfractionation, and ultraperformance MS to identify environmental hazard factors, serving for broader environmental governance.

Soil's Hidden Power: The Stable Soil Organic Carbon Pool Controls the Burden of Persistent Organic Pollutants in Background Soils.

Persistent organic pollutants (POPs) tend to accumulate in cold regions by cold condensation and global distillation. Soil organic matter is the main storage compartment for POPs in terrestrial ecosystems due to deposition and repeated air-surface exchange processes. Here, physicochemical properties and environmental factors were investigated for their role in influencing POPs accumulation in soils of the Tibetan Plateau and Antarctic and Arctic regions. The results showed that the soil burden of most POPs was closely coupled to stable mineral-associated organic carbon (MAOC). Combining the proportion of MAOC and physicochemical properties can explain much of the soil distribution characteristics of the POPs. The background levels of POPs were estimated in conjunction with the global soil database. It led to the proposition that the stable soil carbon pools are key controlling factors affecting the ultimate global distribution of POPs, so that the dynamic cycling of soil carbon acts to counteract the cold-trapping effects. In the future, soil carbon pool composition should be fully considered in a multimedia environmental model of POPs, and the risk of secondary release of POPs in soils under conditions such as climate change can be further assessed with soil organic carbon models.

First-principles study on the heterogeneous formation of environmentally persistent free radicals (EPFRs) over α-Fe2O3(0001) surface: Effect of oxygen vacancy.

Metal oxides with oxygen vacancies have a significant impact on catalytic activity for the transformation of organic pollutants in waste-to-energy (WtE) incineration processes. This study aims to investigate the influence of hematite surface oxygen point defects on the formation of environmentally persistent free radicals (EPFRs) from phenolic compounds based on the first-principles calculations. Two oxygen-deficient conditions were considered: oxygen vacancies at the top surface and on the subsurface. Our simulations indicate that the adsorption strength of phenol on the α-Fe2O3(0001) surface is enhanced by the presence of oxygen vacancies. However, the presence of oxygen vacancies has a negative impact on the dissociation of the phenol molecule, particularly for the surface with a defective point at the top layer. Thermo-kinetic parameters were established over a temperature range of 300-1000 K, and lower reaction rate constants were observed for the scission of phenolic O-H bonds over the oxygen-deficient surfaces compared to the pristine surface. The negative effects caused by the oxygen-deficient conditions could be attributed to the local reduction of FeIII to FeII, which lower the oxidizing ability of surface reaction sites. The findings of this study provide us a promising approach to regulate the formation of EPFRs.

Response of ocean acidification to atmospheric carbon dioxide removal.

Artificial CO2 removal from the atmosphere (also referred to as negative CO2 emissions) has been proposed as a potential means to counteract anthropogenic climate change. Here we use an Earth system model to examine the response of ocean acidification to idealized atmospheric CO2 removal scenarios. In our simulations, atmospheric CO2 is assumed to increase at a rate of 1% per year to four times its pre-industrial value and then decreases to the pre-industrial level at a rate of 0.5%, 1%, 2% per year, respectively. Our results show that the annual mean state of surface ocean carbonate chemistry fields including hydrogen ion concentration ([H+]), pH and aragonite saturation state respond quickly to removal of atmospheric CO2. However, the change of seasonal cycle in carbonate chemistry lags behind the decline in atmospheric CO2. When CO2 returns to the pre-industrial level, over some parts of the ocean, relative to the pre-industrial state, the seasonal amplitude of carbonate chemistry fields is substantially larger. Simulation results also show that changes in deep ocean carbonate chemistry substantially lag behind atmospheric CO2 change. When CO2 returns to its pre-industrial value, the whole-ocean acidity measured by [H+] is 15%-18% larger than the pre-industrial level, depending on the rate of CO2 decrease. Our study demonstrates that even if atmospheric CO2 can be lowered in the future as a result of net negative CO2 emissions, the recovery of some aspects of ocean acidification would take decades to centuries, which would have important implications for the resilience of marine ecosystems.

Exogenous Chemicals Impact Virus Receptor Gene Transcription: Insights from Deep Learning.

Despite the fact that coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has been disrupting human life and health worldwide since the outbreak in late 2019, the impact of exogenous substance exposure on the viral infection remains unclear. It is well-known that, during viral infection, organism receptors play a significant role in mediating the entry of viruses to enter host cells. A major receptor of SARS-CoV-2 is the angiotensin-converting enzyme 2 (ACE2). This study proposes a deep learning model based on the graph convolutional network (GCN) that enables, for the first time, the prediction of exogenous substances that affect the transcriptional expression of the ACE2 gene. It outperforms other machine learning models, achieving an area under receiver operating characteristic curve (AUROC) of 0.712 and 0.703 on the validation and internal test set, respectively. In addition, quantitative polymerase chain reaction (qPCR) experiments provided additional supporting evidence for indoor air pollutants identified by the GCN model. More broadly, the proposed methodology can be applied to predict the effect of environmental chemicals on the gene transcription of other virus receptors as well. In contrast to typical deep learning models that are of black box nature, we further highlight the interpretability of the proposed GCN model and how it facilitates deeper understanding of gene change at the structural level.

Source Identification of Organophosphate Esters through the Profiles in Proglacial and Ocean Sediments from Ny-Ålesund, the Arctic.

Little is known about the sources and environmental behavior of organophosphate esters (OPEs) in the Arctic, especially their transformation products. The present study unprecedentedly investigated both 16 tri-OPEs and 8 di-OPEs in proglacial and ocean sediments from Ny-Ålesund, the Arctic. Mean concentrations of tri-OPEs and di-OPEs in proglacial sediments were 487 and 341 pg/g dry weight (dw), respectively, which were significantly lower than those in ocean sediments (1692 and 525 pg/g dw). Ocean sediments might be simultaneously influenced by long-range atmospheric transport (LRAT), oceanic transport, and human activities, whereas proglacial sediments, since they are isolated from human settlements, may be dominantly affected by LRAT. Such source difference was evidenced by the contamination profile of OPEs: chlorinated tri-OPEs with high environmental persistence and high LRAT were dominant in proglacial sediments (66%); however, weakly environmentally persistent and highly hydrophobic aryl tri-OPEs were dominant in ocean sediments (47%), which were plausibly from local emission sources due to their low LRAT potential. Di-OPEs in proglacial and ocean sediments were dominated by groups of parent tri-OPEs with strong photodegradability, such as alkyl (75%) and aryl (58%). A higher mean molar ratio of di-OPE/tri-OPE in the proglacial sediment (14) than that in the ocean sediment (2.2) may be related to its higher photodegradation than that of the ocean sediment.

Linking Transthyretin-Binding Chemicals and Free Thyroid Hormones: In Vitro to In Vivo Extrapolation Based on a Competitive Binding Model.

Large numbers of pollutants competitively inhibit the binding between thyroid hormones and transthyretin (TTR) in vitro. However, the impact of this unintended binding on free thyroid hormones in vivo has not yet been characterized. Herein, we established a quantitative in vitro to in vivo extrapolation (QIVIVE) method based on a competitive binding model to quantify the effect of TTR-binding chemicals on free thyroid hormones in human blood. Twenty-five TTR-binding chemicals including 6 hydroxyl polybromodiphenyl ethers (OH-PDBEs), 6 hydroxyl polychlorobiphenyls (OH-PCBs), 4 halogenphenols, 5 per- and polyfluorinated substances (PFASs), and 4 phenols were selected for investigation. Incorporating the in vitro binding parameters and human exposure data, the QIVIVE model could well predict the in vivo effect on free thyroid hormones. Co-exposure to twenty-five typical TTR-binding chemicals resulted in median increases of 0.080 and 0.060% in circulating levels of free thyroxine (FT4) and free triiodothyronine (FT3) in the general population. Individuals with occupational exposure to TTR-binding chemicals suffered 1.88-32.2% increases in free thyroid hormone levels. This study provides a quantitative tool to evaluate the thyroid-disrupting risks of TTR-binding chemicals and proposes a new framework for assessing the in vivo effects of chemical exposures on endogenous molecules.

Disposable Polypropylene Face Masks: A Potential Source of Micro/Nanoparticles and Organic Contaminates in Humans.

We have been effectively protected by disposable propylene face masks during the COVID-19 pandemic; however, they may pose health risks due to the release of fine particles and chemicals. We measured micro/nanoparticles and organic chemicals in disposable medical masks, surgical masks, and (K)N95 respirators. In the breathing-simulation experiment, no notable differences were found in the total number of particles among mask types or between breathing intensities. However, when considering subranges, <2.5 µm particles accounted for ∼90% of the total number of micro/nanoparticles. GC-HRMS-based suspect screening tentatively revealed 79 (semi)volatile organic compounds in masks, with 18 being detected in ≥80% of samples and 44 in ≤20% of samples. Three synthetic phenolic antioxidants were quantified, and AO168 reached a median concentration of 2968 ng/g. By screening particles collected from bulk mask fabrics, we detected 18 chemicals, including four commonly detected in masks, suggesting chemical partition between the particles and the fabric fibers and chemical exposure via particle inhalation. These particles and chemicals are believed to originate from raw materials, intentionally and nonintentionally added substances in mask production, and their transformation products. This study highlights the need to study the long-term health risks associated with mask wearing and raises concerns over mask quality control.

Association between serum levels of TSH and free T4 and per- and polyfluoroalkyl compounds concentrations in pregnant women.

Many per- and polyfluoralkyl substances (PFASs) may disrupt maternal thyroid hormone homeostasis in pregnancy. Concerns should be raised regarding the PFASs exposure in pregnant women because thyroid hormones are involved in the early development of the fetus. In this study, we measured the concentrations of 13 PFASs, including five novel short-chain PFASs, in serum from 123 pregnant women in Beijing, China. Linear regression models were used to investigate the association between thyroid-stimulating hormone (TSH) or free thyroxine (FT4) levels and PFASs concentrations under consideration of the impacts of pregnancy-induced physiological factors. We found that perfluorobutanoic acid (PFBA) (ß=0.189, 95%CI=-0.039, 0.417, p=0.10) and perfluorodecanoic acid (PFDA) (ß=-0.554, 95%CI=-1.16, 0.049, p=0.071) were suggestive of significant association with TSH in thyroid peroxidase antibody (TPOAb) negative women. No association was observed between all PFASs and FT4 levels after controlling for these confounding factors, such as BMI, gestational weight gain and maternal age. These findings suggest that it should pay more attention to the association between thyroid hormone levels and short-chain PFASs concentrations. Future studies could consider a greater sample and the inclusion of other clinical indicators of thyroid function, such as free T3 and total T3.

Computational Insights into the Allosteric Effect and Dynamic Structural Features of the SARS-COV-2 Spike Protein.

COVID-19 caused by SARS-COV-2 is continuing to surge globally. The spike (S) protein is the key protein of SARS-COV-2 that recognizes and binds to the host target ACE2. In this study, molecular dynamics simulation was used to elucidate the allosteric effect of the S protein. Binding of ACE2 caused a centripetal movement of the receptor-binding domain of the S protein. The dihedral changes in Phe329 and Phe515 played a key role in this process. Two potential cleavage sites S1/S2 and S2' were exposed on the surface after the binding of ACE2. The binding affinity of SARS-COV-2 S protein and ACE2 was higher than that of SARS-COV. This was mainly due to the mutation of Asp480 in SARS-COV to Ser494 in SARS-COV-2, which greatly weakened the electrostatic repulsion. The result provides a theoretical basis for the SARS-COV-2 infection and aids the development of biosensors and detection reagents.

Computational Insights into the Allosteric Effect and Dynamic Structural Features of the SARS-COV-2 Spike Protein.

Invited for the cover of this issue are Aiqian Zhang, Jianjie Fu, Guibin Jiang, and co-workers at the Chinese Academy of Sciences. The image depicts the molecular recognition of human angiotensin-converting enzyme 2 by the SARS-COV-2 spike protein. Read the full text of the article at 10.1002/chem.202104215.

Tracing the Transboundary Transport of Mercury to the Tibetan Plateau Using Atmospheric Mercury Isotopes.

Deposition of atmospheric mercury (Hg) is the most important Hg source on the high-altitude Himalayas and Tibetan Plateau. Herein, total gaseous Hg (TGM) at an urban and a forest site on the Tibetan Plateau was collected respectively from May 2017 to October 2018, and isotopic compositions were measured to clarify the influences of landforms and monsoons on the transboundary transport of atmospheric Hg to the Tibetan Plateau. The transboundary transported anthropogenic emissions mainly originated over Indo-Gangetic Plain and carried over the Himalayas by convective storms and mid-tropospheric circulation, contributing over 50% to the TGM at the Lhasa urban site, based on the binary mixing model of isotopes. In contrast, during the transport of TGM from South Asia with low altitude, the uptake by evergreen forest in Yarlung Zangbo Grand Canyon largely decreased the TGM level and shifted isotopic compositions in TGM at the Nyingchi forest site, which are located at the high-altitude end of the canyon. Our results provided direct evidence from Hg isotopes to reveal the distinct patterns of transboundary transport to the Tibetan Plateau shaped by landforms and climates, which is critical to fully understand the biogeochemical cycling of Hg in the high-altitude regions.

Evaluation of the binding performance of flavonoids to estrogen receptor alpha by Autodock, Autodock Vina and Surflex-Dock.

Molecular docking is a widely used method to predict the binding modes of small-molecule ligands to the target binding site. However, it remains a challenge to identify the correct binding conformation and the corresponding binding affinity for a series of structurally similar ligands, especially those with weak binding. An understanding of the various relative attributes of popular docking programs is required to ensure a successful docking outcome. In this study, we systematically compared the performance of three popular docking programs, Autodock, Autodock Vina, and Surflex-Dock for a series of structurally similar weekly binding flavonoids (22) binding to the estrogen receptor alpha (ERα). For these flavonoids-ERα interactions, Surflex-Dock showed higher accuracy than Autodock and Autodock Vina. The hydrogen bond overweighting by Autodock and Autodock Vina led to incorrect binding results, while Surflex-Dock effectively balanced both hydrogen bond and hydrophobic interactions. Moreover, the selection of initial receptor structure is critical as it influences the docking conformations of flavonoids-ERα complexes. The flexible docking method failed to further improve the docking accuracy of the semi-flexible docking method for such chemicals. In addition, binding interaction analysis revealed that 8 residues, including Ala350, Glu353, Leu387, Arg394, Phe404, Gly521, His524, and Leu525, are the key residues in ERα-flavonoids complexes. This work provides reference for assessing molecular interactions between ERα and flavonoid-like chemicals and provides instructive information for other environmental chemicals.

Effect of polar/non-polar facets on the transformation of nanoscale ZnO in simulated sweat and potential impacts on the antibacterial activity.

The use of nanoscale zinc oxide (n-ZnO) in the personal care products would cause interactions between n-ZnO and human sweat. Facet engineering has been applied to n-ZnO to improve its activity. Nevertheless, it is not clear whether the exposed facet would affect transformation of n-ZnO in sweat. Herein, we prepared ZnO nanoneedles with the dominant (1010) non-polar facet (i.e., ZnO-1010) and ZnO nanoflakes with the dominant (0001) polar facet (i.e., ZnO-0001), respectively. We found that n-ZnO can undergo chemical transformation in the simulated sweat within 168 h or 24 h, transforming into amorphous materials and Zn3(PO4)20.4 H2O and/or Na(ZnPO4)·H2O. Given the rate constant (e.g., 0.093 h-1 for ZnO-0001 vs. 0.033 h-1 for ZnO-1010) of ZnO depletion and components of the precipitate from the simulated sweat, nevertheless, the transformation is highly dependent on the dominant exposed facet of n-ZnO. The ZnO-0001 relative to ZnO-1010 would likely undergo chemical transformation, demonstrating that the (0001) polar facet compared to (1010) non-polar facet had a superior activity to the dihydrogen phosphate anions in the simulated sweat, which is supported by density functional theory calculations. The chemical transformation can affect the antibacterial activity of n-ZnO to E. coli, moderating the toxicity due to a great decrease in the concentration of the dissolved zinc. In total, our findings provided insights into the facet-dependent transformation for n-ZnO in the simulated sweat, improving our understanding of potential risk of n-ZnO.

SepPCNET: Deeping Learning on a 3D Surface Electrostatic Potential Point Cloud for Enhanced Toxicity Classification and Its Application to Suspected Environmental Estrogens.

Deep learning (DL) offers an unprecedented opportunity to revolutionize the landscape of toxicity prediction based on quantitative structure-activity relationship (QSAR) studies in the big data era. However, the structural description in the reported DL-QSAR models is still restricted to the two-dimensional level. Inspired by point clouds, a type of geometric data structure, a novel three-dimensional (3D) molecular surface point cloud with electrostatic potential (SepPC) was proposed to describe chemical structures. Each surface point of a chemical is assigned its 3D coordinate and molecular electrostatic potential. A novel DL architecture SepPCNET was then introduced to directly consume unordered SepPC data for toxicity classification. The SepPCNET model was trained on 1317 chemicals tested in a battery of 18 estrogen receptor-related assays of the ToxCast program. The obtained model recognized the active and inactive chemicals at accuracies of 82.8 and 88.9%, respectively, with a total accuracy of 88.3% on the internal test set and 92.5% on the external test set, which outperformed other up-to-date machine learning models and succeeded in recognizing the difference in the activity of isomers. Additional insights into the toxicity mechanism were also gained by visualizing critical points and extracting data-driven point features of active chemicals.