Age-Dependent Abundance of CYP450 Enzymes Involved in Metronidazole Metabolism: Application to Pediatric PBPK Modeling.

The expression of cytochrome P450 (CYP) enzymes is highly variable and associated with factors, such as age, genotype, sex, and disease states. In this study, quantification of metronidazole metabolizing CYP isoforms (CYP2A6, CYP2E1, CYP3A4, CYP3A5, and CYP3A7) in human liver microsomes from 115 children and 35 adults was performed using a quantitative proteomics method. The data confirmed age-dependent increase in CYP2A6, CYP2E1, and CYP3A4 abundance, whereas, as expected, CYP3A7 abundance showed postnatal decrease with age. In particular, the fold difference (neonatal to adulthood levels) in the protein abundance of CYP2A6, CYP2E1, and CYP3A4 was 14, 11, and 20, respectively. In contrast, protein abundance of CYP3A7 was > 125-fold higher in the liver microsomes of neonates than of adults. The abundance of CYP2A6 and CYP3A5 was associated with genotypes, rs4803381 and rs776746, respectively. A proteomics-informed physiologically based pharmacokinetic (PBPK) model was developed to describe the pharmacokinetics of metronidazole and its primary metabolite, 2-hydroxymethylmetronidazole. The model revealed an increase in the metabolite-to-parent ratio with age and showed a strong correlation between CYP2A6 abundance and metabolite formation (r2 = 0.75). Notably, the estimated contribution of CYP3A7 was ~ 75% in metronidazole clearance in neonates. These data suggest that variability in CYP2A6 and CYP3A7 in younger children poses the risk of variable pharmacokinetics of metronidazole and its active metabolite with a potential impact on drug efficacy and safety. No sex-dependent difference was observed in the protein abundance of the studied CYPs. The successful integration of hepatic CYP ontogeny data derived from a large liver bank into the pediatric PBPK model of metronidazole can be extended to other drugs metabolized by the studied CYPs.

Influence of novel CYP2C-haplotype on proton pump inhibitor pharmacokinetics in children.

In this brief report, we provide an analysis of the influence of a novel CYP2C haplotype (CYP2C:TG) on proton pump inhibitor (PPI) pharmacokinetics (PK) in children. The CYP2C:TG haplotype has been proposed to be associated with increased CYP2C19 activity. We sought to determine if this CYP2C:TG haplotype resulted in similar alterations in metabolism for proton pump inhibitors, which are primarily metabolized by CYP2C19. In a cohort of 41 children aged 6-21 participating in a PPI pharmacokinetic study, effects of the CYP2C:TG allele were assessed by fitting two linear regression models for each of the six PK outcomes assessed, the second of which accounted for the presence of the CYP2C:TG allele. The difference in R2 values between the two models was computed to quantify the variability in the outcome that could be accounted for by the CYP2C:TG allele after adjustment for the CYP2C19 genotype. We found the CYP2C:TG haplotype to have no measurable additive impact on CYP2C19-mediated metabolism of PPIs in vivo in older children and adolescents. The findings of this study do not support the clinical utility of routine testing for the CYP2C:TG haplotype to guide PPI dose adjustments in children.

Integrating end-to-end learning with deep geometrical potentials for ab initio RNA structure prediction.

RNAs are fundamental in living cells and perform critical functions determined by their tertiary architectures. However, accurate modeling of 3D RNA structure remains a challenging problem. We present a novel method, DRfold, to predict RNA tertiary structures by simultaneous learning of local frame rotations and geometric restraints from experimentally solved RNA structures, where the learned knowledge is converted into a hybrid energy potential to guide RNA structure assembly. The method significantly outperforms previous approaches by >73.3% in TM-score on a sequence-nonredundant dataset containing recently released structures. Detailed analyses showed that the major contribution to the improvements arise from the deep end-to-end learning supervised with the atom coordinates and the composite energy function integrating complementary information from geometry restraints and end-to-end learning models. The open-source DRfold program with fast training protocol allows large-scale application of high-resolution RNA structure modeling and can be further improved with future expansion of RNA structure databases.

Ontogeny of Scaling Factors for Pediatric Physiologically Based Pharmacokinetic Modeling and Simulation: Cytosolic Protein Per Gram of Liver.

Scaling factors are necessary for translating in vitro drug biotransformation data to in vivo clearance values in physiologically-based pharmacokinetic modeling and simulation. Values for microsomal protein per gram of liver are available from several sources for use as a scaling factor to estimate hepatic clearance from microsomal drug biotransformation data. However, data regarding the distribution of cytosolic protein per gram of liver (CPPGL) values across the lifespan are limited, and sparse pediatric data have been published to date. Thus, CPPGL was determined in 160 liver samples from pediatric (n = 129) and adult (n = 31) donors obtained from multiple sources: the University of Maryland Brain and Tissue Bank, tissue retrieval services at the University of Minnesota and University of Pittsburgh, and Sekisui-XenoTech. Tissues were homogenized and subjected to differential centrifugation to isolate cytosolic fractions. Cytosolic protein content was determined by BCA assay. CPPGL varied from two- to sixfold within each age group/developmental stage. Tissue source and sex did not contribute substantially to variability in protein content. Regression analyses revealed minimal change in CPPGL over the first two decades of life (logCPPGL increases 0.1 mg/g per decade). A mean ± S.D. CPPGL value of 44.4 ± 17.4 mg/g or median 41.0 mg/g is representative of values observed between birth and early adulthood (0-18 years, n = 129). SIGNIFICANCE STATEMENT: Cytosolic protein per gram of liver (CPPGL) is a scaling factor required for physiologically based pharmacokinetic modeling and simulation of drug biotransformation by cytosolic enzymes, but pediatric data are limited. Although CPPGL varies from two- to sixfold within developmental stages, a value of 44.4 ± 17.4 mg/g (mean ± S.D.) is representative of the pediatric period (0-18 years, n = 129).

De novo protein fold design through sequence-independent fragment assembly simulations.

De novo protein design generally consists of two steps, including structure and sequence design. Many protein design studies have focused on sequence design with scaffolds adapted from native structures in the PDB, which renders novel areas of protein structure and function space unexplored. We developed FoldDesign to create novel protein folds from specific secondary structure (SS) assignments through sequence-independent replica-exchange Monte Carlo (REMC) simulations. The method was tested on 354 non-redundant topologies, where FoldDesign consistently created stable structural folds, while recapitulating on average 87.7% of the SS elements. Meanwhile, the FoldDesign scaffolds had well-formed structures with buried residues and solvent-exposed areas closely matching their native counterparts. Despite the high fidelity to the input SS restraints and local structural characteristics of native proteins, a large portion of the designed scaffolds possessed global folds completely different from natural proteins in the PDB, highlighting the ability of FoldDesign to explore novel areas of protein fold space. Detailed data analyses revealed that the major contributions to the successful structure design lay in the optimal energy force field, which contains a balanced set of SS packing terms, and REMC simulations, which were coupled with multiple auxiliary movements to efficiently search the conformational space. Additionally, the ability to recognize and assemble uncommon super-SS geometries, rather than the unique arrangement of common SS motifs, was the key to generating novel folds. These results demonstrate a strong potential to explore both structural and functional spaces through computational design simulations that natural proteins have not reached through evolution.

Combination Treatment of Withalongolide a Triacetate with Cisplatin Induces Apoptosis by Targeting Translational Initiation, Migration, and Epithelial to Mesenchymal Transition in Head and Neck Squamous Cell Carcinoma.

Treatment regimens for head and neck squamous cell carcinoma (HNSCC) typically include cisplatin and radiotherapy and are limited by toxicities. We have identified naturally derived withalongolide A triacetate (WGA-TA) from Physalis longifolia as a lead compound for targeting HNSCC. We hypothesized that combining WGA-TA with cisplatin may allow for lower, less toxic cisplatin doses. HNSCC cell lines were treated with WGA-TA and cisplatin. After treatment with the drugs, the cell viability was determined by MTS assay. The combination index was calculated using CompuSyn. The expression of proteins involved in the targeting of translational initiation complex, epithelial to mesenchymal transition (EMT), and apoptosis were measured by western blot. Invasion and migration were measured using the Boyden-chamber assay. Treatment of MDA-1986 and UMSCC-22B cell lines with either WGA-TA or cisplatin alone for 72 h resulted in a dose dependent decrease in cell viability. Cisplatin in combination with WGA-TA resulted in significant synergistic cell death starting from 1.25 µM cisplatin. Combination treatment with WGA-TA resulted in lower cisplatin dosing while maintaining the downregulation of translational initiation complex proteins, the induction of apoptosis, and the blockade of migration, invasion, and EMT transition. These results suggest that combining a low concentration of cisplatin with WGA-TA may provide a safer, more effective therapeutic option for HNSCC that warrants translational validation.

Fast and accurate Ab Initio Protein structure prediction using deep learning potentials.

Despite the immense progress recently witnessed in protein structure prediction, the modeling accuracy for proteins that lack sequence and/or structure homologs remains to be improved. We developed an open-source program, DeepFold, which integrates spatial restraints predicted by multi-task deep residual neural-networks along with a knowledge-based energy function to guide its gradient-descent folding simulations. The results on large-scale benchmark tests showed that DeepFold creates full-length models with accuracy significantly beyond classical folding approaches and other leading deep learning methods. Of particular interest is the modeling performance on the most difficult targets with very few homologous sequences, where DeepFold achieved an average TM-score that was 40.3% higher than trRosetta and 44.9% higher than DMPfold. Furthermore, the folding simulations for DeepFold were 262 times faster than traditional fragment assembly simulations. These results demonstrate the power of accurately predicted deep learning potentials to improve both the accuracy and speed of ab initio protein structure prediction.

I-TASSER-MTD: a deep-learning-based platform for multi-domain protein structure and function prediction.

Most proteins in cells are composed of multiple folding units (or domains) to perform complex functions in a cooperative manner. Relative to the rapid progress in single-domain structure prediction, there are few effective tools available for multi-domain protein structure assembly, mainly due to the complexity of modeling multi-domain proteins, which involves higher degrees of freedom in domain-orientation space and various levels of continuous and discontinuous domain assembly and linker refinement. To meet the challenge and the high demand of the community, we developed I-TASSER-MTD to model the structures and functions of multi-domain proteins through a progressive protocol that combines sequence-based domain parsing, single-domain structure folding, inter-domain structure assembly and structure-based function annotation in a fully automated pipeline. Advanced deep-learning models have been incorporated into each of the steps to enhance both the domain modeling and inter-domain assembly accuracy. The protocol allows for the incorporation of experimental cross-linking data and cryo-electron microscopy density maps to guide the multi-domain structure assembly simulations. I-TASSER-MTD is built on I-TASSER but substantially extends its ability and accuracy in modeling large multi-domain protein structures and provides meaningful functional insights for the targets at both the domain- and full-chain levels from the amino acid sequence alone.

A longitudinal study of cytochrome P450 2D6 (CYP2D6) activity during adolescence.

CYP2D6 substrates are among the most highly prescribed medications in teenagers and also commonly associated with serious adverse events. To investigate the relative contributions of genetic variation, growth, and development on CYP2D6 activity during puberty, healthy children and adolescents 7-15 years of age at enrollment participated in a longitudinal phenotyping study involving administration of 0.3 mg/kg dextromethorphan (DM) and 4-h urine collection every 6 months for 3 years (7 total visits). At each visit, height, weight, and sexual maturity were recorded, and CYP2D6 activity was determined as the urinary molar ratio of DM to its metabolite dextrorphan (DX). A total of 188 participants completed at least one visit, and 102 completed all seven study visits. Following univariate analysis, only CYP2D6 activity score (p < 0.001), urinary pH (p < 0.001), weight (p = 0.018), and attention-deficit/hyperactivity disorder (ADHD) diagnosis (p < 0.001) were significantly correlated with log(DM/DX). Results of linear mixed model analysis with random intercept, random slope covariance structure revealed that CYP2D6 activity score had the strongest effect on log(DM/DX), with model-estimated average log(DM/DX) being 3.8 SDs higher for poor metabolizers than for patients with activity score 3. A moderate effect on log(DM/DX) was observed for sex, and smaller effects were observed for ADHD diagnosis and urinary pH. The log(DM/DX) did not change meaningfully with age or pubertal development. CYP2D6 genotype remains the single, largest determinant of variability in CYP2D6 activity during puberty. Incorporation of genotype-based dosing guidelines should be considered for CYP2D6 substrates given the prevalent use of these agents in this pediatric age group.

Utility of the 13 C-pantoprazole breath test as a CYP2C19 phenotyping probe for children.

The 13 C-pantoprazole breath test (PAN-BT) is a safe, noninvasive, in vivo CYP2C19 phenotyping probe for adults. Our objective was to evaluate PAN-BT performance in children, with a focus on discriminating individuals who, according to guidelines from the Clinical Pharmacology Implementation Consortium (CPIC), would benefit from starting dose escalation versus reduction for proton pump inhibitors (PPIs). Children (n = 65, 6-17 years) genotyped for CYP2C19 variants *2, *3, *4, and *17 received a single oral dose of 13 C-pantoprazole. Plasma concentrations of pantoprazole and its metabolites, and changes in exhaled 13 CO2 (termed delta-over-baseline or DOB), were measured 10 times over 8 h using high performance liquid chromatography with ultraviolet detection and spectrophotometry, respectively. Pharmacokinetic parameters of interest were generated and DOB features derived using feature engineering for the first 180 min postadministration. DOB features, age, sex, and obesity status were used to run bootstrap analysis at each timepoint (Ti ) independently. For each iteration, stratified samples were drawn based on genotype prevalence in the original cohort. A random forest was trained, and predictive performance of PAN-BT was evaluated. Strong discriminating ability for CYP2C19 intermediate versus normal/rapid metabolizer phenotype was noted at DOBT30 min (mean sensitivity: 0.522, specificity: 0.784), with consistent model outperformance over a random or a stratified classifier approach at each timepoint (p < 0.001). With additional refinement and investigation, the test could become a useful and convenient dosing tool in clinic to help identify children who would benefit most from PPI dose escalation versus dose reduction, in accordance with CPIC guidelines.

Ontogeny of Scaling Factors for Pediatric Physiology-Based Pharmacokinetic Modeling and Simulation: Microsomal Protein Per Gram of Liver.

Microsomal protein per gram of liver (MPPGL) is an important scaling factor for bottom-up physiology-based pharmacokinetic modeling and simulation, but data in pediatrics are limited. Therefore, MPPGL was determined in 160 liver samples from pediatric (n = 129) and adult (n = 31) donors obtained from four sources: the University of Maryland Brain and Tissue Bank (UMBTB), tissue retrieval services at the University of Minnesota and University of Pittsburgh, and Sekisui-Xenotech. Tissues were homogenized and subjected to differential centrifugation to prepare microsomes, and cytochrome c reductase activities in tissue homogenates and microsomes were used to estimate cytochrome P450 reductase (POR) activity as a marker of microsomal recovery; microsomal POR content was also assessed by quantitative proteomics. MPPGL values varied 5- to 10-fold within various age groups/developmental stages, and tissue source was identified as a contributing factor. Using a "trimmed" dataset comprised of samples ranging from 3 to 18 years of age common to the four sources, POR protein abundance and activity in microsomes and POR activity in homogenates was lower in UMBTB samples (autopsy) compared with other sources (perfused/flash-frozen). Regression analyses revealed that the UMBTB samples were driving an apparent age effect as no effect of age on log-transformed MPPGL values was observed when the UMBTB samples were excluded. We conclude that a mean±SD MPPGL value of 30.4±1.7 mg/g is representative between one month postnatal age and early adulthood. Potential source effects should be considered for studies involving tissue samples from multiple sources with different procurement and processing procedures. SIGNIFICANCE STATEMENT: Microsomal protein per gram of liver (MPPGL) is an important scaling factor for bottom up PBPK modeling and simulation, but data in pediatrics are limited. Although MPPGL varies 5- to 10-fold at a given developmental stage, a value of 30.4 ± 1.7 mg/g (mean ± SD) is representative between one month postnatal age and early adulthood. However, when tissue samples are obtained from multiple sources, different procurement and processing procedures may influence the results and should be taken into consideration.

The Impact of the CYP2D6 "Enhancer" Single Nucleotide Polymorphism on CYP2D6 Activity.

rs5758550 has been associated with enhanced transcription and suggested to be a useful marker of CYP2D6 activity. As there are limited and inconsistent data regarding the utility of this distant "enhancer" single nucleotide polymorphism (SNP), our goal was to further assess the impact of rs5758550 on CYP2D6 activity toward two probe substrates, atomoxetine (ATX) and dextromethorphan (DM), using in vivo urinary metabolite (DM; n = 188) and pharmacokinetic (ATX; n = 70) and in vitro metabolite formation (ATX and DM; n = 166) data. All subjects and tissues were extensively genotyped, the "enhancer" SNP phased with established CYP2D6 haplotypes either computationally or experimentally, and the impact on CYP2D6 activity investigated using several linear models of varying complexity to determine the proportion of variability in CYP2D6 activity captured by each model. For all datasets and models, the "enhancer" SNP had no or only a modest impact on CYP2D6 activity prediction. An increased effect, when present, was more pronounced for ATX than DM suggesting potential substate-dependency. In addition, CYP2D6*2 alleles with the "enhancer" SNP were associated with modestly higher metabolite formation rates in vitro, but not in vivo; no effect was detected for CYP2D6*1 alleles with "enhancer" SNP. In summary, it remains inconclusive whether the small effects detected in this investigation are indeed caused by the "enhancer" SNP or are rather due to its incomplete linkage with other variants within the gene. Taken together, there does not appear to be sufficient evidence to warrant the "enhancer" SNP be included in clinical CYP2D6 pharmacogenetic testing.

The Impact of Age and Genetics on Naltrexone Biotransformation.

Naltrexone, an opioid antagonist primarily metabolized by aldo-keto reductase 1C4 (AKR1C4), treats pediatric conditions involving compulsiveness (e.g., autism spectrum, Prader-Willi, eating disorders, non-suicidal self-injury). Pharmacokinetic variability is apparent in adults, yet no data are available for children. This study aimed to examine the impact of age and genetic variation on naltrexone biotransformation. Human liver cytosol (HLC) samples (n = 158) isolated from children and adult organ donors were incubated with therapeutically relevant concentrations of naltrexone (0.1, 1 µM). Naltrexone biotransformation was determined by ultraperformance mass spectrometry quantification of the primary metabolite, 6-beta-naltrexol (6ßN), and 6ßN formation rates (pmol/mg protein/min) were calculated. HLCs from organ donors, age range 0-79 y (mean 16.0 ± 18.2 y), 37% (n = 60) female, 20% (n = 33) heterozygous and 1.2% (n = 2) homozygous for co-occurring AKR1C4 variants (S145C/L311V) showed >200-fold range in 6ßN formation (0.37-76.5 pmol/mg protein/min). Source of donor samples was found to be a substantial contributor to variability. Model estimates for a trimmed data set of source-adjusted pediatric samples (aged 0-18 y) suggested that AKR1C4 genetic variation, age, and sex explained 36% of the variability in 6ßN formation. Although activity increased steadily from birth and peaked in middle childhood (2-5 years), genetic variation (S145C/L311V) demonstrated a greater effect on activity than did age. Naltrexone biotransformation is highly variable in pediatric and adult livers and can be partly accounted for by individual factors feasible to obtain (e.g., genetic variability, age, sex). These data may inform a precision therapeutics approach (e.g., exposure optimization) to further study Naltrexone responsiveness in children and adults. SIGNIFICANCE STATEMENT: Biotransformation of the commonly used opioid antagonist naltrexone is highly variable and may contribute to reduced therapeutic response. Age, sex, and genetic variation in the drug-metabolizing enzyme, AKR1C4, are potential factors contributing to this variability. In pediatric samples, genetic variation (S145C/L311V) demonstrates a greater impact on activity than age. Additionally, the source of donor samples was identified as an important contributor and must be accounted for to confidently elucidate the biological variables most impactful to drug biotransformation.

Comparative Secretomics Analysis Reveals the Major Components of Penicillium oxalicum 16 and Trichoderma reesei RUT-C30.

In this study, the major secretome components of Penicillium oxalicum 16 and Trichoderma reesei RUT-C30 under wheat bran (WB) and rice straw (RS) solid-state fermentation were systematically analyzed. The activities of the major components, e.g., cellulase, hemicellulase, and amylase, were consistent with their abundance in the secretomes. P. oxalicum 16 secreted more abundant glycoside hydrolases than T. reesei RUT-C30. The main up-regulated proteins from the induction of WB, compared with that from RS, were amylase, pectinase, and protease, whereas the main down-regulated enzymes were cellulase, hemicellulase, swollenin, and lytic polysaccharide monooxygenase (LPMO). Specifically, WB induced more ß-1,4-glucosidases, namely, S8B0F3 (UniProt ID), and A0A024RWA5 than RS, but RS induced more ß-1,4-exoglucanases and ß-1,4-endoglucanases, namely, A0A024RXP8, A024SH76, S7B6D6, S7ZP52, A024SH20, A024S2H5, S8BGM3, S7ZX22, and S8AIJ2. The P. oxalicum 16 xylanases S8AH74 and S7ZA57 were the major components responsible for degrading soluble xylan, and S8BDN2 probably acted on solid-state hemicellulose instead of soluble xylan. The main hemicellulase component of T. reesei RUT-C30 in RS was the xyloglucanase A0A024S9Z6 with an abundance of 16%, but T. reesei RUT-C30 lacked the hemicellulase mannanase and had a small amount of the hemicellulase xylanase. P. oxalicum 16 produced more amylase than T. reesei RUT-C30, and the results suggest amylase S7Z6T2 may degrade soluble starch. The percentage of the glucoamylase S8B6D7 did not significantly change, and reached an average abundance of 5.5%. The major auxiliary degradation enzymes of P. oxalicum 16 were LPMOs S7Z716 and S7ZPW1, whereas those of T. reesei RUT-C30 were swollenin and LPMOs A0A024SM10, A0A024SFJ2, and A0A024RZP7.

Folding non-homologous proteins by coupling deep-learning contact maps with I-TASSER assembly simulations.

Structure prediction for proteins lacking homologous templates in the Protein Data Bank (PDB) remains a significant unsolved problem. We developed a protocol, C-I-TASSER, to integrate interresidue contact maps from deep neural-network learning with the cutting-edge I-TASSER fragment assembly simulations. Large-scale benchmark tests showed that C-I-TASSER can fold more than twice the number of non-homologous proteins than the I-TASSER, which does not use contacts. When applied to a folding experiment on 8,266 unsolved Pfam families, C-I-TASSER successfully folded 4,162 domain families, including 504 folds that are not found in the PDB. Furthermore, it created correct folds for 85% of proteins in the SARS-CoV-2 genome, despite the quick mutation rate of the virus and sparse sequence profiles. The results demonstrated the critical importance of coupling whole-genome and metagenome-based evolutionary information with optimal structure assembly simulations for solving the problem of non-homologous protein structure prediction.

Improving fragment-based ab initio protein structure assembly using low-accuracy contact-map predictions.

Sequence-based contact prediction has shown considerable promise in assisting non-homologous structure modeling, but it often requires many homologous sequences and a sufficient number of correct contacts to achieve correct folds. Here, we developed a method, C-QUARK, that integrates multiple deep-learning and coevolution-based contact-maps to guide the replica-exchange Monte Carlo fragment assembly simulations. The method was tested on 247 non-redundant proteins, where C-QUARK could fold 75% of the cases with TM-scores (template-modeling scores) ≥0.5, which was 2.6 times more than that achieved by QUARK. For the 59 cases that had either low contact accuracy or few homologous sequences, C-QUARK correctly folded 6 times more proteins than other contact-based folding methods. C-QUARK was also tested on 64 free-modeling targets from the 13th CASP (critical assessment of protein structure prediction) experiment and had an average GDT_TS (global distance test) score that was 5% higher than the best CASP predictors. These data demonstrate, in a robust manner, the progress in modeling non-homologous protein structures using low-accuracy and sparse contact-map predictions.

Identification of 13 Guanidinobenzoyl- or Aminidinobenzoyl-Containing Drugs to Potentially Inhibit TMPRSS2 for COVID-19 Treatment.

Positively charged groups that mimic arginine or lysine in a natural substrate of trypsin are necessary for drugs to inhibit the trypsin-like serine protease TMPRSS2 that is involved in the viral entry and spread of coronaviruses, including SARS-CoV-2. Based on this assumption, we identified a set of 13 approved or clinically investigational drugs with positively charged guanidinobenzoyl and/or aminidinobenzoyl groups, including the experimentally verified TMPRSS2 inhibitors Camostat and Nafamostat. Molecular docking using the C-I-TASSER-predicted TMPRSS2 catalytic domain model suggested that the guanidinobenzoyl or aminidinobenzoyl group in all the drugs could form putative salt bridge interactions with the side-chain carboxyl group of Asp435 located in the S1 pocket of TMPRSS2. Molecular dynamics simulations further revealed the high stability of the putative salt bridge interactions over long-time (100 ns) simulations. The molecular mechanics/generalized Born surface area-binding free energy assessment and per-residue energy decomposition analysis also supported the strong binding interactions between TMPRSS2 and the proposed drugs. These results suggest that the proposed compounds, in addition to Camostat and Nafamostat, could be effective TMPRSS2 inhibitors for COVID-19 treatment by occupying the S1 pocket with the hallmark positively charged groups.

Protein structure prediction using deep learning distance and hydrogen-bonding restraints in CASP14.

In this article, we report 3D structure prediction results by two of our best server groups ("Zhang-Server" and "QUARK") in CASP14. These two servers were built based on the D-I-TASSER and D-QUARK algorithms, which integrated four newly developed components into the classical protein folding pipelines, I-TASSER and QUARK, respectively. The new components include: (a) a new multiple sequence alignment (MSA) collection tool, DeepMSA2, which is extended from the DeepMSA program; (b) a contact-based domain boundary prediction algorithm, FUpred, to detect protein domain boundaries; (c) a residual convolutional neural network-based method, DeepPotential, to predict multiple spatial restraints by co-evolutionary features derived from the MSA; and (d) optimized spatial restraint energy potentials to guide the structure assembly simulations. For 37 FM targets, the average TM-scores of the first models produced by D-I-TASSER and D-QUARK were 96% and 112% higher than those constructed by I-TASSER and QUARK, respectively. The data analysis indicates noticeable improvements produced by each of the four new components, especially for the newly added spatial restraints from DeepPotential and the well-tuned force field that combines spatial restraints, threading templates, and generic knowledge-based potentials. However, challenges still exist in the current pipelines. These include difficulties in modeling multi-domain proteins due to low accuracy in inter-domain distance prediction and modeling protein domains from oligomer complexes, as the co-evolutionary analysis cannot distinguish inter-chain and intra-chain distances. Specifically tuning the deep learning-based predictors for multi-domain targets and protein complexes may be helpful to address these issues.

Toward the solution of the protein structure prediction problem.

Since Anfinsen demonstrated that the information encoded in a protein's amino acid sequence determines its structure in 1973, solving the protein structure prediction problem has been the Holy Grail of structural biology. The goal of protein structure prediction approaches is to utilize computational modeling to determine the spatial location of every atom in a protein molecule starting from only its amino acid sequence. Depending on whether homologous structures can be found in the Protein Data Bank (PDB), structure prediction methods have been historically categorized as template-based modeling (TBM) or template-free modeling (FM) approaches. Until recently, TBM has been the most reliable approach to predicting protein structures, and in the absence of reliable templates, the modeling accuracy sharply declines. Nevertheless, the results of the most recent community-wide assessment of protein structure prediction experiment (CASP14) have demonstrated that the protein structure prediction problem can be largely solved through the use of end-to-end deep machine learning techniques, where correct folds could be built for nearly all single-domain proteins without using the PDB templates. Critically, the model quality exhibited little correlation with the quality of available template structures, as well as the number of sequence homologs detected for a given target protein. Thus, the implementation of deep-learning techniques has essentially broken through the 50-year-old modeling border between TBM and FM approaches and has made the success of high-resolution structure prediction significantly less dependent on template availability in the PDB library.

Deep learning techniques have significantly impacted protein structure prediction and protein design.

Protein structure prediction and design can be regarded as two inverse processes governed by the same folding principle. Although progress remained stagnant over the past two decades, the recent application of deep neural networks to spatial constraint prediction and end-to-end model training has significantly improved the accuracy of protein structure prediction, largely solving the problem at the fold level for single-domain proteins. The field of protein design has also witnessed dramatic improvement, where noticeable examples have shown that information stored in neural-network models can be used to advance functional protein design. Thus, incorporation of deep learning techniques into different steps of protein folding and design approaches represents an exciting future direction and should continue to have a transformative impact on both fields.