Mathematical model explains differences in Omicron and Delta SARS-CoV-2 dynamics in Caco-2 and Calu-3 cells.

Within-host infection dynamics of Omicron dramatically differs from previous variants of SARS-CoV-2. However, little is still known about which parameters of virus-cell interplay contribute to the observed attenuated replication and pathogenicity of Omicron. Mathematical models, often expressed as systems of differential equations, are frequently employed to study the infection dynamics of various viruses. Adopting such models for results of in vitro experiments can be beneficial in a number of aspects, such as model simplification (e.g., the absence of adaptive immune response and innate immunity cells), better measurement accuracy, and the possibility to measure additional data types in comparison with in vivo case. In this study, we consider a refinement of our previously developed and validated model based on a system of integro-differential equations. We fit the model to the experimental data of Omicron and Delta infections in Caco-2 (human intestinal epithelium model) and Calu-3 (lung epithelium model) cell lines. The data include known information on initial conditions, infectious virus titers, and intracellular viral RNA measurements at several time points post-infection. The model accurately explains the experimental data for both variants in both cell lines using only three variant- and cell-line-specific parameters. Namely, the cell entry rate is significantly lower for Omicron, and Omicron triggers a stronger cytokine production rate (i.e., innate immune response) in infected cells, ultimately making uninfected cells resistant to the virus. Notably, differences in only a single parameter (e.g., cell entry rate) are insufficient to obtain a reliable model fit for the experimental data.

Development of a novel mathematical model that explains SARS-CoV-2 infection dynamics in Caco-2 cells.

Mathematical modeling is widely used to study within-host viral dynamics. However, to the best of our knowledge, for the case of SARS-CoV-2 such analyses were mainly conducted with the use of viral load data and for the wild type (WT) variant of the virus. In addition, only few studies analyzed models for in vitro data, which are less noisy and more reproducible. In this work we collected multiple data types for SARS-CoV-2-infected Caco-2 cell lines, including infectious virus titers, measurements of intracellular viral RNA, cell viability data and percentage of infected cells for the WT and Delta variants. We showed that standard models cannot explain some key observations given the absence of cytopathic effect in human cell lines. We propose a novel mathematical model for in vitro SARS-CoV-2 dynamics, which included explicit modeling of intracellular events such as exhaustion of cellular resources required for virus production. The model also explicitly considers innate immune response. The proposed model accurately explained experimental data. Attenuated replication of the Delta variant in Caco-2 cells could be explained by our model on the basis of just two parameters: decreased cell entry rate and increased cytokine production rate.

Alterations in SARS-CoV-2 Omicron and Delta peptides presentation by HLA molecules.

The T-cell immune response is a major determinant of effective SARS-CoV-2 clearance. Here, using the recently developed T-CoV bioinformatics pipeline (https://t-cov.hse.ru) we analyzed the peculiarities of the viral peptide presentation for the Omicron, Delta and Wuhan variants of SARS-CoV-2. First, we showed the absence of significant differences in the presentation of SARS-CoV-2-derived peptides by the most frequent HLA class I/II alleles and the corresponding HLA haplotypes. Then, the analysis was limited to the set of peptides originating from the Spike proteins of the considered SARS-CoV-2 variants. The major finding was the destructive effect of the Omicron mutations on PINLVRDLPQGFSAL peptide, which was the only tight binder from the Spike protein for HLA-DRB1*03:01 allele and some associated haplotypes. Specifically, we predicted a dramatical decline in binding affinity of HLA-DRB1*03:01 and this peptide both because of the Omicron BA.1 mutations (N211 deletion, L212I substitution and EPE 212-214 insertion) and the Omicron BA.2 mutations (V213G substitution). The computational prediction was experimentally validated by ELISA with the use of corresponding thioredoxin-fused peptides and recombinant HLA-DR molecules. Another finding was the significant reduction in the number of tightly binding Spike peptides for HLA-B*07:02 HLA class I allele (both for Omicron and Delta variants). Overall, the majority of HLA alleles and haplotypes was not significantly affected by the mutations, suggesting the maintenance of effective T-cell immunity against the Omicron and Delta variants. Finally, we introduced the Omicron variant to T-CoV portal and added the functionality of haplotype-level analysis to it.

ExhauFS: exhaustive search-based feature selection for classification and survival regression.

Feature selection is one of the main techniques used to prevent overfitting in machine learning applications. The most straightforward approach for feature selection is an exhaustive search: one can go over all possible feature combinations and pick up the model with the highest accuracy. This method together with its optimizations were actively used in biomedical research, however, publicly available implementation is missing. We present ExhauFS-the user-friendly command-line implementation of the exhaustive search approach for classification and survival regression. Aside from tool description, we included three application examples in the manuscript to comprehensively review the implemented functionality. First, we executed ExhauFS on a toy cervical cancer dataset to illustrate basic concepts. Then, multi-cohort microarray breast cancer datasets were used to construct gene signatures for 5-year recurrence classification. The vast majority of signatures constructed by ExhauFS passed 0.65 threshold of sensitivity and specificity on all datasets, including the validation one. Moreover, a number of gene signatures demonstrated reliable performance on independent RNA-seq dataset without any coefficient re-tuning, i.e., turned out to be cross-platform. Finally, Cox survival regression models were used to fit isomiR signatures for overall survival prediction for patients with colorectal cancer. Similarly to the previous example, the major part of models passed the pre-defined concordance index threshold 0.65 on all datasets. In both real-world scenarios (breast and colorectal cancer datasets), ExhauFS was benchmarked against state-of-the-art feature selection models, including L1-regularized sparse models. In case of breast cancer, we were unable to construct reliable cross-platform classifiers using alternative feature selection approaches. In case of colorectal cancer not a single model passed the same 0.65 threshold. Source codes and documentation of ExhauFS are available on GitHub: https://github.com/s-a-nersisyan/ExhauFS.

Low expression of CD24 is associated with poor survival in colorectal cancer.

In this study we analyzed expression of CD24 in a cohort of colorectal cancer patients using immunohistochemistry staining of CD24. We found a significant association between absence or low expression of CD24 (10% of membranous and 55% of cytoplasmic staining) and shortened patient survival. Protein localization played a crucial role in the prognosis: membranous form was the major and prognostic one in primary tumors, while cytoplasmic expression was elevated in liver metastases compared to the primary tumors and contained prognostic information. Then, using The Cancer Genome Atlas Colon Adenocarcinoma (TCGA-COAD) RNA-seq data, we showed that CD24 mRNA level was two-fold decreased in primary colorectal cancers compared to adjacent normal mucosa. Like the protein staining data, ten percent of patients with the lowest mRNA expression levels of CD24 in primary tumors had reduced survival compared to the ones with higher expression. To explain these findings mechanistically, shRNA-mediated CD24 knockdown was performed in HT-29 colorectal cancer cells. It resulted in the increase of cell migration in vitro, no changes in proliferation and apoptosis, and a slight decrease in cell invasion. As increased cell migration is a hallmark of metastasis formation, this finding corroborates the association of a decreased CD24 expression with poor prognosis. Differential gene expression analysis revealed upregulation of genes involved in cell migration in the group of patients with low CD24 expression, including integrin subunit α3 and α3, ß3 subunits of laminin 332. Further co-expression analysis identified SPI1, STAT1 and IRF1 transcription factors as putative master-regulators in this group.

Hypoxia-Induced miR-148a Downregulation Contributes to Poor Survival in Colorectal Cancer.

Hypoxia is an extensively investigated condition due to its contribution to various pathophysiological processes including cancer progression and metastasis formation. MicroRNAs (miRNAs) are well-known post-transcriptional gene expression regulators. However, their contribution to molecular response to hypoxia is highly dependent on cell/tissue types and causes of hypoxia. One of the most important examples is colorectal cancer, where no consensus on hypoxia-regulated miRNAs has been reached so far. In this work, we applied integrated mRNA and small RNA sequencing, followed by bioinformatics analysis, to study the landscape of hypoxia-induced miRNA and mRNA expression alterations in human colorectal cancer cell lines (HT-29 and Caco-2). A hypoxic microenvironment was chemically modeled using two different treatments: cobalt(II) chloride and oxyquinoline. Only one miRNA, hsa-miR-210-3p, was upregulated in all experimental conditions, while there were nine differentially expressed miRNAs under both treatments within the same cell line. Further bioinformatics analysis revealed a complex hypoxia-induced regulatory network: hypoxic downregulation of hsa-miR-148a-3p led to the upregulation of its two target genes, ITGA5 and PRNP, which was shown to be a factor contributing to tumor progression and poor survival in colorectal cancer patients.

miRGTF-net: Integrative miRNA-gene-TF network analysis reveals key drivers of breast cancer recurrence.

Analysis of regulatory networks is a powerful framework for identification and quantification of intracellular interactions. We introduce miRGTF-net, a novel tool for construction of miRNA-gene-TF networks. We consider multiple transcriptional and post-transcriptional interaction types, including regulation of gene and miRNA expression by transcription factors, gene silencing by miRNAs, and co-expression of host genes with their intronic miRNAs. The underlying algorithm uses information on experimentally validated interactions as well as integrative miRNA/mRNA expression profiles in a given set of samples. The latter ensures simultaneous tissue-specificity and biological validity of interactions. We applied miRGTF-net to paired miRNA/mRNA-sequencing data of breast cancer samples from The Cancer Genome Atlas (TCGA). Together with topological analysis of the constructed network we showed that considered players can form reliable prognostic gene signatures for ER-positive breast cancer. A number of signatures demonstrated remarkably high accuracy on transcriptomic data obtained by both microarrays and RNA sequencing from several independent patient cohorts. Furthermore, an essential part of prognostic genes were identified as direct targets of transcription factor E2F1. The putative interplay between estrogen receptor alpha and E2F1 was suggested as a potential recurrence factor in patients treated with tamoxifen. Source codes of miRGTF-net are available at GitHub (https://github.com/s-a-nersisyan/miRGTF-net).

Association of HLA Class I Genotypes With Severity of Coronavirus Disease-19.

Human leukocyte antigen (HLA) class I molecules play a crucial role in the development of a specific immune response to viral infections by presenting viral peptides at the cell surface where they will be further recognized by T cells. In the present manuscript, we explored whether HLA class I genotypes can be associated with the critical course of Coronavirus Disease-19 by searching possible connections between genotypes of deceased patients and their age at death. HLA-A, HLA-B, and HLA-C genotypes of n = 111 deceased patients with COVID-19 (Moscow, Russia) and n = 428 volunteers were identified with next-generation sequencing. Deceased patients were split into two groups according to age at the time of death: n = 26 adult patients aged below 60 and n = 85 elderly patients over 60. With the use of HLA class I genotypes, we developed a risk score (RS) which was associated with the ability to present severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) peptides by the HLA class I molecule set of an individual. The resulting RS was significantly higher in the group of deceased adults compared to elderly adults [p = 0.00348, area under the receiver operating characteristic curve (AUC ROC = 0.68)]. In particular, presence of HLA-A*01:01 allele was associated with high risk, while HLA-A*02:01 and HLA-A*03:01 mainly contributed to low risk. The analysis of patients with homozygosity strongly highlighted these results: homozygosity by HLA-A*01:01 accompanied early deaths, while only one HLA-A*02:01 homozygote died before 60 years of age. Application of the constructed RS model to an independent Spanish patients cohort (n = 45) revealed that the score was also associated with the severity of the disease. The obtained results suggest the important role of HLA class I peptide presentation in the development of a specific immune response to COVID-19.