Advertisement
Research Article
Transparent Process
Open Access

Suppressed transcript diversity and immune response in COVID-19 ICU patients: a longitudinal study

Priyanka Mehta, Partha Chattopadhyay, Ramakant Mohite, Ranit D’Rozario, Purbita Bandopadhyay, Jafar Sarif, Yogiraj Ray, Dipyaman Ganguly, Rajesh Pandey  Correspondence email
Published 2 November 2023. DOI: 10.26508/lsa.202302305

Abstract

Understanding the dynamic changes in gene expression during Acute Respiratory Distress Syndrome (ARDS) progression in post-acute infection patients is crucial for unraveling the underlying mechanisms. Study investigates the longitudinal changes in gene/transcript expression patterns in hospital-admitted severe COVID-19 patients with ARDS post-acute SARS-CoV-2 infection. Blood samples were collected at three time points and patients were stratified into severe and mild ARDS, based on their oxygenation saturation (SpO2/FiO2) kinetics over 7 d. Decline in transcript diversity was observed over time, particularly in patients with higher severity, indicating dysregulated transcriptional landscape. Comparing gene/transcript-level analyses highlighted a rather limited overlap. With disease progression, a transition towards an inflammatory state was evident. Strong association was found between antibody response and disease severity, characterized by decreased antibody response and activated B cell population in severe cases. Bayesian network analysis identified various factors associated with disease progression and severity, viz. humoral response, TLR signaling, inflammatory response, interferon response, and effector T cell abundance. The findings highlight dynamic gene/transcript expression changes during ARDS progression, impact on tissue oxygenation and elucidate disease pathogenesis.

Introduction

Severe cases of COVID-19 often result in acute respiratory distress syndrome (ARDS), a life-threatening condition characterized by severe lung inflammation and compromised oxygenation, as with many other respiratory infections (1). ARDS poses significant challenges in patient management and requires a comprehensive understanding of the underlying host response at a molecular level. However, it is often underdiagnosed in intensive care settings (2). A key clinical implication of ARDS is the fluctuating oxygenation status, as indicated by the oxygen saturation (SpO2/FiO2) ratio (3). This can serve as a critical parameter to stratify patients and study the heterogeneity of the disease (1, 4). To unravel the intricate mechanisms driving disease progression, a thorough characterization of the host transcriptome is essential.

Transcriptomic data analysis is one of the common methods that provide a comprehensive understanding of the host response and multiple studies have explored the host immune response during COVID-19 infection (5, 6). Although this method allows for high-resolution dynamic gene expression profiling, the bulk of RNA-seq based investigations focus only on genes rather than individual/specific transcripts per gene. This is in part because of the fact that gene-level studies are thought to be more robust and produce more empirically actionable results (7, 8). However, it is important to recognize that transcript-level analysis provides a granular understanding of alternative splicing, isoform expression, and posttranscriptional modifications, which can uncover novel insights into gene regulation and biological processes (9). This is especially important as collaborative international efforts have highlighted pervasive transcription with evidence of average five transcripts/gene (10). By capturing the dynamic nature of gene expression at the transcript level, transcriptomic data provide a deeper understanding of the complexity and heterogeneity of the host response.

An important aspect of host response is its dynamic nature. Numerous studies have focused on the time-dependent variations in immune and inflammatory responses in COVID-19 (11); however, a longitudinal transcriptomic understanding on the progression of ARDS in intensive care unit (ICU)-admitted COVID-19 patients is essential to better understand the underlying molecular mechanisms driving ARDS in COVID-19 patients. By analyzing gene and transcript expression patterns, we can identify dysregulated pathways and potential therapeutic targets and the underlying mechanism driving such changes. A transcriptome analysis over time can aid in the prediction of disease severity, aiding in early identification of individuals at high risk and the decision making towards relevant therapies. Thus, in this investigation, we looked at the longitudinal alterations in expression patterns in severe COVID-19 patients with ARDS, admitted to the ICUs of ID and BG Hospital in Kolkata, India. Longitudinal blood samples were collected from these patients at three distinct time points after the onset of ARDS: day 1 (T1), 3–4 d later (T2), and 7 d after the first collection (T3). The collection intervals allowed us to capture gene expression dynamics during disease progression and clinical outcome.

In our investigation, we have examined both transcript and gene-level differential expression patterns. By exploring these two levels of complementary analysis, we aim to assess which analytical approach provides a comprehensive understanding of the biological processes at play delineating host immune response. The transcriptomic data are integrated with clinical and biochemical parameters to capture the complex interactions and relationships between immune cell types, cytokines, clinical parameters, and gene expression patterns. In summary, our study highlights the reduction in transcript diversity as time and severity increase, association of immunoglobulin genes with disease severity and a shift towards pro-inflammatory state overtime (Fig 1).

Figure 1.
Figure 1. Conceptual summary of the study design, illustrating the integration of multi-variable omics data for longitudinal analysis to investigate gene- and transcript-level expression dynamics over time, with a focus on understanding the severity of ICU-admitted severe COVID-19 in patients with ARDS.

Results

Clinical demographics and patients classification

In this study, longitudinal blood samples were collected from COVID-19 patients admitted to ICU at the ID and BG Hospital in Kolkata, India (Fig S1A). During the period of hospitalization, blood was collected at three time-points at an interval of 3–4 d after admission: day 1 (T1), 4 d later (T2), and 7 d after first collection (T3) (Fig 2A). All the patients suffered from severe COVID-19, according to Indian Council of Medical Research guidelines, symptoms and showed evidence of ARDS as measured by their oxygen saturation (ratio of oxygen saturation [SpO2] and fraction of inspired oxygen [FiO2]) which ranged between 100–300 mmHg. To evaluate the respiratory status across the ARDS patients, the area under the curve (AUC) was computed as described earlier from the SpO2/FiO2 ratio kinetics curve over 7 d post admission (4, 12). Despite all the patients having severe COVID-19, their SpO2 curve varied from 100 to 2,000 mmHg. The patients were segregated into two groups based on AUC values. Patients with an AUC ≤771.7 were categorized as Low AUC and had median SpO2/FiO2 ratios between 90–150 mmHg, whereas those with an AUC > 771.7 were designated as High AUC with median SpO2/FiO2 ratios between 100–350 mmHg (Fig 2B). Between the two groups, there was no difference in the median age of the patients (low = 59 [22–82] yr; high = 56.5 [46–84] yr). Whereas all the patients in the High AUC group recovered, three patients in the Low AUC group resulted in mortality (Table S1). In the Low AUC group, only one patient received convalescent plasma therapy (CPT), whereas in the High AUC group, six patients underwent the same treatment. In addition, a few patients in both groups also received remdesivir, ivermectin, and corticosteroids such as dexamethasone, hydrocortisone along with other drugs as part of their treatment regimen (Table 1). Type 2 diabetes and hypertension were amongst the common comorbidities observed in both the groups. Furthermore, there were no significant differences observed in the biochemical parameters between the patients in both the groups.

Figure S1.
Figure S1. Patient statification and comparison of DE genes and transcripts between time point T3 and T1.

(A) The patient stratification from a randomized clinical trial for longitudinal transcriptome expression study from standard of care and convalescent plasma therapy groups. (B) The bar plot represents biotypes proportions for the significantly up-regulated (in red) and down-regulated (in blue) transcripts between T3 versus T1 comparison. (C) The bar plot represents significantly up-regulated (in red) and down-regulated (in blue) genes biotype proportions between T3 versus T1. (D) Volcano plot representing differential gene expression between T3 and T1. (E) The log2 fold change of overlapping genes from gene-level and transcript-level expression analysis.

Figure 2.
Figure 2. Differential transcript expression between ICU-admitted COVID-19 ARDS patients across longitudinal profiling between T3 and T1.

(A) Blood PBMCs were collected at three time-points–T1, T2, and T3, for patients and classified into two groups of Low AUC and High AUC. (B) The SpO2/FiO2 ratio curve for patients across 7 d from the first day of sample collection divides the patients into Low AUC and High AUC groups. (C) Global transcriptome diversity across different biotypes of genes for the human genome, representing higher transcript proportion to genes. (D) Volcano plot representing differentially expressed transcripts between T3 and T1. (E) Circular heatmap representing the log2 fold change for the differentially expressed transcripts (*represents the significantly expressed log2 fold changes). (F) The dot plot represents transcripts with multiple significantly expressed isoforms; the color corresponds to the log twofold change between T3 and T1. (G) The slope plot represents the number of nonsignificant isoforms expressed for each significant isoform for the up-regulated and down-regulated transcripts between T3 and T1. Multiple genes having the same number of transcripts/expressing the same number of transcripts are clubbed together. (H) Venn diagram represents the overlap between differential expressed genes (in blue) and transcripts (in red). (I) The plot represents the significantly enriched reactome pathways for DGE (as blue triangles) and DTEs (as red dots). The size of the icon represents the number of genes involved in the pathway.

View this table:
Table 1.

Clinical characteristics of the patient cohort.

Transcript expression patterns reveal suppressed transcript diversity and B-cell response during disease progression

According to the Ensembl (GRCh38) genome reference, there are ∼20 k protein coding and ∼17 k noncoding genes in humans which transcribe more than 150 and 50 k distinct transcripts, respectively (Fig 2C). During infection, the transcriptome dynamics has been shown to vary depending on the type of infection and the severity of the disease condition. Therefore, we examined the expression dynamics across the time points from the RNA-seq data at both the gene and transcript isoform levels to derive granular understanding of the patient transcriptome through disease progression. We begin by looking at the transcript-level expression patterns. We observed that 133 transcript isoforms varied significantly between T3 and T1, 107 transcripts down-regulated and 26 up-regulated at Padj < 0.05 and log2FC |1| (Fig 2D) (Table S2). However, across other time point comparisons (T1 versus T2 and T2 versus T3), no significant transcripts were found. We filtered any unannotated transcripts and pseudogenes (n = 20) for further consideration. We then look at the biotypes of each transcript because various isoforms of the same gene were found to be significantly expressed. We find most of the transcripts fall into protein-coding biotypes (n = 67), whereas 15 transcripts were annotated with retained introns, 10 transcripts with nonsense mediated decay, four were lncRNAs and nine processed transcripts. We also observed the presence of several immunoglobulin heavy and light chain transcripts (IG), two for constant (C) and seven for variable regions (V) in the significant isoforms (Fig 2E). Interestingly, all the immunoglobulin transcripts are down-regulated at T3 compared with T1.

It is important to note that we observed multiple isoforms for some of the differentially expressed transcripts (IL18R1, IFI27, SEMA4D, SIGLEC1, FKBP5, TTN, SERPINA1, DHRS13, SYNE1, NAIP, MAP2K6, KIAA1958, TLR5, and CDK5RAP2) (Fig 2F). Majority of these transcripts are down-regulated at T3, and the decreased level of these gene isoforms are involved functionally in decreased B cell and T helper cell activation, type I interferon function, and increased inflammatory cytokine response mediated by TNF-α and TLR signaling pathways (13, 14, 15, 16, 17, 18, 19). Independent investigations have also found that genes related with some of these transcripts (IGI27, SEMA4D, SIGLEC1, and SERPINA1) are repressed in severe COVID-19, underlining their potential as early predictors of severity (20, 21, 22). Of the total differentially expressed transcripts, 61% of the transcripts were protein coding transcript at T3, compared with the 48% protein coding transcript at T1 (Fig S1B and C). We also observed an overall decreased transcript diversity (measured by the number of significantly expressed transcripts of the gene versus the number of nonsignificant transcripts expressed) at T3 compared with T1 (Fig 2G). As the infection progresses from T1 to T3, the findings show an offset in transcript diversity focused at transcription of more protein-coding transcripts in response to the infection.

Next, we performed gene-level differential expression analysis across time points to compare with transcript-level analysis. We found 118 genes to be differentially expressed between T3 and T1 (Padj < 0.05 and log2FC |1.5|), with 78 being down-regulated, whereas 40 genes were up-regulated at T3 (Fig S1D) (Table S3). Overall, there was an overlap of 27 genes between gene and transcript level DE, with two genes, KIAA1958 and HIC1, having multiple significantly expressed isoforms (Figs 2H and S1E). Even though the direction of expression was comparable at the gene and transcript levels, only 20% of the differentially expressed transcripts (DTEs) were captured at the gene level. We used gene set enrichment analysis to investigate if the functional implications of all the DE genes and transcripts overlapped. We found immune response-related pathways like anti-inflammatory response, FC receptor-mediated NF-κB, MAPK and phagocytosis pathways, adaptive immune response, and BCR activation pathways to be negatively enriched at both the gene and transcript levels. However, several pathways like cell cycle, scavenging of heme from plasma, vesicle-mediated transport, cytokine and interleukin signaling pathways, and RNA pol II transcription, ESR-mediated signaling pathways were negatively enriched at only the transcript level (Fig 2I). This suggests that transcript level DEs highlight the alteration of cellular processes vital to maintaining homeostasis and strengthening immunological response during SARS-CoV-2 infection, in addition to the host immune response.

To validate our analysis of transcript-level expression, we conducted Leafcutter analysis comparing T1, T2, and T3 time points. We identified 133 clusters with significant differential splicing between T3 and T1. Out of these, nine clusters were found to overlap with differentially expressed transcripts (CDK5RAP2, EPSTI1, FBXO9, FCGR1B, NAIP NAIPP2, SERPINA1, SHISA5, and SYNE1). Interestingly, no transcripts were found to overlap between the T2 and T3 time point comparison. However, we did observe significant differential splicing in 30 genes between T2 and T1 (Table S4). The CDK5RAP2 gene exhibits five significantly differentially expressed transcripts with distinct biotypes (protein-coding, retained intron, processed transcripts) (Fig 2E). Our investigation into the junctions for this transcript revealed differential splicing across three exons (Fig 3A). Specifically, junction “a” demonstrated an up-regulation at T3 (with a dPSI of 0.325), whereas both junctions “b” and “c” exhibited down-regulation at T3 (with dPSIs of −1.40 and −0.135, respectively) (Fig 3B). Notably, junctions “b” and “c” corresponded to the processed transcript biotype isoforms, whereas junction “a” corresponded to the biotype associated with protein-coding or retained intron (Fig 3C). Despite all these transcripts experiencing down-regulation at T3, we observed a higher occurrence of exon-skipping events at T3. This suggests an inclination towards the production of protein-coding transcripts at the T3 time point compared with T1.

Figure 3.
Figure 3. Alternative splicing serves as a mechanism influencing the expression of distinct isoforms.

(A) Leafcutter visualization illustrates the differential splicing of the CDK5RAP2 gene exon junctions between the T3 and T1 time points. (A, B) The statistics presents junctions, indicating dPSI (Delta Percent Spliced In) values for the three exon junctions of CDK5RAP2 as depicted in (A). (C) An outline of the expressed transcripts for CDK5RAP2 includes their biotypes, with emphasis on the junction featuring reads subject to differential splicing.

Transcript expression indicates dysregulated T-cell activation in severe ARDS patients

As we observed a significant difference in the SpO2 between the high and low AUC patients, we performed differential transcript and gene expression analysis between high and low AUC within each time-point (Fig 4A). The patient subgroups did not differ at T1, but we observed 63 transcript isoforms to be differentially expressed at T2 (12 up-regulated and 51 down-regulated, Padj < 0.05 and log2FC |1|) (Fig 4B) (Table S5). Whereas at T3, 105 transcript isoforms were differentially expressed (50 up-regulated and 55 down-regulated, Padj < 0.05 and log2FC |1|) (Fig 4C) (Table S6). Interestingly, there was an overlap of only eight transcripts (ABI3, BCL11B, CD5, CR1, EST1, PVRIG, SLC25A38, and TRBC1) between the AUC comparisons at T2 and T3 (Fig 4D). The down-regulation of CD5, BCL11B, and PVRIG transcripts in the low AUC group of both T2 and T3 suggest a decreased T cell activation, TCR signaling, and an increased T cell exhaustion (Fig 4E) (23, 24, 25, 26). When the transcripts are classified by biotype, we discover that most of the transcripts are protein coding (79% DTE at T2 and 68% DTE at T3) (Fig 4F). Interestingly, we observed all the significant isoforms of T cell receptor genes to be down-regulated in low AUC patients over time, pointing towards a decreased T cell response in the severe COVID-19 patients (27, 28, 29). Similar to our previous finding, we observed a decreased transcript diversity for the differentially expressed transcripts in the low AUC group at both T2 and T3 (P-value = 0.0015 at T2 and P-value = 0.014 at T3) (Fig 4G and H). However, the protein coding transcript isoforms was less abundant in the low AUC group compared to the High AUC at both T2 (25% protein coding transcript in low AUC, 92.15% protein transcripts coding in High AUC; P-value 0.001) and T3 (58% protein coding transcript in low AUC, 78.18% protein transcripts coding in High AUC; P-value 0.026) (Fig S2A and B).We also see an increase in lncRNA at T3 in the Low AUC group (P-value 0.01). We hypothesized that the decreased transcript diversity and transcription of protein coding isoforms are associated with a more severe form of the SARS-CoV-2 infection.

Figure 4.
Figure 4. Differential transcript expression between Low AUC and High AUC patients across the time points.

(A) The SpO2/FiO2 ratio of ICU-admitted patients fluctuated over time, and each time point was investigated for DE of transcripts to elucidate variations between the two severity sub-groups. (B, C) The volcano plots represent the DE transcripts between Low AUC and High AUC groups at (B) T2 and (C) T3 time points. (D) The Venn diagram represents the overlap of DTEs between Low and High AUC groups between T2 (yellow) and T3 (red). (E) The dot plot represents the log2fold change of the isoforms that are overlapping between the two time-points. (F) The circular heatmap represents the different biotypes of the isoforms expressed between Low and High AUC across time points (*represents significance in that group). (G, H) The violin plots represent the ratio of significant to nonsignificant isoforms between the up-regulated and down-regulated transcripts for (G) T2 Low versus High AUC group, and (H) T3 Low versus High AUC groups. (I) The Venn diagram represents the overlap of genes between the gene-level (violet) and transcript-level (orange) DE comparisons between T2 Low versus High AUC group. (J) The Venn diagram represents the overlap of genes between the gene-level (violet) and transcript-level (orange) DE comparison between T3 Low versus High AUC. (K) The dot plot represents the overlap between gene- and transcript-level comparisons of T2 Low versus High AUC with the color of the dot representing the log2 fold change. (L) The dot plot represents the overlap between gene and transcript-level comparison of T3 Low versus High AUC with the color of the dot representing the log2 fold change. (M) The plot represents the significantly enriched reactome pathways for DGEs (as triangles) and DTEs (as dots) at T2 (yellow) and T3 (red). The size of the icon represents the number of genes involved in the pathway.

Figure S2.
Figure S2. Comparison of differentially expressed gene and transcript between high and low AUC across time points. The bar plots represent the biotype proportions between significantly up-regulated (in purple) and down-regulated (in orange) transcripts between.

(A, B) T2 Low versus High AUC groups, and (B) T3 Low versus High AUC groups. (C, D) The volcano plot represents differentially expressed genes between (C) T2 Low versus High AUC groups, and (D) T3 Low versus High AUC groups. (E, F) The bar plots represent the biotype proportions between significantly up-regulated (in purple) and down-regulated (in orange) genes between (E) T2 Low versus High AUC groups, and (F) T3 Low versus High AUC groups.

When we performed differentially expressed genes (DGE) analysis between the Low and High AUC groups at both T2 and T3, we observed 242 DE genes at T2 (34 up-regulated and 208 down-regulated in the Low AUC) and 363 DE genes at T3 (150 up-regulated and 213 down-regulated in the Low AUC) (Padj 0.05 and log2FC |1.5|) (Fig S2C and D) (Tables S7 and S8). The genes followed a similar trend in the abundance of protein coding genes in the low AUC group compared with High AUC at both T2 and T3 (P-value 0.001) (Fig S2E and F). Despite having little overlap with the differentially expressed transcripts at T2 and T3 (Fig 4I and J), the expression profile of the overlapping genes and transcripts that were differentially expressed were similar at both T2 and T3 (Fig 4K and L). To understand the function of the DE genes and transcripts between the high and low AUC groups at both the time points, we performed pathway enrichment analysis which highlighted distinct pathways associated with transcripts that were not associated with the DE genes otherwise (Fig 4M). For example, pathways related to translation and amino acid metabolism were positively enriched in the low AUC group at T2, suggesting a dysregulated translation and amino acid metabolism process in the severe COVID-19 patients (30). We also observed a positive enrichment of the neutrophil degranulation pathway, a hallmark of severe COVID-19 infection, in the low AUC group at T2 (31). On the other hand, the PD-1 pathway and CD28-mediated co-stimulatory pathway were negatively enriched in the low AUC group at T3, indicating a dysregulated T cell activation in the severe COVID-19 patients (32). The TCR signaling pathway was also negatively enriched at both T2 and T3 in the low AUC group, further suggesting the same. Overall, the pathway enrichment analysis highlights dysregulated metabolic function and T cell-mediated immune response in the low AUC group at T2 and T3. At the same time, it also highlights that transcript level analysis of the host response can provide deeper insights about the disease state during an infection (33). Interestingly, two pathways, adaptive immune response and IL-1 signaling pathway were negatively enriched based on transcript expression, but positively enriched based on the gene expression profile in the low AUC group. To gain a better understanding of this disparity, we investigated the cytokine and immune cell profiles.

Integrated analysis reveals immune response dynamics and key associations with disease progression

Because we have differential abundance of the immune/inflammatory pathways and counter-acting enrichment patterns of two immune-related pathways at the same time, we wanted to characterize the immune/inflammatory response further (Fig 4M). We quantified the plasma cytokines of these patients across the time points. The cytokine expression across time points are available as Table S9. The B cell, T cell, and dendritic cell subsets were also quantified using specific antibodies across time points (Fig 5A). The complete cell type abundance for each individual is available as Table S10. We observed IL-18 to be significantly decreased (P-value = 0.029) at T3 compared with T1 (Fig 5B), whereas macrophage migration inhibitory factor (MIF) level was significantly high (P-value = 0.0034) at T3 compared with T1 (Fig 5C). SCGF-b, IL-16, and IL-6 were also differentially expressed at different time points (Fig 5D–F). The activated B cell and T helper cell populations were also decreased at T3 compared with the T1 (Fig 5G–I), suggesting a decreased B and T cell response at T3, conforming to our initial finding. Whereas the higher expression of MIF was reported to be associated with severe COVID-19 (34), down-regulation of IL-18 is possibly because of the decreased T cell function (35).

Figure 5.
Figure 5. Longitudinal dynamics of cytokines and cell types and association with gene/transcript expression.

(A) The longitudinal dynamics of gene/transcripts and their association with immune cells and cytokines is explored using a Bayesian network. (B, C, D, E, F) The raincloud plots represent cytokine levels across time points (in pg/ml) T1 (in blue), T2 (in yellow), and T3 (in red) for (B) IL-18, (C) MIF, (D) SCGF-b, (E) IL-16, and (F) IL-6. (G, H, I) The raincloud plots represent cell types differentially abundant across time points for (G) CD27−CD38+ activated B cells, (H) CD4+CCR6-CXCR3- helper T cells (I) CD4+CCR6+ helper T cell populations. (J) The Bayesian network represents the connections among clusters of cytokines, cell types, differentially expressed (DE) genes/isoforms, and clinical parameters between T3 and T1. Circular nodes represent DE transcripts, whereas triangle nodes represent DE genes. The color-coded nodes represent various biotypes of genes and transcripts pink for protein-coding, blue for lncRNA, light pink for pseudogenes, orange for immunoglobulins; green nodes representing clinical parameters, purple nodes represent cytokines, and yellow nodes represent cell types.

To summarize the findings related to the progression of the disease from T1 to T3, we constructed a Bayesian network model. This model incorporated the differential expression of genes and transcripts between T1 and T3, cytokine expression, cell type abundance, and clinical parameters including AUC and outcomes. We observed two prominent nodes consisting of AUC and time, suggesting that the AUC is a time-independent clinical variable, whereas the outcome is dependent on AUC (Fig 5J). It is noteworthy that we identified an association between CNGA4, a nucleotide-gated channel, and time. CNGA4 was found to be associated with transcripts of MAP2K6, CDK5RAP2, TLR5, and CLEC4D, and the IL-8 cytokine and CD4+ T cell abundance. This association indicates the involvement of TLR signaling and inflammatory response, and it exhibits variability over time. Another interesting finding is the presence of a node comprising primarily T helper cells, which is associated with time. This suggests a potential association between the T helper cell population and disease progression, supporting our previous observation of decreased T helper cell function at time point (T3). We also observed a correlation between time and the pro-inflammatory cytokine growth-related oncogene-α (GRO-α). Intriguingly, specific transcripts of SIGLEC1, IFI27, EPSTI1, and KIAA1958 genes were found to be associated with GRO-α, and these genes are also involved in regulating the interferon response. In addition, we noticed a cluster containing immunoglobulin genes and transcripts, particularly the variable regions of Immunoglobulin Heavy/Kappa/Lambda chains, which showed an association with the severity (AUC). Three cytokines, IL-4, IL-13, and G-CSF were also found to be associated with the immunoglobulin genes, and are known to be involved in the immunoglobulin response cascade during an infection (36, 37, 38, 39). The Bayesian network revealed the association of specific genes and transcripts with time (TLR signaling, inflammatory response, effector T cell function, and interferon signaling) and AUC (immunoglobulin response).

Bayesian network reveals gene/transcript clusters associated with severity

Similar to the time point-based investigation of immune response, we also investigated the cytokine and cell type abundance between high and low AUC groups at both T2 and T3 (Fig 6A). We observed an increase in IL-6 expression in the low AUC group at T2 (Fig 6B). IL-6 has already been reported to positively correlate with infectious disease severity (40, 41). Two pro-inflammatory cytokines, eotaxin and IL-8, were also significantly differentially expressed between the high and low AUC groups at T1 but not T2 or T3 (Fig 6C and D). Importantly, IL-8, up-regulated in the low AUC group at T1, may act as early biomarkers of COVID-19 disease progression and severity (42, 43).

Figure 6.
Figure 6. Cytokines and cell type characterization in relation to severity.

(A) Integrating differentially expressed genes and transcripts with differences in immune cell abundance and subsequent changes in cytokine responses to examine the interplay through time in distinct severity groups. (B, C, D) The raincloud plots represent cytokine levels across time points for high (in orange) and low (in purple) AUC (in pg/ml) for (B) IL-8, (C) eotaxin, and (D) IL-6. (E, F, G, H) The raincloud plots represent cell types differentially abundant between high (in orange) and low AUC (in purple) for (E) memory B cells CD27+, (F) dendritic cells pDC, (G) CD4+ T cell, and (H) CD8+ T cell. (I) The Bayesian network represents the connections among the clusters of cytokines, cell types, DE genes/isoforms, and clinical parameters between Low and High AUC at T2 and T3. Circular nodes represent DE transcripts, whereas triangle nodes represent DE genes. The color-coded nodes represent various biotypes of genes and transcripts pink for protein-coding, blue for lncRNA, light pink for pseudogenes, orange for T cell receptor (TR); green nodes representing clinical parameters, purple nodes represent cytokines, and yellow nodes represent cell types.

We also observed an overall decrease in the key immune cell types in the low AUC group. For instance, CD27+ memory B cell population decreased (P-value = 0.0004) in the Low AUC at T2 compared with the High AUC (Fig 6E). Similarly, the plasmacytoid dendritic cell population was also significantly decreased (P-value = 0.019) in the low AUC at T2 (Fig 6F). We found the CD4+ and CD8+ T cell populations to be significantly decreased (P-value = 0.002 and 0.01 respectively) in the Low AUC compared with the High AUC (Fig 6G and H). The CD4+ T cell population was also significantly decreased (P-value = 0.004) in the low AUC at T3. Whereas the depleted T cell population is a hallmark of severe COVID-19 and also conforms to our previous finding (Fig 4M), only a handful of studies reported a depleted dendritic cell population in severe COVID-19 and needs further investigation (44). Importantly, few studies reported a decrease in memory B cell population in vaccination breakthrough infections (45); however, there is no direct evidence of depleted B cells in severe COVID-19 in unvaccinated individuals.

Finally, we combined the differential gene and transcript expression between the two AUC groups, the cytokine profile, immune cells abundance, and the clinical parameters to build a Bayesian network to get a comprehensive picture of the differential host response between the high and low AUC patients (Fig 6I). Similar to our previous finding, we observed that the infection outcome is directly associated with AUC, but time is not associated with AUC. We observed a cluster of genes mainly regulated by STAT4 to be associated with AUC, which plays a role in the innate immune response pathway. Another cluster controlled by the HAPLN3, was also associated with AUC. HAPLN3 is involved in ERK signaling and ECM regulation, two important biological pathways altered during infection. The key transcripts present in this nodes are CR1, IL1R1, IRAK3, TLR4, and MAPK14, and helper CD4+ T cell subtypes, and are involved in the regulation of MyD88:MAL signaling cascade, TLR signaling, and interleukin signaling. Interestingly, the AUC was also associated with NDRG2, a target of HIF-1a signaling and regulator of apoptosis, suggesting NDRG2 as a possible regulator of AUC in the COVID-19 patients.

In addition, we have identified other unique clusters that play significant roles in disease-associated functions, despite their lack of association with the AUC. One cluster primarily consists of T cell receptor genes, along with other T cell-specific genes such as CD3E, CD3G, IFI6, CD28, CD40LG, IFIT1, and two subtypes of CD8+ helper T cells. Another cluster is primarily regulated by the protein coding transcript of ZAP70 and includes multiple transcripts of adaptive immune response genes like CD2, CD247, CD5, CD6, DUSP2, CD8B, and cytokines such as FGF-basic and IL-1B, along with subtypes of B cells. We also observed a third cluster consisting of transcripts of B and T cell receptor genes, and other cell type-specific genes, regulated by a protein coding transcript of LDHB. Notably, this cluster includes multiple protein coding transcripts of ribosomal proteins that are known to be involved in both host and viral translational regulation. These findings further support our initial discovery of dysregulated translational regulation in the low AUC patients at T2 (Fig 4M).

In summary, the Bayesian network analysis highlighted that specific signaling pathways, including interleukin signaling, MyD88:MAL signaling, TLR signaling, and innate immune response, are associated with the AUC in severe COVID-19. On the other hand, pathways related to B/T cell response, adaptive immune response, and translational regulation are found to be dysregulated, independent of their association with AUC. These findings highlight the complex interplay of biological pathways in severe COVID-19, with some pathways directly linked to disease severity and others showing dysregulation irrespective of AUC values.

Discussion

Conventional RNA sequencing studies typically focus on gene-level expression differences, overlooking the specific expression patterns of individual transcripts. In this study, we took a comprehensive approach by analyzing both gene expression and transcript level heterogeneity in ICU-admitted severe COVID-19 patients with ARDS over time. Despite being admitted at different time points, all patients had severe COVID-19 and developed ARDS. As the disease advanced, a noticeable pattern in the SpO2/FiO2 ratio emerged among the patients. High AUC values implied improved oxygenation (Fig 2B).

It is worth noting that 7 out of the 23 patients (six from the High AUC group, one from the Low AUC group) in this cohort received CPT as part of the randomized clinical trial. It is important to note that the impact of the intervention was not evaluated in the current study. This is primarily because of the fact that only a small subset of individuals received the intervention, making it difficult to draw definitive and reliable conclusions regarding the long-term effects of the treatment. Besides, the randomized clinical trial involving the same patients and several other studies have reported no significant impact of CPT on the COVID-19 disease severity and outcomes (46, 47, 48, 49). Therefore, the effect of CPT on the gene and transcript expression has not been assessed in this study.

Our transcript-level analysis unveils a decline in transcript diversity over the course of 7 d in our patient cohort. This reduction is particularly significant in the higher severity group, as evidenced by Low AUC values, compared with the High AUC patients. Previous research has already indicated dysregulated transcriptional regulation and disruption of protein coding genes in COVID-19 patients (50). However, the observed decrease in transcriptome diversity in our study adds a novel aspect to our understanding of the disease and its impact on transcriptional processes. In our previous study, we also noted a similar decrease in transcript diversity among COVID-19 patients, further emphasizing the association between transcript diversity and disease severity (P-value 0.032) (Fig S3A) (51). This finding suggests a potential role of transcriptome dysregulation in the development of severe COVID-19 outcomes.

Figure S3.
Figure S3. Timepoint and severity specific transcript expression patterns for specific genes.

(A) Transcript diversity between mild and mortality groups measured by ratio of number of isoforms expressed/total number of isoforms for the differentially expressed genes captured between mortality and mild comparison groups. (B, C, D) The heatmap represents the number of isoforms of (B) CDK5RAP2, (C) SERPINA1, and (D) SEMA4D captured in the study along with the log2 fold change of all the transcripts between T3 versus T1 comparison (*marks significantly expressed isoforms). (E) Expression of NDRG2 gene with respect to the AUC groups (orange for High AUC and purple for Low AUC group samples) across all the three time points.

Interestingly, comparing gene- and transcript-level analyses showed limited overlap between them across the time points and severity groups (Figs 2H and 3I and J). This discrepancy can be attributed to factors such as the loss of transcript-specific information during gene-level quantification, variations in transcript isoform expression patterns, and technical differences in sequencing depth and sensitivity (Fig S3B–D) (33). However, gene set enrichment analysis revealed that transcript-level expression patterns effectively capture important cellular processes related to cell growth, metabolism, and maintenance of cellular functions (Figs 2I and 4M). This highlights the significance of analyzing gene expression at the transcript level, as it provides a more comprehensive understanding of the underlying cellular processes during the host response compared with gene-level analysis alone (52).

As the disease advances, there is a down-regulation of inflammatory pathways, including NF-κB signaling, MAPK signaling, phagocytosis, and BCR activation at T3 (Fig 2I). This down-regulation suggests a reduction in cytokines. These observations highlight the potential therapeutic value of targeting these pathways in severe COVID-19 cases to promote recovery (53, 54, 55). In our analysis of immune and inflammatory responses, we examined both plasma cytokines and immune cell abundance. Out of the 48 cytokines initially measured, we identified 36 cytokines present in plasma. Significant changes were observed in IL-18 and MIF levels between day 1 (T1) and day 7 (T3). In addition, a dynamic expression of IL-6 was observed as the disease progressed, indicating elevated cytokine expression in the severe (low AUC) group on day 4 (42, 43, 56). In addition, we noticed a decline in activated B cells and helper T cells on day 7, aligning with our gene and transcript level findings, indicating impaired innate and adaptive immune responses and TCR signaling (Fig 4M). Similar patterns were observed in the severe group (low AUC), suggesting that persistent viral stimulation might contribute to T cell exhaustion, leading to reduced cytokine production and impaired immune function (57, 58). Furthermore, on day 1 (T1), eotaxin and IL-8 levels were notably lower in the severe (low AUC) group. Whereas IL-8 has been linked to severe COVID-19 in some studies, the association between eotaxin expression and disease severity remains unclear (42, 43, 56). Our findings underscore the importance of gene/transcript-based analysis in gaining a better immunological understanding, surpassing the limitations of plasma cytokine quantification, which is more commonly used in clinical and research settings (59).

The Bayesian network analysis reinforced our findings on differential expression, highlighting the association between disease progression (time) and factors such as TLR signaling, inflammatory response, interferon response, and effector T cell abundance. Moreover, a group of cytokines (IL-16, MIF, monokine induced by IFN-γ, IL-10, GRO-a, IL-18, IFN-g, IP-10) demonstrated close connections to disease progression, although only IL-18 exhibited significant differences between T3 and T1 in the conventional analysis. This emphasizes the need for further investigation to fully understand the role of these cytokines in disease progression. In addition, the analysis revealed new findings, including the independence of disease severity (AUC) on time and the dependence of disease outcome on the severity (AUC) (Figs 5J and 6I). This suggests that measuring the SpO2 curve (AUC) can serve as a valuable predictor of disease outcome, especially for COVID-19 patients with ARDS (60). Notably, the association between immunoglobulin response and disease severity/AUC was stronger than the association with time. Contrary to expectations, there was a decrease in immunoglobulin response and activated B cell population at T3 (Figs 2E and 5G). This could potentially be attributed to the administration of corticosteroids (dexamethasone and hydrocortisone) to the ARDS patients as part of their treatment regimen (Tables 1 and S1). Nearly 75% of the patients in both the high and low AUC groups were administered corticosteroid treatment. Studies have shown that corticosteroids modulate release of immunoglobulins, by affecting the earlier stages of B cell proliferation (61, 62). Therefore, the observed decrease in immunoglobulin response and activated B cell population at T3 may be attributed to the immunosuppressive effects of corticosteroids administered to the patients, as these drugs can directly impact B cell function and the overall immune response (63, 64).

The Bayesian network model identified clusters of genes associated with the B and T cell responses to be co-expressed in the time-dependent AUC comparison group (Fig 5I). Interestingly, both the B and T cell response pathways were associated with a calcium-binding protein family gene, NCALD. Although this gene is known to encode a neuronal calcium sensor, limited evidence shows its expression in immune cells including the T cell subtypes and the NK cells (65, 66). An interferon inducible gene, PYHIN was found to be directly associated with T cell-specific gene (67), CD3E, which in turn, was associated with a series of genes including MAF, IL12R, CD3G, CD28, CD40LG, IFI6, OAS2, cytokines such as IL6, and CD8+ T cells, regulating a cascade of T cell mediated immune and inflammatory response (68, 69, 70, 71, 72, 73). On the other hand, LDHB was identified as a possible regulator of B and T cell response (74). It was significantly associated with T cell receptor genes TRAC and TRBC which control the antigen sensing by T cell, CCL5, which regulates the proliferation and IgM secretion by the B cell, and ETS1, which regulates the proliferation and differentiation of both B and T cells. The network suggests NCALD, PYHIN, and LDHB as possible master regulators of the B and T cell responses. However, the existing literature provides only limited knowledge on their role in regulating B and T cell response and further studies in this direction is warranted, either in different cohorts of the same disease or other infectious disease.

Furthermore, the network analysis revealed additional insights. It identified specific genes and transcripts, such as CNGA4-201 as regulators of TLR signaling and inflammatory response, KIAA1958-202 and GRO-a interferon response, providing novel associations that require further exploration (Fig 6I). The analysis also indicated that the observed decrease in T cell response and TCR signaling in severe COVID-19 may not be directly influenced by the AUC, but rather by factors such as interleukin response, MyD88:MAL signaling, TLR signaling, and innate immune response, which are regulated by specific genes and transcripts. In addition, the adaptive immune response, B/T cell response, and translational regulation pathways were found to be independent of the AUC and regulated by other genes and transcripts. The direct role of the NDRG2 gene in immune response remains unclear, but previous studies have suggested its involvement in NF-B inhibition and its potential relevance to severe COVID-19 (Fig S3E) (75). Further research is needed to fully understand the contribution of NDRG2 and other identified genes and transcripts in modulating the immune response in COVID-19.

The patient stratification method in this study, based on the SpO2/FiO2 ratio sheds light on the transcriptomic distinctions among COVID-19 ARDS patients with varying levels of SpO2. However, clinical utility of such a stratification method is not validated in an external cohort, presenting a potential limitation in the current study. Further investigations with validation cohorts are warranted to potentially overcome/strengthen the classification method used.

In conclusion, our study highlights the importance of transcript-level analysis in capturing specific expression patterns and the association of reduced transcript diversity with disease severity (Fig 7). Gene and transcript-based investigations of the immune response provide valuable immunological insights, surpassing the limitations of plasma cytokine quantification. The Bayesian network analysis served to identify potential regulators of immune response and revealed AUC (indicative of disease severity) as a predictor of COVID-19 outcome in ARDS patients. Finally, the comprehensive analysis of gene and transcript expression in severe COVID-19 patients provides valuable insights into disease pathogenesis and offers potential for improved clinical treatment strategies in the ICU.

Figure 7.
Figure 7. Threading together the comprehensive gene- and transcript-level inferences in ICU-admitted COVID-19 patients at three time points.

The dynamic longitudinal analysis emphasized the significance of transcript-level inferences in complementing our understanding of the disease progression and the interplay between gene expression, cytokine levels, and immune cell activity.

Materials and Methods

Sample collection and classification

The sample was collected from RT–PCR positive COVID-19 patients admitted to the Infectious Diseases & Beleghata General Hospital, Kolkata, India, with mild to moderate ARDS, recruited during a previously published clinical trial on CPT (Clinical Trial Registry of India No. CTRI/2020/05/025209) between May 31, 2020, to October 12, 2020 (4, 48). The enrolled patients were aged ≥ 18, had severe COVID-19 as per the Indian Council of Medical Research guidelines (fever or suspected respiratory infection, plus one of the following; respiratory rate >30 breaths/min, severe respiratory distress, SpO2 <90% at room air) with either mild ARDS (SpO2/FiO2 ratio between 200–300) or moderate ARDS (SpO2/FiO2 ratio between 100–200), without requiring mechanical ventilation and were within 5–10 d from initial onset of symptoms. Pregnant and breastfeeding mothers, and individuals predisposed to clinical conditions were excluded. The trial recruited 40 individuals in each arm: one which received only standard-of-care therapy as per the advisory in 2020 and CPT which received additional two consecutive doses of ABO-matched 200 ml convalescent plasma on two consecutive days, the first transfusion being on the day of enrolment, in addition to standard of care (48) (Fig S1A). At least 5 ml of blood was collected in BD Vacutainer EDTA tubes at three time points, viz. on the day of enrolment (T1), on the third or fourth day after enrollment (T2) and 7 d post enrollment (T3). Furthermore, plasma was isolated by centrifuging the blood at 2,000g for 10 min at 4°C. The study was designed in accordance with the Declaration of Helsinki and was approved by the Institutional Ethics Committee of CSIR-Institute of Genomics and Integrative Biology (CSIR-IGIB), Delhi, India (Ref No: CSIR-IGIB/IHEC/2020 21/01), CSIR-Indian Institute of Chemical Biology, Kolkata, India (IICB/IRB/2020/3P), and Infectious Diseases and Beleghata General Hospital (ID and BG Hospital), Kolkata, India (IDBGH/Ethics/2429). In addition, records of the medications administered during the COVID-19 treatment for each patient were curated (Table S1). This encompassed a range of drugs, including remdesivir, and corticosteroids like dexamethasone and hydrocortisone. However, IL-6 receptor blockers and baricitinib, which are recommended in WHO guidelines, were not administered because of unavailability during the study period (76).

Evaluation of SpO2 in blood

To assess blood oxygenation levels over a span of 7 d after enrollment, we calculated the AUC for the ratio of SpO2 in capillary blood to the FiO2, denoted as the S/F ratio or SFR. This metric, referred to as SFR7dAUC, provided a comprehensive view of blood oxygenation trends during this 7-d period. In instances where data were not recorded on either the first or seventh day (in the event of patient survival), we imputed values from the nearest available day, giving precedence to the lower value when two such values were accessible. The resulting SFR7dAUC values were computed for each patient and used for correlational analyses with other parameters. In case of missing data between the first and seventh days (the mortality patients), the closest available value was used and replicated when calculating the SFR ratio 7-d kinetics curve. This metric was demonstrated as an effective predictor of disease severity in our cohort, with median value exhibiting robust stratification capabilities as detailed in the prior study by (4). Thus, based on the AUC values, patients were categorized into two groups. Patients with an AUC ≤ 771.7 were categorized as Low AUC and had median SpO2/FiO2 ratios between 90–150 mmHg, whereas those with an AUC > 771.7 were designated as High AUC with median SpO2/FiO2 ratios between 100–350 mmHg representing distinct levels of oxygenation as surrogate for disease severity.

All individuals in the High AUC group made a full recovery, whereas three out of the nine patients in the low AUC group fatal outcomes to the SARS-CoV-2 infection (Table S1). Importantly, despite all patients having severe COVID-19, the SpO2 curves exhibited considerable variability, ranging from 100 to 2,000. Notably, the low AUC group demonstrated poorer outcomes, with a higher mortality rate compared with the High AUC.

Library preparation and sequencing

We used Illumina TruSeq Stranded Total RNA Library Prep Gold (cat. no 20020599) to prepare the sequencing libraries from 250 ng of total RNA extracted from the PBMC collected from patients at the three time points, as per the manufacturer’s reference guide (1000000040499 v00). In summary, cytoplasmic and mitochondrial rRNA have been depleted using target-specific biotinylated oligos and Ribo-Zero rRNA removal beads, and pure RNA was fragmented using a divalent cation at a high temperature. Random primers and SuperScript IV reverse transcriptase were used to create the first strand cDNA from the fragmented RNA. After RNaseH-mediated degradation of the RNA strand from the previous phase, DNA polymerase 1 was used to synthesize the second strand of cDNA. Adenylation of the double-stranded cDNA’s 3′ blunt end was done before indexing and amplification. AMPure XP bead with sample to bead ratio of 1:1 (A63881; Beckman Coulter) was used for cleaning the final library. The library quality was checked using Agilent 2100 bioanalyzer followed by sequencing on NovaSeq 6000 using NovaSeq S2 v1.5 reagents with a 2 × 101 read length and 450 pM loading concentration.

Differential gene and transcript expression analysis

The raw sequencing reads were quality checked using FastQC and trimmed with Trimmomatic v.0.39 to remove low-quality bases (77). The reads were then reassessed with FastQC to confirm quality improvements (78). The filtered reads were then quantified against the human reference genome (GRCh38.106 primary assembly from Ensembl) using Salmon (v.1.8.0) (79). For differential gene expression analysis, the Salmon-quantified reads were analyzed using the DESeq2 package (80). For differential transcript expression analysis--numGibbsSamples parameter with 20 inferential replicates was used in Salmon to generate bootstrap abundance estimates for each sample using posterior Gibbs sampling to estimate the variability in transcript abundance. The quantification files generated by Salmon were imported to the R environment using the tximport for differential transcript-level expression analysis of RNA-seq using inferential replicate counts with Swish method in Bioconductor package fishpond (v.2.0.1) (81, 82). Differential transcript and gene-level analysis was performed for different time points (T3 versus T1, T2 versus T3 and T2 versus T1) and within the time point between different AUC groups of different severity (T3 Low versus High AUC, T2 Low versus High and T1 Low versus High). The Benjamini–Hochberg correction was used to correct for multiple comparisons (with an FDR cut-off <0.05). DGEs with Padj value < 0.05 and log2foldchange (FC) |1.5| were considered as significant, whereas for DTE Padj value < 0.05 and log2FC|1| was considered significant. A lower log fold change threshold for transcript-level analysis was selected to ensure capturing of more subtle differences in expression of individual transcripts, especially for detecting isoform-level changes.

Plasma cytokine quantification

The plasma samples were obtained from the peripheral blood of the patients and collected in EDTA vials. Cytokine levels (measured in pg/ml) were determined using the Bio-Plex Pro Human Cytokine Screening Panel, 48-Plex Assay from Bio-Rad. This assay allows for the quantification of 48 different cytokines, including FGF basic, eotaxin, G-CSF, GM-CSF, IFN-γ, IL-1β, IL-1ra, IL-1α, IL-2Rα, IL-3, IL-12 (p40), IL-16, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, GRO-α, hepatocyte growth factor, IFN-α2, leukemia inhibitory factor, monocyte chemo-attractant protein-3 (MCP-3), IL-10, IL-12 (p70), IL-13, IL-15, IL-17A, IL-18, IFN-γ–inducible protein-10 (IP-10), MCP-1, monokine induced by IFN-γ, NGF-β, stem cell factor, stem cell growth factor-β, stromal cell-derived factor-1α, macrophage-inflammatory protein-1α, macrophage-inflammatory protein-1β, PDGF-BB, regulated upon activation normal T-cell expressed and secreted, tumor necrosis factor-α, tumor necrosis factor-β, VEGF, cutaneous T cell-attracting chemokine, MIF, TNF-related apoptosis-inducing ligand, and M-CSF. To perform the assay, the patient plasma samples were diluted 1:3 in sample diluent. The assay was carried out following the instructions provided by the manufacturer. The Bio-Plex 200 System from Bio-Rad was used to run and analyze the plate.

Cell type quantification

After plasma collection, the whole blood sample was treated with RBC lysis buffer. The leukocyte pellet thus obtained was subsequently fixed using 1% paraformaldehyde and stained using fluorochrome tagged antibodies (BD Biosciences) to analyze the different immune cell subsets based on surface marker expression. The stained samples were then acquired in a FACSAria III flow cytometer. Further analysis was performed on FlowJo Software to obtain the cell type quantification data.

Bayesian network analysis

The data from RNA expression for both differentially expressed transcripts/genes, cytokines, and cell type abundance were integrated with AUC, outcome, and time of sampling. The integrative modeling analysis was conducted using the wiseR package, which uses end-to-end Bayesian network learning and inference techniques (83). To facilitate biological interpretation, all continuous variables within the integrated dataset were discretized into three quartile ranges: low, middle, and high. A discrete Bayesian network was then learned from the data using hill climbing optimization to identify the directed acyclic graph that represents the structural dependencies between the variables. Subsequently, the structure of the network was parameterized using Monte Carlo Markov Chain approximate inference method to determine the conditional probability distributions.

Gene set enrichment analysis

Function enrichment of DGEs and DTEs was performed using the fgsea R package against the reactome database using c2.cp.reactome.v2023.1.Hs.symbols.gmt (84 Preprint). Pathways with statistically significant P-value cutoff <0.05 and with at least three genes were considered. The pathways were plotted using the ggplot2 R package against the combined score and number of genes involved in the pathways. Redundant pathways were excluded (85).

Differential splicing analysis

The initial step involved aligning the pre-processed fastq files against the reference human genome (GRCh38.106) using the splice-aware aligner STAR (v2.7.11) (86). Subsequently, the resulting bam files were transformed into junction files using regtools (87 Preprint). After this, intron clustering and analysis of differential splicing were conducted using Leafcutter (v.0.2.9), with allowed intron sizes set between 50–500 kb (88). The Ensembl human reference (GRCh38.106 primary assembly from Ensembl) was employed for annotation. The groups for differential splicing analysis were structured in the same manner as those used for the comparison in the differential gene/transcript level analysis. The outcomes were then visualized using Leafviz, and a report was generated using RStudio (https://github.com/jackhump/leafviz) (Table S4).

Statistical analysis and data visualization

The immune cell type and cytokine expression levels between all the time points and AUC groups were compared using the nonparametric two-way mixed-effect ANOVA model with Tukey’s correction to account for unbalanced/missing data and account for individual-level variability in the dataset (89). Wherever appropriate, we compared the differences between data points using the two-tailed Mann–Whitney U test, and Chi-square testing. The statistical tests were performed using a licensed version of GraphPad Prism. The ggbio (v.1.44.1) (90), GenomicFeatures (v.1.48.3) (91), ggplot2 (v.3.3.3) (85), EnhancedVolcano (v.1.14.0) (92) Gviz (v.1.40.1) (93) R packages were used for data visualization. The P-value < 0.05 was considered statistically significant.

Data Availability

The datasets presented in this study can be found online at the NCBI-SRA under the BioProject accession numbers PRJNA816679.

Ethics statement

The study was designed in accordance with the Declaration of Helsinki and was approved by the Institutional Ethics Committee of CSIR-Institute of Genomics and Integrative Biology, Delhi, India (Ref No: CSIR-IGIB/IHEC/2020-21/01), CSIR-Indian Institute of Chemical Biology, Kolkata, India (IICB/IRB/2020/3P), and Infectious Diseases and Beleghata General Hospital (ID and BG Hospital), Kolkata, India (IDBGH/Ethics/2429). The patients/participants provided their written informed consent before participation in this study.

Acknowledgements

The authors duly acknowledge all the COVID-19 patients who participated in the study. Authors acknowledge the help and support from Dr. Aradhita Baral and Dr Bharti Kumari towards facilitation as research manager and coordination with the funders. Authors acknowledge the support of Anil Kumar and Nisha Rawat towards COVID-19 sample transport and sample management. P Chattopadhyay acknowledges the CSIR for his Research Fellowship. This research was funded by Bill and Melinda Gates Foundation, grant numbers INV-033578 to R Pandey and MLP-129 from Council of Scientific and Industrial Research, India, to D Ganguly.

Author Contributions

  • P Mehta: data curation, software, formal analysis, investigation, visualization, methodology, and writing—original draft.

  • P Chattopadhyay: data curation, investigation, writing—original draft and sequencing.

  • R Mohite: investigation and writing—review and editing.

  • R D’Rozario: resources and data curation.

  • P Bandopadhyay: resources and data curation.

  • J Sarif: resources and data curation.

  • Y Ray: resources and data curation.

  • D Ganguly: resources, data curation, supervision, and writing—review and editing.

  • R Pandey: conceptualization, supervision, funding acquisition, investigation, visualization, project administration, and writing—review and editing.

Conflict of Interest Statement

The authors declare that they have no conflict of interest.

  • Received August 3, 2023.
  • Revision received October 21, 2023.
  • Accepted October 23, 2023.
Creative Commons logoCreative Commons logohttps://creativecommons.org/licenses/by/4.0/

This article is available under a Creative Commons License (Attribution 4.0 International, as described at https://creativecommons.org/licenses/by/4.0/).

References

  1. 1.↵
    1. Li X,
    2. Ma X
    (2020) Acute respiratory failure in COVID-19: Is it “typical” ARDS? Crit Care 24: 198. doi:10.1186/s13054-020-02911-9
    OpenUrlCrossRefPubMed
  2. 2.↵
    1. Bellani G,
    2. Laffey JG,
    3. Pham T,
    4. Fan E,
    5. Brochard L,
    6. Esteban A,
    7. Gattinoni L,
    8. van Haren F,
    9. Larsson A,
    10. McAuley DF, et al.
    (2016) Epidemiology, patterns of care, and mortality for patients with acute respiratory distress syndrome in intensive care units in 50 countries. JAMA 315: 788–800. doi:10.1001/jama.2016.0291
    OpenUrlCrossRefPubMed
  3. 3.↵
    1. Srinivasan S,
    2. Panigrahy AK
    (2021) COVID-19 ARDS: Can systemic oxygenation utilization guide oxygen therapy? Indian J Crit Care Med 25: 115–116. doi:10.5005/jp-journals-10071-23740
    OpenUrlCrossRef
  4. 4.↵
    1. Raychaudhuri D,
    2. Bandopadhyay P,
    3. D’Rozario R,
    4. Sarif J,
    5. Ray Y,
    6. Paul SR,
    7. Singh P,
    8. Chaudhuri K,
    9. Bhaduri R,
    10. Pandey R, et al.
    (2022) Clinical trial subgroup analyses to investigate clinical and immunological outcomes of convalescent plasma therapy in severe COVID-19. Mayo Clin Proc Innov Qual Outcomes 6: 511–524. doi:10.1016/j.mayocpiqo.2022.09.001
    OpenUrlCrossRef
  5. 5.↵
    1. Daamen AR,
    2. Bachali P,
    3. Owen KA,
    4. Kingsmore KM,
    5. Hubbard EL,
    6. Labonte AC,
    7. Robl R,
    8. Shrotri S,
    9. Grammer AC,
    10. Lipsky PE
    (2021) Comprehensive transcriptomic analysis of COVID-19 blood, lung, and airway. Sci Rep 11: 7052. doi:10.1038/s41598-021-86002-x
    OpenUrlCrossRefPubMed
  6. 6.↵
    1. Wargodsky R,
    2. Dela Cruz P,
    3. LaFleur J,
    4. Yamane D,
    5. Kim JS,
    6. Benjenk I,
    7. Heinz E,
    8. Irondi OO,
    9. Farrar K,
    10. Toma I, et al.
    (2022) RNA Sequencing in COVID-19 patients identifies neutrophil activation biomarkers as a promising diagnostic platform for infections. PLoS One 17: e0261679. doi:10.1371/journal.pone.0261679
    OpenUrlCrossRef
  7. 7.↵
    1. Kale P,
    2. Gupta E,
    3. Bihari C,
    4. Patel N,
    5. Rooge S,
    6. Pandey A,
    7. Bajpai M,
    8. Khillan V,
    9. Chattopadhyay P,
    10. Devi P, et al.
    (2021) Vaccine breakthrough infections by SARS-CoV-2 variants after ChAdOx1 nCoV-19 vaccination in healthcare workers. Vaccines (Basel) 10: 54. doi:10.3390/vaccines10010054
    OpenUrlCrossRef
  8. 8.↵
    1. Maurya R,
    2. Shamim U,
    3. Mishra P,
    4. Swaminathan A,
    5. Raina A,
    6. Tarai B,
    7. Budhiraja S,
    8. Pandey R
    (2023) Intertwined dysregulation of ribosomal proteins and immune response delineates SARS-CoV-2 vaccination breakthroughs. Microbiol Spectr 11: e0429222. doi:10.1128/spectrum.04292-22
    OpenUrlCrossRef
  9. 9.↵
    1. Wang C,
    2. Chen L,
    3. Chen Y,
    4. Jia W,
    5. Cai X,
    6. Liu Y,
    7. Ji F,
    8. Xiong P,
    9. Liang A,
    10. Liu R, et al.
    (2022) Abnormal global alternative RNA splicing in COVID-19 patients. PLoS Genet 18: e1010137. doi:10.1371/journal.pgen.1010137
    OpenUrlCrossRef
  10. 10.↵
    1. Hangauer MJ,
    2. Vaughn IW,
    3. McManus MT
    (2013) Pervasive transcription of the human genome produces thousands of previously unidentified long intergenic noncoding RNAs. PLoS Genet 9: e1003569. doi:10.1371/journal.pgen.1003569
    OpenUrlCrossRefPubMed
  11. 11.↵
    1. Zeng Z,
    2. Yu H,
    3. Chen H,
    4. Qi W,
    5. Chen L,
    6. Chen G,
    7. Yan W,
    8. Chen T,
    9. Ning Q,
    10. Han M, et al.
    (2020) Longitudinal changes of inflammatory parameters and their correlation with disease severity and outcomes in patients with COVID-19 from Wuhan, China. Crit Care 24: 525. doi:10.1186/s13054-020-03255-0
    OpenUrlCrossRef
  12. 12.↵
    1. Sarif J,
    2. Raychaudhuri D,
    3. D’Rozario R,
    4. Bandopadhyay P,
    5. Singh P,
    6. Mehta P,
    7. Hoque MA,
    8. Sinha BP,
    9. Kushwaha M,
    10. Sahni S, et al.
    (2021) Plasma gradient of soluble urokinase-type plasminogen activator receptor is linked to pathogenic plasma proteome and immune transcriptome and stratifies outcomes in severe COVID-19. Front Immunol 12: 738093. doi:10.3389/fimmu.2021.738093
    OpenUrlCrossRef
  13. 13.↵
    1. Alvarez F,
    2. Istomine R,
    3. Da Silva Lira Filho A,
    4. Al-Aubodah T-A,
    5. Huang D,
    6. Okde R,
    7. Olivier M,
    8. Fritz JH,
    9. Piccirillo CA
    (2023) IL-18 is required for the TH1-adaptation of TREG cells and the selective suppression of TH17 responses in acute and chronic infections. Mucosal Immunol 16: 462–475. doi:10.1016/j.mucimm.2023.05.004
    OpenUrlCrossRef
  14. 14.↵
    1. Shojaei M,
    2. Shamshirian A,
    3. Monkman J,
    4. Grice L,
    5. Tran M,
    6. Tan CW,
    7. Teo SM,
    8. Rodrigues Rossi G,
    9. McCulloch TR,
    10. Nalos M, et al.
    (2022) IFI27 transcription is an early predictor for COVID-19 outcomes, a multi-cohort observational study. Front Immunol 13: 1060438. doi:10.3389/fimmu.2022.1060438
    OpenUrlCrossRef
  15. 15.↵
    1. Maleki KT,
    2. Cornillet M,
    3. Björkström NK
    (2016) Soluble SEMA4D/CD100: A novel immunoregulator in infectious and inflammatory diseases. Clin Immunol 163: 52–59. doi:10.1016/j.clim.2015.12.012
    OpenUrlCrossRefPubMed
  16. 16.↵
    1. Clancy RM,
    2. Halushka M,
    3. Rasmussen SE,
    4. Lhakhang T,
    5. Chang M,
    6. Buyon JP
    (2019) Siglec-1 macrophages and the contribution of IFN to the development of autoimmune congenital heart block. J Immunol 202: 48–55. doi:10.4049/jimmunol.1800357
    OpenUrlAbstract/FREE Full Text
  17. 17.↵
    1. Dutta AK,
    2. Goswami K
    (2021) Host genomics of COVID-19: Evidence point towards Alpha 1 antitrypsin deficiency as a putative risk factor for higher mortality rate. Med Hypotheses 147: 110485. doi:10.1016/j.mehy.2021.110485
    OpenUrlCrossRef
  18. 18.↵
    1. Khiem D,
    2. Cyster JG,
    3. Schwarz JJ,
    4. Black BL
    (2008) A p38 MAPK-MEF2C pathway regulates B-cell proliferation. Proc Natl Acad Sci U S A 105: 17067–17072. doi:10.1073/pnas.0804868105
    OpenUrlAbstract/FREE Full Text
  19. 19.↵
    1. Zheng W,
    2. Xu Q,
    3. Zhang Y,
    4. E X,
    5. Gao W,
    6. Zhang M,
    7. Zhai W,
    8. Rajkumar RS,
    9. Liu Z
    (2020) Toll-like receptor-mediated innate immunity against herpesviridae infection: A current perspective on viral infection signaling pathways. Virol J 17: 192. doi:10.1186/s12985-020-01463-2
    OpenUrlCrossRef
  20. 20.↵
    1. Shapira G,
    2. Shomron N,
    3. Gurwitz D
    (2020) Ethnic differences in alpha-1 antitrypsin deficiency allele frequencies may partially explain national differences in COVID-19 fatality rates. FASEB J 34: 14160–14165. doi:10.1096/fj.202002097
    OpenUrlCrossRef
  21. 21.↵
    1. Doehn J-M,
    2. Tabeling C,
    3. Biesen R,
    4. Saccomanno J,
    5. Madlung E,
    6. Pappe E,
    7. Gabriel F,
    8. Kurth F,
    9. Meisel C,
    10. Corman VM, et al.
    (2021) CD169/SIGLEC1 is expressed on circulating monocytes in COVID-19 and expression levels are associated with disease severity. Infection 49: 757–762. doi:10.1007/s15010-021-01606-9
    OpenUrlCrossRefPubMed
  22. 22.↵
    1. Rosendal E,
    2. Mihai IS,
    3. Becker M,
    4. Das D,
    5. Frängsmyr L,
    6. Persson BD,
    7. Rankin GD,
    8. Gröning R,
    9. Trygg J,
    10. Forsell M, et al.
    (2022) Serine protease inhibitors restrict host susceptibility to SARS-CoV-2 infections. MBio 13: e0089222. doi:10.1128/mbio.00892-22
    OpenUrlCrossRef
  23. 23.↵
    1. Abboud G,
    2. Stanfield J,
    3. Tahiliani V,
    4. Desai P,
    5. Hutchinson TE,
    6. Lorentsen KJ,
    7. Cho JJ,
    8. Avram D,
    9. Salek-Ardakani S
    (2016) Transcription factor Bcl11b controls effector and memory CD8 T cell fate decision and function during poxvirus infection. Front Immunol 7: 425. doi:10.3389/fimmu.2016.00425
    OpenUrlCrossRef
  24. 24.↵
    1. Whelan S,
    2. Ophir E,
    3. Kotturi MF,
    4. Levy O,
    5. Ganguly S,
    6. Leung L,
    7. Vaknin I,
    8. Kumar S,
    9. Dassa L,
    10. Hansen K, et al.
    (2019) PVRIG and PVRL2 are induced in cancer and inhibit CD8+ T-cell function. Cancer Immunol Res 7: 257–268. doi:10.1158/2326-6066.CIR-18-0442
    OpenUrlAbstract/FREE Full Text
  25. 25.↵
    1. Wada T
    (2018) Downregulation of CD5 and dysregulated CD8+ T-cell activation. Pediatr Int 60: 776–780. doi:10.1111/ped.13636
    OpenUrlCrossRef
  26. 26.↵
    1. Birnbaum DJ,
    2. Picard M,
    3. Da Costa Q,
    4. Delayre T,
    5. Finetti P,
    6. Cabaud O,
    7. Agavnian E,
    8. De Rauglaudre B,
    9. Denicolaï E,
    10. Bertucci F, et al.
    (2023) PVRIG expression is an independent prognostic factor and a new potential target for immunotherapy in hepatocellular carcinoma. Cancers (Basel) 15: 447. doi:10.3390/cancers15020447
    OpenUrlCrossRef
  27. 27.↵
    1. Wang J,
    2. Jiang M,
    3. Chen X,
    4. Montaner LJ
    (2020) Cytokine storm and leukocyte changes in mild versus severe SARS-CoV-2 infection: Review of 3939 COVID-19 patients in China and emerging pathogenesis and therapy concepts. J Leukoc Biol 108: 17–41. doi:10.1002/JLB.3COVR0520-272R
    OpenUrlCrossRefPubMed
  28. 28.↵
    1. Sekine T,
    2. Perez-Potti A,
    3. Rivera-Ballesteros O,
    4. Strålin K,
    5. Gorin J-B,
    6. Olsson A,
    7. Llewellyn-Lacey S,
    8. Kamal H,
    9. Bogdanovic G,
    10. Muschiol S, et al.
    (2020) Robust T cell immunity in convalescent individuals with asymptomatic or mild COVID-19. Cell 183: 158–168.e14. doi:10.1016/j.cell.2020.08.017
    OpenUrlCrossRefPubMed
  29. 29.↵
    1. Chen G,
    2. Wu D,
    3. Guo W,
    4. Cao Y,
    5. Huang D,
    6. Wang H,
    7. Wang T,
    8. Zhang X,
    9. Chen H,
    10. Yu H, et al.
    (2020) Clinical and immunological features of severe and moderate coronavirus disease 2019. J Clin Invest 130: 2620–2629. doi:10.1172/JCI137244
    OpenUrlCrossRefPubMed
  30. 30.↵
    1. Gassen NC,
    2. Papies J,
    3. Bajaj T,
    4. Emanuel J,
    5. Dethloff F,
    6. Chua RL,
    7. Trimpert J,
    8. Heinemann N,
    9. Niemeyer C,
    10. Weege F, et al.
    (2021) SARS-CoV-2-mediated dysregulation of metabolism and autophagy uncovers host-targeting antivirals. Nat Commun 12: 3818. doi:10.1038/s41467-021-24007-w
    OpenUrlCrossRef
  31. 31.↵
    1. Muralidharan A,
    2. Wyatt TA,
    3. Reid SP
    (2022) SARS-CoV-2 dysregulates neutrophil degranulation and reduces lymphocyte counts. Biomedicines 10: 382. doi:10.3390/biomedicines10020382
    OpenUrlCrossRef
  32. 32.↵
    1. Coleman MJ,
    2. Zimmerly KM,
    3. Yang XO
    (2021) Accumulation of CD28null senescent T-cells is associated with poorer outcomes in COVID19 patients. Biomolecules 11: 1425. doi:10.3390/biom11101425
    OpenUrlCrossRef
  33. 33.↵
    1. Yi L,
    2. Pimentel H,
    3. Bray NL,
    4. Pachter L
    (2018) Gene-level differential analysis at transcript-level resolution. Genome Biol 19: 53. doi:10.1186/s13059-018-1419-z
    OpenUrlCrossRef
  34. 34.↵
    1. Bleilevens C,
    2. Soppert J,
    3. Hoffmann A,
    4. Breuer T,
    5. Bernhagen J,
    6. Martin L,
    7. Stiehler L,
    8. Marx G,
    9. Dreher M,
    10. Stoppe C, et al.
    (2021) Macrophage migration inhibitory factor (MIF) plasma concentration in critically ill COVID-19 patients: A prospective observational study. Diagnostics (Basel) 11: 332. doi:10.3390/diagnostics11020332
    OpenUrlCrossRef
  35. 35.↵
    1. Vecchié A,
    2. Bonaventura A,
    3. Toldo S,
    4. Dagna L,
    5. Dinarello CA,
    6. Abbate A
    (2021) IL-18 and infections: Is there a role for targeted therapies? J Cell Physiol 236: 1638–1657. doi:10.1002/jcp.30008
    OpenUrlCrossRef
  36. 36.↵
    1. Vaz de Paula CB,
    2. de Azevedo MLV,
    3. Nagashima S,
    4. Martins APC,
    5. Malaquias MAS,
    6. Miggiolaro AFRDS,
    7. da Silva Motta Júnior J,
    8. Avelino G,
    9. do Carmo LAP,
    10. Carstens LB, et al.
    (2020) IL-4/IL-13 remodeling pathway of COVID-19 lung injury. Sci Rep 10: 18689. doi:10.1038/s41598-020-75659-5
    OpenUrlCrossRef
  37. 37.↵
    1. Sasson J,
    2. Moreau GB,
    3. Petri WA
    (2023) The role of interleukin 13 and the type 2 immune pathway in COVID-19: A review. Ann Allergy Asthma Immunol 130: 727–732. doi:10.1016/j.anai.2023.03.009
    OpenUrlCrossRef
  38. 38.↵
    1. Morawetz RA,
    2. Gabriele L,
    3. Rizzo LV,
    4. Noben-Trauth N,
    5. Kühn R,
    6. Rajewsky K,
    7. Müller W,
    8. Doherty TM,
    9. Finkelman F,
    10. Coffman RL, et al.
    (1996) Interleukin (IL)-4-independent immunoglobulin class switch to immunoglobulin (Ig)E in the mouse. J Exp Med 184: 1651–1661. doi:10.1084/jem.184.5.1651
    OpenUrlAbstract/FREE Full Text
  39. 39.↵
    1. Van der Pouw Kraan TC,
    2. Van der Zee JS,
    3. Boeije LC,
    4. De Groot ER,
    5. Stapel SO,
    6. Aarden LA
    (1998) The role of IL-13 in IgE synthesis by allergic asthma patients. Clin Exp Immunol 111: 129–135. doi:10.1046/j.1365-2249.1998.00471.x
    OpenUrlCrossRefPubMed
  40. 40.↵
    1. Aziz M,
    2. Fatima R,
    3. Assaly R
    (2020) Elevated interleukin-6 and severe COVID-19: A meta-analysis. J Med Virol 92: 2283–2285. doi:10.1002/jmv.25948
    OpenUrlCrossRefPubMed
  41. 41.↵
    1. Belletti A,
    2. Campochiaro C,
    3. Marmiere M,
    4. Likhvantsev V,
    5. Yavorovskiy A,
    6. Dagna L,
    7. Landoni G,
    8. Zangrillo A,
    9. Hajjar LA
    (2021) Efficacy and safety of IL-6 inhibitors in patients with COVID-19 pneumonia: A systematic review and meta-analysis of multicentre, randomized trials. Ann Intensive Care 11: 152. doi:10.1186/s13613-021-00941-2
    OpenUrlCrossRef
  42. 42.↵
    1. D’Rozario R,
    2. Raychaudhuri D,
    3. Bandopadhyay P,
    4. Sarif J,
    5. Mehta P,
    6. Liu CSC,
    7. Sinha BP,
    8. Roy J,
    9. Bhaduri R,
    10. Das M, et al.
    (2023) Circulating interleukin-8 dynamics parallels disease course and is linked to clinical outcomes in severe COVID-19. Viruses 15: 549. doi:10.3390/v15020549
    OpenUrlCrossRef
  43. 43.↵
    1. Li L,
    2. Li J,
    3. Gao M,
    4. Fan H,
    5. Wang Y,
    6. Xu X,
    7. Chen C,
    8. Liu J,
    9. Kim J,
    10. Aliyari R, et al.
    (2020) Interleukin-8 as a biomarker for disease prognosis of coronavirus disease-2019 patients. Front Immunol 11: 602395. doi:10.3389/fimmu.2020.602395
    OpenUrlCrossRef
  44. 44.↵
    1. Venet M,
    2. Ribeiro MS,
    3. Décembre E,
    4. Bellomo A,
    5. Joshi G,
    6. Nuovo C,
    7. Villard M,
    8. Cluet D,
    9. Perret M,
    10. Pescamona R, et al.
    (2023) Severe COVID-19 patients have impaired plasmacytoid dendritic cell-mediated control of SARS-CoV-2. Nat Commun 14: 694. doi:10.1038/s41467-023-36140-9
    OpenUrlCrossRef
  45. 45.↵
    1. Tay MZ,
    2. Rouers A,
    3. Fong S-W,
    4. Goh YS,
    5. Chan Y-H,
    6. Chang ZW,
    7. Xu W,
    8. Tan CW,
    9. Chia WN,
    10. Torres-Ruesta A, et al.
    (2022) Decreased memory B cell frequencies in COVID-19 delta variant vaccine breakthrough infection. EMBO Mol Med 14: e15227. doi:10.15252/emmm.202115227
    OpenUrlCrossRef
  46. 46.↵
    1. Yang P,
    2. Wang J,
    3. Zheng R,
    4. Tan R,
    5. Li X,
    6. Liu X,
    7. Li Y,
    8. Yuan Z,
    9. Wang Y,
    10. Chen Q, et al.
    (2022) Convalescent plasma may not be an effective treatment for severe and critically ill covid-19 patients: A systematic review & meta-analysis of randomized controlled trials. Heart Lung 53: 51–60. doi:10.1016/j.hrtlng.2022.01.019
    OpenUrlCrossRef
  47. 47.↵
    1. Klassen SA,
    2. Senefeld JW,
    3. Senese KA,
    4. Johnson PW,
    5. Wiggins CC,
    6. Baker SE,
    7. van Helmond N,
    8. Bruno KA,
    9. Pirofski L-A,
    10. Shoham S, et al.
    (2021) Convalescent plasma therapy for COVID-19: A graphical mosaic of the worldwide evidence. Front Med (Lausanne) 8: 684151. doi:10.3389/fmed.2021.684151
    OpenUrlCrossRef
  48. 48.↵
    1. Ray Y,
    2. Paul SR,
    3. Bandopadhyay P,
    4. D’Rozario R,
    5. Sarif J,
    6. Raychaudhuri D,
    7. Bhowmik D,
    8. Lahiri A,
    9. Vasudevan JS,
    10. Maurya R, et al.
    (2022) A phase 2 single center open label randomised control trial for convalescent plasma therapy in patients with severe COVID-19. Nat Commun 13: 383. doi:10.1038/s41467-022-28064-7
    OpenUrlCrossRef
  49. 49.↵
    1. Agarwal A,
    2. Mukherjee A,
    3. Kumar G,
    4. Chatterjee P,
    5. Bhatnagar T,
    6. Malhotra P, PLACID Trial Collaborators
    , (2020) Convalescent plasma in the management of moderate covid-19 in adults in India: Open label phase II multicentre randomised controlled trial (PLACID trial). BMJ 371: m3939. doi:10.1136/bmj.m3939
    OpenUrlAbstract/FREE Full Text
  50. 50.↵
    1. Vijayakumar A,
    2. Park A,
    3. Steitz JA
    (2022) Modulation of mRNA 3'-end processing and transcription termination in virus-infected cells. Front Immunol 13: 828665. doi:10.3389/fimmu.2022.828665
    OpenUrlCrossRef
  51. 51.↵
    1. Mehta P,
    2. Chattopadhyay P,
    3. Ravi V,
    4. Tarai B,
    5. Budhiraja S,
    6. Pandey R
    (2023) SARS-CoV-2 infection severity and mortality is modulated by repeat-mediated regulation of alternative splicing. Microbiol Spectr 11: e0135123. doi:10.1128/spectrum.01351-23
    OpenUrlCrossRef
  52. 52.↵
    1. Soneson C,
    2. Love MI,
    3. Robinson MD
    (2015) Differential analyses for RNA-seq: Transcript-level estimates improve gene-level inferences. F1000Res 4: 1521. doi:10.12688/f1000research.7563.2
    OpenUrlCrossRef
  53. 53.↵
    1. Grimes JM,
    2. Grimes KV
    (2020) p38 MAPK inhibition: A promising therapeutic approach for COVID-19. J Mol Cell Cardiol 144: 63–65. doi:10.1016/j.yjmcc.2020.05.007
    OpenUrlCrossRefPubMed
  54. 54.↵
    1. Attiq A,
    2. Yao LJ,
    3. Afzal S,
    4. Khan MA
    (2021) The triumvirate of NF-κB, inflammation and cytokine storm in COVID-19. Int Immunopharmacol 101: 108255. doi:10.1016/j.intimp.2021.108255
    OpenUrlCrossRef
  55. 55.↵
    1. Farahani M,
    2. Niknam Z,
    3. Mohammadi Amirabad L,
    4. Amiri-Dashatan N,
    5. Koushki M,
    6. Nemati M,
    7. Danesh Pouya F,
    8. Rezaei-Tavirani M,
    9. Rasmi Y,
    10. Tayebi L
    (2022) Molecular pathways involved in COVID-19 and potential pathway-based therapeutic targets. Biomed Pharmacother 145: 112420. doi:10.1016/j.biopha.2021.112420
    OpenUrlCrossRef
  56. 56.↵
    1. Li J,
    2. Rong L,
    3. Cui R,
    4. Feng J,
    5. Jin Y,
    6. Chen X,
    7. Xu R
    (2021) Dynamic changes in serum IL-6, IL-8, and IL-10 predict the outcome of ICU patients with severe COVID-19. Ann Palliat Med 10: 3706–3714. doi:10.21037/apm-20-2134
    OpenUrlCrossRefPubMed
  57. 57.↵
    1. Diao B,
    2. Wang C,
    3. Tan Y,
    4. Chen X,
    5. Liu Y,
    6. Ning L,
    7. Chen L,
    8. Li M,
    9. Liu Y,
    10. Wang G, et al.
    (2020) Reduction and functional exhaustion of T cells in patients with coronavirus disease 2019 (COVID-19). Front Immunol 11: 827. doi:10.3389/fimmu.2020.00827
    OpenUrlCrossRefPubMed
  58. 58.↵
    1. Fenwick C,
    2. Joo V,
    3. Jacquier P,
    4. Noto A,
    5. Banga R,
    6. Perreau M,
    7. Pantaleo G
    (2019) T-cell exhaustion in HIV infection. Immunol Rev 292: 149–163. doi:10.1111/imr.12823
    OpenUrlCrossRefPubMed
  59. 59.↵
    1. Di Nicola M,
    2. Cattaneo A,
    3. Hepgul N,
    4. Di Forti M,
    5. Aitchison KJ,
    6. Janiri L,
    7. Murray RM,
    8. Dazzan P,
    9. Pariante CM,
    10. Mondelli V
    (2013) Serum and gene expression profile of cytokines in first-episode psychosis. Brain Behav Immun 31: 90–95. doi:10.1016/j.bbi.2012.06.010
    OpenUrlCrossRefPubMed
  60. 60.↵
    1. Roozeman J-P,
    2. Mazzinari G,
    3. Serpa Neto A,
    4. Hollmann MW,
    5. Paulus F,
    6. Schultz MJ,
    7. Pisani L, PRoVENT-COVID Collaborative Group
    , (2022) Prognostication using SpO2/FiO2 in invasively ventilated ICU patients with ARDS due to COVID-19 - insights from the PRoVENT-COVID study. J Crit Care 68: 31–37. doi:10.1016/j.jcrc.2021.11.009
    OpenUrlCrossRef
  61. 61.↵
    1. Annane D
    (2021) Corticosteroids for COVID-19. J Intensive Med 1: 14–25. doi:10.1016/j.jointm.2021.01.002
    OpenUrlCrossRef
  62. 62.↵
    1. Cupps TR,
    2. Gerrard TL,
    3. Falkoff RJ,
    4. Whalen G,
    5. Fauci AS
    (1985) Effects of in vitro corticosteroids on B cell activation, proliferation, and differentiation. J Clin Invest 75: 754–761. doi:10.1172/JCI111757
    OpenUrlCrossRefPubMed
  63. 63.↵
    1. Kovacs WJ
    (2014) To B or not to B? Glucocorticoid impact on B lymphocyte fate and function. Endocrinology 155: 339–342. doi:10.1210/en.2013-2085
    OpenUrlCrossRef
  64. 64.↵
    1. Saxon A,
    2. Stevens RH,
    3. Ramer SJ,
    4. Clements PJ,
    5. Yu DT
    (1978) Glucocorticoids administered in vivo inhibit human suppressor T lymphocyte function and diminish B lymphocyte responsiveness in in vitro immunoglobulin synthesis. J Clin Invest 61: 922–930. doi:10.1172/JCI109017
    OpenUrlCrossRefPubMed
  65. 65.↵
    1. Feng L-Y,
    2. Li L
    (2020) Low expression of NCALD is associated with chemotherapy resistance and poor prognosis in epithelial ovarian cancer. J Ovarian Res 13: 35. doi:10.1186/s13048-020-00635-6
    OpenUrlCrossRef
  66. 66.↵
    1. Monaco G,
    2. Lee B,
    3. Xu W,
    4. Mustafah S,
    5. Hwang YY,
    6. Carré C,
    7. Burdin N,
    8. Visan L,
    9. Ceccarelli M,
    10. Poidinger M, et al.
    (2019) RNA-seq signatures normalized by mRNA abundance allow absolute deconvolution of human immune cell types. Cell Rep 26: 1627–1640.e7. doi:10.1016/j.celrep.2019.01.041
    OpenUrlCrossRefPubMed
  67. 67.↵
    1. Ding J-M,
    2. Lin W-R,
    3. Fei Z-D,
    4. Chen C-B
    (2022) PYHIN1 correlates with CD8+ T cells infiltration and confers good patient survival in oral cancer. J Dent Sci 17: 551–559. doi:10.1016/j.jds.2021.06.014
    OpenUrlCrossRef
  68. 68.↵
    1. Imbratta C,
    2. Hussein H,
    3. Andris F,
    4. Verdeil G
    (2020) c-MAF, a Swiss army knife for tolerance in lymphocytes. Front Immunol 11: 206. doi:10.3389/fimmu.2020.00206
    OpenUrlCrossRef
  69. 69.↵
    1. Michée-Cospolite M,
    2. Boudigou M,
    3. Grasseau A,
    4. Simon Q,
    5. Mignen O,
    6. Pers J-O,
    7. Cornec D,
    8. Le Pottier L,
    9. Hillion S
    (2022) Molecular mechanisms driving IL-10- producing B cells functions: STAT3 and c-MAF as underestimated central key regulators? Front Immunol 13: 818814. doi:10.3389/fimmu.2022.818814
    OpenUrlCrossRef
  70. 70.↵
    1. Goplen NP,
    2. Saxena V,
    3. Knudson KM,
    4. Schrum AG,
    5. Gil D,
    6. Daniels MA,
    7. Zamoyska R,
    8. Teixeiro E
    (2016) IL-12 signals through the TCR to support CD8 innate immune responses. J Immunol 197: 2434–2443. doi:10.4049/jimmunol.1600037
    OpenUrlAbstract/FREE Full Text
  71. 71.↵
    1. Linsley PS,
    2. Brady W,
    3. Grosmaire L,
    4. Aruffo A,
    5. Damle NK,
    6. Ledbetter JA
    (1991) Binding of the B cell activation antigen B7 to CD28 costimulates T cell proliferation and interleukin 2 mRNA accumulation. J Exp Med 173: 721–730. doi:10.1084/jem.173.3.721
    OpenUrlAbstract/FREE Full Text
  72. 72.↵
    1. Elgueta R,
    2. Benson MJ,
    3. de Vries VC,
    4. Wasiuk A,
    5. Guo Y,
    6. Noelle RJ
    (2009) Molecular mechanism and function of CD40/CD40L engagement in the immune system. Immunol Rev 229: 152–172. doi:10.1111/j.1600-065X.2009.00782.x
    OpenUrlCrossRefPubMed
  73. 73.↵
    1. Ho W-HJ,
    2. Law AMK,
    3. Masle-Farquhar E,
    4. Castillo LE,
    5. Mawson A,
    6. O’Bryan MK,
    7. Goodnow CC,
    8. Gallego-Ortega D,
    9. Oakes SR,
    10. Ormandy CJ
    (2022) Activation of the viral sensor oligoadenylate synthetase 2 (Oas2) prevents pregnancy-driven mammary cancer metastases. Breast Cancer Res 24: 31. doi:10.1186/s13058-022-01525-z
    OpenUrlCrossRef
  74. 74.↵
    1. Decking S-M,
    2. Bruss C,
    3. Babl N,
    4. Bittner S,
    5. Klobuch S,
    6. Thomas S,
    7. Feuerer M,
    8. Hoffmann P,
    9. Dettmer K,
    10. Oefner PJ, et al.
    (2022) LDHB overexpression can partially overcome T cell inhibition by lactic acid. Int J Mol Sci 23: 5970. doi:10.3390/ijms23115970
    OpenUrlCrossRef
  75. 75.↵
    1. Lee KW,
    2. Lim S,
    3. Kim KD
    (2022) The function of N-myc downstream-regulated gene 2 (NDRG2) as a negative regulator in tumor cell metastasis. Int J Mol Sci 23: 9365. doi:10.3390/ijms23169365
    OpenUrlCrossRef
  76. 76.↵
    1. Lamontagne F,
    2. Agarwal A,
    3. Rochwerg B,
    4. Siemieniuk RA,
    5. Agoritsas T,
    6. Askie L,
    7. Lytvyn L,
    8. Leo Y-S,
    9. Macdonald H,
    10. Zeng L, et al.
    (2020) A living WHO guideline on drugs for covid-19. BMJ 370: m3379. doi:10.1136/bmj.m3379
    OpenUrlAbstract/FREE Full Text
  77. 77.↵
    1. Bolger AM,
    2. Lohse M,
    3. Usadel B
    (2014) Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics 30: 2114–2120. doi:10.1093/bioinformatics/btu170
    OpenUrlCrossRefPubMed
  78. 78.↵
    (n.d.) Babraham Bioinformatics - FastQC A Quality Control Tool for High Throughput Sequence Data. https://www.bioinformatics.babraham.ac.uk/projects/fastqc/
  79. 79.↵
    1. Patro R,
    2. Duggal G,
    3. Love MI,
    4. Irizarry RA,
    5. Kingsford C
    (2017) Salmon provides fast and bias-aware quantification of transcript expression. Nat Methods 14: 417–419. doi:10.1038/nmeth.4197
    OpenUrlCrossRefPubMed
  80. 80.↵
    1. Love MI,
    2. Huber W,
    3. Anders S
    (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15: 550. doi:10.1186/s13059-014-0550-8
    OpenUrlCrossRefPubMed
  81. 81.↵
    1. Soneson C,
    2. Love MI,
    3. Robinson MD
    (2015) Differential analyses for RNA-Seq: Transcript-level estimates improve gene-level inferences. F1000Res 4: 1521. doi:10.12688/f1000research.7563.1
    OpenUrlCrossRefPubMed
  82. 82.↵
    1. Zhu A,
    2. Srivastava A,
    3. Ibrahim JG,
    4. Patro R,
    5. Love MI
    (2019) Nonparametric expression analysis using inferential replicate counts. Nucleic Acids Res 47: e105. doi:10.1093/nar/gkz622
    OpenUrlCrossRefPubMed
  83. 83.↵
    1. GitHub - SAFE-ICU/wiseR
    (n.d.) A R/Shiny Application for Learning, Reasoning, Inference, Validation, Visualization and Web-Deployment of Bayesian Network Models. https://github.com/SAFE-ICU/wiseR
  84. 84.↵
    1. Korotkevich G,
    2. Sukhov V,
    3. Budin N,
    4. Shpak B,
    5. Artyomov MN,
    6. Sergushichev A
    (2016) Fast gene set enrichment analysis. BioRxiv doi:10.1101/060012(Preprint posted February 01, 2021).
    OpenUrlAbstract/FREE Full Text
  85. 85.↵
    1. Wickham H
    (2016) ggplot2 - Elegant Graphics for Data Analysis. New York, NY: Springer-Verlag New York.
  86. 86.↵
    1. Dobin A,
    2. Davis CA,
    3. Schlesinger F,
    4. Drenkow J,
    5. Zaleski C,
    6. Jha S,
    7. Batut P,
    8. Chaisson M,
    9. Gingeras TR
    (2013) STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 29: 15–21. doi:10.1093/bioinformatics/bts635
    OpenUrlCrossRefPubMed
  87. 87.↵
    1. Cotto KC,
    2. Feng Y-Y,
    3. Ramu A,
    4. Skidmore ZL,
    5. Kunisaki J,
    6. Conrad DF,
    7. Lin Y,
    8. Chapman W,
    9. Uppaulri R,
    10. Govindan R, et al.
    (2023) RegTools: Integrated analysis of genomic and transcriptomic data for discovery of splicing variants in cancer. BioRxiv doi:10.1101/436634(Preprint posted April 08, 2021).
    OpenUrlCrossRef
  88. 88.↵
    1. Li YI,
    2. Knowles DA,
    3. Humphrey J,
    4. Barbeira AN,
    5. Dickinson SP,
    6. Im HK,
    7. Pritchard JK
    (2018) Annotation-free quantification of RNA splicing using LeafCutter. Nat Genet 50: 151–158. doi:10.1038/s41588-017-0004-9
    OpenUrlCrossRefPubMed
  89. 89.↵
    1. Gueorguieva R,
    2. Krystal JH
    (2004) Move over ANOVA: Progress in analyzing repeated-measures data and its reflection in papers published in the archives of general psychiatry. Arch Gen Psychiatry 61: 310–317. doi:10.1001/archpsyc.61.3.310
    OpenUrlCrossRefPubMed
  90. 90.↵
    1. Yin T,
    2. Cook D,
    3. Lawrence M
    (2012) ggbio: an R package for extending the grammar of graphics for genomic data. Genome Biol 13: R77. doi:10.1186/gb-2012-13-8-r77
    OpenUrlCrossRefPubMed
  91. 91.↵
    1. Lawrence M,
    2. Huber W,
    3. Pagès H,
    4. Aboyoun P,
    5. Carlson M,
    6. Gentleman R,
    7. Morgan MT,
    8. Carey VJ
    (2013) Software for computing and annotating genomic ranges. PLoS Comput Biol 9: e1003118. doi:10.1371/journal.pcbi.1003118
    OpenUrlCrossRefPubMed
  92. 92.↵
    EnhancedVolcano: Publication-ready volcano plots with enhanced colouring and labeling. Available at: https://bioconductor.org/packages/devel/bioc/vignettes/EnhancedVolcano/inst/doc/EnhancedVolcano.html.
  93. 93.↵
    1. Hahne F,
    2. Ivanek R
    (2016) Visualizing genomic data using gviz and bioconductor. Methods Mol Biol 1418: 335–351. doi:10.1007/978-1-4939-3578-9_16
    OpenUrlCrossRefPubMed
Back to top
Download PDF
Email Article
Citation Tools
Transcription dynamics during Severe COVID-19 ARDS
Priyanka Mehta, Partha Chattopadhyay, Ramakant Mohite, Ranit D’Rozario, Purbita Bandopadhyay, Jafar Sarif, Yogiraj Ray, Dipyaman Ganguly, Rajesh Pandey
Life Science Alliance Nov 2023, 7 (1) e202302305; DOI: 10.26508/lsa.202302305
Twitter logo Facebook logo Mendeley logo
Issue Cover

In this Issue

Volume 7, No. 1
January 2024
Advertisement

Jump to section

Subjects

Related Articles

  • No related articles found.

Cited By...

  • No citing articles found.

More in this TOC Section

Similar Articles