Immune inhibitory receptor-mediated immune response, metabolic adaptation, and clinical characterization in COVID-19 patients

doi:10.21203/rs.3.rs-3316286/v1

Download PDF

Article

Immune inhibitory receptor-mediated immune response, metabolic adaptation, and clinical characterization in COVID-19 patients

https://doi.org/10.21203/rs.3.rs-3316286/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 06 Nov, 2023

Read the published version in Scientific Reports →

You are reading this latest preprint version

Immune inhibitory receptors (IRs) have been demonstrated to play a critical role in the regulation of immune response to various respiratory viral infection. However, in COVID-19, the potential roles of the integrated effect of these IRs in immune modulation, metabolic reprogramming, and clinical characterization remains to be determined. Through the consensus clustering analysis of IR transcription in the peripheral blood of COVID-19 patients, we identified two distinct IR patterns in COVID-19 patients. And we demonstrated that IR_cluster2 patients characterized by lower expression of most IRs presented suppressed immune response, lower nutrient metabolism, and worse clinical manifestations or prognosis. To quantify and assess the IR patterns of individual COVID-19 patients, we established a scoring system named IRscore based on principal component analysis algorithms. Similar to IR_cluster2 patients, patients with high IRscore had a longer hospital-free days at day 45, required ICU admission and mechanical ventilatory support, and presented higher Charlson comorbidity index score and SOFA score. Moreover, high IRscore was also linked to high viral load, acute infection phase, and absence of drug intervention. Our investigation comprehensively elucidates the potential role of IR patterns in regulating immune response, modulating metabolic processes, and shaping clinical manifestations for COVID-19. All these evidences suggest the essential role of prognostic stratification and biomarker screening based on IR patterns in the clinical management and drug development of future emerging infectious diseases such as COVID-19.

immune inhibitory receptors

immunomodulation

metabolic adaptation

clinical characterization

COVID-19

In 2019, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) caused coronavirus disease 2019 (COVID-19) and rapidly escalated to the twenty-first century’s deadliest pandemic. At present, the number of infected cases has exceeded 755 million, while more than 6.8 million people have died from this pandemic [1]. Clinical symptoms of SARS-CoV-2 infection range from asymptomatic infection to life-threatening disease. Approximately 5% patients develop into severe disease that requires intensive care with mechanical ventilation [2, 3]. Epidemiological studies attempt to identify the risk factors for severe disease. Immune imbalance, such as dysregulated innate immune responses, early immunosuppression, lymphopenia, and cytokine storm, has been considered as one of the causes of life-threatening COVID-19 [4].

Homeostasis is key for biological systems to keep health. When SARS-CoV-2 viruses disturb the homeostasis of our physiological system, immune system needs to rapidly recognize this disturbance and restore homeostasis as quickly as possible. Otherwise, organism will develop immunopathology. In COVID-19 patients, delayed early antiviral immune response leads to the uncontrolled virus replication. The persistent viral exposure and antigen presentation might result in the excessive production of cytokines and complement, remarkably decreased immune cell numbers and exhaustion of host cellular immune responses [5]. This is deleterious for COVID-19 patients for the increasing risk of developing immunopathology.

Therefore, our immune system has evolved an extensive immune regulatory network tightly regulated by both the activation and inhibitory immune receptors (IRs). The physiological role of IRs is primarily to maintain immune homeostasis and tolerance [6, 7]. More importantly, IRs can fine-tune immune responses in a highly specific manner to prevent the occurrence of immunopathology [8]. Increasing studies show that IRs are indispensable modulators of immune system during the development of COVID-19. Many studies have discovered that programmed cell death-1 (PD-1), cytotoxic T lymphocyte-associated antigen (CTLA-4), and killer cell immunoglobulin like receptor 2DL1 (KIR2DL1) are upregulated during the process of developing severe COVID-19 disease [9–12]. Another study suggests the expressions of T-cell immunoglobulin mucin-3 (TIM-3, also named as HAVCR2) and lymphocyte-activation gene-3 (LAG3) in more severe COVID-19 patients are much higher than that in mild patients. Whereas, both inhibitory receptors rapidly decline to normal levels during convalescence [13]. In addition, upregulation of TIM-3 can also induce depletion of NKT cells, which is related to disease severity and prognosis of COVID-19 patients. Compared with healthy individuals, TIM-3⁺ NKT cells in COVID-19 patients present high levels of co-inhibitory receptors such as PD-1, CTLA4, and LAG3 [14]. To clarify the expression profile of more IRs during SARS-CoV-2 infection, Narjes et al. performed bioinformatics analyses to determine the expression levels of specific IRs in different types of samples, and identified eight upregulated IRs in both nasopharyngeal swabs and autopsies, and determined their correlations with viral loads [15]. These researches indicate that the elevated expression of IR in severe COVID-19 patients might contribute to the exhaustion of effector immune cells and the impairment of their specific immune activity. However, some studies reported that CD8⁺ T cells with high expression of PD-1 and CTLA-4 are not exhausted but functional [12, 16]. Some researchers have also proposed that these IRs are upregulated in effector T cells is to protect vital organs from the excessive inflammatory environment triggered by SARS-CoV-2 infection [17], which suggests the immune activation-induced negative feedback mechanism of IRs. Therefore, the mechanism by which IR regulates the immune system to control infection in COVID-19 remains controversial.

Although the above findings provide the clue of the correlation between dysregulated IRs and severity of COVID-19, most of these studies primarily focus on only one or two IRs and cell types. However, severe disease is characterized by immune disorder tuned by the integrated effect of multiple IRs. Therefore, a comprehensive cognition of the role of IR regulatory network in modulating immune response and shaping clinical manifestations will enhance our understanding of COVID-19 progression and provide avenues for discovering effective therapeutic targets.

In this study, we utilized the transcriptome sequencing data of COVID-19 peripheral blood mononuclear cell (PBMC) samples to comprehensively evaluate the correlation between IR expression patterns and patients’ prognosis, immune and metabolic characteristics. We also established a set of scoring system (IRscore) to quantify the IR expression pattern in individual COVID-19 patients. Finally, we verified the reliability and stability of IRscore in predicting the disease state, clinical characteristics, and immunotherapeutic efficacy of COVID-19.

2.1. The source of COVID-19 dataset

The datasets analyzed during the current study are available in the GEO repository, GSE157103 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE157103), GSE198256 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE198256), GSE152418 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE152418), and GSE163317 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE163317). The basic information of these datasets was summarized in Table S1.

To assess the stability of IRscore to predict the disease state of COVID-19, we included GSE198256 and GSE152418 datasets. There were 11 healthy controls, 7 acute infection patients, 16 convalescents in GSE198256. Monocytes isolated from PBMCs of these patients were sequenced by NovaSeq 6000 [19]. The normalized counts were downloaded. In GSE152418, RNA-seq analysis of PBMCs from 17 COVID-19 subjects and 17 healthy controls was included. Among COVID-19 patients, 1 were convalescent patient, 4 were moderate patients, and 12 were severe patients [20]. The counts were downloaded.

To determine the relationship between IRscore and therapeutic effects, we collected GSE163317 dataset with treatment information. GSE163317 included RNA-seq data from whole blood samples of 4 severe COVID-19 patients before and 7 days after anakinra (IL-1 receptor blocker) administration [21].

2.2 Principal component analysis

The principal component analysis (PCA) was performed based on transcriptome expression to distinguish distinct IR patterns. PCA was grouped according to IR stratification.

2.3. Unsupervised clustering for 42 immune inhibitory receptors

We utilized createDataPartition function in “caret” R package to split samples in GSE157103 dataset into training cohort with 75 samples and test sets with 25 samples [22]. In an attempt to balance the sample distributions within the splits, we conducted the random sampling within the levels of HFD-45 outcomes. The training cohort with 75 COVID-19 patients was utilized for model training and the test cohort with 25 samples was applied to evaluate model performance.

There are more than 300 potential inhibitory receptor genes in human genome. And only more than 60 of them have been functionally demonstrated over the past few decades [7, 8]. We checked the available literature on functionally characterized inhibitor receptor-ligand pairs and selected 42 immune inhibitory receptors (IRs) for identifying distinct IR patterns. We applied unsupervised clustering analysis based on the expression of 42 IRs to classify COVID-19 patients in training cohort for further analysis. The number of clusters and the stability of grouping were determined by the consensus clustering algorithm performed through the “ConsensusClusterPlus” R package [23]. For guaranteeing the stability of classification, we assigned the number of subsamples as 1000.

2.4. Gene set variation analysis (GSVA) and functional annotation

To identify the discrepancy on biological process between IR patterns, we executed GSVA enrichment analysis through the “GSVA” R package. GSVA is commonly applied to estimate the variations in signaling pathway and biological process over a population in a non-parametric and unsupervised method [24]. The gene sets of “c2.cp.kegg.v7.1.symbols” and “c5.go.v7.5.1.symbols” were downloaded from MSigDB database for GSVA analysis (http://www.gsea-msigdb.org/gsea/downloads.jsp). The pathway with an adjusted P value less than 0.01 was considered statistically significant. The “clusterProfiler” R package was utilized to annotate the function of IR-related differentially expressed genes (DEGs), with 0.01 as the FDR cutoff value [25].

2.5 Estimation of immune cell fractions

To quantify the proportions of innate and adaptive immune cells in COVID-19 samples, we used CIBERSORT algorithm to estimate the abundances of member cell types in a mixed cell population (https://cibersort.stanford.edu). CIBERSORT is a deconvolution algorithm that uses support vector regression and a set of reference gene-expression values representing several human immune cell phenotypes to infer member cell proportions from gene expression data of blood samples with mixed cell types. The LM22 signature gene matrix and RNA-seq data of samples were used as the inputs [26].

2.6 Metabolic expression subtype classification

According to the method of metabolic expression subtype classification developed by Xinxin Peng et al. [27]. The gene sets of seven metabolic pathways were based on the latest Reactome annotations [28]. The seven metabolic pathways were lipid metabolism (766 genes), amino acid metabolism (348 genes), carbohydrate metabolism (286 genes), vitamin cofactor metabolism (168 genes), TCA cycle (148 genes), integration of energy (110 genes), and nucleotide metabolism (90 genes). We firstly normalized gene expression across samples by Z score to obtain a rank value for each gene (~ 19000 coding genes) within each sample. Then, we conducted gene set enrichment analysis (GSEA) based on the gene set of a specific metabolic pathway. And we classified samples into three subtypes according the resulting rank values. The classification criterion is: (1) samples with the high Z scores for a specific metabolic pathway were defined as “upregulated subtype” (FDR < 0.05); (2) samples with the low Z scores for a specific metabolic pathway were defined as “downregulated subtype” (FDR < 0.05); and (3) samples showing no significant enrichment pattern were defined as “neutral subtype”. Given the metabolic expression subtypes, we evaluated expression alteration of protein-coding genes by Kruskal-Wallis test and then used Kolmogorov-Smirnov test to determine whether the p values of the metabolic pathway genes were lower than those from other genes (FDR < 0.05). Among seven metabolic pathways, the p value of energy metabolism was not significant. Through the above analysis, each sample was labeled with six kinds of metabolic expression subtypes. Differences in the proportions of each metabolic expression subtype between the two IR clusters were statistically analyzed by the chi-squared test.

2.7 Generation of IR_score

To quantify the IR patterns of individual COVID-19 patients, we established a scoring system and termed as IR_score. The model constructed in the training cohort was validated in the test cohort. The procedures for constructing this model were as follows:

Acquisition of significant DEGs. Firstly, the analysis of RNA-Seq counts of the training cohort was conducted through “DESeq” package to detect differentially expressed genes between the two IR patterns [29]. The DEGs identified from distinct IR patterns were extracted to classify patients. The unsupervised consensus clustering algorithm was used to determine the number of gene clusters as well as the stability of clustering. Secondly, we divided COVID-19 patients according to the days of no hospitalization in 45 days, those with hospitalization days more than 26 days were named as HFD-45 > 26, and the rest were named as the HFD-45 < = 26. The DEGs between the two groups were acquired using the “DESeq” package. Thirdly, we identified DEGs between COVID-19 and non-COVID-19 patients. DEGs in each of the above three comparison groups were determined by significance criteria, which retained the DEGs with adjusted P values less than 0.05 and log2 fold change more than 1 for the following analyses. The overlap analyses were conducted to extract overlapping DEGs from the above three differential expression analysis results (Figure S4M). Finally, the Random Forest algorithm was adopted to perform dimension reduction to reduce noise and redundant genes. The Random Forest classification algorithm was implemented through R package Boruta [30].

Construction of IR_score. The expression of each gene was first transformed into a Z score. Then we performed PCA to construct IR relevant gene signature. Both principal component 1 and 2, which reflected the main characteristics of IR and prognosis related DEGs, were extracted as signature scores. We then applied a method similar to gene expression grade index (GGI) to define the IR_score of each patient [31]:

IR_score = Σ(PC1_i + PC2_i)

where i is the IR phenotype-related genes.

2.8 Lipopolysaccharide-induced inflammatory model

Lipopolysaccharide (LPS)-induced inflammatory model was established as previous report described [32]. 8 weeks old male C57BL/6J mice weighing 22–24 g were purchased from Charles River Laboratories. All mice were maintained in pathogen-free cages at 24\(\text{℃}\) with 12-hour light-dark cycle. We randomly divided the mice into two groups: the control group (N = 6) and the LPS group (6 mg/kg, N = 14). LPS (Escherichia coli O55:B5, Sigma, St.Louis, MO) was dissolved in sterile saline and given via intraperitoneal. The control group received sterile saline. Mice were weighed 24 hours after LPS administration. The method of mice euthanasia was cervical dislocation. Then the lung and spleen were collected. The whole lungs of mice were weighed. The ratio of lung weight to body weight was used to determine the success of the LPS-induced inflammation model. The lung tissues were embedded in paraffin and subjected to hematoxylin-eosin (H&E) staining. No animal anesthesia was involved in any of the experiments. No animal anesthesia was involved in any of the experiments. All mice experiments were conducted according to the standard protocols approved by the animal ethics committee at the Institute of Health Service and Transfusion Medicine (application number: IACUC-DWZX-2023-P657). This study is reported in accordance with ARRIVE guidelines (https://arriveguidelines.org).

2.9 Spleen lymphocyte isolation

The spleens were processed by mechanical disruption on a 40 \(\mu\)M mesh filter. The saline with spleen cells suspended was transferred to 3 mL lymphocyte separation medium (7211011, Dakewe, China). Centrifuge 800 g at room temperature for 30 min, and set a slower lifting and descending speed. After centrifugation, collect the lymphocyte layer and wash the cells with 10 mL saline. The cells were collected by centrifugation at 250 g for 10 min at room temperature.

Total RNA of the isolated spleen lymphocytes was extracted using TRIzol reagent (15596026, Invitrogen, Thermo Fisher Scientific, USA). 2 mg of total RNA was reversely transcribed by All-in-one RT MasterMix (G592, abm, Canada) as the manufacturer instructed. Real-time polymerase chain reaction (real-time PCR) was performed using UltraSYBR Mixture (CW0957S, CWBIO, China) on LightCycler96 (Roche, Switzerland). The real-time PCR primers were listed in supplementary Table S3.

3.1 Hematoxylin-eosin staining

The lung tissues in control and LPS group at 24h post-LPS administration were fixed in 4% formaldehyde. The hematoxylin-eosin (H&E) staining was performed according to the methods reported in previous study [33]. The pathological photos were taken by the microscope (Olympus, Japan).

3.2 Statistical Analysis

Correlation coefficients between the CIBERSORT member cells and expression of IR genes were calculated by Spearman. Mann-Whitney and Student’s t test were applied to conduct difference comparisons between two groups. One-way analysis of variance (ANOVA) with Dunnett's multiple comparison test was utilized to compare differences among three or more groups of data subject to normal distribution. Heatmap of antiviral innate immune responses was constructed on DEGs between two IR patterns belonging to the GO:0140374 annotations (http://geneontology.org, Table S4). Heatmap of major histocompatibility complex was constructed on DEGs between two IR patterns according to a publication by Zhang et al. [34]. In light of the correlation between IRscore and mechanical ventilator support state, the optimal cut point of dataset subgroup was determined by surv_cutpoint function in “survminer” R package. This is an outcome-oriented methods providing a value of a cutpoint that correspond to the most significant relation with outcome. Therefore, we implemented this algorithm to dichotomize patients into low and high IRscore groups based on the maximally selected rank statistics. We adopted a multivariable Cox regression model to ascertain the independent prognostic factors for evaluating COVID-19 patients’ outcomes. The “forestplot” R package was employed to visualize the results of multivariate Cox regression analysis for IRscore in training cohort. The performance (sensitivity and specificity) of IRscore on predicting the COVID-19 patients’ mechanical ventilator support state was evaluated by receiver operating characteristic (ROC) curves, which were plotted through the area under the curve (AUC) scores calculated by “survivalROC”. All statistical analyses were conducted by R (https://www.r-project.org, R version 4.0.4) or SPSS software (version 26.0.0.0). The two-sided P values less than 0.05 were considered as statistically significant.

3.1. The expression level of 42 immune inhibitory receptors was associated with the characteristics of immune cell abundance and disease state in COVID-19 patients

We reviewed the available literature on functionally characterized inhibitor receptor-ligand pairs and selected 42 immune inhibitory receptors (IRs) for identifying distinct IR patterns. We first conducted correlation analysis to identify the interaction patterns of these IRs in the GSE157103. As shown in Fig. 1A, most IRs displayed synergistic expression, whereas LILRB3 exhibited an antagonistic action with most of IRs. Referring to the distribution of inhibitory receptors on immune cells summarized by Narjes et al. [15], we surprisingly found that IRs with strong correlations (coefficient > 0.8) were primarily expressed on the same type of immune cells (Table 1). Specifically, TIGIT, CD244, and CD160 with high correlation coefficients were all distributed on NK cells and T cells (the correlation coefficient of TIGIT and CD244 was 0.883, the correlation coefficient of TIGIT and CD160 was 0.872, the correlation coefficient of CD244 and CD160 was 0.844, Fig. 1A). We also estimated the abundances of member cell types in blood lymphocytes of COVID-19 patients through the CIBERSORT and xCell method and determined the association between IRs and the abundance of immune cells. CD8⁺ T cells were positively correlated with the expression levels of CD160, CD200R1, CD22, CD244, CD5, CD72, CTLA4, KLRB1, KLRC1, KLRG1, LAG3, PDCD1, SIGLEC8, and TIGIT, whereas its abundances showed negative correlation with LILRB3 (Fig. 1B and 1C). Then we utilized a univariate Cox regression model to calculate the hazard ratio of each IR in the peripheral blood mononuclear cells (PBMCs) of COVID-19 patients. The forest plot displayed the 42 IR prognostic values for ventilator-free intervals in COVID-19 patients (Fig. 1D), indicating the possible association between the IR expression levels and disease state. The above results indicated that crosstalk among 42 IRs might play considerable roles in the formation of different IR patterns, the characterization of PBMC immune cell abundance or composition, and the prediction of disease states in individual patients.

Table 1

list of immune inhibitory receptors with correlation coefficients more than 0.800 and their predominant distribution on immune cells
Gene1	Gene2	Correlation coefficient	Predominant Cell Distribution¹
HAVCR2	CD33	0.800	monocytes, macrophages
LILRB1	HAVCR2	0.816	monocytes, macrophages, NK cells
LILRB4	LILRB1	0.824	monocytes, macrophages, neutrophils, DCs
TIGIT	KLRB1	0.827	NK cells
KLRC1	KLRB1	0.827	NK cells
KLRB1	CD5	0.830	not on the same cells
KLRG1	CD244	0.831	NK cells
KLRB1	CD160	0.843	NK cells
CD244	CD160	0.844	NK cells, T cells
TIGIT	KLRG1	0.850	NK cells
KLRG1	CD160	0.856	NK cells
LILRB1	CD33	0.860	monocytes, macrophages
TIGIT	CD160	0.872	NK cells, T cells
TIGIT	CD5	0.880	T cells
TIGIT	CD244	0.883	NK cells, T cells
CD72	CD22	0.909	B cells

¹ The predominant cell distribution was taken from a publication by Narjes et al [15].

3.2. IR Patterns mediated by 42 immune inhibitory receptors and clinical characteristics of each pattern

Regarding the crosstalk of 42 IRs in the PBMCs of COVID-19 patients, we performed consensus clustering to classify patients into the appropriate subgroups with different IR patterns based on the expression levels of 42 IRs. Two distinct patterns were identified, with 43 cases in cluster 1 and 32 cases in cluster 2. We termed these patterns as IR_cluster1 and IR_cluster2, respectively. The cluster heatmaps displayed the stability of the two clusters (Figure S1A-S1C). Principal component analysis (PCA) revealed a clear distinction between the two patterns (Figure S1D). A unique IR transcriptional profile was observed between the two IR patterns. IR_cluster2 displayed significantly high expressions of CEACAM1 and LILRB3, while IR_cluster1 was characterized by markedly elevated expression of most other IRs (Fig. 2A).

Then we investigated the differences in disease state between the two IR patterns. As shown in Fig. 2B, the average hospital-free days at day 45 (HFD-45) of patients in IR_cluster2 was significantly shorter than that of patients in IR_cluster1, suggesting a worse prognosis for IR_cluster2. Consistent with the poor disease state of IR_cluster2, more patients in IR_cluster2 required the intensive care unit (ICU) admission (20.9% versus 81.3%) and mechanical ventilatory support (20.9% versus 75.0%, Fig. 2H-2I). And 28.1% patients in IR_cluster2 had Charlson Comorbidity Index (CCI) greater than 6, whereas only 7.0% in IR_cluster1 (Fig. 2J), indicating the worse overall burden of comorbidities in IR_cluster2. Moreover, the higher levels of biomarkers for the diagnosis and monitoring of COVID-19 [such as D-dimer, C-reactive protein (CRP), lactate (LDH), ferritin] suggested the severe disease in IR_cluster2 patients (Fig. 2C-2F). The high Sequential Organ Failure Assessment (SOFA) scores for IR_cluster2 patients also indicated their more severe organ dysfunction or failure (Fig. 2G). We also observed that IR_cluster2 was predominated by male (female vs male, 55.8% vs 65.6%) and elderly patients (Figure S1E-S1F). All the evidences mentioned above suggested the essential role of IR expression patterns to predict the prognosis and severity of COVID-19 patients, and revealed a prominent survival disadvantage in IR_cluster2.

3.3. Immunological features in distinct IR patterns of COVID-19 patients

To explore the biological functions of these distinct IR patterns, we performed GSVA enrichment analysis. As expected, the significant discrepant enrichments primarily focused on immune signaling pathways. Specifically, IR_cluster1 presented significant enrichment in pathways associated with immune activation such as T cell receptor signaling pathway, B cell receptor signaling pathway, natural killer cell mediated cytotoxicity, antigen processing and presentation, chemokine signaling pathway and so on (Fig. 3A). The immune system is broadly divided into two arms, the innate immune system and adaptive immune system. During viral infection, the innate immune system rapidly recognizes the infection and triggers the “alarm bells” of type I interferon (IFN) signaling pathway and induces the production of downstream antiviral proteins and other molecules [35]. We analyzed the expression of antiviral innate immune response signaling pathway genes between different IR patterns, and observed a significantly diminished antiviral innate immune response in IR_cluster2 patients (Fig. 3B). In addition, we also discovered the remarkable decreases of major histocompatibility complex (MHC)-related genes in IR_cluster2 patients (Fig. 3C). Huang et al. have documented that most patients of severe COVID-19 displayed enhanced levels of proinflammatory cytokines [36]. Therefore, we evaluated the expression levels of these cytokines between the two IR patterns. As shown in Figure S1G, S100A12, a marker of inflammation in severe sepsis, was dramatically augmented in IR_cluster2 patients, further indicating that IR_cluster2 patients might develop pulmonary injury and endure severe disease burden.

Although innate immune and adaptive immune are linked in multitudinous important ways, they each comprise distinct cell types with specific functions. B cells, CD4⁺ T cells, and CD8⁺ T cells are three major adaptive immune cell types to control and clear the viral infections. Both CIBERSORT and xCell immune cell component analyses displayed patients of IR_cluster2 had lower levels of memory CD4⁺ T cells and CD8⁺ T cells (Fig. 3D and 3E), suggesting the prominently suppressed T cell dependent adaptive immune responses in IR_cluster2. All the results mentioned above indicated the lower levels of innate and adaptive immune response in IR_cluster2 patients, who might be more vulnerable to infection and disease progression. Therefore, the differences in infection control for IR pattern stratification patients might be attributed to the IR pattern mediated discrepancies in landscape of immune response. Meanwhile, COVID-19 patients with lower expression of most IRs presented absence of CD8⁺ T cells and worse prognosis, indicating that IR might shape the clinical characteristics of COVID-19 through immune activation-induced negative feedback mechanism.

3.4. Immunometabolic adaptation characteristics in distinct IR patterns of COVID-19 patients

Besides immune signaling pathways, we also noticed specific metabolic pathway enrichment discrepancies between two IR patterns. To further determine the metabolic differences between distinct IR patterns, we utilized a computational method to classify individual patients into “directional” lipid, amino acid, carbohydrate, vitamin cofactor, TCA cycle, and nucleotide subtypes based on the expression pattern of metabolic pathway genes [27]. As shown in Fig. 4A, fewer patients in IR_cluster2 exhibited upregulated TCA cycle (38.1% versus 6.3%) and amino acid metabolism (41.9% versus 9.7%) compared to IR_cluster1, while more patients in IR_cluster2 presented downregulated nucleotide metabolism (2.3% versus 18.8%). These results suggested that IR_cluster2 might be characterized by downregulated energy production, amino acid metabolism, and nucleotide metabolism compared to IR_cluster1.

A large number of studies have reported that the precise control of cellular metabolic processes was necessary for fueling the fate decision and effector function of immune cells [37]. Therefore, we attempted to examine the crosstalk between immune cell abundances and cellular metabolism through spearman’s correlation analyses. Our results showed that immune cells with significantly different abundances between distinct IR patterns were positively correlated with both tricarboxylic acid (TCA) cycle and aminoacyl tRNA biosynthesis, while negatively correlated with Glycosphingolipid Biosynthesis Lacto and Neolacto Series (Figure S2). Furthermore, IR_cluster2 with lower overall immune cell abundance also exhibited lack of TCA cycle and aminoacyl tRNA biosynthesis pathway, while presented enrichment in Glycosphingolipid Biosynthesis Lacto and Neolacto Series pathway (Fig. 4B). As recent studies suggested that aminoacyl-tRNA synthetases (ARSs) could control antiviral innate immunity and lead to viral clearance [38–41], we analyzed the expression levels of ARSs between two IR patterns. As expected, almost all of these ARSs were obviously downregulated in IR_cluster2 (Fig. 4C). Therefore, the decreased expression of ARSs might also represent one of the metabolic mechanisms that suppressed antiviral immune responses in IR_cluster2 patients, further verifying the potential of ARSs as the therapeutic targets for infectious diseases.

We also compared the expression levels of genes that linked effector T cell activation and metabolic process in distinct IR patterns. These genes were summarized from a publication by Miguel et al. [37]. Notably, patients in IR_cluster2 displayed decreased expression of glycolysis modulators (PDK1, LAT, LCK, ZAP70, CD28, AKT1, NFATC1, NFATC2), nutrient uptake transporters (SLC38A1, SLC3 and SLC7 subfamily members), T cell receptor (TCR)-induced transcription factors (MYC, IRF4), nutrient-sensitive signaling complex (mTOR), regulator of fatty acid biosynthesis (ACACA), and an electron transport chain (ETC) complex III component (UQCRFS1) (Figure. 4D). All the evidences mentioned above suggested a strong association between metabolic process and adaptive immune regulation, and the specific immune metabolic characteristics in different IR patterns, indicating that the low frequency of CD8⁺ T cells in IR_cluster2 might be attributed to the suppressed nutrient metabolism mediated by immune metabolic regulators. Dysregulated metabolic pathways affected CD8⁺ T cells to adequately respond to SARS-CoV-2 infection.

3.5. Construction of IR signature score system for COVID-19 patients

To further figure out the potential biological behavior of each IR pattern, we obtained 1036 IR phenotype-related DEGs through DESeq analysis (Table S5). GO enrichment analysis revealed that these DEGs showed enrichment in immune cell activation and differentiation, cytokine production and secretion, inflammatory and immune response (Figure S3). Then we performed unsupervised clustering analyses based on 1036 DEGs to classify patients into genomic subtypes. As expected, patients in training cohort were stably classified into two distinct IR genomic phenotypes and we named these two subtypes as DEG_cluster1-2, respectively (Figure S4A-S4C and Fig. 5A). 32 patients were clustered in DEG_cluster1, which was related to IR_cluster1, while 43 were clustered in DEG_cluster2, which was related to IR_cluster2. We also observed a worse prognosis in DEG_cluster2 patients, similar to patients in IR_cluster2. Compared to DEG_cluster1, patients in DEG_cluster2 had shorter HFD-45 (Figure S4D). Meanwhile, DEG_cluster2 exhibited higher concentrations of D-dimer, CRP, lactate, ferritin, and presented higher SOFA scores (Figure S4E-S4I). Furthermore, patients with ICU requirement or mechanical ventilatory support were mainly concentrated in DEG_cluster2 (ICU: 12.5% versus 72.1%; mechanical ventilatory support: 12.5% versus 67.4%; Figure S4J-S4K). And more patients in DEG_cluster2 had Charlson comorbidity index scores greater than 6 (6.3% versus 23.3%, Figure S4L).

The above results again showed that IRs were essential to predict the prognosis of COVID-19 patients. However, these evidences were based on patient populations and could not directly predict the IR patterns of individual patients. To indicate the individual heterogeneity of IR patterns, we extracted overlapping differential genes and further removed redundant genes through Random Forest analysis. Finally, 10 signature genes, including CLEC10A, CYSTM1, EPAS1, FCER1A, PTH2R, RSPO2, SCN5A, SEZ6L, TMEM52B, and TMEM92, were obtained to construct a PCA-based scoring system. We termed this model as the IRscore. Using the cutoff value determined by survminer package, we divided patients into two groups, named IRscore_high and IRscore_low, respectively. The alluvial diagram displayed the attribute alterations of individual patients (Fig. 6A). The expression levels of 42 IRs between IRscore_high and IRscore_low group were analyzed through Mann-Whitney U test. IRscore_low was characterized by markedly elevated expressions of most IRs (Figure S5B). Mann-Whitney U test revealed the IRscore of IR_cluster1 was significantly lower than that of IR_cluster2 (Fig. 5B). Meanwhile, DEG_cluster1 also presented obviously lower IR_score compared with DEG_cluster2 (Fig. 5C), suggesting that high IRscore might be strongly linked to poor prognosis and immune inactivation-associated signatures. Therefore, we sought to determine the value of IRscore in predicting COVID-19 patients’ outcome. As expected, patients with high IRscore had shorter HFD-45 (Fig. 5D). And there were considerable increases in the concentrations of D-dimer, CRP, lactate, ferritin, and procalcitonin in IRscore_high (Fig. 5E-5I). In addition, patients with ICU requirement displayed higher IRscore compared to patients with no need for ICU (Fig. 5J). Consistent with ICU alteration trends, patients exhibiting worse prognostic clinical features (such as mechanical ventilation support, high Charlson comorbidities index score, or high SOFA score) had a higher IRscore (Fig. 5K-5M). We also utilized multivariate Cox regression model to investigate whether the IRscore could serve as an independent prognostic biomarker for COVID-19. As shown in Figure S5A, IRscore could be considered as a robust and independent prognostic biomarker for evaluating COVID-19 patients’ outcomes (HR 1.20 (1.05–1.38)). Moreover, the areas under the receiver operating characteristic curve (ROC) of training cohort on day 7, day 14, and day 26 were 0.795, 0.782, and 0.817, respectively, indicating the scoring model possessed good predictive capacity (Fig. 6B). All these results demonstrated that IRscore can reflect the IR patterns and predict clinical characteristics (such as the HFD-45, need for admission into the ICU, days spent on mechanical ventilator support, SOFA score, and Charlson comorbidity index score) of COVID-19 patients.

3.6. The potential role of IRscore in predicting disease state and the therapeutic effect of COVID-19 patients

To identify the stability of IRscore model, we applied IRscore signature established in training cohort to testing cohort. Results consistent with the training set were obtained, indicating the stability of IRscore to predict outcomes of COVID-19 patients (Fig. 6C-6H). IR patterns with lower expressions of IRs exhibited lower levels of CD4⁺ T cells and CD8⁺ T cells in testing cohort, further confirmed the association between IR transcriptional profile and immune cell fraction changes in training cohort (Figure S5C and S5D). In addition, the association between IRscore and IR patterns was also validated in testing cohort (Figure S5E). The areas under the ROC of testing cohort on day 7, day 14, and day 26 was 0.85, 0.877, 0.913, respectively (Fig. 6I), indicating the stability and reliability of IRscore model.

To further confirm the capability of IRscore to evaluate the clinical characteristics of COVID-19, we also applied IRscore to LPS-induced inflammation model and other independent COVID-19 cohorts with clinical information such as disease stage and disease state. In LPS-induced inflammation model, the higher lung-to-body weight (L/B) ratios and weight losses indicated the occurrence of pulmonary edema and poor physical condition in LPS-treated mice (Fig. 7A and 7B). LPS group also displayed the thickened alveolar septum and marked neutrophil accumulation which indicated the successful establishment of inflammatory model (Fig. 7C). More importantly, we observed the IRscore of LPS-treated mice was significantly higher than that of control mice (Fig. 7D). Through applying IRscore to GSE198256, we validated that IRscore could reflect the COVID-19 progress (Fig. 7E). In addition, we reconfirmed the positive correlation between IRscore and COVID-19 severity in GSE152418 (Fig. 7F). These data further verified the positive correlation between high IRscore and multiple poor clinical characteristics of COVID-19.

Considering the close association between IRscore and disease state of COVID-19 patients, we examined the power of IRscore to predict the patients’ responses to immunotherapies. As shown in Fig. 7G, after 7-day treatment with IL-1 receptor blocker anakinra, every COVID-19 patient presented a decrease in IRscore, suggesting that the established IRscore model might be useful for therapeutic efficacy prediction in COVID-19.

Immune homeostasis tuned by IRs is critical during virus infection. The association between individual dysregulated IRs and severity of COVID-19 has been mentioned in a few studies. However, few studies have comprehensively identified the immune modulation, immune-associated metabolic reprogramming, and clinical characterization of COVID-19 mediated by integrated effects of multiple IRs. Furthermore, the mechanism by which IR regulates the immune system to control infection remains controversial. In this study, we demonstrated that integrated effects of multiple IRs defined immune and metabolic characteristics, which determined the clinical outcomes of COVID-19 patients. Our study also indicated that the integrated effect of IRs influenced the disease state of COVID-19 patients through mechanisms other than inhibiting T cell activation. Identifying the correlation of IR patterns with immune cell infiltration and COVID-19 clinical features contribute to enhancing our understanding of IR in modulating immune responses and determining patient outcome, as well as developing more effective therapeutic strategies for COVID-19 patients.

In this study, we identified two distinct IR patterns based on 42 IRs. The two patterns had significantly distinct disease severity, immune response characterization, and metabolic adaption. Previous studies have proposed that lymphopenia as well as the higher serum levels of CRP, D-dimers, ferritin, and lactate can be considered as the risk factors of severe COVID-19 [42]. Consistent with these evidences, IR_cluster2 patients with higher serum levels of CRP, D-dimers, ferritin, lactate and lower frequencies of CD8⁺ T cells required longer hospital stay, ICU, and mechanical ventilatory support. Arunachalam et al. find that the myeloid cells of severe COVID-19 patients present a reduced human leukocyte antigen class DR (HLA-DR) expression [20]. As expected, an analogous reduction of HLA-DRA, HLA-DRB1, HLA-DRB3, and HLA-DRB5 levels was observed in PBMCs of IR pattern with poorer prognosis. Another feature of most severe COVID-19 patients is their enhanced levels of proinflammatory cytokines in the plasma. Among these proinflammatory cytokines, only S100A12, the gene encoding EN-RAGE, is substantially enhanced in the PBMCs of COVID-19 patients. And its gene expression in PBMCs has been confirmed to be consistent with the protein level in plasma. Furthermore, previous study suggests that S100A12 is a biomarker of pulmonary damage involved in pathogenesis of sepsis-induced ARDS [43–45]. Our discoveries on the significant upregulation of S100A12 in IR_cluster2 implied that IR_cluster2 patients might have developed the pulmonary injury, thus resulting in poor survival of these patients. Therefore, IR_cluster2 patients presented the suppressed immune state and were prone to pulmonary damage, which might explain their poor disease outcomes.

Through metabolic expression subtype classification analysis, we clarified that IR_cluster1 was characterized by upregulated energy production, amino acid metabolism, and nucleotide metabolism compared with IR_cluster2. In IR_cluster1, the increased levels of PDK1, LAT, LCK, and ZAP70 indicated the initiation of faster energy production route (glycolysis) to activate T cell in the absence of increased glucose uptake [46, 47]. Previous studies suggest that activated CD8⁺ T cells rely on glycolysis to break down glucose to fuel different kinds of metabolic synthesis pathways. The upregulation of NFATC1 and NFATC2 implied the enhancement of glycolytic metabolism of patients in IR_cluster1 [48, 49]. The elevated expression of CD28 and AKT1 in IR_cluster1 also indicated the maintenance of enhanced glycolysis by PI3K-AKT signaling downstream of CD28 co-stimulation in IR_cluster1 patients [46, 50, 51]. To further satisfy the nutrient demands, activated CD8⁺ T cells need to increase the expression of solute carrier (SLC) transporters. We discovered the upregulation of SLC38A1, SLC3 and SLC7 subfamily members in IR_cluster1. SLC38A1, known as a glutamine transporter, is upregulated upon T cell activation in a CD28-dependent manner to supply carbons for TCA cycle, which breaks down acetyl-CoA to carbon intermediates for producing energy and synthesizing new metabolic products [52]. SLC3 and SLC7 subfamily members regulate the uptake of large neutral amino acids to sustain protein synthesis in CD8⁺ T cells [53, 54]. Therefore, the upregulation of these SLC transporters suggested the notion that IR_cluster1 patients might satisfy the bioenergetic demands of CD8⁺ T cells for proliferation and differentiation by boosting glutamine uptake and its downstream metabolism.

Glutamine is another important nutrient that dictates the activation and function of immune cells. Glutaminolysis fueled by MYC is demanded to promote nucleotide synthesis to facilitate CD8⁺ T cell growth and proliferation [48]. We found MYC was strengthened in IR_cluster1, indicating the initiation of glutaminolysis. Although MYC plays a crucial role in establishing the metabolic reprogramming for T cell activation, another TCR-induced transcription factor, IRF4, is necessary for activated T cells to maintain their metabolic activity [55]. Similar to MYC, the expression of IRF4 was also remarkably ascending in IR_cluster1, indicating the sustaining of the metabolic activity for T cell activation. To synchronize nutrient availability with demand, cells utilize evolutionarily conserved nutrient-sensitive signaling complex mTOR to effectively control their growth, survival, and metabolism [56]. Congruent with its essential role in coordinating metabolic adaptation and immune cell fate, we noticed the increased expression of mTOR in IR_cluster1. Besides glucose, amino acid, and nucleotide metabolism, CD8⁺ T cells also require fatty acid synthesis regulated by the activity of acetyl-CoA carboxylase alpha (ACACA) to support their proliferation and survival [57, 58]. As expected, patients in IR_cluster1 displayed noticeable upregulation of ACACA. Previous study proposes that mitochondrial ROS production, T cell expansion, and effector function are suppressed by the lack of UQCRFS1 in vivo [59]. The high expression of UQCRFS1 in IR_cluster1 might contribute to facilitating mitochondrial ROS production, leading to the expansion and effector function of T cell, thus stimulating the effective immune response of IR_cluster1 patients to SARS-CoV-2 infection. In this study, we observed that IR_cluster1 upregulated the regulators that regulated both effector T cell activation and metabolic pathways. The upregulation of these regulators contributed to the enhanced nutrient metabolism for T cell activation and the increased infiltration of CD8⁺ T cells, thus determining IR_cluster1 patients’ better prognostic outcomes. Therefore, the immune response was intertwined with metabolic processes to ultimately determine the capability of organism to provide effective immune response and restore homeostasis.

In our study, the distinct disease severity, immune landscape, and metabolic characteristics between two IR patterns were undoubtedly closely associated with the expression profile of 42 IRs. A striking increase of most IRs in IR_cluster1 with better prognosis was seemingly at odds with the literature describing an obvious upregulation of specific IRs in severe COVID-19 patients. But under our scenario, patients in IR_cluster1 displayed not only an enhanced expression of IRs but also an increased frequency of immune effector cells (such as CD8⁺ T cells and activated CD4⁺ T memory cells) and an enhanced enrichment of antiviral immune pathways, such as T and B cell receptor signaling pathway, NK cell mediated cytotoxicity, antigen processing and presentation pathway, chemokine signaling pathway, cytokine-cytokine receptor interaction pathway, interferon signaling pathway, and NF\(\kappa\)B signaling pathway. Moreover, IR_cluster1 patients also showed significant inhibition of the biomarker of pulmonary damage, S100A12. All these results indicated that the physiological function of IRs during SARS-CoV-2 infection was not to simply dampen the innate or adaptive immune activation triggered by viral infection, but to maintain homeostasis to rapidly clear pathogen and effectively prevent immunopathology. Previous evidence proposes that the impact of IRs on cellular function is dependent on the strength of the inhibitory signal relative to the activation signal given within a certain time window [60]. Suppression of immune response by IRs might have a disadvantageous effect on pathogen clearance, but without activation-induced negative feedback mechanism regulated by IRs, the termination of effector mechanism activation will be delayed, thus increasing the chance of developing immunopathology. Therefore, the additional regulation through IR-induced negative feedback ensures pathogen clearance with less affect owing to timely inhibition of the late activation signal. By analogy, the vigorous upregulation of most IRs in IR_cluster1 might reflect an immune activation-induced negative feedback mechanism to prevent rampant systemic immune hyperactivation. It was this negative feedback that sustained more effective pathogen clearance, maintained organism homeostasis, and decreased the cost of the immune response, which guaranteed the favorable prognosis of patients in IR_cluster1. This also suggests that clinicians should consider whether cancer patients infected with SARS-CoV-2 are at increased risk of COVID-19 related immunopathology mediated by IR suppression while benefiting from immune checkpoint drug administration.

After determining the correlation between IR patterns and disease severity, immune status and metabolic characteristics of COVID-19 patients, we further extracted the differentially expressed genes between distinct IR patterns, namely IR-related signature genes. Based on these genes, we identified two genomic subtypes correlated with the IR patterns and COVID-19 patients’ prognosis, further confirming the existence of two IR-associated subtypes in COVID-19. Considering the heterogeneity of IR regulation in individuals, we therefore constructed a scoring system to evaluate the IR pattern of individual patients and termed this scoring system as IRscore. The IR pattern with more severe disease presented a higher IRscore, while the pattern with better prognostic outcome exhibited a lower IRscore. Subsequently, we well verified the reliability and stability of IRscore model in predicting the disease severity and staging in the testing cohort, LPS-induced inflammation model and other datasets (GSE198256 and GSE152418).

Considering the close association between IRscore and disease severity, we further investigated the ability of IRscore to evaluate the therapeutic effect. Immunotherapies have been used to treat patients with moderate COVID-19 admitted to general medicine wards [61]. Anakinra is considered for patients who do not require oxygen therapy but have high expression of inflammation-related biomarkers [62]. In GSE163317 dataset, IRscore decreased slightly after anakinra administration. We speculated that a more significant reduction in IRscore might be observed if the treatment sample size was expanded. This also indicated that IRscore has the potential to be a reliable and powerful tool to guide precise treatment for COVID-19 patients. In summary, IRscore showed predicative advantages in both COVID-19 disease status assessment and immunotherapy efficacy evaluation.

Although the importance of the integrated effect of 42 IRs in predicting COVID-19 disease state has been elucidated, there were still some inadequacies in our study. One limitation was the requirements of further validating our results in a prospective cohort to more scientifically define the cutoff value that could divide the high and low IRscore of COVID-19 patients. Additionally, as the IRscore model was established according to the transcriptomic levels of IRs in PBMCs of COVID-19 patients, the model might not be able to use protein levels of other sample types to make accurate and stable predictions of disease state in COVID-19 patients. Therefore, confirmatory experiments of protein levels from other sample types were needed to determine the universality of this model and further validate our results.

Our work demonstrated an extensive regulatory mechanism by which IRs affected immune response, immune-associated metabolic adaption and clinical features of COVID-19. The difference in IR patterns was a non-negligible factor leading to the different clinical manifestations in individuals infected with SARS-CoV-2. Our work highlights the important role of IR integrated effect in infection immunomodulation, disease status surveillance, and administration efficacy assessment, offers a theoretical basis that may aid future studies, and will facilitate the development of personalized and effective approaches for infectious diseases such as COVID-19.

ACKNOWLEDGEMENTS

We thank all the researchers involved in this work. This research was supported by grants from the research on key technologies for the prevention and treatment of important viral diseases funded by Institute of Health Service and Transfustion Medicine.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

WHO Coronavirus (COVID-19) dashboard, Mar. 2023 [online database] (https://covid19.who.int).
D. A. Berlin, R. M. Gulick and F. J. Martinez, Severe Covid-19, N Engl J Med. 2020. 383 (25): 2451-2460.
T. P. Velavan, S. R. Pallerla, J. Ruter, Y. Augustin, P. G. Kremsner, S. Krishna and C. G. Meyer, Host genetic factors determining COVID-19 susceptibility and severity, EBioMedicine. 2021. 72 103629.
S. A. Lowery, A. Sariol and S. Perlman, Innate immune and inflammatory responses to SARS-CoV-2: Implications for COVID-19, Cell Host Microbe. 2021. 29 (7): 1052-1062.
M. Zheng, Y. Gao, G. Wang, G. Song, S. Liu, D. Sun, Y. Xu and Z. Tian, Functional exhaustion of antiviral lymphocytes in COVID-19 patients, Cell Mol Immunol. 2020. 17 (5): 533-535.
M. Colonna, Immunoglobulin superfamily inhibitory receptors: from natural killer cells to antigen-presenting cells, Res Immunol. 1997. 148 (3): 169-71.
M. Daeron, S. Jaeger, L. Du Pasquier and E. Vivier, Immunoreceptor tyrosine-based inhibition motifs: a quest in the past and future, Immunol Rev. 2008. 224 11-43.
M. Rumpret, J. Drylewicz, L. J. E. Ackermans, J. A. M. Borghans, R. Medzhitov and L. Meyaard, Functional categories of immune inhibitory receptors, Nat Rev Immunol. 2020. 20 (12): 771-780.
M. Saresella, D. Trabattoni, I. Marventano, F. Piancone, F. La Rosa, A. Caronni, A. Lax, L. Bianchi, P. Banfi, J. Navarro, NK cell subpopulations and receptor expression in recovering SARS-CoV-2 Infection, Mol Neurobiol. 2021. 58 (12): 6111-6120.
H. Y. Zheng, M. Zhang, C. X. Yang, N. Zhang, X. C. Wang, X. P. Yang, X. Q. Dong and Y. T. Zheng, Elevated exhaustion levels and reduced functional diversity of T cells in peripheral blood may predict severe progression in COVID-19 patients, Cell Mol Immunol. 2020. 17 (5): 541-543.
Y. Kong, Y. Wang, X. Wu, J. Han, G. Li, M. Hua, K. Han, H. Zhang, A. Li and H. Zeng, Storm of soluble immune checkpoints associated with disease severity of COVID-19, Signal Transduct Target Ther. 2020. 5 (1): 192.
S. Shahbaz, L. Xu, W. Sligl, M. Osman, N. Bozorgmehr, S. Mashhouri, D. Redmond, E. Perez Rosero, J. Walker and S. Elahi, The Quality of SARS-CoV-2-Specific T Cell Functions Differs in Patients with Mild/Moderate versus Severe Disease, and T Cells Expressing Coinhibitory Receptors Are Highly Activated, J Immunol. 2021. 207 (4): 1099-1111.
M. Herrmann, S. Schulte, N. H. Wildner, M. Wittner, T. T. Brehm, M. Ramharter, R. Woost, A. W. Lohse, T. Jacobs and J. Schulze Zur Wiesch, Analysis of co-inhibitory receptor expression in COVID-19 Infection compared to acute plasmodium falciparum Malaria: LAG-3 and TIM-3 correlate With T cell activation and course of disease, Front Immunol. 2020. 11 1870.
J. Yang, T. Chang, L. Tang, H. Deng, D. Chen, J. Luo, H. Wu, T. Tang, C. Zhang, Z. Li, Increased expression of Tim-3 Is associated with depletion of NKT cells In SARS-CoV-2 Infection, Front Immunol. 2022. 13 796682.
N. Saheb Sharif-Askari, F. Saheb Sharif-Askari, B. Mdkhana, S. Al Heialy, H. S. Alsafar, R. Hamoudi, Q. Hamid and R. Halwani, Enhanced expression of immune checkpoint receptors during SARS-CoV-2 viral infection, Mol Ther Methods Clin Dev. 2021. 20 109-121.
M. S. Rha, H. W. Jeong, J. H. Ko, S. J. Choi, I. H. Seo, J. S. Lee, M. Sa, A. R. Kim, E. J. Joo, J. Y. Ahn, PD-1-Expressing SARS-CoV-2-Specific CD8(+) T Cells Are Not Exhausted, but Functional in Patients with COVID-19, Immunity. 2021. 54 (1): 44-52 e3.
A. A. Mohammad A. Al-Mterin, Eyad Elkord, Inhibitory Immune Checkpoint Receptors and Ligands as Prognostic Biomarkers in COVID-19 Patients, Front Immunol. 2022. 13 870283.
K. A. Overmyer, E. Shishkova, I. J. Miller, J. Balnis, M. N. Bernstein, T. M. Peters-Clarke, J. G. Meyer, Q. Quan, L. K. Muehlbauer, E. A. Trujillo, Large-scale multi-omic analysis of COVID-19 severity, Cell Syst. 2021. 12 (1): 23-40 e7.
E. Brauns, A. Azouz, D. Grimaldi, H. Xiao, S. Thomas, M. Nguyen, V. Olislagers, I. Vu Duc, C. Orte Cano, V. Del Marmol, Functional reprogramming of monocytes in patients with acute and convalescent severe COVID-19, JCI Insight. 2022. 7 (9):
P. S. Arunachalam, F. Wimmers, C. K. P. Mok, R. Perera, M. Scott, T. Hagan, N. Sigal, Y. Feng, L. Bristow, O. Tak-Yin Tsang, Systems biological assessment of immunity to mild versus severe COVID-19 infection in humans, Science. 2020. 369 (6508): 1210-1220.
A. Bertoni, F. Penco, H. Mollica, P. Bocca, I. Prigione, A. Corcione, D. Cangelosi, F. Schena, G. Del Zotto, A. Amaro, Spontaneous NLRP3 inflammasome-driven IL-1-beta secretion is induced in severe COVID-19 patients and responds to anakinra treatment, J Allergy Clin Immunol. 2022. 150 (4): 796-805.
M. Kuhn, Building predictive models in R using the caret package, Journal of statistical software. 2008. 28 (5):
M. D. Wilkerson and D. N. Hayes, ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking, Bioinformatics. 2010. 26 (12): 1572-3.
S. Hanzelmann, R. Castelo and J. Guinney, GSVA: gene set variation analysis for microarray and RNA-seq data, BMC Bioinformatics. 2013. 14 7.
G. Yu, L. G. Wang, Y. Han and Q. Y. He, clusterProfiler: an R package for comparing biological themes among gene clusters, OMICS. 2012. 16 (5): 284-7.
A. M. Newman, C. L. Liu, M. R. Green, A. J. Gentles, W. Feng, Y. Xu, C. D. Hoang, M. Diehn and A. A. Alizadeh, Robust enumeration of cell subsets from tissue expression profiles, Nat Methods. 2015. 12 (5): 453-7.
X. Peng, Z. Chen, F. Farshidfar, X. Xu, P. L. Lorenzi, Y. Wang, F. Cheng, L. Tan, K. Mojumdar, D. Du, Molecular characterization and clinical relevance of metabolic expression subtypes in human cancers, Cell Rep. 2018. 23 (1): 255-269 e4.
A. Fabregat, K. Sidiropoulos, P. Garapati, M. Gillespie, K. Hausmann, R. Haw, B. Jassal, S. Jupe, F. Korninger, S. McKay, The Reactome pathway Knowledgebase, Nucleic Acids Res. 2016. 44 (D1): D481-7.
S. Anders and W. Huber, Differential expression analysis for sequence count data, Genome Biol. 2010. 11 (10): R106.
M. B. Kursa and W. R. Rudnicki, Feature selection with the boruta package, Journal of Statistical software. 2010. 36 (11):
C. Sotiriou, P. Wirapati, S. Loi, A. Harris, S. Fox, J. Smeds, H. Nordgren, P. Farmer, V. Praz, B. Haibe-Kains, Gene expression profiling in breast cancer: understanding the molecular basis of histologic grade to improve prognosis, J Natl Cancer Inst. 2006. 98 (4): 262-72.
H. Xian, Y. Liu, A. Rundberg Nilsson, R. Gatchalian, T. R. Crother, W. G. Tourtellotte, Y. Zhang, G. R. Aleman-Muench, G. Lewis, W. Chen, Metformin inhibition of mitochondrial ATP and DNA synthesis abrogates NLRP3 inflammasome activation and pulmonary inflammation, Immunity. 2021. 54 (7): 1463-1477 e11.
A. T. Feldman and D. Wolfe, Tissue processing and hematoxylin and eosin staining, Methods Mol Biol. 2014. 1180 31-43.
B. Zhang, Q. Wu, B. Li, D. Wang, L. Wang and Y. L. Zhou, m(6)A regulator-mediated methylation modification patterns and tumor microenvironment infiltration characterization in gastric cancer, Mol Cancer. 2020. 19 (1): 53.
A. Sette and S. Crotty, Adaptive immunity to SARS-CoV-2 and COVID-19, Cell. 2021. 184 (4): 861-880.
C. Huang, Y. Wang, X. Li, L. Ren, J. Zhao, Y. Hu, L. Zhang, G. Fan, J. Xu, X. Gu, Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China, Lancet. 2020. 395 (10223): 497-506.
M. Reina-Campos, N. E. Scharping and A. W. Goldrath, CD8(+) T cell metabolism in infection and cancer, Nat Rev Immunol. 2021. 21 (11): 718-738.
E. Y. Lee, S. Kim and M. H. Kim, Aminoacyl-tRNA synthetases, therapeutic targets for infectious diseases, Biochem Pharmacol. 2018. 154 424-434.
E. Y. Lee, H. C. Lee, H. K. Kim, S. Y. Jang, S. J. Park, Y. H. Kim, J. H. Kim, J. Hwang, J. H. Kim, T. H. Kim, Infection-specific phosphorylation of glutamyl-prolyl tRNA synthetase induces antiviral immunity, Nat Immunol. 2016. 17 (11): 1252-1262.
D. Liang, L. Tian, R. You, M. M. Halpert, V. Konduri, Y. C. Baig, S. Paust, D. Kim, S. Kim, F. Jia, AIMp1 Potentiates TH1 Polarization and Is Critical for Effective Antitumor and Antiviral Immunity, Front Immunol. 2017. 8 1801.
Y. H. Ahn, S. Park, J. J. Choi, B. K. Park, K. H. Rhee, E. Kang, S. Ahn, C. H. Lee, J. S. Lee, K. S. Inn, Secreted tryptophanyl-tRNA synthetase as a primary defence system against infection, Nat Microbiol. 2016. 2 16191.
T. P. Velavan and C. G. Meyer, Mild versus severe COVID-19: Laboratory markers, Int J Infect Dis. 2020. 95 304-307.
Z. Zhang, N. Han and Y. Shen, S100A12 promotes inflammation and cell apoptosis in sepsis-induced ARDS via activation of NLRP3 inflammasome signaling, Mol Immunol. 2020. 122 38-48.
F. Zhao, B. Hoechst, A. Duffy, J. Gamrekelashvili, S. Fioravanti, M. P. Manns, T. F. Greten and F. Korangy, S100A9 a new marker for monocytic human myeloid-derived suppressor cells, Immunology. 2012. 136 (2): 176-83.
O. M. Pena, D. G. Hancock, N. H. Lyle, A. Linder, J. A. Russell, J. Xia, C. D. Fjell, J. H. Boyd and R. E. Hancock, An Endotoxin Tolerance Signature Predicts Sepsis and Organ Dysfunction at Initial Clinical Presentation, EBioMedicine. 2014. 1 (1): 64-71.
A. V. Menk, N. E. Scharping, R. S. Moreci, X. Zeng, C. Guy, S. Salvatore, H. Bae, J. Xie, H. A. Young, S. G. Wendell, Early TCR Signaling Induces Rapid Aerobic Glycolysis Enabling Distinct Acute T Cell Effector Functions, Cell Rep. 2018. 22 (6): 1509-1521.
C. Cammann, A. Rath, U. Reichl, H. Lingel, M. Brunner-Weinzierl, L. Simeoni, B. Schraven and J. A. Lindquist, Early changes in the metabolic profile of activated CD8(+) T cells, BMC Cell Biol. 2016. 17 (1): 28.
R. Wang, C. P. Dillon, L. Z. Shi, S. Milasta, R. Carter, D. Finkelstein, L. L. McCormick, P. Fitzgerald, H. Chi, J. Munger, The transcription factor Myc controls metabolic reprogramming upon T lymphocyte activation, Immunity. 2011. 35 (6): 871-82.
S. Klein-Hessling, K. Muhammad, M. Klein, T. Pusch, R. Rudolf, J. Floter, M. Qureischi, A. Beilhack, M. Vaeth, C. Kummerow, NFATc1 controls the cytotoxicity of CD8(+) T cells, Nat Commun. 2017. 8 (1): 511.
K. A. Frauwirth, J. L. Riley, M. H. Harris, R. V. Parry, J. C. Rathmell, D. R. Plas, R. L. Elstrom, C. H. June and C. B. Thompson, The CD28 signaling pathway regulates glucose metabolism, Immunity. 2002. 16 (6): 769-77.
S. R. Jacobs, C. E. Herman, N. J. Maciver, J. A. Wofford, H. L. Wieman, J. J. Hammen and J. C. Rathmell, Glucose uptake is limiting in T cell activation and requires CD28-mediated Akt-dependent and independent pathways, J Immunol. 2008. 180 (7): 4476-86.
E. L. Carr, A. Kelman, G. S. Wu, R. Gopaul, E. Senkevitch, A. Aghvanyan, A. M. Turay and K. A. Frauwirth, Glutamine uptake and metabolism are coordinately regulated by ERK/MAPK during T lymphocyte activation, J Immunol. 2010. 185 (2): 1037-44.
D. Fotiadis, Y. Kanai and M. Palacin, The SLC3 and SLC7 families of amino acid transporters, Mol Aspects Med. 2013. 34 (2-3): 139-58.
L. V. Sinclair, J. Rolf, E. Emslie, Y. B. Shi, P. M. Taylor and D. A. Cantrell, Control of amino-acid transport by antigen receptors coordinates the metabolic reprogramming essential for T cell differentiation, Nat Immunol. 2013. 14 (5): 500-8.
K. Man, M. Miasari, W. Shi, A. Xin, D. C. Henstridge, S. Preston, M. Pellegrini, G. T. Belz, G. K. Smyth, M. A. Febbraio, The transcription factor IRF4 is essential for TCR affinity-mediated metabolic programming and clonal expansion of T cells, Nat Immunol. 2013. 14 (11): 1155-65.
J. D. Powell and G. M. Delgoffe, The mammalian target of rapamycin: linking T cell differentiation, function, and metabolism, Immunity. 2010. 33 (3): 301-11.
J. Lee, M. C. Walsh, K. L. Hoehn, D. E. James, E. J. Wherry and Y. Choi, Regulator of fatty acid metabolism, acetyl coenzyme a carboxylase 1, controls T cell immunity, J Immunol. 2014. 192 (7): 3190-9.
S. A. Ibitokou, B. E. Dillon, M. Sinha, B. Szczesny, A. Delgadillo, D. Reda Abdelrahman, C. Szabo, L. Abu-Elheiga, C. Porter, D. Tuvdendorj, Early Inhibition of Fatty Acid Synthesis Reduces Generation of Memory Precursor Effector T Cells in Chronic Infection, J Immunol. 2018. 200 (2): 643-656.
L. A. Sena, S. Li, A. Jairaman, M. Prakriya, T. Ezponda, D. A. Hildeman, C. R. Wang, P. T. Schumacker, J. D. Licht, H. Perlman, Mitochondria are required for antigen-specific T cell activation through reactive oxygen species signaling, Immunity. 2013. 38 (2): 225-36.
J. V. Ravetch and L. L. Lanier, Immune inhibitory receptors, Science. 2000. 290 (5489): 84-9.
F. L. van de Veerdonk, E. Giamarellos-Bourboulis, P. Pickkers, L. Derde, H. Leavis, R. van Crevel, J. J. Engel, W. J. Wiersinga, A. P. J. Vlaar, M. Shankar-Hari, A guide to immunotherapy for COVID-19, Nat Med. 2022. 28 (1): 39-50.
E. Kyriazopoulou, G. Poulakou, H. Milionis, S. Metallidis, G. Adamis, K. Tsiakos, A. Fragkou, A. Rapti, C. Damoulari, M. Fantoni, Early treatment of COVID-19 with anakinra guided by soluble urokinase plasminogen receptor plasma levels: a double-blind, randomized controlled phase 3 trial, Nat Med. 2021. 27 (10): 1752-1760.

No competing interests reported.

Download PDF

Journal Publication

published 06 Nov, 2023

Read the published version in Scientific Reports →

Editorial decision: Major revision
24 Sep, 2023
Reviews received at journal
18 Sep, 2023
Reviewers agreed at journal
13 Sep, 2023
Reviewers invited by journal
04 Sep, 2023
Editor assigned by journal
04 Sep, 2023
Editor invited by journal
04 Sep, 2023
Submission checks completed at journal
04 Sep, 2023
First submitted to journal
01 Sep, 2023

You are reading this latest preprint version

Immune inhibitory receptor-mediated immune response, metabolic adaptation, and clinical characterization in COVID-19 patients

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

METHODS

2.1. The source of COVID-19 dataset

2.2 Principal component analysis

2.3. Unsupervised clustering for 42 immune inhibitory receptors

2.4. Gene set variation analysis (GSVA) and functional annotation

2.5 Estimation of immune cell fractions

2.6 Metabolic expression subtype classification

2.7 Generation of IR_score

2.8 Lipopolysaccharide-induced inflammatory model

2.9 Spleen lymphocyte isolation

RNA extraction and real-time polymerase chain reaction

3.1 Hematoxylin-eosin staining

3.2 Statistical Analysis

Results

3.2. IR Patterns mediated by 42 immune inhibitory receptors and clinical characteristics of each pattern

3.3. Immunological features in distinct IR patterns of COVID-19 patients

3.4. Immunometabolic adaptation characteristics in distinct IR patterns of COVID-19 patients

3.5. Construction of IR signature score system for COVID-19 patients

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1