Featured immune characteristics of COVID-19 and systemic lupus erythematosus revealed by multidimensional integrated analyses

Coronavirus disease 2019 (COVID-19) shares similar immune characteristics with autoimmune diseases like systemic lupus erythematosus (SLE). However, such associations have not yet been investigated at the single-cell level. We integrated and analyzed RNA sequencing results from different patients and normal controls from the GEO database and identified subsets of immune cells that might involve in the pathogenesis of SLE and COVID- 19. We also disentangled the characteristic alterations in cell and molecular subset proportions as well as gene expression patterns in SLE patients compared with COVID-19 patients. Key immune characteristic genes (such as CXCL10 and RACK1) and multiple immune-related pathways (such as the coronavirus disease-COVID-19, T-cell receptor signaling, and MIF-related signaling pathways) were identified. We also highlighted the differences in peripheral blood mononuclear cells (PBMCs) between SLE and COVID-19 patients. Moreover, we provided an opportunity to comprehensively probe underlying B-cell‒cell communication with multiple ligand–receptor pairs (MIF-CD74+CXCR4, MIF-CD74+CD44) and the differentiation trajectory of B-cell clusters that is deemed to promote cell state transitions in COVID-19 and SLE. Our results demonstrate the immune response differences and immune characteristic similarities, such as the cytokine storm, between COVID-19 and SLE, which might pivotally function in the pathogenesis of the two diseases and provide potential intervention targets for both diseases.


Introduction
Coronavirus disease 2019 (COVID- 19) is an infectious disease, which is attribute to infection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and has recently spread rapidly worldwide and led to great human hardship.Systemic lupus erythematosus (SLE) is a chronic relapsing multisystem autoimmune disease of unknown etiology, and patients with SLE may have a high risk of acquiring SARS-CoV-2 infection [1].From a public health perspective, both diseases are significant as they affect a large number of people worldwide.Clinically, SLE requires long-term management with immune-suppressive drugs, while COVID-19 requires immediate attention and treatment to prevent severe complications such as pneumonia and acute respiratory distress syndrome.For pharmaceutical companies, developing effective treatments and vaccines for both diseases are crucial to help manage and prevent their spread.
The coronavirus shares a common host cell entry receptor, angiotensin converting enzyme 2 (ACE2), of which the aberrant upregulation has been observed in T cells from SLE patients [2].Patients with SLE are susceptible to SARS-CoV-2 infection due to innate immune disorders and increased inflammation [3].Once infected with coronavirus, SLE patients exhibit severely weakened and altered immune systems [4,5].Moreover, SARS-CoV-2 infections can induce other autoimmune diseases including autoimmune kidney disease as well as some rheumatic complications [4,[6][7][8].
Several studies have reported that coronaviruses may be correlated with the pathogenesis of autoimmune diseases.Some specific inflammatory and immune responses are involved in the pathogenesis of both COVID-19 and autoimmune disease [5].Several similarities between COVID-19 and autoimmune diseases such as SLE mainly include aberrant immune responses such as proinflammatory cytokines, lymphopenia, and aberrant T-and B-cell responses [9].To some extent, organ injury in COVID-19 patients correlates with host immune status, as observed in patients with SLE [10].In addition, SLE and COVID-19 share some of the same complications including leukopenia, lymphocytopenia, thrombocytopenia, and vasculitis [11].Some autoantibodies associated with SLE can also be detected in SARS-CoV-2-infected patients [7].Furthermore, the microbiome characteristics of COVID-19 patients are similar to those of SLE patients [12].All these findings suggest that crosstalk exists between implicated inflammatory pathways/mechanisms and clinical characteristics in both SLE and COVID-19 patients.
However, there are few reports on the common molecular mechanisms of COVID-19 and SLE [13].The mechanisms underlying SARS-CoV-2 infection modulates the risk of autoimmune disease including SLE has not been fully elucidated [14].Human leukocyte antigens (HLA) molecules have been shown to play an important role in protective immunity as well as in disease-causing autoimmune responses [15].New evidences have shown that the HLA gene sets were associated with SLE [16].The HLA gene sets, as part of host genetic factors, contribute to disease susceptibility and prognosis.HLA is crucial for appropriate immune responses in SARS CoV-2 infections [17].A recent study provides evidence that HLA genotype influences clinical outcome in COVID-19 [18].We thus investigated the association of HLA gene sets with SLE and COVID-19 in this study.Moreover, the efficiency and safety of COVID-19 vaccines for the prevention of COVID-19 in SLE patients is unclear, especially in the patients receiving immunosuppressive treatment.In many cases, the current understanding of peripheral blood immune cells in SLE patients infected with COVID-19, which may be modulated during disease, is limited to flow cytometry assay, but the specific status of immune cells such as T and B cells, remains unclear.Singlecell transcriptome analysis facilitates discrimination of the disease specific immune status.Single-cell RNA sequencing (scRNA-seq) methods through the unbiased clustering of cell populations can obviously capture potential pathogenic drivers and biological targets [19], which have been utilized to discriminate specific inflammatory cell states and identify specific immune cell populations in healthy subjects or diseased patients such as those with SLE [20,21].Therefore, investigating the potential associations and molecular mechanisms between COVID-19 and SLE by analyzing high-throughput sequencing data is particularly urgent to help in diagnosing and treating both diseases.
In this study, the distinct immune subtypes and molecular subtype pattern characteristics of COVID-19 and SLE cohorts were recognized.We also explored the transcriptional features of human peripheral blood mononuclear cells (PBMCs) from COVID-19 and SLE patients at the single-cell level to reveal the molecular mechanisms of specific disease processes.We then analyzed the single-cell immunological landscape to reveal immune cell types that might contribute to immunotherapy development.The key immune-related gene regulators in our results may serve as good biomarkers for diagnosing and monitoring the efficacy of immunotherapy.The common molecular pathway characteristics of the two diseases revealed in our study may also help scientists develop specific COVID-19 therapies and vaccines.

Verification of immunity-related subtypes and differentially expressed genes
By using single-sample gene set enrichment (ssGSEA) of " Gene set variation analysis (GSVA)" R package to quantify the enrichment levels of the 29 immune-associated maker gene sets in each sample, which covered diverse immune cell types, functions, and pathways, and performed hierarchical clustering of samples, we evaluated the immune infiltration landscape of SLE and COVID-19 patients from datasets and the immune status of the subjects was divided into two immunophenotype groups (low immunity (immune_L) and high immunity (immune_H)) [22,23].The immune-related differentially expressed genes (DEGs) were determined by expression comparison between the two immunophenotype groups using the limma package of R software (adjusted P < 0.05 and |log2FC|> 1) [24].The gene sets with immunerelation were downloaded from the Immunology Database and Analysis Portal (ImmPort) database (http:// www.immpo rt.org) [25].Furthermore, the relationship between the immune microenvironment and HLA gene expression in the two groups was analyzed.

WGCNA
Weighted gene coexpression network analysis (WGCNA) is an R package for weighted correlation network analysis [26].We conducted hierarchical clustering of samples to cluster modules.The dynamic cutting algorithm was used to validate the gene modules.Subsequently, the characteristic genes were clustered and merged into modules.The gene network can be divided into different modules according to the similarity of expression, the leftmost color block represents the different gene modules and the rightmost color bar represents the correlation range.In the middle part of the heat map, the darker the color, the higher the correlation.Red indicates positive correlation, and green indicates negative correlation.The numbers in each cell indicate relevance and significance.Module-trait relationships between each module and the clinical characteristics were identified based on Pearson correlation.

Functional annotation with molecular subtypes
Three COVID-19 and SLE cohorts were used to develop the molecular subtype classification.Principal component analysis (PCA) was used for data dimensionality reduction and visualization.

Gene set variation analysis (GSVA)
We used the GSVA R package to study the activity of biological pathways among molecular subtypes [23].The 'c2. cp.kegg.v7.0.symbols' gene sets were obtained from Release 6.0 of the Molecular Signatures Database (MSigDB) for GSVA analyses [27].The clusterProfiler R package was used to conduct Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses between distinct subtypes [28].

Cell clustering annotation and visualization
The Seurat v4.0.0 package of R software (v4.0.3) was used to perform the unsupervised clustering [29].The data were normalized to transcript copies per 10,000 and log-normalized to reduce sequencing depth variability.After quality control and filtering, the variation coefficient of genes was analyzed with Seurat, and 2000 genes were chosen for further analysis.Uniform Manifold Approximation and Projection (UMAP) plots was adopted for cluster identification and dimensionality reduction of visualization.The FindConservedMarkers was adopted for identification of the clusters.The parameters setting criteria are logFC.threshold> 0.25, minPct > 0.25 and Padj ≤ 0.05.Then, clusters were annotated based on the DEGs and the canonical cellular markers according to the Cell Marker database [30].Major cell type annotation section also referring to the published literatures [31][32][33].

Gene set enrichment analysis (GSEA) and Single-sample GSEA (ssGSEA)
The immune status in COVID-19 and SLE was evaluated by ssGSEA.The enrichment of a gene set was indicated by the enrichment fraction in each sample.The pathway enrichment analysis was performed using the GSEA in patients with different subtypes of COVID-19 or SLE according to the MSigDB.GO and KEGG pathway analyses were also conducted with GSEA and a false discovery rate (FDR) ≤ 0.05 was defined as obviously enriched [34].The circus plots were drawn by using R v4.0.3.

Intercellular communication analysis
To study potential intercellular communication between the B-cell subsets, we used Cell Chat package in R to evaluate ligand-receptor expression and distribution of B-cell subsets through standard pipelines as described previously [35].We used Cell Phone DB (ligand-receptor pair list, www.cellp honedb.org) [36] to identify cell-cell interactions between B-cell subsets.

Trajectory analysis
The cellular dynamic processes of B-cell subpopulation differentiation [37] and the origination were modeled using a trajectory inference method.Specifically, the differential GeneTest [38] was used to calculate the top 150 signature genes.Monocle default parameters were used to infer B-cell differentiation trajectories after dimensionality reduction and cell ordering.The 'DDRTree' and 'plot_cell_trajectory' were utilized for dimensionality reduction and for visualization, respectively.

Statistical analysis
The software R v4.0.3 and corresponding R packages (http:// www.bioco nduct or.org/) were used to perform statistical analysis.Spearman's correlation was utilized to analyze correlations.The comparisons between two groups were examined by means of Wilcoxon test.All statistical P values (P < 0.05) represents statistical significance.

Immune status and immune-related DEGs in COVID-19 and SLE patients
We searched six gene expression datasets representing samples from COVID-19 and SLE patients and samples from normal controls to start the analyses.We analyzed the immune status using the ssGSEA method, and performed clustering by dividing the COVID-19 patients into two groups, the immune_H group and immune_L group.
Analysis of the difference in the distributions between both groups indicated that the stromal scores of the immune_H group were higher (P < 0.01) based on the Mann-Whitney U test, consistent with the estimate scores and immune scores, as shown in Fig. 1A and B. We also analyzed the relationship between both immune types and HLA gene sets.The results demonstrated that the samples in the immune_H group showed 12 genes (HLA-DMA, HLA-DMB, HLA-DOA, HLA-DPA1, HLA-DPB1, HLA-DPB2, HLA-DQA1, HLA-DQA2, HLA-DQB2, HLA-DRA, HLA-DRB6, HLA-L) with higher expression and 6 genes (HLA-A, HLA-C, HLA-E, HLA-B, HLA-F and HLA-J) with significantly lower expression than those in the immune_L group (Fig. 1C).Furthermore, the immune_L group had more neutrophils while the immune_H group had more resting NK cells and CD8 + T cells (Fig. 1D).Ninety-eight immune-related DEGs were found between the both groups (Supplementary Figure S1 (Figure S1) and Fig. 1E).In addition, GSEA analysis of the immune-related genes showed that COVID-19 patients were positively associated with allograft rejection biological processes (Fig. 1F).WGCNA was used to describe the associations between the module eigengenes and clinical characteristics.The analysis confirmed that the pink module had significant positive correlation with hospital-free days and that the blue module had negative correlation with mechanical ventilation (Fig. 1G).
We also divided the SLE patients into immune-high and -low groups by hierarchical clustering.That is the immune_H group has higher immune scores and vice versa (Fig. 2A and B).The relationship between both immune types and HLA gene sets demonstrated that the samples in the immune_H group had significantly higher expression of 12 genes (HLA-DMA, HLA-DOA, HLA-DMB, HLA-DPB1, HLA-DPA1, HLA-DPB2, HLA-DQA2, HLA-DQB2, HLA-DQA1, HLA-DRB5, HLA-DRA, HLA-DRB6) and 6 significantly lower expression of genes (HLA-A, HLA-C, HLA-E, HLA-B, HLA-L and HLA-F) compared with the immune_L group (Fig. 2C).Furthermore, the immune_L group had more plasma cells and naive CD4 + T cells, while the immune_H group had more regulatory T cells (Treg), memory B cells, CD8 + T cells, neutrophils and resting NK cells (Fig. 2D).Total 112 immune-related DEGs between both groups were obtained (Fig. 2E and Figure S1).WGCNA revealed that the associations between the module eigengenes and clinical characteristics were not significant (Fig. 2F).In addition, GSEA analysis of the immune-associated genes showed that SLE was negatively correlated with propanoate metabolism, and valine leucine and isoleucine degradation (Fig. 2G).There are several studies demonstrated that the possible etiological mechanisms of SLE and COVID-19 include immunometabolism dysfunction [39][40][41][42][43].Our results provide evidence that metabolites may be involved in the development of SLE, which might provide novel insights into investigating causal role of blood metabolites in development of SLE through a comprehensive genetic pathway.To identify immune-related DEGs that contribute to SLE and COVID-19 pathogenesis, we analyzed the six cohorts to identify immune-related DEGs with significant differences (P < 0.01) between the immune_H group and immune_L group (Figure S2).We selected and identified 4 upregulated immune-related differential gene expression (DEGs) (|fold change|> 2, P < 0.01) that were significantly related to SLE and COVID-19 (Fig. 2H), among which C-X-C motif chemokine ligand 10 (CXCL10) showed a higher level of expression in patients of both groups.The CXCL10 was increased in various autoimmune diseases like SLE [44], which may also be a key regulator of the 'cytokine storm' immune response to SARS-CoV-2 infection [45].The common immune-related DEGs (CD3-TCR complex (CD3D), CXCL10, SH2 domain-containing 1A (SH2D1A), SH2 domain-containing 1B (SH2D1B)) of the six cohorts were obtained by the expression comparison of the DEGs in both groups with different immune status (Fig. 2H and Figure S2).All of these genes were upregulated in patients with COVID-19 and SLE (Table S4).

Immune-related molecular subtype and pathway characteristics of SLE and COVID-19 patients
We further analyzed the immune-related molecular subtype and pathway characteristics of the SLE and COVID-19 patients.The GSE157103, GSE161731 and GSE163151 cohorts were enrolled into one COVID-19 cohort.Based on the gene expression status, three subtypes were obtained by unsupervised clustering (Fig. 3A).The proportion of males was a clear different between subgroups I and II (Fig. 3B).The proportion of age was also significantly distinction between subgroups I and II (Fig. 3C).Subsequently, the relationship heatmap between eigengenes and clinical characteristics (Fig. 3D) and the module eigengene dendrogram (Fig. 3E) were obtained.Moreover, Gene Ontology (GO) and KEGG analyses were conducted to indicate the biological behavior among the three subgroups (Figure S3 and Fig. 3F).The different pathways associated with COVID-19 included the spliceosome, ferroptosis, tryptophan metabolism, T-cell receptor signaling, coronavirus disease-COVID-19, ribosome, malaria, EGFR tyrosine kinase inhibitor resistance, insulin signaling and cell cycle (Fig. 3F).The GSE72509, GSE49454 and GSE110169 cohorts were enrolled into one SLE cohort for comparison with the COVID-19 cohorts (Fig. 3G).There was no difference in the proportion of females between subgroups I and II (Fig. 3H).The associations between eigengenes and clinical characteristics are shown in Fig. 3I.The GO and KEGG pathways associated with SLE included coronavirus disease-COVID-19, T-cell receptor signaling, ribosome, graft-versus-host disease, ECM-receptor interaction, nitrogen metabolism and spliceosome (Figure S4 and Fig. 3J).Therefore, the different pathways associated with both SLE and COVID-19 included coronavirus disease-COVID-19, T-cell receptor signaling, ribosome, graft-versus-host disease, extracellular matrix (ECM)-receptor interaction and spliceosome.

Comparison of the single-cell immune transcription atlas of PBMCs between SLE and COVID-19 patients
To explore the transcriptional features of PBMCs from SLE and COVID-19 patients at the single-cell level, we characterized the cell and molecular mechanisms of specific disease processes.ScRNA-seq was used to identify specific immune cell subsets of COVID-19 and SLE patients.After stringent quality control and filtering by multiple criteria.Transcriptomes of 25,059 and 23,902 single cells from COVID-19 and SLE samples belonging to four single-cell datasets, namely, GSE135779 (PBMCs, SLE), GSE142016 (PBMCs, SLE), GSE155222 (PBMCs, COVID-19) and GSE166992 (PBMCs, COVID-19), were obtained, with about 17,474 and 16,559 genes per cell detected, respectively.Then, the single-cell datasets were combined for further systematic comparison between SLE and COVID-19 samples.Based on PCA, the similar expression patterns of cells were cluster by using the UMAP algorithm.The analysis results further classified the two group samples into 21 clusters (Fig. 4A, B).We then performed a comparative study of the composition of PBMC subpopulations between the SLE and COVID-19 samples, we found that some cell clusters (including cluster 2, B cells 1; cluster 11, B cells 2; cluster 17, B cells 3; cluster 14, platelet cells) were decreased in SLE patients (Fig. 4B) and that some cell subpopulations (cluster 0, CD14 + mono1 cells; cluster 9, CD8 + naive T cells; cluster 8, CD4 + memory T cells; cluster 3, CD8 + effector T cells; cluster 6, CD8 + T cells; cluster 10, CD14 + monocyte 2 cells) were increased in SLE samples (Fig. 4B).Considering the high dropout rate of scRNA-seq data, we here selected the dropClust, which is a novel algorithm for clustering and visualization of ultra-large single cell RNAseq (scRNA-seq) data [46].We classified main cells subpopulations based on various cell markers.Finally, 21 cell clusters were identified (Fig. 4C-F).The cell type-specific gene expressions were shown in the dot plot (Fig. 4C) and the UMAP plot for cell clustering visualization (Fig. 4D).No specified unique cell subset in either group was detected based on the expression level of gene markers (refer to the database of CellMarker) (Fig. 4E).The top 3 DEGs of each subset are listed in Fig. 4F.CD14 + mono1 cells, naive T cells, B cells 1, CD8 + effector T cells and CD4 + T cells were the most abundant cell types (Fig. 4G).For the SLE patients, the dominant cell clusters were CD14 + monocytes, naive T cells, CD8 + T cells, CD8 + effector T cells, CD4 + T cells, and CD4 + memory T cells.However, in COVID-19 patients, B cells 1, naive T cells, naive B1 cells, CD4 + T cells, B cells 2, and naive CD4 + T cells were the top 6 dominant cell clusters.The results suggested a significant difference in the B-cell region in the two diseases.Therefore, we selected the B-cell datasets of the two diseases to further explore the underlying differences in characteristics.
The transcriptomic features of each cluster were displayed based on gene expression levels across all PBMCs in the COVID-19 and SLE samples (Fig. 5A).The cell populations of the SLE and COVID-19 samples were also enriched in expressing CD3D, granulysin (GNLY) and CD79a molecule (CD79A) in the dot plot (Fig. 5B).As the unifying marker in COVID-19 patients, the ATP synthase (ATP5E) gene was downregulated while receptor for activated protein C kinase 1 (RACK1) gene was upregulated compared with SLE patients (Fig. 5C, D).In addition, we collected PBMC from ten SLE and COVID-19 patients respectively to verify the expression of ATP5E and RACK1.The experimental results were consistent with those observed in bioinformatic analysis (Fig. 5E).We also found several differentially expressed genes, including RACK1 and ATP synthase F1 subunit epsilon (ATP5F1E), in SLE patients compared with COVID-19 patients from various cell clusters (Fig. 5F-K and Figure S5).

Differential immune characteristics of B-cell subpopulations in SLE and COVID-19 patients
To compare the immune characteristics of B-cell subpopulations between SLE and COVID-19 samples, the recently published B-cell single-cell datasets (GSE163121 (B cells, SLE), GSE164379 (B cells, COVID-19)) were merged for further systematic comparison between SLE and COVID-19 samples.Transcriptomes of 14,685 and 13,060 single B cells from the COVID-19 and SLE patients were obtained, with 15,763 and 14,590 genes per cell detected, respectively.The analysis results classified the two group samples into 15 clusters using the UMAP plot (Fig. 6A).Compared to COVID-19, the results showed that immune-related B-cell clusters, such as cluster 4, MT + B cells; cluster 6, CD83 + B cells; and cluster 12, STAT4 + B cells, were decreased (Fig. 6A, B).These B-cell subpopulations (cluster 1, CXCR5 + B cells; cluster 5, HLA-DRB5 + B cells) in the SLE samples were increased (Fig. 6A, B).Among the B cell subpopulations, TCL1A + naive B1 cells, CXCR5 + B cells, CD80 + B cells, TCL1A + naive B2 cells, MT + B cells, HLA-DRB5 + B cells, CD83 + B cells and CCL5 + B cells were the most abundant B-cell types (Fig. 6B).For the SLE patients, the dominant cell clusters were CXCR5 + B and HLA-DRB5 + B. However, in the COVID-19 patients, MT + B cells, CD83 + B cells, CCL5 + B cells, and STAT4 + B cells were the top 4 dominant cell clusters.Based on the expression of main gene markers (refer to the Database of CellMarker), we identified 15 cell subsets, as shown in Fig. 6C-E.The dot plot shows the expression levels of marker genes of well-known cell type (Fig. 6C).There was no specified unique cell subset in either group was detected based on the expression level of gene markers (Fig. 6D), the top 3 DEGs of each subset are listed in Fig. 6E.
The cell populations of the SLE and COVID-19 samples were both enriched in the expression of the marker genes CD79A, IL7R and membrane spanning 4-domains A1(MS4A1) (Fig. 7A).As a unifying marker in SLE patients, there was higher expression of AC090498.1 and lower expression of RACK1 compared to COVID-19 patients (Fig. 7B, C).We found several differentially expressed genes, including RACK1, ATP5F1E, ATP5E and AC090498.1, in the SLE samples compared with the COVID-19 samples from various B-cell clusters (Fig. 7D-M).
To investigate the role of B cells in SLE and COVID-19, the pseudotime methods were used to simulate the differentiation trajectory of B cells.We divided the samples into multiple cell populations (states) under differentiation states according to the gene expression status, and generate an intuitive lineage development tree diagram to predict the differentiation and development track of cells.A total of 11 cell types and 12 states were identified in the SLE samples (Fig. 8A), and 15 cell types and 5 states were subsequently identified in the COVID-19 samples (Fig. 8B).Here, genes that were markedly differentially expressed along the pseudotime axis were further clustered into 11 modules according to their expression patterns in the SLE samples in a heatmap (Fig. 8C).Previous findings suggested that cluster 2 was mainly composed of CD83 + B cells, we thus used the branch of state 9 as the starting point (Fig. 8D).For COVID-19, the genes that were markedly differentially expressed along the pseudotime axis were further clustered into 15 modules according to the expression patterns (Fig. 8E).State 5 contained more B cells from the COVID-19 group compared with the SLE group (Fig. 8F).Previous findings indicated that cluster 5 was primarily composed of S100A10 + B cells, we that used the branch of state 5 as the starting point (Fig. 8F).Then, along the pseudotime axis, the expression of encoding genes was examined in the SLE and COVID-19 groups (Fig. 8G, H).In our study, the comparisons of B cells by single-cell RNAseq provided evidence that B cells in diverse infectious diseases and in systemic autoimmune diseases are highly related and conducive to exploration share common drivers of differentiation and expansion.
The above results suggested that the B cells expanded in different directions in the SLE and COVID-19 groups and that the B-cell clusters may drive different heterogeneity and cell state transitions in COVID-19 and SLE.To investigate this hypothesis, we applied CellChat functionalities to B-cell scRNA-seq datasets of COVID-19 and SLE.CellChat identifies communication patterns and predicts functions [35].Our results showed that B-cell clusters may communicate with other B-cell clusters via the MIF signaling pathway, thereby regulating cell function (Fig. 9A, B).Moreover, the MIF signaling pathway networks (Fig. 9C, D) showed not only high signaling redundancy but also high target promiscuity (for instance, B-cell clusters can act as MIF targets).Cells have unique communication modes, including senders, mediators, receivers and influencers.Complex signaling networks were dissected by explicitly assigning sender and receiver cells to distinguish the MIF signaling pathway network.It demonstrated that the MIF signaling pathway network in the COVID-19 and SLE patients is of high redundancy, and multiple ligand sources can target most of B-cell clusters (Fig. 9E, F).Further analysis of ligand-receptor demonstrated that immune signaling axes (MIF-CD74+CXCR4 and MIF-CD74+CD44) might participate in the intercellular crosstalk between B-cell subsets (Fig. 9G, H).Together, these results provide an opportunity to deeply probe underlying B-cell-cell communication with multiple ligand-receptor pairs (MIF-CD74+CXCR4, MIF-CD74+CD44) that are often deemed to drive heterogeneity and cell state transitions in COVID-19 and SLE.

Discussion
The COVID-19 pandemic has become a major concern worldwide [7].Patients with SLE, an autoimmune disease, are susceptible to COVID-19, which has attracted much attention in the context of the current pandemic.There are many similarities and differences between COVID-19 and SLE.SLE patients are at high risk of becoming infected with the coronavirus.Further examination of the molecular mechanisms in COVID-19 and SLE patients may contribute to finding better therapeutic options.Therefore, we systematically conducted a comparative analysis of SLE and COVID-19 from multiple perspectives.First, we focused on the molecular characterization of the two diseases as it relates to the immune system, and we identified the common upregulated immune-related genes CD3D, CXCL10, SH2 domain containing 1B (SH2D1B)) by comparing two immunity clusters.Among the above immune-related gene regulators, CXCL10 is a proinflammatory chemokine that is involved in COVID-19 that leads to acute respiratory distress syndrome (ARDS) [47].CXCL10 could be a key candidate gene related to the cytokine storm of COVID-19 ARDS patients [48].One study has shown that CXCL10 is upregulated in blood samples of COVID-19 patients [49].It is expected to become a therapeutic target [45].A previous study also demonstrated that CXCL10 is upregulated in PBMCs and B lymphocytes, serum and/or tissue of patients with SLE [44,50].Moreover, CXCL10 responds to IFN-γ pathway inhibition [51].Therefore, we speculated that targeting CXCL10 may be a promising strategy for treating SLE patients who also have COVID-19.
COVID-19 vaccines are urgently needed to control the pandemic, for patients with SLE, scientific vaccination helps to reduce the risk of infection in order to achieve an adequate and durable immune response [52].The association study about the association of HLA gene sets with immune status between SLE and COVID-19 could help screen potential targets for the immune system and explore the impact of the vaccine on protective immunity.Individual HLA alleles can influence the risk and the severity of viral infections.For COVID-19, these analyses may contribute to assessing the susceptibility of SARS-CoV-2 and the epidemiological level [53].Clarification of these specific differences in HLA subtype between different groups may contribute to predicting Considering that gene regulation is a complicated multisystem adjustment process, we also studied the integrated role of genes in the molecular subtypes of COVID-19 and SLE.Our study demonstrated that the signaling pathways of coronavirus disease-COVID-19, T-cell receptor signaling, ribosome, graft-versus-host disease, ECM-receptor interaction, and spliceosome were significantly activated in COVID-19 and SLE molecular subtypes.Significantly distinct pathway characteristics were observed in the different molecular subtypes, which also confirmed that the specific regulation pathways were remarkably associated with COVID-19 and SLE.Our results may contribute to facilitating the study of the relationships between molecular subtypes and especially pathways among COVID-19 and SLE.
Previous studies have revealed that SLE with high heterogeneity and complexity features [54] is very challenging regarding both precision diagnosis and treatment when SLE patients also have COVID-19.According to these findings, the examination of immune cell dysfunction in COVID-19 and SLE patients may help to discover new pathogenesis mechanisms.It is also important for the clinical management of SLE and COVID-19 patients.Therefore, it is urgent to reveal the underlying molecular mechanism by high cellular resolution strategy in SLE and COVID-19 patients.
Herein, scRNA-seq methods were used to identify the molecular signature and cellular features to dissect their role in these two diseases at the single-cell level.There are 21 and 15 predominant subpopulations of cells with UMAP clustering were identified in the PBMCs and B cells of COVID-19 and SLE patients, respectively.Notably, the features of the cells with distinct transcriptomic patterns were recognized.For the PBMCs of SLE patients, the dominant cell clusters were CD14 + monocytes, CD8 + effector T cells, CD8 + T cells, naive T cells, CD4 + T cells, and CD4 + memory T cells.However, for COVID-19 patients, unknown 1 cells, naive T cells, naive B1 cells, CD4 + T cells, unknown 2 cells, and naive CD4 + T cells were the top 6 dominant cell clusters.Moreover, the results suggested that the significant difference in B cells restored the pathogenesis of the two diseases.Thus, we selected the B-cell datasets of the two diseases to further explore the underlying differences in characteristics.Based on comparative analysis, for the SLE patients, the dominant cell clusters were CXCR5 + B and HLA-DRB5 + B cells.However, in the COVID-19 patients, MT + B cells, CD83 + B cells, CCL5 + B cells, and STAT4 + B cells were the top 4 dominant cell clusters.In general, our results demonstrated that differences in the B-cell subpopulations existed between the two groups.The differences in the B-cell subpopulations existed between the two groups showed that B-cell clusters might drive cell state heterogeneity and transitions in COVID-19 and SLE.B-cell clusters in diverse infectious diseases and in systemic autoimmune diseases are highly related, which share respective drivers of differentiation and expansion.However, B-cell clusters in different diseases are not identical and even show discrete disease-specific features.As we know, B cells can recognize the antigen (foreign body) and produce antibodies against it.B cells involves in humoral-mediated immunity or antibodymediated immunity (AMI).They fight and protect the body from the virus that enters the bloodstream [55].A better understanding of the commonality and differences in the B cell responses in these two diseases may provide critical insights into the development of vaccines that drive pathogen-specific antibody responses and avoid autoimmunity.
It is well established that B cells are one of the main immune cells with abnormal differentiation in SLE patients [56].Analyzing the transcription atlas of relevant B-cell subsets in SLE patients would facilitate new ideals to targeting pathogenic B cells.In our study, pseudotime trajectory analyses indicated that B cells may drive heterogeneity and cell state transitions in COVID-19 and SLE.A total of 11 cell types and 12 states were identified in the SLE group, and a total of 15 cell types and 5 states were subsequently identified in the COVID-19 group.Based on our findings that cluster 2 primarily consisted of CD83 + B cells, the branch of state 9 was the starting point.State 5 comprised more S100A10 + B cells from the COVID-19 group compared with the SLE group; thus, the branch of state 5 was set as the starting point.These results suggested that B cells in the SLE and COVID-19 groups were expanded in different directions.Moreover, different intimate cell-cell communications among B-cell clusters were identified in the two diseases.We also found that the high expression of receptor-ligand complexes, such as MIF-CD74+CXCR4 and MIF-CD74+CD44, may play crucial roles in SLE and COVID-19 patients.The activation of MIF-related pathways is participated in the formation of B-cell communication in the two diseases.Our findings provide novel insight to support that MIFrelated signaling pathways might potentially serve as an underlying molecular mechanism of B cells in SLE and COVID-19.Thus, this finding suggests that the B-cell populations could be the candidate targets for MIF inhibitors.In addition, we also discovered several unidentified differentially expressed genes, such as RACK1, ATP5F1E and AC090498.1, in the SLE samples compared to the COVID-19 samples from various fine-sorted B-cell clusters.The differential expression of the key genes we identified in PBMCs and B cells might facilitate our understanding of SLE and COVID-19.Through database analysis, our results showed that these genes that are related to immune cells and involved in the regulation of immune cell activity.For instances, RACK1 was identified as a novel host factor required for Zika virus (ZIKV) replication.The experiments of depletion of RACK1 demonstrated that RACK1 is important for replication of SARS-CoV-2 [57].A better understanding of the commonality and differences about these key genes in the PBMC and B cell in these two diseases might provide critical insights into the development of vaccines that drive pathogen-specific antibody responses and avoid autoimmunity.
A recent study reported that Omicron RBD memory B cell recognition was substantially reduced to 42% compared with other variants in subjects 6 months post-vaccination [58].Understanding the SARS-CoV-2 RBD-specific naive repertoire may inform potential responses capable of recognizing future SARS-CoV-2 variants or emerging coronaviruses, enabling the development of pan-coronavirus vaccines aimed at engaging protective germline response [59].Memory B cell reserves can generate protective antibodies against repeated SARS-CoV-2 infections, but with unknown reach from original infection to antigenically drifted variants.Although emerging SARS-CoV-2 variants of concern escape binding by many members of the groups associated with the most potent neutralizing activity, some antibodies in each of those groups retain affinity, suggesting that otherwise redundant components of a primary immune response are important for durable protection from evolving pathogens [60].Scheid JF et al. found that the SARS-CoV-2-specific B cell repertoire consists of transcriptionally distinct B cell populations with cells producing potently neutralizing antibodies (nAbs) localized in two clusters that resemble memory and activated B cells [61].Together, our results characterize transcriptional differences among SARS-CoV-2-specific and SLE B cells and would provide reference for immunogen and therapeutic design against coronaviruses.
Several limitations are existed in our study, and more comprehensive studies are needed.First, we used data from different studies based on the GEO database, which might cause unavoidable heterogeneity.In addition, population heterogeneity is also important to be considered before conducting any analysis on the data.The parameters such as age, gender, race, activity of disease and medication history can significantly impact the results of any analysis.For example, if the dataset includes individuals from different age groups, the results of the analysis may not be representative of any particular age group.Similarly, if there are significant differences in medication history or disease activity between individuals, it may be difficult to compare and combine their data.Therefore, it is important to ensure that the data being analyzed is consistent in terms of these important parameters in order to have roughly mergeable data and accurate results.Secondly, the exact molecular mechanisms of the genetic signatures need to be further identified by experiments and clinical practice in the future.Further explores are urged to confirm the functions of these DEGs in SLE and COVID-19.Our primary objective was to discuss the multifaceted impacts on the COVID-19 pandemic for SLE patients in transcriptomic and single-cell analyses.These theoretical perspectives lay a foundation for designing specific and effective therapeutic methods for fighting SARS-CoV-2.

Conclusions
In summary, we analyzed the difference in COVID-19 and SLE from multiple dimensions.The systematic evaluation of characteristic differences between different subgroups in our study makes sense for the heterogeneity and treatment complexity of the two diseases.While the exact role of CXCL10 in SLE and SARS-CoV-2 infection is still being investigated, evidence suggests that it may play a significant role by affecting immune system and could be a potential target for therapeutic interventions.Immune cell disorder appears in COVID-19 and SLE patients at the single-cell level and in blood tests.The significantly upregulated and downregulated genes in PBMCs and B-cell clusters of SLE and COVID-19 patients mainly included RACK1, ATP5E and AC090498.1.Moreover, we provided an opportunity to comprehensively probe underlying B-cell-cell communication with multiple ligand-receptor pairs (MIF-CD74+CD44, MIF-CD74+CXCR4) and the differentiation trajectory of B-cell clusters that can drive cell state heterogeneity and transitions in COVID-19 and SLE.Therefore, this study revealed the cellular and molecular associations of COVID-19 with SLE, especially for the synergistic complexities of COVID-19 and SLE, which might provide clues for personalized immunotherapy for SLE patients infected with COVID-19 in the future.

Fig. 1
Fig. 1 Identification of immunological characteristics and immunerelated DEGs in GSE157103.A Unsupervised clustering of immune cell infiltration between the immune_H and immune_L subtypes (blue: low expression; red: high expression).The upper columns include ESTIMATEScore, ImmuneScore, StromalScore, and Subtype.B TME scores between the two groups.C Expression of HLA genes between both groups.D Immune cell fraction between both groups.E The overlapped genes between DEGs and immunity-related

Fig. 2
Fig. 2 Identification of immunological characteristics and immunityrelated DEGs in GSE72509.A Unsupervised clustering of immune cell infiltration between the immune_H and immune_L subtypes (the red represents high expression and the blue represents low expression).The upper columns include ESTIMATEScore, ImmuneScore, StromalScore, and Subtype.B TME score between both groups.C Expression of HLA genes between both groups.D Immune cell fraction between both groups.E The overlapping genes between DEGs and immune-related genes.F Heatmap of the correlation between

Fig. 3
Fig. 3 Identification of molecular subtype characteristics in the COVID-19 and SLE cohorts.A PCA was performed to distinguish the GEO cohorts of COVID-19 (GSE157103, GSE161731 and GSE163151).B, C Differences in the proportion of males (B) and age (C) between 3 subgroups of COVID-19 cohorts (Wilcoxon test) are shown.D Gene dendrogram revealing the 13 modules identified by average linkage hierarchical clustering by WGCNA.E Heatmap of the correlation between the clinical characteristics and module eigengenes.Red: strongly correlated modules (|Cor|> 0.5, P < 0.05); Blue: weakly correlated modules (|Cor|< 0.5, P < 0.05).F Biological pathways in distinct subgroups of COVID-19 was shown in the heatmap.The upper columns consist of subtype and term.The k = 3 was

Fig. 4
Fig. 4 Single-cell immune transcription atlas of PBMCs in SLE and COVID-19 samples.A UMAP plot of 48,961 single cells from PBMCs.Each cell type is highlighted with a different color.B UMAP plot of SLE and COVID-19 groups.C Expression of selective marker genes for cell clusters and the cell positions are in the UMAP plot of panels.D UMAP plot of 48,961 single cells from PBMCs.Each cell type is highlighted with a different color.E Dot plot of cell speci-

Fig. 5
Fig. 5 Transcriptomic features of each cluster.A The scatter plot displays gene expression across all PBMCs in COVID-19 vs. SLE samples.B UMAP plots of classical cell markers in cell clusters and the color indicated expression levels (gray:low, red: high).C, D The expression levels of representative signatures in 21 cell subtypes.Violin plots show the differential expression of RACK1 and ATP5E genes in each cluster of cells from SLE and COVID-19 patients.E

Fig. 6
Fig. 6 Single-cell transcription atlas of B cells in SLE and COVID-19.A UMAP plot of 27,745 single cells from all B cells.Different cell types are highlighted with different color.B The cell counts and percent of each cluster were detected in B cells of the SLE and COVID-19 groups.C Expression levels of the selected marker genes

Fig. 7
Fig. 7 Transcriptomic features of each B-cell cluster.A UMAP plots of cell clusters labeling with cell markers, color represented expression levels (gray: low, red: high).B, C The expression levels of representative signatures in the two subtypes.D-M Volcano plots showing the differentially expressed genes of B-cell clusters in COVID-19

Fig. 8
Fig. 8 Pseudotime trajectory analysis of B cells among SLE and COVID-19 patients based on Monocle.A, B Potential differentiation routines among SLE and COVID-19.C Heatmap shows that the genes were markedly differentially expressed along the pseudotime axis in the SLE samples.These genes were further clustered into 11 modules according to the expression patterns.D Pseudotemporal expression dynamics of specific representative genes in the SLE

Fig. 9
Fig. 9 Visualization and analysis of B-cell-cell communication of COVID-19 and SLE.A Circle plot: SLE.B Circle plot: COVID-19.C, D Hierarchical plot of cell-cell communication network about the MIF signaling pathway.Solid circles indicate the source and open circles indicate the target.E, F Heatmap showing the relative impor-