Single-cell RNA transcriptomics divulged altered gene expressions in monocytes and HLA variations in decompensated cirrhosis patients with sepsis Authors: Rashi Sehgal, Pramod Gautam, Mojahidul Islam, Jayesh Kumar Sevak, Maryam Javed, Navkiran Kaur, Gayatri Ramakrishna, Rakhi Maiwall, Shiv Kumar Sarin, Nirupama Trehanpati Published: 2025 ABSTRACT AbstractImmune dysfunction and systemic inflammation are hallmarks of decompensated liver cirrhosis (DLC), associated with an increased risk of developing sepsis. Rapid immune degradation is poorly understood in DLC patients; therefore we performed single-cell RNA (scRNA) transcriptomics in DLC patients. Twenty-one DLC patients (all males, 42 ± 7 year) with sepsis (n = 10) and without sepsis (w/o, n = 11) and ten healthy controls PBMCs, were analysed for scRNA transcriptomics using the BD Rhapsody. Cell clustering and cell types were determined using gene expression data and validation of specific genes was done by qRT-PCR. Retrospectively analyzed proteomics data from same patients was used for RNA-Protein interactions. Ten clusters and 7 cell types were detected, with high heterogeneity in the monocyte cluster in DLC patients. All DLC patients irrespective of sepsis showed down regulation of O6-methylguanine-DNA methyltransferase (MGMT) mediated DNA damage reversal, mitochondrial transcription termination, and melanin biosynthesis, with upregulated lactose synthesis, hydroxycarboxylic acid, FGFR1b and FGFR1c receptors. DLC-sepsis showed down regulation of NF-kβ, TNF-α, and IL-17 signaling in classical monocytes compared to w/o sepsis. Serotonin and DNA repair genes were significantly increased in DLC sepsis (P < 0.05). In addition to HLA-DR, HLA-A, HLA-B, and HLA-DQA1 were also decreased but HLA-DRB1, HLA-C, HLAE, HLA-DRA, and HLA-DQA2 were raised in sepsis. RNA-protein interaction revealed the down regulation of PROS1, CDC42, CD62L, FCGR3B, CX3CR1, Rpl31 genes in sepsis. Chances of sepsis increased in DLC patients due to monocytic defects in genes and signaling pathways. These markers further can be explored for diagnostic purposes and early detection of sepsis.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-89084-z.Keywords: Single cell RNA transcriptomic, Immune modulation, Liver cirrhosis, Sepsis, BD rhapsody, RNA-Protein interaction, MonocytesSubject terms: Medical research, Translational research Introduction Patients with cirrhosis with immunological dysfunction are more vulnerable to infections and sepsis, which lead to increased morbidity and mortality1,2. There are approximately 19 million sepsis patients worldwide, with a 25–30% death rate that rises to 50% in the context of septic shock3. The primary cause of sepsis is a malfunctioning immune response that includes neutrophils, monocytes, dendritic cells, lymphocytes, and cytokine storm3,4. Despite an era of significant research and multiple clinical trials, therapies for sepsis patients remain restricted and involve supportive care with invasive mechanical ventilation and efforts to reverse the underlying etiology4. Immunoparalysis in sepsis patients is associated with altered cellular and tissue metabolism5. The question is whether the cytokine storm impacts immune cells or whether immune cells cause the cytokine storm, resulting in immunoparalysis. In our previous work6, we found a higher number of neutrophils but with poor functioning and reduced HLA-DR expression on monocytes, particularly intermediate monocytes, in decompensated liver cirrhosis patients with sepsis. Other studies proposed that in sepsis, circulating immature neutrophils differentiate into monocytic cells7. Furthermore, in sepsis patients, monocytes exhibited decreased expression of inflammatory response genes while increased expression of apoptosis-related genes8. Though we detected decreased CD4 T-cells in sepsis2, a clinical retrospective analysis of 28 days revealed that survivors had a higher lymphocyte count than non-survivors3 and that IFN-I activated B cells had a role in bacterial sepsis9. Apart from these immune cells, MDSCs were discovered to be elevated in human and animal models of sepsis, with the ability to inhibit the immunological response2,9. Patients with complicated courses showed a mix of hyper-inflammatory and immunosuppressive transcriptional alterations, consistent with long-term immune dysregulation10,11. As a result, both the hyper-inflammatory and immune suppression phases are being targeted in sepsis patients to improve immunity during sepsis and to reduce the immunocompromised phase in septic shock11. Despite so much research on sepsis, the genetic and cellular foundation of fast immunological degradation and sepsis-induced systemic immune dysregulation in DLC patients is only partially understood. Current innovations in single-cell RNA-seq (scRNA-seq) allow the identification of immune subsets and their relevance in health and illness12–15. By harnessing single-cell RNA-seq technology in PBMCs, we have elucidated the transcriptional changes in specific immune cells in DLC patients with or without sepsis. scRNA transcriptomics data identified seven cell types and subtypes of monocytes based on the expression of several canonical markers. Our study found changes in cell composition between DLC patients with and without sepsis, as well as differences in monocyte clusters with distinct HLA allele expression in sepsis. Differential gene expressions also revealed distinct altered pathways in sepsis patients. These markers can be used in future as diagnostic markers and for therapeutic purposes. Methods Study group Total twenty-one DLC patients with sepsis (n = 10) and without sepsis (n = 11) (age 42 ± 7 year) and age matched healthy controls were recruited in this study at the Institute of liver and biliary sciences (ILBS), New Delhi, India. This study was approved by the research and institutional ethics committee with IEC No IEC/2016/45/NA/C2 and informed consent was obtained from all the subjects enrolled in the study. Patients were subjected to scRNA transcriptomics (Fig. S1). This work was done in accordance with the 1975 declaration of Helsinki. DLC patients were classified into sepsis and without sepsis according to the following inclusion criteria. DLC with sepsis were diagnosed based on the presence of SIRS, SOFA (sequential organ failure assessment) score with an infection, confirmed either by cultures or by imaging. Following, four criteria for the diagnosis of SIRS were used16: (a) Temperature > 38.0 °C or < 36.0 °C (b) TLC (Total leukocyte count) < 4 or > 12 × 109/l (c) Pulse > 90 beats/minutes (d) RR (Respiratory rate) > 20 or < 32 beats/minutes. DLC without (w/o) sepsis was diagnosed when there was no evidence of SIRS, or infection on any of the cultures and ultrasonography. Patients with viral hepatitis infection (HEV, HAV, HBV, HCV, and HIV), with HCC, any other malignancy and co-morbidities, and no consent given were excluded from the study. Single-cell RNA transcriptomics scRNA transcriptomics was done using BD Rhapsody (BD Biosciences, #633701) as per the manufacturer protocol followed by sequencing. In brief, 10–15 mL of peripheral blood was collected from patients at the time of recruitment. Plasma devoid blood was used for isolation of peripheral blood mononuclear cells (PBMCs). Isolated PBMCs were cryopreserved and later used for scRNA experiments. Cryopreserved PBMCs were thawed at 37 0C until a tiny ice crystal was left in the tube, and then carefully transferred in 5-10mL of pre-warmed complete RPMI (RPMI 1640 - GIBCO, #22400-089) with 10% FBS (GIBCO, #10270-106),1% penicillin-streptomycin (HIMEDIA, #A001) in 15 ml conical tube. Further, cells were centrifuged at 250 xg for 5 minutes and supernatant was discarded. An additional 10 mL of pre-warmed complete RPMI were added drop by drop, followed by centrifugation for 5 minutes at 400 xg and resuspension in 5 mL of complete RPMI. Cells counting was done using trypan blue to ascertain cell viability. 5 × 105 live cells per sample was used for transcriptomic experiment. Each sample was tagged with a barcode using BD Single-Cell Multiplexing Kit (BD Biosciences, Cat. No. 633781) and washed twice in FACS buffer and resuspended in cold BD Sample Buffer (BD Biosciences). Samples from the same group were pooled in equal amounts to achieve approximately 15-25000 cells in 620 µl of sample buffer and loaded onto a BD Rhapsody cartridge for an incubation of 20 min at room temperature. Then, cell capture Beads (BD Biosciences) were added to the cartridge and incubated at room temperature for 3 min, and thereafter cartridges were washed. Cells were lysed, and the released mRNAs were captured by cell capture beads. mRNAs were then retrieved, and washed before performing reverse transcription and treatment with Exonuclease I. Further, cDNA libraries were prepared by using the BD Rhapsody Whole Transcriptome Analysis Amplification kit. Quality of final libraries was assessed by using Agilent 2100 Bioanalyzer (Agilent Technologies) and quantified using a Qubit Fluorometer using the Qubit dsDNA HS Kit (ThermoFisher, #Q32854). Final libraries were diluted and multiplexed for paired end (150 bp) sequencing on a NovaSeq 6000 sequencer (Illumina) using version 1.5. Quality control of raw sequencing data Post sequencing, raw bcl files obtained from all samples were demultiplexed to fastq files using bcl2fastq. The paired-end fastq files were then pre-processed using locally installed seven bridges WTA (Whole Transcriptome Analysis) pipeline using CWL-runner (developed by BD Rhapsody) resulting in normalized combined count matrix. The whole transcriptome analysis (WTA) pipeline takes the FASTQ files, a reference genome (fasta file), a transcriptome annotation file (GTF file). Briefly, the paired end reads stored in Fastq format after demultiplexing of bcl files were filtered for quality. Both the paired-end reads were filtered for high sequence-quality reads with dropping the too short (less than 64 bases for R2) reads or reads with less than 20 base quality score. Valid read-pairs then collapse to generate raw molecules. The reads with the same cell label, same UMI sequence, and same gene collapsed into a single raw molecule and the alignment was done with STAR index reference transcriptome (GRCh38). The artifact molecules generated due to UMI or PCR errors were removed using recursive substitution error correction (RSEC). The putative cells were then determined employing a two-step algorithm for filtering cell labels. The sample of origin for each putative cell was then determined using sample tags. For sample tag allocation, a cell is called a singlet if the minimum read count for a given sample tag was reached, and more than 75% of the sample tag reads were derived from a single sample tag. Both multiplets and undetermined cells were excluded to generate the single cell gene expression matrix containing normalized molecule counts per cell. The downstream analysis steps for the resulting gene expression count data were performed using the R software package Seurat17, including de-multiplexing, quality control, normalization, dimension reduction, clustering and differential gene testing and cell type identification. Quality control, filtering of count matrix The experimental count matrix from our experiment (sepsis and without sepsis) was first combined with the publicly available data (https://www.10xgenomics.com/resources/datasets/3-k-pbm-cs-from-a-healthy-donor-1-standard-1-1-0) from PBMCs of a healthy donor Sequenced on Illumina NextSeq 500. Annotated data from cell count using EnsDb.Hsapiens.v94 was used. Quality control and filtering of the raw count data Quality metrics related to percentage of mitochondrial and ribosomal genes per cell were calculated. Standard detection-based filtering approach was performed to filter cells with low amounts of reads as well as genes present in at least a certain number of cells. Cells with at least 200 detected genes were considered which need to be expressed in at least 3 cells. Cells with high proportion of mitochondrial and low proportion (lesser than 5%) of ribosomal reads were removed and data was further regressed out for ribosomal genes during scaling. Dimensionality reduction Using the QC-filtered dataset, important features/genes were selected to distinguish cell types. For this purpose, the top 2000 genes were selected to provide a good separation of the cell clusters. Next, the PCA was performed on 50 PCs, additionally, we also used regression to remove any unwanted sources of variation from the dataset, such as cell cycle (cell cycle score was calculated based on the difference of mean expression of the given genes and the mean expression of reference genes), sequencing depth, fraction of mitochondrial genes. The data was corrected for batch-effects using integration modality in Seurat workflow. Cell embeddings were visualized using UMAP as it arguably preserves the global structure with superior run time performance and no computational restrictions on embedding dimensions. Batch correction and clustering Standard pre-processing (log-normalization) was performed for identification of variable features individually for each dataset based on a variance stabilizing transformation (“vst”). Additionally, K-means and Hierarchical clustering were performed, and differentially expressed genes based on multiple functionalities were identified. Gene set enrichment analysis (GSEA) and cell-type prediction Based on differentially expressed genes, their combined function was checked using a hypergeometric test using Reactome pathways database source. Cell-types on cluster level were assigned using gene set enrichment among the DEGs with known markers for different cell-types. For this, top DEGs are extracted from the dataset and checked for top DEGs enrichment in the reference MsigDB dataset containing cell type gene sets. DEG between each group condition in each cell-type was then identified. Integrated analysis of differentially expressed RNAs from scRNA sequencing and differentially expressed proteins from MS proteomics data The differentially expressed RNAs and proteins of sepsis and w/o sepsis groups monocytes were used for integrated analysis. A total of 440 Uniprot IDs for DE proteins and 273 DE RNA were selected based on significant p value (< 0.05) and logFC of 1.5. The matched entries were searched from BIOGRID database for subnetwork search. Then enrichment analyses on the identified subnetworks were performed identifying enriched pathways. Then enrichment analysis was done using Reactome database (v2022). A similar analysis approach was repeated for protein data. The functionally similar terms/pathways were clustered together. We combined the active-subnetwork-oriented enrichment analysis results from both the datasets to further investigate common and distinct terms between them. The term-gene network graphs were made combining the enriched functional terms from RNA and protein datasets which depicts Up/Downregulated genes/proteins and common functional terms. The analysis was done using the pathfindR package (version 2.3.0.9000)18. Monocyte isolation and qRT-PCR analysis Total monocytes were isolated from whole blood using an adherent method followed by CD14+CD16+ staining. Total RNA was extracted from the isolated monocytes using TRIzol reagent (Thermo Fisher) according to the manufacturer’s protocol and further quantified using NanoDrop 2000 (ThermoScientifc). 2 µg of total RNA was used to synthesize cDNA using with high-capacity cDNA kit (ABI, Thermo Fisher Scientific, Waltham, MA, USA; #4368813). For analysing the quantitative expression of the genes, the cDNA prepared was mixed with SYBR Green Master Mix (Thermo Fisher, #A25742) (Table S1; Gene primer list) and the reactions were carried out in a CFX-96 Biorad thermal cycler (Biorad, CA, USA). Statistical analysis Data was analyzed using the statistical software Prism (version 6; GraphPad Software, San Diego, CA) and SPSS version 22 (IBM Corp Ltd. Armonk NY. USA). Comparison for continuous data is carried by using One-way ANOVA/ Kruskal–Wallis test followed by probability adjustment by the Mann–Whitney test and it is represented as mean ± standard deviation (SD). The p-values < 0.05 were considered as statistically significant. Results Single-cell transcriptomic profiling of PBMCs Clinical and biochemical characteristics of all DLC patients with sepsis (n = 11, age 44 (29–60) years); w/o sepsis (n = 10, age 48 (22–62) years) were listed in Table S2. Sepsis patients showed a significantly higher levels of total bilirubin, AST levels, INR, PCT, lactate, MELD Na, and creatinine than those w/o sepsis. These patients had lymphopenia with increased neutrophils but decreased monocytes in sepsis patients (our published data)6. scRNA transcriptomics assay was performed using BD Rhapsody protocol for scRNA sequencing (Fig. 1A), which was further analyzed using R Seurat package version 4.1 (Fig. S1, S2). Percentages of mitochondrial and ribosomal genes per cell were calculated for quality related metrics. Cells with high proportion of mitochondrial and low proportion of ribosomal reads were removed and further regressed out the ribosomal genes data during scaling. Further, the number of genes per sample was calculated post quality check (QC) (Fig. 1B–C) and quality filters were applied on the raw count matrix for case and control to get the final median number of genes/cells in the study groups for further analysis (Fig. 1C). Quality control of scRNA sequencing data. (A) Graphical representation of a scRNA sequencing data in PBMCs of sepsis and without sepsis patients (B) Violin plots depicting nfeature and count RNA, percentage mitochondria, ribosomal, hemoglobin and platelet genes (C) Bar diagram showing median number of genes / cells. Identification of clusters and cell types in DLC patients UMAP plots were used to demonstrate dimensionality reduction and clustering, identifying 10 distinct clusters and 7 distinct cell types with differentially expressed genes (DEGs) between the groups. (Fig. 2A). We found mainly 7 types of immune cells including non-classical (nc) monocytes, classical (c) monocytes, cDCs, CD4 T-cells, CD8 T-cells, NK cells and B-cells in the study groups (Fig. 2B). Classical canonical markers were used to identify each cell types like CD3, CD4, CD8A and CD8B for T cells; CD14, FCGR3A, FCGR3B, HLA-DRA, HLA-DRB1 and HLA-DRB5 for monocytes; LYZ and S100A8 identified activated monocytes and neutrophils; CD19 and MS4A1 for B-cells; NKG7 and GNLY for identifying NK cells population (Fig. 2B,C). A heatmap was produced for each cell type using the top 50 DEGs (Fig. 2D). Clusters 0, 1 and 4 of heatmap represented activated CD4 T-cells and cytotoxic CD8 T-cells and their effector functions; Cluster 2 represented B-cell development and its differentiation; Clusters 3 and 5 represented monocyte development and non-classical monocyte activity; Cluster 6 represented activated NK cells with cytotoxic activity and their memory function while Cluster 9 represented mDCs and cDC2 functions. However, clusters 8, 10, and 11 were not specified with any specific cell type (Fig. 2D). (A, B) UMAP plots representing the distribution of 11 clusters and 7 cell types in three groups – healthy control, sepsis and without sepsis patients (C) Identification of precise cell type using canonical markers for each cell type (D) Heatmap represents the top 50 genes in 7 cell types of clusters. Identification of altered pathways using transcriptomic signatures in DLC patients with and without sepsis According to the DEGs, multiple pathways were downregulated in DLC patients with sepsis including NTF4 activated NTRK2 signaling; FMO oxidizes nucleophiles, fatty acids and COX reactions compared to w/o sepsis patients and healthy. While all DLC patients, regardless of sepsis, showed downregulation of O6-Methylguanine-DNA Methyltransferase (MGMT)-mediated DNA damage reversal, mitochondrial transcription termination and melanin biosynthesis compared to HC. While Lactose synthesis, hydroxycarboxylic-acid binding receptors, FGFR1b and FGFR1c ligand binding and activation were significantly increased in all DLC patients (sepsis and w/o sepsis) compared to HC (Fig. 3A). Heatmap represents the significantly enriched pathways after Gene Set Enrichment Analysis (A) healthy control, sepsis and without sepsis patients (B) significantly enriched pathways in immune cells in all three groups (C) mRNA expression of specific genes i.e. 5-HRT3A, TREX-1, HRH-1 and HRH-2 in all subjects. P value was calculated using t- test and p < 0.05 was considered as significant. Heatmaps were generated in R software (ver. 4.2.3, URL: https://www.R-project.org/) using package GSVA R package from Bioconductor repository (ver. 1.46.0) depicting significantly enriched pathways in compared categories47,48. Furthermore, gene set enrichment analysis (GSEA) was carried out in each cell type. The majority of the cells (NK, B, T cells, cMono, ncMono) showed enhanced downregulation of acetylcholine, alanine metabolism, FGCR1c-Klotho binding ability and FGR3b ligand binding and activation. Particularly, NK cells and ncMono showed decrease in FGFR1c and FGR3b ligand binding and activation pathway while regulation of TREX1, lactose synthesis and serotonin metabolism were significantly increased in ncMono, cMono, cDCs, and B-cells (Fig. 3B). The Heatmaps shown in Fig. 3A,B were generated in R software (ver. 4.2.3, URL: https://www.R-project.org/) using package GSVA R package from Bioconductor repository (ver. 1.46.0) depicting significantly enriched pathways in compared categories. We observed that histamine receptors i.e., HRH-1 and HRH-2; serotonin receptors i.e., 5-HTR3A; DNA repair gene i.e. TREX-1 genes were associated with the activity and functions of monocytes. Further validation was performed using qRT-PCR in sorted monocytes and it was found that TREX-1, HRH-1 and HRH-2 expression was significantly increased in sepsis and w/o sepsis patients compared to HC. However, serotonin receptor 5-HTR3A was significantly enhanced in sepsis compared to HC (Fig. 3C). Single-cell RNA transcriptomics revealed variations in monocyte subsets in sepsis As monocytes play a significant role in development of sepsis in DLC patients19,20, therefore we have specifically analyzed DEGs in monocytes including c-mono and nc-mono. Our study found that top 25 genes associated with NF-kB, TNF-α and IL-17 signaling pathways were down regulated in classical monocytes of sepsis patients compared to w/o sepsis group (Fig. 4A,B and Table S3). The Heatmap in Fig. 4A was plotted using Seurat (ver. 4.3.0) function “DoHeatmap()” depicting topmost 25 genes differentially expressed between Sepsis vs. w/o Sepsis group. Further analysis revealed that expression of FGFR2, FGFR3, FGFR3c and FGFR4 were downregulated in w/o sepsis compared to HC, which showed augmented expression in sepsis. Differentially expressed genes in classical monocytes (cmonocytes). (A) Heatmap represents the top 25 up and down DEGs in classical monocytes (cMono) in sepsis and w/o sepsis patients; (B) Bar plot represents the significantly enriched reactome pathways for up and down DGEs in sepsis cMono. Differential gene expression was performed using the Seurat pipeline with Wilcoxon rank sum test, p value < 0.05 was considered as significant. Heatmap was plotted using Seurat (ver. 4.3.0) function “DoHeatmap()” depicting topmost 25 genes differentially expressed between Sepsis vs. w/o Sepsis group. In addition to cMono and ncMono, other cell types such as B-cells, NK cells, CD8 T- and CD4 T-cells were also observed inour study groups. The top 25 differentially expressed up and down regulated genes of each cells type in sepsis patients provided information regarding their linked up and down-regulated pathways (Figs. S3–S6). Many published studies found attenuated response of HLA-DR expression on monocytes in DLC patients with or without sepsis (6, 20 (Fig. S7). As a result, the primaryfocus was on HLA expression on combined both classical (c) and non-classical (nc) monocyte population. Unique expression of HLA subtypes in sepsis monocytes Our scRNA transcriptomics data identified 20 HLA subsets (Fig. 5A). HLA-A, HLA-B, HLA-C, HLA-DPA1, HLA-DPB1, HLA-DRA and HLA-DRB1 had upregulated expression while HLA-DMA, HLA-DMB, HLA-DOA, HLA-DOB, HLA-DQA1, HLA-DQA2, HLA-DOB1, HLA-DOB2, HLA-G and HLA-DOB1-AS1 showed decreased expression in all study groups (Fig. S8). (A) UMAP plot showing the cell types and focusing on classical (blue) and non-classical (purple) cluster; (B) Expression of different HLA subtypes in classical and non-classical monocytes in healthy control, w/o sepsis and sepsis (C) Violin plot of different HLA subtypes in healthy control, w/o sepsis and sepsis. Further analysis found that overall expression of HLA-C, HLA-E, HLA-DRA and DRB1 were augmented in sepsis and w/o sepsis compared to healthy. HLA-DRB1, HLA-E and HLA-DQB1-AS1 were enhanced in w/o sepsis compared to sepsis and HC. Whereas expression of HLA-C and HLA-DMA were enhanced in sepsis compared to healthy and w/o sepsis but HLA-A, HLA-B and HLA-DQA1 were reduced in sepsis (Fig. 5B). Violin plots of HLA sub-types stated above clearly showed a significant decrease in the expression of HLA_DRB1 (p = 0.042), HLA-DRB5 (p = 0.0037), HLA-F (p = 0.0014) and HLA-DQB1 (p = 0.012) in sepsis compared to without sepsis patients. While both sepsis and without sepsis patients showed significant increase in HLA-DQB1, HLA-A, HLA-B, HLA-C, HLA-E compared to HC (Fig. 5C). HLA-DRB5, HLA-DPA1, and HLA-DPB1 were highly expressed in healthy individuals compared to all DLC patients (sepsis and without sepsis), although HLA-DRA was elevated compared to without sepsis. RNA-Protein networking revealed genes responsible for Monocytes activity To uncover the common genes of interest that are competent in restoring cMono activity in DLC patients, both differentially expressed RNA and protein in sepsis and w/o sepsis were further integrated. The RNA-Protein network revealed up- and down-regulated genes involved in multiple processes. PROS1, CDC42, CD62L, FCGR3B, CX3CR1, RPL31, PSMC1, INPP5D, SELL, SDC4, RPS25, PPIA, RPS2, MDH2, COX5B, XBP1, SIRPA, IGKV3-20, CS, SDHA, CYCS, ATF4, RXRA, GSK3B, PKLR, TRIM25, EP300 and MAP1LC3B were downregulated while APOB, IGKV1-5, CALR, and IGHA1 were up-regulated (Fig. 6 and Table S4). The immunological relevance of some of the above down- and up-regulated genes is described in Table S5. Whereas enrichment of protein and RNA data using reactome analysis involving top 5 clusters revealed that binding and uptake of ligands by Scavenger receptors was highly enriched at protein level but RAG1-3, GTPase cycle, VEGFA-VEGFR2 pathway and TNF signaling were strongly enriched at the RNA level (Fig. S9–12). RNA-Protein networking in classical monocytes Common network of protein and RNA data of classical monocytes. Integration of protein and RNA was performed after selecting differentially express proteins/ RNA with p < 0.05. Discussion Single-cell transcriptomics sequencing has lately expanded rapidly due to its broad application across multiple domains since it delivers an accurate depiction of every cell type and its associated gene expression patterns21–23. In the current study, we analyzed the biological characteristics of peripheral blood mononuclear cells (PBMCs) in decompensated liver cirrhosis patients with or without sepsis. Using single-cell RNA sequencing (scRNA-seq), we attempted a unique transcriptome study of circulating immune cells that cause sepsis in patients with decompensated liver cirrhosis. In our scRNA transcriptome analysis revealed 7 distinct cell types and 11 clusters, including B-cells, CD4 T-cells, CD8 T-cells, cDC, monocytes, and NK cells. These findings support the significance of these cells in sepsis and are consistent with previous research by Wen et al., 202024, who examined 6 distinct cell types in the peripheral blood monolayer cells (PBMCs) of sepsis patients: T-cells, monocytes, B-cells, NK cells, DCs, and NKT cells. They discovered that NEAT1, SELL, CD163, and CCL3L1 were among the few monocyte markers present in PBMCs24. Reyes et al., 2020, on the other hand, identified subgroups of monocytes according to different markers, such as inflammatory cytokines (MS4), non-classical CD16hi monocytes that resembled class II MHC, and class II MHC (MS2)25. In our study, we distinguished between classical and non-classical monocytes and their associated genes because prior studies on monocytes found a decrease in both classical and non-classical monocytes in sepsis patients. In sepsis, we observed reduced melanin synthesis, termination of mitochondrial transcription, and MGMT-mediated repair of DNA damage. These genes revealed critical pathways in sepsis. The enzyme MGMT, commonly known as a DNA “suicide” repair enzyme, is predominantly active in the liver. It protects cell DNA against cytotoxic or mutagenic agents that may cause O6-guanine alkylation26. Apart from classical DCs (cDC), it was shown that the majority of cells (NK, B, T cells, cMono, ncmono) had increased downregulation of acetylcholine, alanine metabolism, FGCR1c-Klotho binding capacity, and FGR3b ligand binding and activation. FGCR1c and FGR3b ligand binding and activation pathway was reduced, particularly in NK cells and ncMono, but TREX1, lactose synthesis, and serotonin metabolism were markedly upregulated in ncMono, cMono, cDCs, and B-cells. Many immune cells including monocytes express serotonin receptors (5-HTR2A and 5-HTR3A) and their activation helps serotonin in regulating the cytokine secretion such as TNF-α and IL-1β27. Validation of specific genes such as TREX-1, 5-HTR3A, HRH1 and HRH2 showed enhanced expression in sepsis patients, and these can be used as markers for the monocytes activity. The primary source of STING-mediated IFN-α production is monocytes, whereas TREX1 is recognized as the gatekeeper enzyme of the cGAS-STING pathway28,29. Downregulated genes in sepsis patients were linked to several pathways, including the TNF and NF-kB signaling pathways30. Previous research suggests that alterations in monocyte NF-κB signaling may contribute to both inflammation and aging. However, TNF-α receptors exhibit both pro- and anti-inflammatory activity in monocytes, allowing for selective inhibition of TNFR130. TNF-α production by human monocytes activates Raf-1/MEK1-MEK2/ERK1-ERK2 Pathway, even in response to LPS20. Decompensated liver cirrhosis patients with sepsis had decreased HLA-DR compared to w/o sepsis and HC. This study aimed to clarify HLA subtype expression in DLC patients, as the relevance of HLA-DR has been investigated in decompensated liver cirrhosis. We observed 20 HLA subtypes were differently expressed on monocytes in patients and healthy controls. Sepsis patients had higher expressions of HLA-C and HLA-DMA, whereas lower expressions of HLA-A, HLA-B, and HLA-DQA1. According to the violin plots, HLA subtypes i.e. HLA-DRB1, HLA-DRB5, HLA-F and HLA-DQB1 were significantly decreased in sepsis patients compared to without sepsis patients. Patients’ outcomes with HCV were influenced by HLA-C’s interactions with NK inhibitory receptor genes1. HLA-B*57 was linked to higher mortality in HIV/HCV co-infected individuals with an increased risk of mortality from bacterial infections and sepsis31, whereas HLA-A*31 was recognized as the marker for the genetic predisposition to sepsis32. Enrichment analysis in RNA and protein networking revealed enriched binding and uptake by Scavenger receptors ligands at protein level but VEGF-VEGFR2 pathway and TNF signaling were highly enriched at RNA level. These pathways are associated with a few of the significant genes that were previously identified in pathways related to the monocyte functions and activity. Further, RNA protein work showed down-regulated genes like PROS1, PSMC1, CDC42, RPL31, INPP5D, SELL, SDC4, RPS25, PPIA, RPS2, MDH2, COX5B, XBP1, SIRPA, IGKV3-20, CS, SDHA, CYCS, ATF4, RXRA, GSK3B, PKLR, TRIM25, EP300 and MAP1LC3B while up-regulated genes were APOB, IGKV1-5, CALR, and IGHA1. Out of these above genes, many are already known for their association with LPS or sepsis such as PROS1, GSK-3B, COX5B, TRIM25, SDC4 and CALR. From down-regulated genes especially PROS1 and Mer/Tyro3 Receptor Tyrosine Kinase Axis activate human monocyte growth in response to LPS33. Apart from this, they show positive correlation with neutrophils activity and number along with the oxidative burst activity6. During LPS stimulation, GSK-3B works as an inhibitory role in cardiomyocyte TNF-alpha expression34. While experimental sepsis showed significant correlation of COX5B35. However, TRIM25 is known for diagnostic standards for discriminating septic shock blood sample and sepsis development36. Vascular endothelial cell dysfunction involving syndecan (SDC) proteoglycans contributes to acute sepsis-associated lung injury while SDC4 is involved in LPS-induced sepsis-associated ALI37. Up-regulated gene CALR was found to be positively correlated with the severity of the sepsis and increased CALR predicts poor prognosis of sepsis patients. It was found as highly differentially expressed protein in patients who subsequently develops AKI38,39. On the other hand, SELL (CD62L), an early indicator of progenitor cell commitment, is expressed on classical monocytes along with CD14, and CCR2, whereas FCGR3B, CX3CR1, and ITGAL are non-classical genes. TLR4 signaling in human monocytes positively regulates TLR4-mediated cytokine production and directly recruits ATF440,41. Cdc42, present in mature blood cells regulates a variety of immune cell functions, including monocyte transmigration, macrophage chemotaxis, etc. Rpl31 and Rpl37 expressions were upregulated by macrophages/monocytes during DSS inflammation and inhibit p53-mediated apoptosis41,42. In the mice model of NASH, lysosomal-associated protein transmembrane 5 (LAPTM5) enhances the degradation of CDC42 and promotes liver injury. Activated form of small GTPase Cdc42 stimulates apoptosis in Jurkat T-cells by activating protein kinase cascade leading to stimulation of c-Jun amino terminal kinase (JNK). Whereas both XBP1 and PERK activate ER stress pathway by triggering downstream ATF4 which upregulates pro-apoptotic proteins such as CHOP. While XBP1 interaction with HDAC3 protects endothelial cells from oxidative stress. Other than these genes, both the retinoid X receptors (RXRs) and peroxisome proliferator-activated receptors (PPARs) are transcription factors which plays role in various processes i.e. energy balance, including triglyceride metabolism, fatty acid handling and storage, and glucose homeostasis. Transcriptomic analysis revealed that PPARδ activation enhances genes involved in glycolysis and the TCA cycle43–46. All these up- and down regulated genes can be used as therapeutic targets in DLC patients with sepsis. Role of new HLA subsets, various signaling pathways (NF-kB, TNF-α, and IL-17) and their associated genes in the early discovery of sepsis in mice models which can be further used in clinical setup, can be considered as the limitations of this study. Therefore, these genes and pathways such as NF-kB, TNF-α, and IL-17 signaling can serve as biomarkers for future studies in DLC patients with sepsis. We conclude that monocytes from decompensated liver cirrhosis patients with sepsis play a crucial role, as evidenced by the discovery of novel markers which can be utilized in future. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We sincerely acknowledge all patients recruited in this study. Abbreviations Decompensated liver cirrhosis Without Single cell RNA) Peripheral blood mononuclear cells Systemic inflammatory response syndrome Sequential organ failure assessment Total leukocyte count Respiratory rate RNA protein interaction Whole transcriptome analysis Base call Recursive substation error correction Unique molecular identifiers Uniform manifold approximation and projection Quality check Principal component analysis Differentially expressed genes Gene set enrichment analysis Standard deviation Aspartate aminotransferase International normalized ratio Procalcitonin O6 Methylguanine-DNA Methyltransferase Human leukocyte antigen Classical monocytes Non classical monocytes Classical dendritic cells Human immunodeficiency virus Hepatitis C virus Myeloid derived suppressor cells Human leukocyte antigen Author contributions RS performed all experiments after inputs from NT and did the initial analysis and validation. PG performed all the bioinformatic analysis. RM and SS have characterized and recruited patients. MI, JK, RK and MJ helped in performing the experiments. An initial draft of the manuscript was written by RS. SS, RM, NK, and GR have provided initial inputs in the manuscript. NT designed the study, revised, corrected, and finalized the manuscript. All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication. Funding This study experiments were partially supported by research funds from the Department of Biotechnology (DBT) [BT/PR27792/BRB/10/1668/2018] and fellowship was provided to Rashi sehgal by Indian Council of Medical Sciences (ICMR). However, major reagents were arranged by corresponding author self-driven projects. Data availability Data is analyzed and presented in manuscript and supplementary file. However, on the request it’s available from corresponding author. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rashi Sehgal and Pramod Gautam contributed equally to this work. Contributor Information Shiv Kumar Sarin, Email: shivsarin@gmail.com, Email: shivsarin@ilbs.in. Nirupama Trehanpati, Email: trehanpati@gmail.com, Email: trehanpati@ilbs.in. References Associated Data Data Citations Supplementary Materials Supplementary Material 1 Data Availability Statement Data is analyzed and presented in manuscript and supplementary file. However, on the request it’s available from corresponding author.