A TLR8 Variant Identified From Whole Exome Sequencing as a Sepsis‐Prone Mutation

ABSTRACT

Sepsis remains a leading cause of morbidity and mortality worldwide, with outcomes highly influenced by host immune responses. While environmental and pathogen‐related factors are well recognized, the contribution of host genomic variants to sepsis susceptibility and severity is increasingly appreciated. Because approximately 85% of known disease‐causing mutations reside in the exome, whole exome sequencing (WES) offers a powerful strategy to uncover pathogenic variants in critically ill patients and to identify potential inborn errors of immunity that may modulate disease course. In the present study, we performed WES on 31 sepsis patients across different age groups, stratified into pre‐school‐aged children, school‐aged children, and adults, and identified multiple genes harboring high‐ or medium‐impact variants. Of particular interest, a high‐impact TLR8 variant (rs3764880: A>G; p.Met1Val) was observed across all the age groups and predominantly in individuals with bacterial sepsis. Single‐cell RNA sequencing of peripheral blood mononuclear cells demonstrated that TLR8 was highly expressed in non‐classical monocytes, with transcription levels markedly elevated in carriers of the variant. Functional studies revealed that this TLR8 variant enhanced IFN‐β secretion upon ligand stimulation, suggesting that dysregulated TLR8 signaling might modulate host inflammatory responses during bacterial sepsis. Given the established role of IFN‐β in exacerbating sepsis severity, these findings support a model in which the TLR8 rs3764880 variant contributes to sepsis pathophysiology by amplifying IFN‐β‐mediated monocyte responses. This study underscores the importance of integrating genomic and functional immunologic analyses to identify host determinants of sepsis, highlights TLR8 as a potential biomarker and therapeutic target, and provides a framework for precision medicine approaches to predict and modulate outcomes in bacterial sepsis.

Whole‐exome sequencing of sepsis patients identified a recurrent high‐impact TLR8 rs3764880 variant enriched in bacterial sepsis. Single‐cell transcriptomics localized elevated TLR8 expression to non‐classical monocytes, while bulk RNA‐seq and functional assays demonstrated enhanced IFN‐β responses following TLR8 stimulation. These findings link host genetic variation to dysregulated innate immune signaling and identify a TLR8‐dependent interferon pathway that contributes to susceptibility and inflammatory responses in sepsis.

1. Introduction

Sepsis is a life‐threatening syndrome characterized by a dysregulated host response to infection, leading to systemic inflammation, immune dysfunction, and multi‐organ failure. It remains a leading cause of mortality worldwide, particularly in critically ill patients, accounting for an estimated 11 million deaths annually [1]. Despite advances in critical care, treatment remains largely supportive, as no specific targeted therapy currently exists to counteract the underlying pathophysiological mechanisms of sepsis [2].

The role of host genomic factors in sepsis susceptibility and outcomes has been increasingly recognized. Family, twin and adoption studies suggest that genetic susceptibility contributes to familial aggregation of infectious diseases or to death from infection [3]. Additionally, genome‐wide association studies (GWAS) and sequencing approaches have identified multiple genetic variants associated with immune dysregulation, inflammatory response, and sepsis severity [4]. However, the genetic underpinnings of individual susceptibility to sepsis remain incompletely understood. Occult immune deficiencies, genetic predispositions that may remain asymptomatic until triggered by a severe physiological stressor such as sepsis, are increasingly recognized as contributing factors to disease severity. Inborn errors of immunity (IEIs), formerly referred to as primary immunodeficiency disorders (PIDs), represent a broad spectrum of genetic conditions that impair immune function [5]. Many of these disorders present atypically and may go undiagnosed in adults, only manifesting upon exposure to severe infections. Studies suggest that defects in pathways related to Toll‐like receptors (TLRs), nuclear factor kappa B (NF‐κB) signaling, and cytokine production may contribute to impaired pathogen recognition and immune dysregulation in sepsis patients [6].

Because approximately 85% of known disease‐causing mutations reside in the exome [7], whole exome sequencing (WES) has become an invaluable tool for identifying pathogenic variants in critically ill patients. WES enables the detection of rare or novel genetic variants affecting host immune responses, including defects in key immune‐modulating genes such as TLR4, NLRP3 (nucleotide‐binding domain, leucine‐rich repeat, and pyrin domain‐containing protein 3), CECR1 (cat eye syndrome chromosome region, candidate 1), and STAT1 (signal transducer and activator of transcription 1), which have been implicated in sepsis susceptibility [8]. Previous WES studies in pediatric sepsis cohorts, particularly in populations with a high rate of consanguinity, have uncovered previously undiagnosed primary immunodeficiency contributing to sepsis susceptibility [9]. In one study, nearly 20% of pediatric patients with life‐threatening infections harbored pathogenic or likely pathogenic variants in genes related to immune functions, suggesting that genetic screening may be essential for identifying patients at high risk of severe sepsis [9].

Building on these findings, in the present study we conducted exome sequencing of peripheral blood mononuclear cells (PBMCs) from a cohort of sepsis patients from a tertiary medical center in the U.S., aiming to identify functionally significant DNA variants associated with immune dysfunction and sepsis susceptibility. To assess the biological relevance of these variants, we performed functional validation studies to determine their impact on immune signaling pathways, cytokine production, and pathogen clearance. By uncovering genetic variants linked to immune dysregulation, our study contributes to a growing body of evidence that sepsis is, in part, a genetically influenced condition. The identification of genetic predispositions to sepsis could pave the way for precision medicine approaches, such as early genetic screening of at‐risk individuals, targeted immunomodulatory therapies, and personalized treatment strategies based on a patient's genetic profile. Understanding the genetic basis of sepsis may ultimately lead to the development of novel therapeutic interventions aimed at mitigating immune dysfunction and improving patient outcomes.

2. Methods

2.1. Enrollment of Sepsis Patients and for Whole Exome Sequencing

Adult and pediatric patients diagnosed with sepsis were previously enrolled in a biobank protocol for sample collection at Boston Children's Hospital (BCH). Among these patients, individuals who met the organ dysfunction‐based definition of sepsis according to the Pediatric Sequential Organ Failure Assessment (pSOFA) or adult SOFA score [10] were selected for exome sequencing. The study was approved by the BCH Institutional Review Board (IRB)(IRB‐P00034801).

2.2. Whole Exome Sequencing Analysis

Genomic DNA was extracted using a bead‐based method for high purity and yield [11], and its quantity was assessed fluorometrically [12]. Exome sequencing analysis was performed at Quest Diagnostics (Secaucus, NJ). The DNA was then enzymatically fragmented [13], and sequencing libraries were prepared by ligating adapters to both ends of the fragments, followed by bead‐based size selection for optimal sequencing efficiency [14]. PCR amplification enriched the target sequences while minimizing bias [15]. Exonic and intronic regions of interest were captured via a hybridization‐based method [16]. Quality control (QC) ensured correct template size, concentration, and the removal of contaminants such as adapter dimers [17]. High‐quality libraries were sequenced using Illumina's sequencing‐by‐synthesis (SBS) technology [18], with paired‐end (150 bp × 150 bp) sequencing to enhance variant detection [19]. Data processing, including base calling and quality scoring, was performed using Illumina's proprietary software, generating CBCL files [20]. Whole exome sequencing (WES) analyzes all protein‐coding regions (~20,000 genes) and ~1500 non‐coding disease‐associated variants [21]. Whole Exome targets all protein‐coding regions of all genes (~20,000) as well as ~1500 non‐coding disease‐causing variants.

2.3. Bioinformatics Analysis of Whole Exome Sequencing

Raw sequencing data were first subjected to quality control (QC) to remove low‐quality reads and adapter contamination using standard QC tools. Cleaned reads were then aligned to the human reference genome (GRCh38/hg38) using a high‐accuracy aligner BWA‐MEM (v0.7.15), and alignment files were converted into Variant Call Format (VCF) for downstream analysis. Variant calling was performed using GATK (v4.1.0.0), a haplotype‐based variant detector, to identify single nucleotide variants (SNVs) and small insertions/deletions (indels). To ensure high‐confidence calls, we applied both quality thresholds and standard hard filters: minimum Phred‐scaled quality score (QUAL ≥ 30), minimum read depth (DP ≥ 10×), and retention of variants present in ≥ 90% of samples. Additional GATK hard‐filter parameters included: Quality by Depth (QD < 2.0), Fisher Strand Bias (FS > 60.0), Mapping Quality (MQ < 40.0), Mapping Quality Rank Sum (MQRankSum < −12.5), and Read Position Rank Sum (ReadPosRankSum < −8.0). To facilitate downstream annotation and comparison, all individual sample VCF files were annotated with SnpEff (v4.3 + T.galaxy2) using human reference genome (GRCh38/hg38), assigning predicted functional impacts (e.g., synonymous, missense, nonsense) based on known gene models. All single files were then merged using BCFtools merge (v1.15.1 + galaxy3), producing a unified, uncompressed VCF file. Annotated VCF files were subsequently processed with GEMINI load (v0.20 + galaxy2) [22] to populate a searchable database structure using the annotation source. GEMINI integrates data from the 1000 Genomes Project to provide minor allele frequency (MAF) and other annotations; GEMINI annotations w/GERP & CADD (2019‐01‐12 snapshot). Variant data were then extracted using GEMINI query (v0.20 + galaxy2), enabling efficient filtering and classification of variants based on predicted functional impact, allele frequency, and other relevant annotations.

2.4. Enrollment of Sepsis Patients and Healthy Controls for Single‐Cell RNA Sequencing

In this study, we performed scRNA‐seq of PBMCs from control and sepsis subjects recruited between May 2021 and January 2023. The study was approved by the Institutional Review Board (IRB) at Boston Children's Hospital (IRB‐P00031930), and written informed consent was obtained from all patients. If indicated, assent was also obtained. The study was registered in ClinicalTrials.gov (Trial number NCT04103268) and carried out in accordance with the Declaration of Helsinki.

2.5. Sample Collection and Leukocyte Purification for Single‐Cell RNA‐Seq Analysis

As control subjects, we enrolled otherwise healthy patients who underwent elective procedures. For sepsis patients, we enrolled patients with documented or suspected infection and a score of 2 or more in at least one of Sequential Organ Failure Assessment (SOFA) system sub‐scores at the time of intensive care unit (ICU) admission [23]. We excluded patients with congenital heart diseases, malignancy, autoimmune diseases, transplantation, human immunodeficiency virus (HIV) infection, and on steroid treatment. 1 mL of blood was collected into heparin anticoagulant tube at each blood collection. Leukocytes were purified by combining polymorphonuclear and mononuclear cell layers using Polymorphprep reagent (ProteoGenix, Schiltigheim, France) at room temperature. Leukocytes were immediately fixed with 10× fixing reagent (10× Genomics; Pleasanton, CA) and stored in liquid nitrogen until sequencing.

2.5.1. ’RNA Library Preparation and Sequencing on 10× Genomics Platform

3’RNA library & preparation, and single cell RNA sequencing analysis were performed as we previously described [24].

2.6. Single‐Cell RNA‐Seq Data Processing and Analysis

Raw sequencing data were processed using the Cell Ranger Software Suite (v3.1.0, 10× Genomics). Base call (BCL) files were converted to FASTQ format and demultiplexed using the mkfastq command. Unique molecular identifiers (UMIs) and cell barcodes were extracted, and reads were aligned to the GRCh38 human reference genome to generate digital gene expression (DGE) matrices. Downstream analysis was conducted in R using the Seurat package (v5.2.1). Cells expressing fewer than 200 genes and genes detected in fewer than 3 cells were excluded. Cells with > 10% mitochondrial gene content were also filtered out to eliminate low‐quality or apoptotic cells. After metadata curation, individual Seurat objects were merged and normalized. Dimensionality reduction, clustering, and visualization were performed following the standard Seurat pipeline. A clustering resolution of 0.1 was applied to delineate major leukocyte populations.

To correct for batch effects and technical variability, we applied Harmony (v1.2.3) for integration, adjusting for donor identity, sequencing batch, gender, and age. Functional phenotypes and module activity scores were computed using the AddModuleScore function, incorporating gene sets curated from the Gene Ontology database.

2.7. TLR8 Functional Analysis

THP‐1 cells were used to examine the role of TLR8. THP‐1 cells were cultured at 37°C in RPMI1640 supplemented with 10% FBS and 1% penicillin/streptomycin. TLR8 deletion was performed using CRIPR/Cas9 editing. Briefly, electroporation was performed 1 day after passing cells when they were in log phase of growth using the Lonza 4D nucleofector. Ribonucleoprotein (RNP) complex was made by combining 100 pmol Cas9 (IDT; Newark, NJ) and 100 pmol modified sgRNA (Synthego; Redwood City, CA) targeting TLR* using either sg1 (UGAAGGAACAUGUUUUCCUA) or sg2 (AGGUCAGCAUUGACGACUGA). As a control, sgRNA targeting AAVS1 was used (GGGGCCACUAGGGACAGGAU). 2 × 10⁵ cells were suspended in 20 μL SF cell line solution (Lonza; Basel, Switzerland), mixed with RNP and subjected to nucleofection with program FF‐100 per the company recommendation. TLR8 expression level was tested on western blot analysis. Then, THP‐1 cells were stimulated with human TLR8 agonist ssRNA40/LyoVec (5′‐GCCCGUCUGUUGUGUGACUC‐3′) (10 μg/mL) for 24 h to examine IFN‐β levels by ELISA (R&D). In a separate experiment, whole blood was stimulated with ssRNA40/LyoVec (10 μg/mL) for 6 h and its IFN‐β levels were determined by ELISA.

2.8. Polymerase Chain Reaction‐Restriction Fragment Length Polymorphism (PCR‐RFLP)

DNA isolation from whole blood was performed using a DNA extraction kit (Qiagen). Then TLR8 Met|Val (rs3764880) was determined by PCR‐RFLP method, performing PCR using the previously reported primer sequences (F: 5′‐GTGTGTGTCTGATTTGGGTTG‐3′, R: 5′‐TTTCTAGGCTCACACCATTTG‐3′), followed by digestion with NiaIII. Digestion product was run on agarose gel [25].

2.9. Statistical Analysis

Data were analyzed as indicated in the corresponding figure legends. Statistical significance was defined as p < 0.05. All the statistical calculations were performed using PRISM 9 software (GraphPad Software; La Jolla, CA, USA).

3. Results

3.1. A TLR8 Gene Variant Was One of the Major Mutations Observed Across All Age Groups in the Sepsis Cohort

We performed whole exome sequencing (WES) on DNA from 31 pediatric and adult sepsis patients (Figure 1a). The demographics of the cohort is shown in Table 1. The median age was 13 years (range 0.8–29 years). To account for age‐related variability in immune responses, patients were categorized into pre‐school (< 5 years), school‐aged (5–17 years), and adults (≥ 18 years), reflecting commonly recognized developmental stages associated with differences in immune system function that may influence host responses to sepsis [26]. This stratification enabled the identification of genetic variants and host‐response pathways associated with sepsis that are conserved across developmental stages and therefore more likely to represent fundamental mechanisms of disease susceptibility with relevance to broader patient populations. Most patients were White, and bacterial infections were the predominant cause of sepsis.

FIGURE 1.

Whole‐Exome Sequencing (WES) study design and variant landscape in the sepsis cohort. (a) Overview of the WES study design. (b) Bar plots showing the distribution of high‐ and medium‐impact variant subclasses across different age groups. (c) Bar plots of single nucleotide polymorphisms (SNPs) and insertions/deletions (INDELs) across age groups. (d) Bar plots showing the counts of transition (Ts) and transversion (Tv) variants across age groups. (e) UpSet plot illustrating unique and shared genes affected by high‐impact variants across the cohort.

TABLE 1.

Demographic data of subjects enrolled for the whole exome sequencing analysis.

Age	13 (0.8, 29.07) year
Gender	17 males, 14 females
Race	White 20, Asian 3, Black 2, Unknown 6
Ethnicity	American 17, European 3, Hispanic/Latino 3, Asian 1, Unknown 7
Infection type	Bacterial 18, Viral 2, Fungal 1, Unknown 10

Analysis of variant impact revealed that high‐impact variants, meaning those predicted to cause complete loss of gene function (e.g., stop‐gained, frameshift indels, canonical splice‐site disruptions), were dominated by frameshift mutations across all age groups, followed by splice‐site mutations (Figure 1b). Medium‐impact variants, primarily nonsynonymous/missense substitutions, were predicted to alter protein function without abolishing it. Single nucleotide polymorphisms (SNPs) were the most abundant variant class, with insertion–deletion (INDEL) variants also detected at lower frequency; among SNPs, transitions were more common than transversions (Figure 1c). Pre‐school‐aged children carried the highest number of SNPs compared to older groups (Figure 1d).

To determine whether disruptive variants were shared across patients and developmental groups, we analyzed genes harboring high‐impact variants and identified both unique and overlapping genes. Among all high‐impact variants, only TLR8 and OR13C7 harbored identical variants across all age groups (Figure 1e), suggesting a potentially important role in sepsis susceptibility. Additional high‐ and medium‐impact variants shared between age groups are presented in Figure S1.

To gain insight into the biological pathways potentially affected by disruptive variants in the cohort and to help prioritize candidate genes for downstream analyses, we performed gene set enrichment analysis (GSEA), using genes harboring high‐impact variants across the sepsis patients. To account for potential developmental differences in immune maturation and identify conserved pathways, enrichment analysis was conducted separately within each age group (pre‐school, school‐aged, and adults). Several pathways were consistently enriched across groups (Figure 2) and were associated with innate immune signaling and host defense mechanisms, including pathways related to infection responses and regulation of interferon‐β production, processes that are consistent with the known role of TLR8 in pathogen sensing and inflammatory signaling [27]. In contrast, enrichment of pathways related to sensory perception of taste likely reflects the presence of variants in OR13C7, a member of the olfactory receptor family, whose role in host defense and sepsis biology remains unclear. Based on these observations, TLR8 was prioritized for further investigation. To further explore potential clinical associations, we examined the relationship between the TLR8 variant (rs3764880) and demographic and clinical parameters. A correlation heatmap illustrating these associations is shown in Figure 3.

FIGURE 2.

Dot plots showing enriched biological pathways associated with genes affected by high‐impact variants across age groups. Pathway enrichment analysis was performed separately for adults, school‐aged children, and preschoolers. Dot plots display the top significantly enriched pathways as P value, highlighting both immune‐related and sensory‐associated processes.

FIGURE 3.

Correlation heatmap of TLR8 SNP rs3764880 with demographic and clinical parameters.

3.2. Single‐Cell Transcriptomic Analysis Identifies Non‐Classical Monocytes as the Primary Cellular Context of TLR8 Expression in Sepsis

Based on the identification of the TLR8 variant rs3764880 in the WES analysis, we next sought to determine which immune cell populations express TLR8 and whether expression patterns differ in sepsis. To address this, we performed single‐cell RNA sequencing (scRNA‐seq) on peripheral blood mononuclear cells (PBMCs) from three pediatric patients with severe sepsis, three with mild sepsis, and four healthy controls, yielding 31,198 cells in total (Figure 4a and Table 2). UMAP visualization identified distinct immune populations, including T cells, NK cells, B cells, plasmablasts, and monocytes (Figures 4b and S2a). Across these populations, TLR8 expression was largely restricted to monocytes (Figure S2b), suggesting that TLR8‐mediated signaling primarily operates within the innate immune compartment during sepsis.

FIGURE 4.

Single‐cell RNA‐seq analysis of PBMCs in sepsis patients and healthy donors. (a) Schematic of the experimental design for single‐cell RNA sequencing of PBMCs. (b) UMAP visualization of all immune cell types within PBMCs. (c) UMAP plot showing monocyte subclusters across study conditions (severe sepsis, mild sepsis, and healthy donors). (d) Dot plot of canonical marker genes defining the three monocyte subtypes. (e) Dot plot showing TLR8 expression across monocyte subtypes. (f) Dot plot showing TLR8 expression across the three study conditions. (g) Violin plot illustrating type I interferon signaling scores across monocyte subtypes. One‐way ANOVA with Bonferroni post hoc analysis was used. ****p < 0.0001. (h) Table summarizing the clinical demographics of patients with the TLR8 rs3764880 polymorphism.

TABLE 2.

Demographic data of subjects enrolled for the single‐cell RNA‐Sequencing analysis.

Subject	Age (years)	Gender	Diagnosis	SOFA
Severe 1	1	Female	Bronchiolitis	6
Severe 2	15	Female	Toxic shock syndrome	12
Severe 3	3	Female	Aspiration pneumonia	7
Mild 1	0.7	Male	RSV infection	3
Mild 2	15	Male	Otomastoiditis	3
Mild 3	15	Male	Candidiasis	8
Healthy 1	0.3	Male	0	0
Healthy 2	10	Female	0	0
Healthy 3	6	Male	0	0
Healthy 4	0.8	Female	0	0

To further refine the cellular context of TLR8 expression, monocytes were subclustered into classical, intermediate, and non‐classical subsets (2026, 1801, and 788 cells, respectively) (Figure 4c,d). TLR8 expression was highest in non‐classical monocytes, where levels were significantly higher in mild sepsis patients compared with severe sepsis and healthy controls (p < 0.001) (Figure 4e,f).

To determine whether this pattern was consistent in independent datasets, we analyzed a published adult sepsis cohort (GEO: GSE167363) [28]. Similar to our cohort, TLR8 expression was enriched in CD14⁺ monocytes (intermediate and non‐classical subsets) and was higher in survivors compared with non‐survivors and healthy controls (Figure S5a–c).

Given the enrichment of TLR8 expression in non‐classical monocytes, we next examined whether TLR8 expression was associated with specific immune activation pathways. Gene set enrichment analysis revealed that type I interferon signaling pathways were significantly enriched in non‐classical monocytes, the subset with the highest TLR8 expression (Figure 4g). Comparative pathway analysis further showed that TLR8^high monocytes were enriched for pathways related to immune activation, phagocytosis, and antigen presentation, whereas TLR8^low monocytes were enriched in IL‐10 signaling, chemotaxis, GPCR ligand binding, and LPS responses (Figure S6). Among the inflammatory pathways examined, type I interferon signaling emerged as the most prominently enriched pathway in TLR8^high cells (Figure 5).

FIGURE 5.

Enrichment of inflammatory signaling pathways in TLR8^high and TLR8^low monocytes. Violin plots show enrichment scores for interferon, IL‐1, IL‐4, IL‐6, IL‐10, and IL‐17 signaling pathways. Student t test was performed. * p < 0.05.

Notably, the TLR8 variant rs3764880 (A>G; p.Met1Val) identified in the WES cohort has been previously reported to influence translation efficiency of two TLR8 isoforms: TLR8v1, predominantly expressed in non‐classical monocytes, and TLR8v2, enriched in classical and intermediate monocytes [29]. The G allele increases TLR8v1 expression by approximately 50% while decreasing TLR8v2 expression by approximately 25%, without altering receptor localization or signaling capacity [29]. This can provide mechanistic context in our observations. Although TLR8 isoforms, including TLR8v1 and TLR8v2, arise from alternative splicing and therefore correspond to distinct transcripts [30], the design of our scRNA‐seq analysis does not allow reliable resolution of these isoforms. Specifically, our single‐cell data were generated using a 3′‐end capture protocol, which preferentially sequences fragments near the poly‐A tail. Because the structural differences between these isoforms, such as the skipping of Exon 2 in the v2 isoform [30], occur at the 5′ end of the transcript, they fall outside the typical coverage range of our sequencing reads. Consequently, our analyses represent total TLR8 transcript abundance rather than isoform‐specific quantification. However, we detected enrichment of TLR8 expression in non‐classical (CD14⁺CD16⁺) monocytes, a subset previously reported to preferentially express TLR8v1²⁹, which provides indirect support for the relevance of this isoform in the context of sepsis. In addition, in our cohort, 15 of 31 patients (48.4%) carried the G allele that increases TLR8v1 translation efficiency [29], and the majority of these individuals presented with bacterial sepsis, suggesting that TLR8‐mediated signaling may contribute to host responses to bacterial pathogens (Figure 4e and Table 3).

TABLE 3.

Frequency of rs3764880 G allele in the population.

Population group	Sample size	A allele frequency	G allele frequency	Homozygous A	Homozygous G	Heterozygous
Global	318,458	70.63%	29.37%	60.18%	18.92%	20.90%
European	269,820	73.50%	26.50%	63.27%	16.26%	20.47%
African	11,613	72.46%	27.54%	59.51%	14.59%	25.90%
African Others	408	71.10%	28.90%	58.33%	16.18%	25.49%
African American	11,205	72.51%	27.49%	59.55%	14.54%	25.91%
Asian	6962	17.68%	82.32%	7.70%	72.34%	19.97%
East Asian	4998	16.33%	83.67%	7.28%	74.63%	18.09%
Other Asian	1964	21.13%	78.87%	8.76%	66.50%	24.75%
Latin American 1	1130	63.19%	36.81%	54.61%	28.19%	17.20%
Latin American 2	7198	50.69%	49.31%	35.29%	33.87%	30.84%
South Asian	5224	49.39%	50.61%	38.55%	39.78%	21.67%

3.3. Functional Perturbation of TLR8 Demonstrates Regulation of Type I Interferon Responses in Monocytes

To determine whether TLR8 directly regulates type I interferon signaling, we performed functional perturbation experiments in THP‐1 monocytes, a widely used model of human monocyte biology. TLR8 knockout cells were generated using CRISPR‐Cas9 targeting the sg_1 guide RNA (Figure 7a), followed by bulk RNA sequencing after ssRNA stimulation with ssRNA40/LyoVec, a known TLR8 agonist [31], that has been shown to be a weak TLR7 and strong TLR8 agonist when tested using cell lines expressing either human or mouse TLR7 or TLR8 (https://www.invivogen.com/ssrna40‐lv). In addition, it has been reported before that TLR8 is the predominant ssRNA‐sensing receptor in human monocytes and myeloid cells, whereas TLR7 expression is generally low in these populations and is primarily associated with plasmacytoid dendritic cells and certain lymphocyte subsets [32]. Prior studies have also demonstrated that TLR8 agonist stimulation in differentiated monocytes selectively induces TLR8v1 and enhances downstream inflammatory signaling [29].

FIGURE 7.

The role of the TLR8 SNP variant (rs3764880) in IFN‐β production. (a) Western blot showing TLR8 knockout in THP‐1 monocyte cells using CRISPR‐Cas9. (b) Dot plot showing expression patterns of IRF7 and CREBBP, key regulators of the TLR8–IFN‐β axis, across TLR8high and TLR8low monocyte subsets (P value < 0.001). (c) IFN‐β production by THP‐1 cells stimulated with the TLR8 agonist ssRNA40 (10 μg/mL); n = 4 per group. Student t test was performed. ***p < 0.001. (d) PCR‐RFLP assay detecting the presence of the TLR8 SNP variant (rs3764880) in whole blood from healthy volunteers. (e) IFN‐β production upon ssRNA40 (10 μg/mL) stimulation in carriers vs. non‐carriers of the rs3764880 variant; n = 3 per group. Student t test was performed. *p < 0.05.

Compared to wild‐type cells, TLR8‐deficient cells showed reduced expression of type I interferon signaling genes, while genes related to monocyte differentiation, chemokine production, and phagocytosis were upregulated (Figure 6a–d). Specifically, in ssRNA40‐stimulated THP‐1 monocytes, we identified 3691 upregulated genes and 2627 downregulated genes relative to controls (Figure 6a). In stimulated TLR8‐knockout THP‐1 cells, there were 2162 upregulated and 912 downregulated genes compared with their respective controls (Figure 6b). Pathway enrichment analysis showed that upregulated genes were associated with monocyte differentiation, chemokine production, phagocytosis, and related processes (Figure 6c). Downregulated genes were enriched for antiviral defense pathways, including type I interferon signaling and antigen processing and presentation (Figure 6d). Interferon‐stimulated genes (ISGs) altered in ssRNA40‐stimulated TLR8‐knockout THP‐1 monocytes relative to stimulated wild‐type controls are listed in Figure 6e.

FIGURE 6.

Bulk RNA‐seq analysis of ssRNA‐stimulated WT and TLR8‐knockout THP‐1 monocytes. (a) Volcano plot of differentially expressed genes (DEGs) in ssRNA‐stimulated THP‐1 monocytes compared with unstimulated controls. (b) Volcano plot of DEGs in ssRNA‐stimulated TLR8‐knockout THP‐1 monocytes relative to stimulated wild‐type controls. (c) Gene Ontology enrichment analysis of upregulated genes in ssRNA‐stimulated THP‐1 monocytes. (d) Gene Ontology enrichment analysis of downregulated genes in ssRNA‐stimulated THP‐1 monocytes. (e) Table of interferon‐stimulated genes (ISGs) in ssRNA‐stimulated TLR8‐knockout THP‐1 monocytes relative to stimulated wild‐type controls.

Consistent with these findings, single‐cell RNA‐Seq analysis revealed that IRF7 and CREBBP, key upstream regulators of IFN‐β and downstream effectors of TLR8 signaling, were significantly upregulated in TLR8^high monocytes compared to TLR8^low monocytes (Figure 7b).

To further validate the role of TLR8 in interferon responses, the levels of IFN‐β were measured in WT and TLR8‐knockout THP‐1 monocytes stimulated with ssRNA40/LyoVec. Wild‐type THP‐1 cells produced significantly higher levels of IFN‐β compared to TLR8‐deficient clones (Figure 7c), underscoring the essential role of TLR8 in driving type I interferon responses to ssRNA40 stimulation.

Finally, to assess whether the TLR8 rs3764880 variant influences interferon responses in human samples, we examined the presence of the TLR8 rs3764880 SNP in blood samples from healthy volunteers and identified individuals carrying the variant as well as non‐carriers (Figure 7d). Following stimulation with ssRNA40/LyoVec, blood samples from variant carriers produced significantly higher levels of IFN‐β compared to non‐carriers at 6 h (Figure 7e). Together, these findings provide functional evidence that TLR8 regulates type I interferon responses and that the rs3764880 variant is associated with enhanced TLR8‐mediated interferon signaling.

4. Discussion

In the present study we performed WES on pediatric and adult sepsis patients to identify conserved host genomic factors associated to sepsis susceptibility. Our results uncovered several genes with potentially high‐ or medium‐impact variants. Notably, a high‐impact TLR8 variant (rs3764880: A>G; p.Met1Val) was detected across all age groups, predominantly in individuals with bacterial sepsis. This discovery provided the rationale to investigate TLR8 further using single‐cell and bulk RNA sequencing and functional assays, aiming to elucidate the mechanistic link between genetic variation and sepsis pathophysiology. The single‐cell analysis identified monocytes as the primary TLR8‐expressing population, with highest expression in non‐classical monocytes. Consistent with this cellular distribution, type I interferon signaling pathways were enriched in this same cell population. Comparative pathway analysis further showed enrichment of immune activation pathways in TLR8^high monocytes, with interferon signaling emerging as the most prominent inflammatory pathway. To further investigate the functional role of TLR8, we performed bulk RNA sequencing in THP‐1 monocytes (WT and CRISPR‐mediated knockout of TLR8), which resulted in reduced expression of type I interferon signaling following ssRNA40 stimulation. Correspondingly, functional analysis showed that TLR8‐deficient cells produced significantly lower levels of IFN‐β compared to wild‐type controls. Importantly, stimulation of whole blood samples from individuals carrying the rs3764880 variant resulted in significantly higher IFN‐β production compared with non‐carriers, providing functional evidence that this variant can influence TLR8‐mediated interferon responses.

TLRs are major pattern recognition receptors (PRRs) in our immune system. Among them, both TLR7 and TLR8 are endosomal PRRs that detect single‐stranded RNA (ssRNA) from various pathogens, including viruses and bacteria [31, 33]. Recently, bacterial RNA recognition by TLR7 and TLR8 has been described [34]. While TLR8 has traditionally been studied in the setting of viral infections, its role in bacterial infections has been less well characterized in contrast to TLR7. TLR8 recognizes uridine (U) and uridine‐rich (UR/URR) motifs in RNA, leading to the activation of downstream signaling pathways and the production of pro‐inflammatory cytokines [33]. Given this central role in pathogen sensing, we aimed to explore whether genetic variation in TLR8 could influence sepsis susceptibility and outcomes. In the present study, we discovered that a single nucleotide polymorphism (SNP) in TLR8, specifically rs3764880, was highly prevalent in patients with bacterial sepsis. This observation suggests that TLR8 and its associated polymorphisms are not only relevant in viral infections [35] but may also play a critical role in bacterial sepsis.

The rs3764880 SNP, located in exon 1 of TLR8 on the X chromosome, involves an A‐to‐G nucleotide substitution, resulting in a methionine‐to‐valine change at position 1 (p.Met>Val), which alters TLR8 protein expression and function [36]. A decrease in the truncated TLR8v2 translation seen with the G allele of rs3764880 may imply reduced TLR8 sensing and activation, leading to a diminished inflammatory response by classical and intermediate monocytes [29, 37, 38, 39]. However, our scRNA‐seq data showed that non‐classical monocytes substantially expand during sepsis, and in this subset, TLR8v1 is predominant and exhibits significantly higher translational efficiency than the wild‐type [29]. These findings highlight a potential cell‐type‐specific compensatory mechanism, whereby TLR8v1 in non‐classical monocytes may amplify immune responses despite reduced activity in classical subsets. In line with these observations, studies of TLR7 have shown that knockout mice exhibit significantly better outcomes in severe experimental sepsis induced by cecal ligation and puncture (CLP) surgery, suggesting that TLR7 signaling may have a detrimental effect in this context [40]. By analogy, understanding TLR8's role in bacterial sepsis could reveal similar immunomodulatory effects.

Historically, TLR8 has been extensively studied in viral infections, serving as a major sensor for viral ssRNA. It has been implicated in immune responses to HIV, hepatitis C virus (HCV), dengue fever, Crimean‐Congo hemorrhagic fever, and COVID‐19 [35, 41, 42, 43, 44, 45]. Polymorphisms in TLR8, particularly rs3764880, have been associated with disease susceptibility and severity in these infections and have been shown to be both sex, age, and ethnic background dependent. Because TLR8 locates on the X chromosome, sex‐bias is particularly important in this context [38, 46]. For example, in HIV, rs3764880 increases susceptibility primarily in adult males and infant females but is also linked to slower progression to AIDS in both sexes [37, 47, 48]. In HCV, the A allele of rs3764880 increases the risk of chronic infection [49, 50, 51]. Similarly, in dengue fever, the TLR8 haplotype (rs3764879‐rs3764880) is more frequent in infected males, with the AA/A genotype associated with warning signs in patients [52, 53, 54]. In Crimean‐Congo hemorrhagic fever, rs3764880 predisposes Turkish individuals to severe disease [44], and in COVID‐19, rs3764880 is associated with disease severity in females, influencing ICU admissions [25], while specific haplotypes (rs3764879‐C/G, rs3764880‐A/G, rs3761624‐A/G) correlate with severe disease in men [55]. Differences in allele frequencies among ethnic groups further support population‐specific immunogenetic effects: Asians have a higher frequency of the G allele, whereas the A allele predominates in African and European populations [56].

Despite the extensive research on TLR8 in viral infections, its involvement in bacterial infections remains underexplored. Our findings raise the possibility that TLR8 variation may also contribute to host responses in bacterial sepsis, expanding the potential relevance of this pathway beyond viral pathogen sensing. We observed that the majority of sepsis patients carrying the rs3764880 variant suffered from bacterial sepsis, suggesting that TLR8 contributes to host defense in bacterial infections. This aligns with prior in vitro studies showing TLR8 recognition of bacterial pathogens such as Staphylococcus aureus and Escherichia coli [33] and highlights the potential for therapeutic modulation of TLR8 in sepsis. Additional evidence supporting a role for TLR8 in bacterial immunity comes from studies of tuberculosis (TB), in which rs3764880 is associated with disease susceptibility across various populations. In Kazakhstan, the A/A genotype increases TB risk [57, 58]; in Russia, the G allele is protective [59]; similar associations have been reported in Moldavia [60], Indonesia [37], Pakistan [61], and Sudan [62]. Pediatric pulmonary TB in Turkish males is strongly associated with the A/− genotype [63]. Furthermore, treatment with TLR8 agonists in humanized TLR8 mice enhanced protection against Mycobacterium tuberculosis [64]. Collectively, these findings underscore the broader immunological relevance of TLR8 in bacterial infections.

In contrast to TB studies, our sepsis cohort predominantly carried the G allele. As TLR8v1 expression has been reported to be significantly upregulated in carriers of the G allele [29] and non‐classical monocytes expanded during sepsis in our analysis and were enhanced in TLR8 expression, it is likely that TLR8v1 exerts a dominant effect in monocyte‐mediated immune responses. In support, it has been reported before that TLR8 is the predominant ssRNA‐sensing receptor in human monocytes and myeloid cells, whereas TLR7 expression is generally low in these populations and is primarily associated with plasmacytoid dendritic cells and certain lymphocyte subsets [32]. Furthermore, in our in vitro studies, TLR8 stimulation induced IFN‐β expression, while CRISPR/Cas9‐mediated TLR8 knockout attenuated IFN‐β production in THP‐1 cells. Moreover, blood from individuals harboring the TLR8 SNP produced significantly higher IFN‐β, reinforcing the functional impact of the variant. The role of IFN‐β in sepsis is context‐dependent as well. In severe sepsis, IFNαβR knockout mice show improved outcomes due to reduced systemic inflammation [65], whereas in milder sepsis, IFNαβR deficiency impairs neutrophil recruitment and phagocytosis [66]. Consistent with this duality, TLR8 expression was highest in mild sepsis cases in our study compared to severe cases, suggesting a nuanced immunoregulatory role. However, higher IFN‐β production in SNP carriers may be unfavorable in certain disease states, potentially exacerbating severe sepsis. Analysis of an adult sepsis cohort (GSE167363) [28] revealed higher TLR8 expression in survivors compared to non‐survivors, supporting a protective role under certain conditions. In our cohort, 5 of 15 patients carrying the TLR8 SNP passed away (Figure 3h). Together, our findings support a model in which the TLR8 rs3764880 variant modulates translation initiation and shifts the relative production of TLR8 isoforms toward TLR8v1. TLR8 expression was enriched in non‐classical monocytes in our scRNA‐seq dataset, a monocyte subset in which TLR8 signaling is particularly prominent. Prior studies have shown that stimulation of differentiated monocytes with TLR8 agonists preferentially induces the TLR8v1 isoform. Consistent with this framework, ssRNA stimulation in our experiments activated interferon‐related transcriptional programs and leukocytes from rs3764880 carriers produced significantly higher levels of IFN‐β, supporting a model in which the rs3764880 variant amplifies TLR8‐dependent interferon responses in monocyte populations.

5. Conclusion and Future Directions

In summary, our study identifies TLR8 and its rs3764880 variant as potential modulators of host immune responses during sepsis through the integration of whole‐exome sequencing, single‐cell transcriptomics, and functional perturbation experiments. WES analysis revealed that rs3764880 is highly prevalent in our sepsis cohort, particularly among patients with bacterial infection. Single‐cell transcriptomic profiling demonstrated that TLR8 expression is enriched in monocytes, especially non‐classical monocytes, potentially through TLR8v1 isoform amplification, and is associated with type I interferon downstream signaling during sepsis. Functional studies further showed that TLR8 signaling regulates IFN‐β production following RNA stimulation, and that individuals carrying the rs3764880 variant exhibit altered interferon responses, supporting a role for TLR8 in shaping innate immune activation. A proposed mechanism is summarized in Figure 8.

FIGURE 8.

Proposed mechanistic model linking the TLR8 rs3764880 variant to enhanced interferon responses in sepsis. (A) In the wild‐type allele (A), translation initiation occurs at the canonical AUG start codon in exon 1 of the TLR8 gene (exons 1–3). Under these conditions, the TLR8v2 isoform is predominantly produced. TLR8v2 lacks exon 1 in the mature transcript and contains exons 2–3. (B) The rs3764880 variant (A → G; Met1Val) alters the canonical translation initiation site in exon 1 (AUG → GUG), modifying translation initiation and favoring production of the TLR8v1 isoform. TLR8v1 contains exons 1–3, resulting in a shift in isoform balance toward increased TLR8v1 relative to TLR8v2. (C) Upon stimulation with bacterial single‐stranded RNA (ssRNA), TLR8 signaling is activated, particularly in monocyte populations enriched for TLR8 expression such as non‐classical monocytes. In this model, the rs3764880‐associated shift toward the TLR8v1 isoform is proposed to enhance downstream TLR8 signaling, leading to increased production of interferon‐β (IFN‐β) and amplified innate immune responses.

Despite these findings, several limitations should be considered. First, while prior studies suggest that rs3764880 influences the balance between the TLR8v1 and TLR8v2 isoforms and our results indirectly support TLR8v1 favoring, our study did not directly evaluate isoform‐specific contributions. Although TLR8 isoforms arise from alternative splicing that corresponds to distinct mRNA transcripts [30], our study design does not allow reliable resolution of these isoforms. Specifically, our scRNA‐seq and bulk RNA‐seq datasets were generated using standard short‐read sequencing approaches. In the case of scRNA‐seq, the inherent 3′ bias limits coverage of the 5′ exon structure where these isoforms differ. Furthermore, while bulk RNA‐seq theoretically covers the full transcript, the library preparation process involves fragmenting the mRNA into short sequences. To distinguish between variants such as TLR8v2, which skips Exon 2, the analysis relies on capturing ’junction reads’ that physically overlap the boundary between Exon 1 and Exon 3. Given our sequencing depth and fragment length, the mathematical probability of capturing a sufficient number of these specific junction‐spanning reads is low. Accordingly, our analyses capture total TLR8 transcript abundance rather than enabling definitive isoform‐specific quantification. Second, the scRNA‐seq subset included severe cases that were female and mild cases that were male, resulting in collinearity between sex and disease severity and preventing formal evaluation of sex‐specific effects. Third, although the broader WES cohort included both sexes and both pediatric and adult patients, the overall cohort size remains modest and most participants were of similar ethnic background, limiting our ability to perform adequately powered analyses across sex, age, or population subgroups.

Future studies in larger and more diverse sepsis cohorts will be necessary to validate these findings and to determine how host factors such as sex, age, and ancestry influence TLR8‐mediated immune responses. In addition, while the use of primary human samples carrying the naturally occurring variant in our study provided physiologically relevant evidence of the functional consequences of this SNP in human immune cells, mechanistic studies using genome editing approaches, such as CRISPR/Cas9 knock‐in models, may help define how the rs3764880 variant alters TLR8 translation, signaling strength, and downstream interferon responses in specific monocyte populations. In addition, isoform‐specific analyses will be important to clarify the relative contributions of TLR8v1 and TLR8v2 in classical, intermediate, and non‐classical monocytes during sepsis.

Together, these findings suggest that TLR8 may represent an important determinant of host susceptibility and immune responses in bacterial sepsis, extending the relevance of this pathway beyond its well‐established role in viral infection. Further investigation of TLR8‐mediated signaling pathways may provide opportunities for biomarker development and targeted immunomodulatory therapies in sepsis.

Author Contributions

S.K. conceived and supervised the study, secured the funding, and performed sample collection and data acquisition. F.A. conducted bioinformatics analysis and performed experiments. S.G. and X.H. conducted experiments. All authors contributed to data interpretation and manuscript drafting. All authors reviewed and approved the final version of the manuscript and agree to be accountable for all aspects of the work.

Funding

This research was supported by (R21HD099194) (S.K.). The funding bodies had no role in the design of the study, data collection, analysis, interpretation, or manuscript preparation.

Disclosure

Institutional Review Board Statement: This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (IRB) of Boston Children's hospital. Informed consent was obtained from the parents or legal guardians of all pediatric participants prior to sample collection and data analysis.

Informed Consent Statement: All participants, or their legal representatives in the case of minors, provided written informed consent for the publication of anonymized data. No identifying information is included in this manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Shared high‐ and medium‐impact variants across age groups. A compiled list of genetic variants with high or medium predicted impact that are shared across the three studied age groups.

Figure S2: Expression patterns of marker genes and TLR8 across PBMC subsets. (a) Dot plot showing the expression of three representative marker genes for each peripheral blood mononuclear cell (PBMC) type, used for cell type identification. (b) Dot plot illustrating the expression pattern of TLR8 across all PBMC subsets.

Figure S3: Proportional distribution of cell populations across severe sepsis, mild sepsis, and healthy controls in (a) PBMCs and (b) monocyte subsets.

Figure S4: (a) Density plot showing the expression distribution of TLR8 within monocytes. (b‐c) Signature genes defining classical, intermediate, and non‐classical monocytes, curated from the Human Gene Atlas, Azimuth Cell Types, and HuBMAP ASCT+B augmented 2022 databases, respectively.

Figure S5: Expression patterns of marker genes and TLR8 in adult sepsis. (a) UMAP visualization of blood monocytes subsets in adult sepsis. (b) Dot plot illustrating the expression pattern of CD14, and CD16, and TLR8 across monocytes clusters. (c) Dot plot showing the expression pattern of TLR8 across adult sepsis survivors (S) and non‐survivors (NS), as well as healthy controls (HC).

Figure S6: Pathway enrichment of TLR8^high/low monocyte subsets.

Distinct biological pathways were enriched in TLR8^high and TLR8^low monocytes, indicating functional differences between the two subsets.

Data Availability Statement

VCF files generated from the whole‐exome sequencing analysis can be obtained from the corresponding author upon reasonable request. The single‐cell RNA‐Seq datasets generated in this study have been deposited in Zenodo under the identifier 10.1016/j.clim.2024.110175. The sepsis subjects included in the analysis were categorized as follows: severe sepsis‐AP1, AP3, and AP5; mild sepsis‐AP2, AP4, and AP6.