The intestinal microbiome and metabolome discern disease severity in cytotoxic T-lymphocyte-associated protein 4 deficiency

Abstract

Background

Cytotoxic T-lymphocyte-associated protein 4 deficiency (CTLA4-D) is an inborn error of immunity (IEI) caused by heterozygous mutations, and characterized by immune cell infiltration into the gut and other organs, leading to intestinal disease, immune dysregulation and autoimmunity. While regulatory T-cell dysfunction remains central to CTLA4-D immunopathogenesis, mechanisms driving disease severity and intestinal pathology are unknown but likely involve intestinal dysbiosis. We determined whether the intestinal microbiome and metabolome could distinguish individuals with severe CTLA4-D and identify biomarkers of disease severity.

Results

The genera Veillonella and Streptococcus emerged as biomarkers that distinguished CTLA4-D from healthy cohorts from both the National Institutes of Health (NIH) Clinical Center, USA (NIH; CTLA-D, n = 32; healthy controls, n = 16), and a geographically distinct cohort from the Center for Chronic Immunodeficiency (CCI) of the Medical Center - University of Freiburg, Germany (CCI; CTLA4-D, n = 25; healthy controls, n = 24). Since IEIs in general may be associated with perturbations of the microbiota, a disease control cohort of individuals with common variable immunodeficiency (CVID, n = 20) was included to evaluate for a CTLA4-D-specific microbial signature. Despite common IEI-associated microbiome changes, the two bacterial genera retained their specificity as biomarkers for CTLA4-D. We further identified intestinal microbiome and metabolomic signatures that distinguished patients with CTLA4-D having severe vs. mild disease. Microbiome changes were associated with distinct stool metabolomic profiles and predicted changes in metabolic pathways. These differences were impacted by the presence of gastrointestinal manifestations and were partially reversed by treatment with abatacept and/or sirolimus.

Conclusions

Loss of intestinal microbial diversity and dysbiosis causing metabolomic changes was observed in CTLA4-D. Albeit some of these features were shared with CVID, the distinct changes associated with CTLA4-D highlight the fact that IEI-associated microbiome changes likely reflect the underlying immune dysregulation. Identified candidate intestinal microbial and metabolic biomarkers distinguishing individuals with CTLA4-D based on severity should be studied prospectively to determine their predictive value, and investigated as potential therapeutic targets.

Download video file ^{(143.2MB, mp4)}

Video Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s40168-025-02028-7.

Introduction

Cytotoxic T-lymphocyte-associated protein 4 (CTLA4) is essential to maintain immune homeostasis. It is expressed by T-cells and outcompetes CD28 for its cognate ligands CD80 and CD86 expressed on effector T-cells (T_eff) and antigen-presenting cells [1]. CTLA4 is therefore critical to regulatory T-cell (T_reg) functions such as inhibiting T_eff cell proliferation and controlling antigen presentation [2], which together help maintain tolerance. Although common CTLA4 gene variants have been implicated in the predisposition to several autoimmune conditions [3–5], it was not until 2014 that the first cases of CTLA4 deficiency (CTLA4-D) were described as a rare autosomal dominant immune dysregulation disorder caused by novel germline uniallelic mutations [6, 7]. Patients with CTLA4-D can develop multiorgan autoimmunity and lymphoproliferation that occur with variable severity even among carriers of the same mutation within the same family [6, 8, 9]. The disease spectrum includes enteropathy, hypogammaglobulinemia, recurrent respiratory infections, and lymphocytic infiltration of nonlymphoid organs including the skin, brain, lung, gut, liver, and bone marrow, which are observed along with lymphadenopathy, splenomegaly, autoimmune cytopenia, lymphoma, and autoimmune endocrinopathies [10]. Several patients with CTLA4-D have received allogeneic hematopoietic cell transplantation (HCT) with variable success [11].

The CTLA4 signaling pathway is exploited to treat the clinical manifestations of CTLA4-D with immunomodulators such as sirolimus (a.k.a. rapamycin, a mammalian target of rapamycin [mTOR] inhibitor) and/or abatacept (i.e., a CTLA4 mimetic) [9, 12]. In contrast, the monoclonal antibody-mediated blockade of CTLA4 (e.g., ipilimumab) drives T-cell antitumor activity and has constituted a major breakthrough in the treatment of malignancies such as metastatic melanoma [13]. Unfortunately, immune-related adverse events have been associated with anti-CTLA4 antibodies including the following: vitiligo, colitis, hepatitis, endocrinopathies, inflammatory polyneuropathies, myasthenia gravis, and immune-mediated cytopenia [14].

While T_reg dysfunction remains central to the immunopathogenesis of CTLA4-D, the factors driving disease penetrance (estimated at 70% [9]) and expressivity are unknown. However, the wide range of inflammatory dysregulation attributed to CTLA4-D highlight the importance of disease modifiers, such as the environment. In fact, the microbiota and its associated metabolome represent key potential environmental drivers of disease severity. Meanwhile, the contribution of the intestinal microbiota in shaping T_reg responses both locally and systemically is well-established [15]. Although several key studies have demonstrated the important role of the microbiota [16, 17] and associated metabolites [18, 19] in mediating response to CTLA4 blockers to treat malignancies, where T_reg activity is intentionally dampened, no studies to date have evaluated the role of the intestinal microbiota in driving disease severity in individuals with CTLA4-D. We present the first study evaluating the composition of the intestinal microbiome and metabolome in individuals with asymptomatic or mild compared to severe CTLA4-D. We identified specific bacterial taxa and metabolites associated with severe symptomatic disease and enteropathy, and validated our findings in a separate, geographically distinct patient cohort. Further, we elucidated a specific microbial signature that distinguishes patients with CTLA4-D from those with other IEIs such as common variable immunodeficiency (CVID).

Methods

Study design and population

Patients were recruited from the National Institutes of Health (NIH) Clinical Center, USA (NIH; total n = 48), between December 2013 and August 2019 and from the Center for Chronic Immunodeficiency (CCI) of the Medical Center of the Faculty of Medicine of Freiburg University Treatment Consortium, Germany (CCI; total n = 69, i.e., geographically distinct cohort) between October 2015 and August 2018. All patients providing consent for their stool sample were included, and each cohort was analyzed separately. Data and samples from one visit per patient were included in this cross-sectional observational study.

NIH cohort

Healthy individuals (n = 16) and patients with CTLA4-D (n = 32) provided informed consent in accordance with the Declaration of Helsinki and were enrolled into the National Institute of Allergy and Infectious Diseases (NIAID) Institutional Review Board (IRB)-approved protocol 93-I-0119 — Detection and Characterization of Host Defense Defects (ClinicalTrials.gov: NCT00001355), a prospective longitudinal observational cohort study.

CCI cohort

Healthy individuals (n = 24) and patients with CTLA4-D (n = 25) or CVID (n = 20) were recruited; patient consent was collected by the treating physicians, and samples were processed at the Uniklinik Freiburg under the ethics-approved protocols 526/14, 295/13_140782, and 60/18.

Study procedures

NIH cohort

Clinical and laboratory data was retrospectively collected from electronic medical records in accordance with our IRB-approved protocol, and all participants completed a self-administered questionnaire which included the numeric rating scale (NRS) [20], the Short Inflammatory Bowel Disease Questionnaire (SIBDQ) [21], and the Patient Simple Clinical Colitis Activity Index (P-SCCAI) [22]. A Clinical Activity Index (CAI) was assigned based on the following criteria: CAI 1 = 0–2 bowel movements (BM)/day, CAI 2 = 3–4 BMs/day, CAI 3 = > 4 BMs/day, and/or the presence of blood or mucus in stool and/or fistulae and/or perianal disease. Stool samples were self-collected in sterile dry tubes and immediately stored at − 80 °C. All fecal samples were evaluated for the presence of occult blood using the Hemoccult II SENSA Single Slides System (Beckman Coulter). Fecal calprotectin was measured using a Calprotectin ELISA Assay Kit (Eagle Biosciences).

CCI cohort

Participant clinical and laboratory data, including medication history, were retrospectively collected from the medical charts as per the IRB-approved protocol. Stools were self-collected in tubes containing a stabilizing liquid (STRATEC) which allows for the conservation of samples up to 3 months at room temperature. Upon arrival, stool samples were homogenized, divided into aliquots, and frozen at − 80 °C.

For both the NIH and CCI cohorts, CTLA4-D disease severity was defined according to the “NIH classification.” Patients with CTLA4-D classified as “mild” had no known clinical manifestation of CTLA4-D or had one or more of the following manifestations: hypogammaglobulinemia, dermatological (e.g., vitiligo or psoriasis), and endocrinological (e.g., type 1 diabetes or Hashimoto’s thyroiditis). Patients with any additional manifestation affecting any other organ system were considered either “severe no GI” or “severe GI” depending on whether they also had any gastrointestinal (GI) complications (i.e., diarrhea lasting more than 3 weeks, celiac disease, enteropathy, lymphocytic colitis, inflammatory bowel disease [IBD]). The choice of clinical manifestations distinguishing patients with CTLA4-D with mild vs. severe disease was based on the observation that patients with manifestations beyond those included in the mild classification also had reduced T_reg counts.

Fecal DNA extraction and amplicon sequencing

NIH cohort

Fecal DNA was extracted using the PowerSoil DNA Isolation Kit (MO BIO Laboratories), with the addition of Proteinase K addition (Qiagen) and incubation at 65 °C for 10 min prior to bead beating using a TissueLyser II (Qiagen) at 30 Hz for 2 min. DNA was eluted in UltraClean® PCR Water (MO BIO Laboratories), and 16S rRNA gene (V4 region) sequencing was performed using primers (515F = GTGCCAGCAGCCGCGGTAA and 806R = GGACTACCAGGGTATCTAAT) that were modified to include a linker sequence, a 12-bp index sequence, and a heterogeneity spacer [23]. PCR was set up with a LA PCR™ Kit (TaKaRa Bio), and libraries were cleaned up (Agencourt AMPure XP Kit; Beckman Coulter), quantified (Quant-IT dsDNA High-Sensitivity Assay Kit; Invitrogen), and sequenced using a 250-bp paired-end sequencing protocol on the MiSeq platform (Illumina Inc.) at the NIH Intramural Sequencing Center.

CCI cohort

Microbial DNA was extracted from stool samples using the QIAamp DNA Stool Mini Kit (Qiagen) with the following modifications: stool lysis was performed at 95 °C, and 400 µl of supernatant (instead of 200 µl) was pipetted into 15 μl of proteinase K, to which 400 µl of AL buffer (instead of 200 µl) was added, followed by mixing and incubation at 70 °C for 10 min. 16S rRNA gene (V3–V4 regions) sequencing was performed using the Illumina 16S metagenomic sequencing library protocol (primers: F = TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG and R = GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC). PCR was set up with 2 × KAPA HiFi HotStart ReadyMix (Roche). PCR products were purified using magnetic beads (Agencourt AMPure XP-Kit; Beckman Coulter). The index PCR was performed with Nextera XT Index Kit v2 Set B and 2 × KAPA HiFi HotStart ReadyMix and quantified on a TapeStation using D1000 High Sensitivity reagents from Agilent Technologies. The library was denatured with NaOH, diluted, mixed with 5% PhiX control, and loaded on a MiSeq platform (Illumina Inc.) for high-throughput sequencing (2 × 300 cycle V3 kit).

Stool metabolomics

Liquid chromatography-mass spectrometry with hybrid metabolomics method

Liquid chromatography (LC)-mass spectrometry (MS) analysis and metabolite extraction were performed on stool samples from the NIH cohort as described [24]. MS analyses were performed by coupling the LC system to a Thermo Q-Exactive HF mass spectrometer operating in heated electrospray ionization mode. The parameters included the following: 30 min duration, 3.5-kV spray voltage, 320 °C capillary temperature, 35 sheath gas rate, 10 aux gas, 100 μA of max spray current, 120,000 resolution with an automatic gain control target of 3e6, a maximum IT of 100 ms, and scan range from 67 to 1000 m/z. Tandem MS spectra used the following: 15,000 resolution, AGC target of 1e5, maximum IT of 50 ms, isolation window of 0.4 m/z, isolation offset of 0.1 m/z, fixed first mass of 50 m/z, and 3-way multiplexed normalized collision energies of 10, 35, and 80. The minimum AGC target was 1e4 with an intensity threshold of 2e5, and all data were acquired in profile mode.

Global metabolomics data processing and relative quantification of metabolites

Using an in-house generated Python script, Thermo RAW files were converted to SQLite format to enable downstream peak detection and quantification. MS/MS spectra were searched against the NIST17MS/MS [25, 26], METLIN [27], and respective Decoy spectral library databases [28, 29] applying a 100% false discovery rate. The peak heights for each sample and metabolite were extracted from the sqlite3 files based on the metabolite retention time ranges and accurate masses. Metabolite peaks were extracted based on the theoretical m/z of the expected ion type (e.g., [M + H] +), with a 15 ppm tolerance ± 0.2 min peak apex retention time tolerance within an initial retention time search window of ± 0.25 min. The final peak detection was calculated based on a signal-to-noise ratio of 3 × compared to blank controls with a floor of 10,000 (arbitrary units). The median metabolite intensity of each sample was used to normalize the intensities to account for inter-batch variations.

Amplicon sequence and metabolomics data analysis

Raw sequence reads were analyzed using QIIME2 (Quantitative Insights Into Microbial Ecology version 2). Sequence quality analysis was performed using FastQC and sample sequence inference with DADA2 v1.10 [30]. The sequences were clustered into operational taxonomic units (OTUs) using an open-reference OTU picking workflow to classify the resulting amplicon sequence variants (ASVs) with the SILVA v138 database and the species identified by SATé-Enabled Phylogenetic Placement. For all analyses, microbial features were filtered based on having at least 0.01% relative abundance and a 20% prevalence filter (that is 20% of values had at least 4 counts). The data was normalized using total sum scaling. The following R packages were used to perform the indicated analyses: Phyloseq for diversity trends, community composition visualizations, and alpha diversity [31] with comparisons using the Kruskal–Wallis test, EdgeR [32] for differential abundance analyses, metacoder [33] for heat tree analyses, and lefser version 1.8.0 [34] for linear discriminant analysis (LDA) effect size (LEfSe) analyses to identify microbiome biomarkers [35]. Beta diversity was evaluated by principal coordinates analyses (PCoA) based on the weighted UniFrac distance method, and significance determined by permutational MANOVA (PERMANOVA, 9999 permutations). Predictive functional profiling of the microbial communities was performed using PICRUSt2 [36]. For integrative analyses, MicrobiomeAnalyst 2.0 was used [37]. The heat tree analyses depicting taxonomic differences utilized the median abundance value and the nonparametric Wilcoxon rank-sum test to leverage the hierarchical structure of taxonomic classifications.

Metabolomics data was processed with an in-house pipeline using Metabolyze version 1.0., and t-tests were performed using the Python SciPy library (1.1.0) [38, 39]. The R-library DESeq2 (1.24.0) [40] was used to test for significant differences. Enrichment pathway analyses and metadata analyses were performed using the MetaboAnalyst R package [41]. For accurate prediction of biomarkers, the random forest machine learning algorithm was applied using the randomForest package in R by averaging 1000 decision trees [42].

All diversity trends and community composition visualizations were created using the ggplot2 package in R. This work utilized the computational resources of the NIH HPC Biowulf cluster (http://hpc.nih.gov) and the Nephele platform from the NIAID Office of Cyber Infrastructure and Computational Biology (OCICB) (http://nephele.niaid.nih.gov, Bethesda, MD, USA). All sequences, metadata files, and supplementary files associated with the data analysis are available in the National Center for Biotechnology Information Sequence Read Archive (BioProject ID: PRJNA996628; https://www.ncbi.nlm.nih.gov/sra/PRJNA996628).

Statistical analyses

Statistical analyses of clinical data were performed using GraphPad Prism Version 9.0. Continuous variables were expressed as means with standard deviations (SD) and categorical variables as absolute values with percentages. Unpaired t-tests with Welch correction were used for two-group comparisons, and one-way ANOVA with Welch correction was used for multiple comparisons. All tests were two-sided, and p < 0.05 was considered significant. The correlogram of clinical and laboratory variables was constructed with corrplot package in R. The R versions 4.1.3 to 4.2.2 were used for data analysis and visualization. Figures were assembled using Adobe Illustrator or Photoshop.

Results

Demographic and clinical characteristics

NIH cohort

Clinical characteristics of patients with CTLA4-D compared to healthy individuals recruited from the NIH and included in the study are shown in Supplementary Table S1. The mean age of participants was similar between both groups, but a larger proportion of healthy participants was female (75% vs. 37.5%). Most patients with CTLA4-D were classified as severe (78.2%), and while 71.9% had a history of GI manifestations, only 34.4% had active symptoms at the time of visit. The most common manifestations of CTLA4-D included anemia (50%), endocrinopathy (53.1%), pulmonary (53.1%) and/or central nervous system (CNS) complications (53.1%), and hypogammaglobulinemia (50%). Regarding medications taken regularly, 37.5% of participants were on antibiotics, 43.8% were on sirolimus, 37.5% were on systemic steroids, and 21.9% were on abatacept.

Patients with CTLA4-D had C-reactive protein (CRP) levels that were above the upper limit of normal. Compared to healthy individuals, patients with CTLA4-D were more likely to have detectable fecal occult blood and elevated fecal calprotectin levels. They also had significantly worse quality of life (NRS, SIBDQ) and disease activity (CAI, P-SCCAI) scores (Supplementary Table S2). Correlations between demographic and clinical characteristics are shown in Supplementary Fig. S1.

CCI cohort

Clinical characteristics of patients with CTLA4-D compared to healthy individuals recruited from the CCI are shown in Supplementary Table S3. Compared to the NIH cohort, participants included in the CCI cohort were slightly older (mid-40 s versus early 30 s), and most patients with CTLA4-D were female (64%). Less patients with CTLA4-D had severe disease (56%) and a history of GI manifestations (56%). However, all patients with a history of GI manifestations had active symptoms at time of visit. The most common manifestations of CTLA4-D included hypogammaglobulinemia (56%), endocrinopathy (52%), dermatological manifestations (52%), pulmonary complications (40%), and cytopenia (40%). Significantly, less patients in the CCI cohort were on active medications at the time of visit with 12% on antibiotics, 12% on systemic steroids, and none on sirolimus or abatacept.

Impact of CTLA4 deficiency on intestinal microbiome signatures

We examined whether CTLA4-D was associated with changes in alpha diversity (Fig. 1A1, A2) and bacterial composition (Fig. 1B1, B2) in participants from the NIH and CCI cohorts. The richness analysis giving weight to rare species (Chao1) and diversity indices for richness and evenness (Shannon and Fisher) are shown. In both cohorts, patients with CTLA4-D, independent of any other variable including disease severity, had significantly decreased alpha diversity and distinct bacterial composition as measured by beta diversity. The heat trees (Fig. 1C1, C2) and LEfSe graphs show specific genera (Fig. 1D1, D2) and species (Fig. 1E1, E2) that distinguish the intestinal microbiome in patients with CTLA4-D from healthy individuals in each cohort. Notably, samples from patients with CTLA4-D from the NIH cohort were enriched in Enterococcus spp., Veillonella spp., Ruminococcus gnavus group, and Streptococcus spp. At the species level, samples were enriched in Butyricicoccus sp., Clostridium sp., Bilophila wadsworthia, Alistipes obesi, and Clostridium asparagiforme. Compared to healthy individuals, patients with CTLA4-D were depleted in 10 genera including Butyricicoccus spp., Ruminococcus gauvreauii group, Eubacterium xylanophilum group. Oscillibacter spp., Bilophila spp., Colidextribacter spp., Phascolarctobacterium spp., and 3 Lachnospiraceae spp. (Fig. 1D1, E1, Supplementary Fig. 2A1).

Fig. 1.

Impact of CTLA4 deficiency on intestinal microbiome signatures. Data for the National Institutes of Health (NIH) Clinical Center and Center for Chronic Immunodeficiency of the Medical Center of the Faculty of Medicine of Freiburg Treatment Consortium (CCI) cohorts are designated by 1 and 2, respectively. Microbiome analyses comparing healthy subjects with patients with CTLA4 deficiency (CTLA4) are shown (NIH cohort: healthy n = 16; CTLA4 n = 32 and CCI cohort: healthy n = 23; CTLA4 n = 23). Alpha diversity measures (Chao1, Shannon, and Fisher) with line and whiskers in the box plot representing the median and inter-quartile range, respectively (A1, A2). Principal coordinates analysis (PCoA) plot of beta diversity based on unweighted UniFrac distances for the two comparison groups with p-values determined by analysis of similarities (ANOSIM) (B1, B2). Heat tree depicting the significant differential abundances of bacterial genera between CTLA4 and healthy (red = higher abundance; blue or teal = lower abundance) (C1 and C2). Linear discriminant analysis (LDA) score determined by the LDA effect size (LEfSe) analysis showing the defining genera (D1, D2) and species (E1, E2) for each cohort. p-values are provided for each comparison with p < 0.05 considered as significant

Similar to the NIH cohort, stool samples from patients with CTLA4-D from the CCI cohort were enriched in Veillonella spp. and Streptococcus spp. while being depleted in A. obesi and Butyricicoccus spp. Compared to healthy individuals, they were enriched in Lactobacillus spp. and Streptococcus pneumoniae while being depleted in several other taxa (e.g., Subdoligranulum spp., Faecalibacterium spp., Alistipes spp., Parabacteroides spp., Eubacterium eligens group, Fusicatenibacter spp., Odoribacter spp., Ruminococcus torques, Blautia luti, Blautia faecis, a Ruminococcus sp., and Ruthenibacterium lactatiformans; Fig. 1D2, E2, Supplementary Fig. 2A2). In both cohorts, the phyla Actinobacteriota and Desulfobacterota were significantly less abundant in CTLA4-D (Supplementary Fig. S2B), with the genera Veillonella and Streptococcus being enriched, as predicted by the generalized linear models to find associations between microbial features and having a CTLA4 mutation (Supplementary Fig. S2C).

Fig. 2.

Distinct intestinal microbiome changes associated with CTLA4 deficiency disease severity in the NIH cohort. Comparisons between healthy individuals and patients with different classifications of CTLA4 deficiency based on disease severity are presented for the NIH cohort (healthy n = 16, mild n = 7, severe without gastrointestinal illness [severe no GI] n = 6, severe with GI illness [severe GI] n = 19). A Relative abundance of phyla. B Alpha diversity analyses (*p < 0.05, **p < 0.01). C Principal coordinates analysis (PCoA) plot of beta diversity based on the Bray-Curtis metric with p-values determined by analysis of similarities (ANOSIM; p = 0.029). D Heat tree depicting the significant differential abundance (p< 0.05) of bacterial genera between severe GI and severe no GI groups (red = higher abundance; blue = lower abundance). Linear discriminant analysis (LDA) score determined by the LDA effect size (LEfSe) analysis showing biomarkers at genus (E) and species (F) levels

Impact of disease severity on intestinal microbiome signatures in patients with CTLA4 deficiency

Since disease severity and GI manifestations appeared to impact intestinal bacterial diversity and composition in both cohorts, we next evaluated each cohort separately to determine whether patients with CTLA4-D had distinct intestinal microbiome signatures based on disease severity (Figs. 2, 3 and Supplementary Fig. S3).

Fig. 3.

Distinct intestinal microbiome changes associated with CTLA4 deficiency disease severity in the CCI Cohort. Comparisons between healthy individuals and patients with different classifications of CTLA4 deficiency based on disease severity are presented for the CCI cohort (healthy n = 23, mild n = 9, severe without gastrointestinal illness [severe no GI] n = 4, severe with GI illness [severe GI] n = 10). A Relative abundance of phyla. B Alpha diversity analyses (*p < 0.05,**p < 0.01, ***p < 0.001). C Principal coordinates analysis (PCoA) plot of beta diversity based on the Bray-Curtis metric with p-values determined by analysis of similarities (ANOSIM; p = 0.001). D Heat tree depicting the significant differential abundance (p < 0.05) of bacterial genera between severe GI and severe no GI groups (red = higher abundance; blue = lower abundance). Linear discriminant analysis (LDA) score determined by the LDA effect size (LEfSe) analysis showing biomarkers at genus (E) and species (F) levels

NIH cohort

Stool samples from patients with CTLA4-D classified as mild or severe no GI had similar phylum distributions and alpha diversity measures to healthy individuals, whereas patients with severe disease and a history of GI manifestations (severe GI) had reduced intestinal microbiome diversity (Fig. 2A, B) which was characterized by a relative increase in Proteobacteria and decrease in Actinobacteriota and Desulfobacterota (Supplementary Fig. S3A) and a distinct bacterial composition (Fig. 2C). More specifically, samples from patients with CTLA4-D classified as severe GI were enriched in Veillonella spp., while samples from patients classified as severe no GI were enriched in Ruminococcus spp. (including R. gnavus group and Ruminococcus bicirculans), Coprococcus spp., Collinsella spp., and Clostridium leptum (Fig. 2D, E, F), indicating potential microbial markers of disease severity. Bacteria from the phyla Verrucomicrobiota and Desulfobacterota were significantly less abundant in severe GI patients compared to mild (Supplementary Fig. S3B). Intestinal microbiome signatures from mild patients were less distinct from healthy individuals (Supplementary Fig. S3C) and were enriched in Dorea spp., Eubacterium hallii, Parabacteroides spp., Intestinimonas spp., Parabacteroides merdae, B. wadsworthia, and Clostridium glycyrrhizinilyticum (Fig. 2E, F).

CCI cohort

In the CCI cohort, patients classified as mild had similar alpha diversity indices as healthy individuals, whereas patients with severe disease, whether they had a history of GI manifestations or not, had the least diverse microbiome (Fig. 3A, B). Unlike samples from the NIH cohort, samples from the CCI cohort had increased relative abundance of Actinobacteriota but not Proteobacteria (Fig. 3A). However, bacterial community composition was distinct between samples from patients with different degrees of disease severity (Fig. 3C). Similar to the NIH cohort, samples from patients with CTLA4-D and classified as severe GI were enriched in Veillonella spp., in addition to S. pneumoniae. Samples from patients classified as severe no GI were enriched in Streptococcus spp., Monoglobus spp., Erysipelatoclostridium spp., and R. lactatiformans. Samples from patients with mild disease were enriched in several taxa, including Subdoligranulum spp., Fusicatenibacter spp., Lactobacillus spp., Alistipes onderdonkii, B. faecis, and R. torques, which was also enriched in mild patients from the NIH cohort (Fig. 3E, F).

Predictive functional profiling by PICRUSt2 analysis of the microbiome in the NIH cohort comparing stool samples from healthy individuals with patients with CTLA4-D by disease severity showed that a subgroup of patients with severe manifestations of CTLA4-D had microbiome profiles predicted to have significantly increased activity in several pathways involved in menaquinol biosynthesis (Supplementary Fig. S4A). In the CCI cohort, a subgroup of healthy individuals also had microbiome profiles predicted to have increased menaquinol biosynthesis (Supplementary Fig. S4B).

Several patients with CTLA4-D had active GI or a history of GI disease at the time of sample collection. We therefore evaluated intestinal dysbiosis in patients with CTLA4-D from both cohorts based on previous GI disease (Supplementary Fig. S5). As observed in several previous studies, a history of GI disease was associated with changes in phylum distribution and beta diversity (Supplementary Fig. S5A, B). Interestingly, Veillonella spp. and Streptococcus spp. were no longer identified by LEfSe as markers of a history of GI disease in the NIH cohort, although the log2FC levels were significantly higher in CTLA4-D with a history of GI disease versus no known previous GI history (Supplementary Fig. S5C1, D1). In the CCI cohort, Veillonella but not Streptococcus spp. were associated with a previous GI history (Supplementary Fig. S5C2, D2).

Microbiome changes together with their predicted functional profiling indicate a clear reduction in microbial diversity associated with increased disease severity, with the Veillonella genus emerging as a potential biomarker for the severe GI phenotype in patients with CTLA4-D.

Microbial signatures specific to CTLA4 deficiency compared to CVID

Our high-level analyses revealed intestinal microbiome signatures that readily differentiate patients with CTLA4-D from healthy individuals. We next examined if these biomarkers are specific to CTLA4-D by comparing the CTLA-D intestinal microbiome to that of a cohort of patients with CVID (i.e., an IEI disease control group). The CVID cohort was diverse in terms of clinical presentation, disease severity, and type of immune deficiency. The microbiome signature observed in this cohort therefore represented “non-specific or generic” dysbiosis associated with having an IEI. In other words, the CVID cohort was considered as a negative control to subtract background microbiome changes resulting from immune deficiency. Since CTLA4-D is an autosomal dominant disease that can often involve a CVID phenotype, several clinical and demographic characteristics overlapped between the CTLA4-D and CVID cohorts (Supplementary Table S3). Gut diseases such as colitis (n = 3), gastritis (n = 5), diarrhea (n = 8), intestinal lymphoid nodular hyperplasia (n = 2), and GERD (n = 3) were observed in patients with CVID. There were no significant differences in beta diversity (Fig. 4A) and phylum distribution (Fig. 4C) between the CTLA4-D and CVID cohorts. However, random forest analyses confirmed the discriminatory power of the genera Veillonella, Streptococcus, and Granulicatella, among others, as biomarkers for CTLA4-D (Fig. 4B). Analysis of genera based on differential abundance fold changes in EdgeR identified 14 genera that can potentially distinguish patients with CTLA4-D from healthy (WT) and patients with CVID (Fig. 4D, F). The common and distinct genera are listed in Supplementary Table S4. The log-transformed counts of Streptococcus spp., Veillonella spp., and Granulicatella spp. showed enrichment specifically in CTLA4-D, while Coprobacter spp. were significantly decreased.

Fig. 4.

Microbial biomarkers that distinguish CTLA4 deficiency from another inborn error of immunity. Comparisons between healthy (WT) individuals, patients with CTLA4 deficiency (CTLA4), and patients with common variable immunodeficiency (CVID) (C) from the CCI cohort (healthy n = 23, CTLA4-D n = 23, CVID n = 20) are shown. A Principal coordinates analysis (PCoA) plot of beta diversity based on the Bray–Curtis metric with p-values determined by analysis of similarities (ANOSIM). B Random forest plots highlighting microbial biomarkers at the genus level (red = high abundance, blue = low abundance). C Relative abundance of phyla. D Venn diagram of differentially abundant genera between comparison groups (genera are listed in Supplementary Table S4). E Genera that are distinct in CTLA4 deficiency. F Log-transformed counts of bacterial genera that are potential biomarkers of CTLA4 deficiency

Intestinal metabolomic profiles in patients with CTLA4 deficiency

To begin to elucidate the metabolic function associated with the described microbiome signatures, we evaluated the fecal metabolome in patients with CTLA4-D from the NIH cohort (Fig. 5). Fecal samples from patients with mild CTLA4-D had similar metabolomic profiles to healthy individuals (Fig. 5A, B), whereas patients with severe CTLA4-D (with or without GI manifestations) had distinct and less diverse metabolomic profiles compared to healthy and mild groups, reflecting similarities with microbiome profiles. Notably, samples from patients with CTLA4-D had significantly increased levels of cis-4,10,13,16-docosatetraenoic acid (p = 0.003) and decreased levels of adenosine, 2′-deoxy (p = 0.009) compared to healthy individuals (Fig. 5C, E), and both metabolites were predicted to be metabolic makers for CTLA4-D by random forest analysis (Fig. 5D). Pathway analysis of metabolic profiles highlighted many of the amino acid metabolism pathways including phenylalanine, tyrosine, and tryptophan biosynthesis as being impacted in patients with CTLA4-D (Fig. 5F). To identify markers of severity, mild and severe patients were compared. The area under receiver operating characteristic curve (AUROC) identified reduced tris (2-butoxyethyl) phosphate levels (Fig. 5G, H) in patients with severe disease. The fecal metabolomic profiles clearly demonstrated that amino acid and nucleotide metabolic pathways were highly affected in severe compared to mild patients (Fig. 5I, upper graph) with tris(2-butoxyethyl) phosphate potentially being a marker for progression to severe disease. Moreover, the observed metabolic profiles have been associated with other inflammatory and/or autoimmune pathologies, including colorectal cancer, inflammatory bowel disease, and enthesitis-related arthritis (Fig. 5I, lower graph).

Fig. 5.

Metabolomic changes in CTLA4 deficiency. Heatmap (A), principal coordinates analysis (PCoA) plot (B), and volcano plot (C) of the metabolomic profiles distinguishing healthy individuals (n = 16) from patients with CTLA4 deficiency (CTLA4, n = 25). In the heatmap, rows display metabolites, and columns represent samples (blue = decreased, red = increased). The brightness of each color corresponds to the magnitude of the difference when compared with average values. The lines in the volcano plots indicate the significance cutoff for p-value (-log10 p-value of 1.3013 corresponding to p < 0.05) and fold change (FC) (log₂FC > 2, log₂FC < −2). All significantly different metabolites are shown. D Random forest plots highlighting metabolic biomarkers (red = high intensities, blue = low intensities). E Box-whisker plots showing metabolite levels with their area under the curve (AUC) for the receiver operating characteristics (ROC) analysis. F Impacted pathways with their respective -log₁₀ p-values. The size of the bubble is proportional to the size of the p-value (the larger the size, the higher the significance). G ROC curves are shown for the individual biomarkers along with the AUC value and 95% confidence interval. H Distribution of individual values is shown in a box-whisker plot. I Enrichment ratio is shown for significant metabolic pathways (top) and associated disease databases (bottom) according to CTLA4 deficiency disease severity. p-values are indicated wherever required, and p < 0.05 is considered significant

Treatment with inhibitors of T-cell- or immune cell-mediated inflammation was associated with improved intestinal dysbiosis in patients with CTLA4 deficiency

To determine other major drivers of intestinal microbiome changes, we compared measures of alpha and beta diversity based on the presence or absence of GI manifestations and current treatments in the CTLA4-D groups from NIH and CCI (Supplementary Fig. S6A, B). Age and sex had modest effects on microbial diversity and community structure in both cohorts, while any history of or active GI disease significantly impacted both alpha and beta diversity in the NIH cohort and had only a mild impact on beta diversity in the CCI cohort.

Patients with CTLA4-D and severe manifestations are often treated with immunomodulators, particularly abatacept (i.e., a CTLA4 mimetic) and/or sirolimus (a.k.a. rapamycin, mTOR inhibitor) (Supplementary Fig. S7), which may alter T_reg function in patients with CTLA4-D and by extension have an important impact on the microbiota [43]. As previously mentioned, an important difference between the NIH and CCI cohorts was the fact that 56% of patients with severe disease in the NIH cohort were on sirolimus and/or abatacept compared to none in the CCI cohort (Supplementary Tables S1 and S3). We further observed that having a history of GI manifestations seemed to be one of the main factors impacting microbiome signature in the NIH cohort (Supplementary Fig. S5), whereas in the CCI cohort, having severe disease independent of GI manifestations altered the microbiota. We therefore hypothesized that treatment with sirolimus and/or abatacept improved intestinal dysbiosis in the NIH cohort, thereby resulting in GI manifestations having more impact on the intestinal microbiome than disease severity. In contrast, the lack of treatment with abatacept and/or sirolimus in the CCI cohort further highlighted the impact of CTLA4-D disease severity on the intestinal microbiome. Thus, we evaluated the impact of treatment with sirolimus and/or abatacept on the microbiome in patients with CTLA4-D from the NIH cohort.

When comparing patients that were on sirolimus and abatacept, which were mostly patients with CTLA4-D classified as severe GI, to other CTLA4-D patients who were not on these drugs, we observed significant increases in certain taxa including Holdemanella spp., Frisingicoccus spp., Slackia spp. and Alistipes spp., and decreased abundance of Lactobacillus spp. (Fig. 6A). Patients on sirolimus (with or without abatacept) had increased relative abundance of Holdemanella spp., Streptococcus anginosus, unidentified species from the genera Odoribacter and Fusobacterium, and uncultured Eubacterium from the Lachnospiraceae family (Fig. 6B). Random forest analyses identified metabolites predicted to be significantly associated with the administration of abatacept and sirolimus (Fig. 6A) or sirolimus (Fig. 6B). Both drugs increased levels of DL-2-methyl glutamic acid, a derivative of glutamic acid, and reduced 4-(dimethylamino)-benzoic acid, 2,5-dioxo-1-pyrrolidinyl ester (DMABA NHS ester), a synthetic derivative of the naturally occurring amino acid, N-dimethylamino-2-butanol (DMAB). Sirolimus treatment was clearly associated with changes in the metabolism of amino acids identified to be altered in CTLA4-D. Each treatment was associated with clinical improvement and restoration of changes in abundance of certain bacteria and metabolites related to amino acid metabolism in patients with CTLA4-D.

Fig. 6.

Effects of abatacept and sirolimus on restoring the intestinal microbiota in patients with CTLA4 deficiency. A, B Linear discriminant analysis (LDA) score determined by LDA effect size (LEfSe) analysis showing bacterial biomarkers at the genus (left) and species levels (middle), and random forest plots (right) showing fecal metabolic biomarkers (red = high abundance, blue = low abundance) for treated compared to untreated patients with CTLA4 deficiency (CTLA4). C Heatmap showing log₂ fold change (FC) of distinguishing bacterial genera for the listed comparisons (i.e., untreated patients with CTLA4 deficiency vs. healthy individuals (WT), sirolimus-treated vs. untreated patients with CTLA4 deficiency, abatacept- and sirolimus-treated vs. untreated patients with CTLA4 deficiency). D Bar graph showing log₂FC of bacterial genera that were restored in treated groups compared to untreated patients with CTLA4 deficiency

The therapeutic effect of these drugs on CTLA4-D-induced intestinal dysbiosis was further evaluated by comparing the differential abundance of genera in untreated patients (i.e., no treatment with abatacept or sirolimus) with CTLA4-D with either 1) healthy individuals (WT), 2) patients with CTLA4-D treated with sirolimus, or 3) patients with CTLA4-D treated with sirolimus and abatacept (Fig. 6C). Of the 6 genera that were decreased in CTLA4-D patients compared to healthy individuals, 3 were restored with treatment, including a 9-fold (i.e., 3.07 log₂ fold) increase in Lachnospiraceae spp. Several of the bacterial genera found to be abundant in untreated patients with CTLA4-D were significantly reduced with immunomodulator treatment. The highest treatment-driven differences were noted for Akkermansia spp., Frisingicoccus spp., and Veillonella spp. Interestingly, the genera Akkermansia, R. gnavus group, and Veillonella, which were found to be enriched in untreated patients with CTLA4-D from both the NIH and CCI cohorts (red bars; Fig. 6D), were reduced > 2.5 log₂ fold in patients with CTLA4-D treated with an immunomodulator from the NIH cohort (yellow and blue bars; Fig. 6D). Treatment with both sirolimus and abatacept showed better reversal of intestinal dysbiosis compared to treatment with sirolimus alone (Fig. 6C, D).

Discussion

To our knowledge, this is the first study to date evaluating the intestinal microbiome and metabolome in patients with CTLA4-D. Our study is also novel insofar as it identified changes in the intestinal microbiome and metabolome that distinguished patients with severe disease, and identified both common and distinct microbial signatures of CTLA4-D compared to other IEIs such as CVID. Our study also included a second cohort of patients recruited from a different continent, which allowed us to identify a core group of bacterial taxa that were altered in the context of severe disease in both cohorts. Moreover, we identified that treatment with sirolimus and/or abatacept improves clinical disease in association with the reduction of certain bacterial taxa, and metabolic changes.

Immune checkpoint inhibitors such as CTLA4 monoclonal antibodies (e.g., ipilimumab, tremelimumab) are increasingly used to treat various cancers. Despite their therapeutic advantages, CTLA4 blockers can result in life-threating immune-related adverse events [44] that are associated with dysregulation of GI immunity [45]. Meanwhile, Ctla4^−/− mice exhibit diffuse infiltration of multiple organs with inflammatory immune cells and peripheral T-cell proliferation resulting in fatal multiorgan tissue destruction [46, 47]. Although the clinical presentation of patients with CTLA4-D shares some similarities with the side effects of treatment with CTLA4 inhibitors, and immune responses observed in Ctla4^−/− mice (e.g., GI illness, lymphoproliferation, tissue destruction etc.), the manifestations of CTLA4-D in both study cohorts ranged from mild to severe even within the same family [6, 9], suggesting that an extrinsic factor may be driving disease severity.

A potential role for the microbiome as a driver of disease severity in CTLA4-D is not only presented in this study but also supported by studies reporting changes in the structure and function of the microbiome with CTLA4 blockade (either alone or in combination with other immune checkpoint inhibitors) and their association with response to therapy [48–51]. Indeed, studies have demonstrated that the efficacy of CTLA4 blockade partly depends on the host microbiota by showing, for example, that fecal microbiota transplantation (FMT) from patients with multiple myeloma treated with CTLA4 blockade into germ-free mice reduced tumor size in a manner that depended on the presence of distinct Bacteroides species [16]. Similarly, two clinical trials demonstrated that FMT can promote therapeutic response in patients with immunotherapy-refractory melanoma [52, 53]. In these studies, FMT promoted favorable changes in proteomic and metabolomic signatures, immune cell infiltrates and CD8 + T-cell activation, and decreased the frequency of interleukin-8-expressing myeloid cells. Other studies have underlined how certain changes in the intestinal microbiota may predispose to colitis associated with anti-CTLA4 and/or anti-PD1 therapy [54–57]. Meanwhile, treatment with certain probiotics [58–60] or FMT [61–63] may help treat immune checkpoint inhibitor-associated colitis by altering the proportion and function of T_regs and other immune cells in the intestinal mucosa in addition to reconstituting the gut microbiota. Our study established the direct impact of CTLA4 on intestinal health through alterations in the microbiome and associated metabolome by demonstrating that immune dysregulation resulting from CTLA4-D leads to altered intestinal homeostasis with reduced bacterial alpha diversity and significant changes in beta diversity in comparison to healthy; the degree of impact likely agree with clinical severity and possibly gene penetrance.

CTLA4-D was associated with significantly reduced relative abundances of Actinobacteriota and Desulfobacterota, and increased Proteobacteria compared to healthy individuals. The genera Veillonella and Streptococcus were significantly enriched in CTLA4-D subjects compared to healthy individuals in both cohorts. There was also a large number of bacterial genera reported to have beneficial intestinal functions that were reduced in patients with CTLA4-D compared to healthy individuals, which included but was not limited to Oscillibacter (i.e., has positive regulatory effects in obesity and chronic inflammation [64]), Odoribacter (i.e., limits colonic inflammation when transplanted for ulcerative colitis [65]), Dorea (Lachnospiraceae family, associated with positive response to CTLA4 blockade [66]), Collinsella (i.e., produces ursodeoxycholate and mitigates COVID-19 [67]), and Butyricicoccus (i.e., produces the short-chain fatty acid (SCFA) butyrate which has anti-inflammatory properties and helps maintain intestinal barrier integrity [68, 69]). In patients with severe GI CTLA4-D, altered abundances of Firmicutes, Bacteroidota, and Proteobacteria were observed in both cohorts. Moreover, samples from patients with severe GI CTLA4-D were enriched with members of the R. torques group but depleted in Erysipelatoclostridium, Peptostreptococcaceae, and Clostridium spp. The genus Veillonella was again significantly enriched in severe GI patients from both cohorts, while the species S. pneumoniae was only enriched in severe GI patients from the CCI cohort. Veillonella was also enriched in patients with CTLA4-D and any previous history of GI disease. However, the temporal relationship needs to be studied in detail to understand if enrichment with Veillonella drives GI severity or vice versa.

While deeper metagenomic sequencing will be required to identify the specific species or strains of Veillonella that are enriched in patients with severe GI CTLA4-D, several studies have identified Veillonella spp. and specifically Veillonella parvula, as markers of intestinal dysbiosis and drivers of several inflammatory conditions when in overabundance. These include IBD, Hirschsprung’s disease-associated enterocolitis, primary sclerosing cholangitis, and systemic lupus erythematosus [70–72]. We further reported enrichment with Holdemanella spp. in association with abatacept and/or sirolimus treatment. Interestingly, Holdemanella biformis (the human homolog of Faecalibaculum rodentium found in rodents) may protect against intestinal tumorigenesis [73] and is known to produce the SCFA butyrate, which helps mitigate inflammation and improve T_reg responses [74–76]. Treatment with abatacept and/or sirolimus was associated with the improvement of intestinal dysbiosis as evidenced by increased diversity in the CTLA4-D-treated group and the absence of significant differences in alpha and beta diversity compared to healthy individuals. Combination therapy with abatacept and sirolimus was effective in reducing Veillonella spp. and increasing Anaerostipes, a bacterial genus associated with a beneficial role in gut health through its conversion of dietary inositol into the SCFAs propionate and acetate [77].

Reduced alpha and beta diversity associated with increased levels of circulating lipopolysaccharide (LPS) has been reported in other IEIs such as CVID [78]. Notably, patients with CVID and non-infectious complications often have worse clinical outcomes associated with intestinal dysbiosis [79]. Similar findings were observed in patients with CTLA4-D, leading us to perform comparative analyses between CVID and CTLA4-D to confirm the specificity of the microbiome signatures identified in CTLA4-D. Like in CVID, the alpha diversity was reduced in patients with CTLA4-D compared to healthy individuals with shared changes in the microbial composition. Nevertheless, the CTLA4-D microbiome signature was distinct from that observed in CVID with significantly increased abundance of Veillonella, Streptococcus, and Granulicatella and almost undetectable levels of Coprobacter. Our observations highlight that although reduced microbial diversity can be observed in other IEIs, specific microbial changes can distinguish certain IEIs while possibly reflecting the underlying immune dysregulation.

T-cell receptor engagement with the major histocompatibility complex accompanied by a costimulatory signal from CD28 leads to T-cell activation, proliferation, and differentiation, all of which are coupled with increased metabolic requirements for substrates such as glucose, and amino acids such as glutamine (Gln), alanine (Ala), serine (Ser), leucine (Leu), methionine (Met), arginine (Arg), cysteine (Cys), and cystine (Cys-Cys) [80]. In patients with CTLA4-D, these metabolic pathways were significantly associated with increased disease severity. It is plausible that microbiome changes directly impacted intestinal metabolism, which by extension, contributed to increased T-cell activity in patients with CTLA4-D. Future studies should address how microbiome-mediated changes in amino acid, glucose, and nucleotide metabolism in the gut impact T-cell phenotypes and function.

Streptococcus spp. and Veillonella spp. can be found in the human small intestine, and their co-occurrence has been demonstrated in various studies [81]. The small intestinal microbiota, including streptococci, is involved in the conversion of simple carbohydrates to lactic acids, which is a preferred substrate for Veillonella [82]. The co-occurrence of these genera together with Granulicatella might depend on their potential for metabolic interactions as shown in the oral cavity [83]. These metabolic interactions could be a key determining factor in the enrichment of certain microbial species, leading to alterations in the intestinal metabolome, which can then impact host immune responses. Interestingly, the predictive pathway analysis associated with microbiome changes in CTLA4-D was characterized by substantial changes in menaquinone (i.e., vitamin K2) biosynthesis, especially in patients with severe disease. Interestingly, vitamin K deficiency has been observed in chronic GI disorders and IBD [84]. However, the underlying mechanisms remain to be elucidated. It is plausible that CTLA4-D-associated intestinal dysbiosis can impactvitamin K2 levels, and its repletion through supplementation may improve both intestinal dysbiosis and inflammation [85].

Undoubtedly, 16S rRNA gene sequencing analysis of the intestinal microbiome combined with metabolomics in patients with CTLA4-D uncovered several interesting new findings. However, future studies should include metagenomic sequencing analysis of fecal and intestinal mucosal samples to identify specific bacterial strains distinguishing patients with CTLA4-D with severe disease from those with asymptomatic or mild disease. Longitudinal studies are also needed to identify microbial biomarkers that will predict who will develop severe disease as these patients may preemptively be treated with immunomodulators or in some cases HCT. Together, these studies would lead to the identification of more specific biomarkers and novel therapeutic strategies for CTLA4-D, including FMT or probiotics, while providing additional insight into the relationship between immune dysregulation, microbial metabolism, and disease penetrance.

Conclusions

Individuals with clinically severe CTLA4-D have distinct intestinal microbiome and metabolome signatures that are impacted by GI pathology and improve with immunomodulator treatment. Our study identified candidate microbes and metabolites to be further investigated as biomarkers that may predict progression to severe CTLA4-D and/or as novel therapeutic targets.

Abbreviations

ASV

Amplicon sequence variants

AUROC

Area under receiver operating characteristic curve

BM

Bowel movement

CAI

Clinical activity index

CCI

Center for Chronic Immunodeficiency

CRP

C-reactive protein

CNS

Central nervous system

CTLA4

Cytotoxic T-lymphocyte antigen 4

CTLA4-D

Cytotoxic T-lymphocyte antigen 4 deficiency

CVID

Common variable immunodeficiency

FOB

Fecal occult blood

GI

Gastrointestinal

HCT

Hematopoietic cell transplantation

IBD

Inflammatory bowel disease

IEI

Inborn errors of immunity

IRB

Institutional Review Board

LC

Liquid chromatography

MS

Mass spectrometry

LDA

Linear discriminant analysis

LEfSe

Linear discriminant analysis effect size

mTOR

Mammalian target of rapamycin

NIH

National Institutes of Health

NIAID

National Institute of Allergy and Infectious Diseases

NRS

Numeric rating scale

OTU

Operational taxonomic units

P-SCCAI

Patient Simple Clinical Colitis Activity Index

PCoA

Principal coordinate analyses

QIIME2

Quantitative Insights Into Microbial Ecology version 2

SD

Standard deviation

SIBDQ

Short Inflammatory Bowel Disease Questionnaire

T_reg

Regulatory T-cell

Authors’ contributions

E.L.F. conceived the project, designed the study and experiments, collected and analyzed data, performed experiments, wrote the manuscript. G.U. and T.H. cared for patients, participated in patient recruitment, performed samples and data collection and edited the manuscript. P.C. analyzed data, produced figures and wrote the manuscript. Y.H., Y.S, D.J, C.D. performed experiments, analyzed data and edited the manuscript. C.G., V.C., P.S. and S.C. performed bioinformatics and/or statistical analyses and edited the manuscript. G.U., S.M.H., J.A.S. and B.G. contributed to study and experimental design, patient care, provided valuable feedback and support, and edited the manuscript. The following authors participated in study design, patient recruitment, sample collection data extraction and analysis, and editing of the manuscript: M.K., N.M., A.G., C.N., M.P.

Funding

This project was supported by the Division of Intramural Research (DIR), the National Institute of Allergy and Infectious Diseases (NIAID), the National Human Genomic Research Institute (NHGRI), and the Bioinformatics and Computational Biosciences Branch (BCBB) Service Contract HHSN3162013000006W of the National Institutes of Health (NIH). This study was funded in part by the NIH Bench-to-Bedside (BtB) Program grant 480707 “Deciphering Immune Dysregulation Due To Germline CTLA4 Mutations.” B. G. is funded by the Deutsche Forschungsgemeinschaft (RESIST — EXC 2155 — Project ID 390874280; CIBSS — EXC-2189 — Project ID 390939984; SFB1160/3_B5; and GR 1617/17–1 — project no. 519635399), the EU-funded PhD program IMMERGE (https://immergeproject.eu), the BMBF rare disease program (GAIN 01GM2206A), and the Wilhelm Sander-Stiftung, Förderantrags-Number. 2023.115.1. This work utilized the computational resources of the National Institutes of Health (NIH) High-Performance Computation (HPC) Biowulf cluster (http://hpc.nih.gov) and the Nephele platform from the NIAID Office of Cyber Infrastructure and Computational Biology (OCICB; http://nephele.niaid.nih.gov). M.K. is supported by the Deutsche Forschungsgemeinschaft SFB1160_2 as a clinician-scientist associated with the IMM-PACT-Program, Faculty of Medicine, University of Freiburg, Freiburg, Germany. E.L.F. is supported by a Canada Research Chair (Tier 2) in “Role of the Microbiome in Primary Immune Deficiencies,” the Canadian Institutes of Health Research (CIHR), Fonds de Recherche du Québec (FRQ), and a J-Louis Lévesque Foundation Research Chair.

Data availability

All sequences, metadata files and supplementary files associated with the data analysis are available in National Center for Biotechnology Information Sequence Read Archive (Bio Project ID: PRJNA996628; https://www.ncbi.nlm.nih.gov/sra/PRJNA996628).

Declarations

Ethics approval and consent to participate

NIH cohort

Healthy individuals and patients with CTLA4-D provided informed consent in accordance with the Declaration of Helsinki and were enrolled into the National Institute of Allergy and Infectious Diseases (NIAID) Institutional Review Board (IRB)-approved protocol 93-I-0119 — Detection and Characterization of Host Defense Defects (ClinicalTrials.gov: NCT00001355), a prospective longitudinal observational cohort study.

CCI cohort

Healthy individuals and patients with CTLA4-D or CVID were recruited; patient consent was collected by the treating physicians, and samples were processed at the Uniklinik Freiburg under the ethics-approved protocols 526/14, 295/13_140782, and 60/18.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.