The participants and study design have previously been presented (16). PWH were recruited from the outpatient clinic of the Department of Infectious Diseases, Oslo University Hospital. INR were defined as < 400 CD4⁺ T cells/µL and IR as having > 600 CD4⁺ T cells/ µL for longer than 3.5 years. All INR (n=4 for RNA-seq and n=5 for proteomics) and IR (n=5, RNA-seq and proteomics) included were male, on continuous efficient ART for > 4 years, < 50 HIV RNA copies/ml plasma > 3.5 years, and the groups were matched on nadir CD4 count (before initiating ART) and age. Men without HIV referred to colonoscopy for control after polyp removal and age-matched with INR and IR, were enrolled as healthy controls (HC, n=5 for RNA-seq and proteomics). The characteristics of the study group are presented in Table 1 .

Single pinch biopsies were obtained from the terminal ileum and the sigmoid colon by colonoscopy as previously described (16). The biopsies were either preserved in RNAlater (Thermo Fisher Scientific, Waltham, MA, USA) at -20°C for transcriptome analyses (RNA-sequencing) or snap frozen in 1.5 mL tubes in liquid nitrogen and stored at -80°C for proteome analyses by liquid chromatography–mass spectrometry (LS-MS/MS).

To integrate the transcriptomic and proteomic datasets from colonic biopsies, we performed a multi-step correlation analysis to identify and characterize concordant molecular features between immunological non-responders (INR) and immunological responders (IR). We compared the DEGs and DEPs lists to identify genes that were significantly dysregulated in both the transcript and protein levels. We then constructed a gene–protein correlation network using the overlapping genes. In this network, node sizes reflected the product of the absolute log₂ fold changes from RNA and protein data, while edge weights indicated the strength of correlation between RNA and protein expression levels. To quantify these relationships, we performed correlation analysis using scatterplots of log₂ fold change values for each overlapping gene and calculated Pearson correlation coefficients (by R built-in function “cor”) to assess the degree of concordance. To create a network plot, the correlation matrix was filtered by coefficient R > 0.5, then further converted into graph matrix by R package igraph using function “graph_from_adjacency_matrix” (parameter weighted = True, mode = “undirected”), further converted as network object by R package Intergraph. The network plot was created from network object using ggplot2 package. Functional annotation of the overlapping genes was carried out using public gene annotation databases, including UniProt, Gene Ontology (GO), and GeneCards, with an emphasis on immune signaling, translational regulation, and mucosal barrier function (by R package clusterProfile and AnnotationDbi). Finally, we performed unsupervised hierarchical clustering based on the integrated RNA and protein expression levels of the overlapping genes (R package p heatmap). Clustering was done using Euclidean distance and complete linkage to assess whether these features could stratify INR and IR samples.

A modified t-test (Wald test) was used in statistical p value analysis for the RNA-seq data. DEGs were defined as adjusted p-value < 0.05. Analyses of DEPs for the proteomics data were performed using Empirical Bayes statistics test. For both RNA-seq and proteomics analyses, the adjusted p-value was used to decrease the false discovery rate (25). The Chi-square test was used to compare regulation in sigmoid colon and terminal ileum (27). For comparisons between paired groups (e.g., INR versus IR within subjects), paired two-tailed t-test was used, while one-way ANOVA was applied to assess differences across more than two paired groups for the deconvolution assay. Gene Ontology (GO) term enrichment was carried out with clusterProfiler, using FDR (adjusted p-value). The correlation between RNA and protein expression levels was assessed by Pearson correlation coefficient. A p-value < 0.05 or adjusted p-value < 0.05 was considered statistically significant throughout.

RNA-seq was performed on endoscopic pinch biopsies of gut mucosa from INR, IR and HC to examine transcriptome regulation in the sigmoid colon and terminal ileum. The participant characteristics are outlined in Table 1 . Differential gene expression analysis, performed using negative binomial GLM fitting and Wald tests, revealed significantly higher expression differences in the sigmoid colon compared to the terminal ileum (p < 0.0001) ( Figures 1A, B ). In the colon, we identified 3326 DEGs when comparing INR to IR, and 202 DEGs when comparing INR to HC ( Figure 1A ). Strikingly, although both INR and IR showed several differences compared with HC (364 and 991 DEGs, respectively) in the TI, there was no difference in the transcriptome between INR and IR in this compartment ( Figure 1B ).

We next examined protein regulation in biopsies from sigmoid colon and terminal ileum by global proteomic analyses, focusing on INR and IR that was the main predefined comparison of the study. We detected a total of 4027 common proteins for all three groups in both tissues after data cleaning and filtering out proteins that were not identified in all replicates for at least one group (INR, IR and/or HC). The data were further evaluated for differential expression by empirical Bayes statistics test. Consistent with our findings at the transcriptional level, more proteins were differentially regulated in sigmoid colon compared to terminal ileum (p < 0.0001; Figures 5A, B ). In the paired comparisons from the three groups (IR, INR and HC), we identified a total of 74 differentially regulated proteins in sigmoid colon ( Figure 5A ) and six in terminal ileum ( Figure 5B ). Twelve proteins were differentially regulated between INR and IR in sigmoid colon but only four in terminal ileum ( Figures 5A, B , respectively).

The most striking differences in gene and protein expression between INR and comparator groups (IR and HC) were found in the sigmoid colon. Transcriptomic analysis identified 3,326 differentially expressed genes (DEGs) between INR and IR in the colon, compared to no significant DEGs in the ileum. Similarly, proteomic analysis revealed 12 differentially expressed proteins (DEPs) in the colon and only four in the ileum. These findings underscore a compartmentalized pattern of immune dysfunction and highlight the distal colon as a uniquely affected site in INR.

The basis for this regional specificity is not fully understood. However, prior studies have reported that the colon may harbor higher levels of residual HIV replication and microbial translocation than the small intestine, which could drive ongoing mucosal immune activation and injury despite ART (3, 62). Our findings support this concept, revealing persistent molecular disturbances in colonic tissue from INR that are absent in the ileum.

The sigmoid colon of INR exhibited extensive transcriptional suppression, particularly affecting key immunological, metabolic, and stress-response pathways. Gene ontology and pathway enrichment analyses revealed significant downregulation of genes involved in protein transport, cytokine signaling, and adaptive immune responses, suggesting a compromised mucosal immune environment. Notably, CD4 mRNA expression was reduced by 37% in INR compared to IR.

We have previously reported that INR have approximately 10% fewer CD4+ T cells in the mucosa of the colon with increased activation and exhaustion of CD4+ T cells, correlating with signs of epithelial damage and mucosal dysfunction (16, 17). The magnitude of CD4 mRNA reduction observed here, suggests that, in addition to the reduced number of CD4+ T cells, INR also have transcriptional downregulation of the CD4 gene itself. This implies that impaired CD4 transcription, in combination with chronic immune activation (2, 3), may contribute to the failure of CD4+ T-cell recovery in INR, extending the mucosal dysfunction initially triggered by CD4⁺ T-cell loss in PWH (16, 17). The HIV Vpu protein and CD4 gene polymorphisms may further suppress CD4 expression (61, 62). Thus, reduced gut mucosal CD4 mRNA is not merely a consequence but also a contributing factor to poor immune reconstitution in INR (3, 28, 29).

Several key immune-regulatory genes were also downregulated in INR, including TGFB1 and IL21R, which play crucial roles in immune modulation, tissue repair, and T/B cell function (33, 34). Similarly, downregulation of HLA-E, a molecule involved in NK cell regulation and epithelial immune tolerance, supports the hypothesis of impaired mucosal immune regulation (35, 36).

In contrast, a limited subset of genes associated with inflammation was upregulated. These include CXCL1, a chemokine known to enhance HIV replication (30, 32), PTPN22, which modulates TCR signaling and T cell activation (37), and GZMA, a granzyme involved in cytotoxic lymphocyte function mediating potential harmful effects on the host (38).

Additional dysregulated genes with recognized roles in HIV pathogenesis included MAP3K5, implicated in stress-induced apoptosis and interactions with HIV proteins (39), and CIITA (–1.138), a master regulator of MHC class II expression and antigen presentation (42). Dysregulation of CIITA further underscores impaired adaptive immune responses in INR, with direct implications for HIV infection.

Proteomic analysis confirmed transcriptomic findings and revealed increased expression of pro-apoptotic and stress-response proteins such as CASP3 and ACE2, the known receptor for SARS-CoV-2 entry, both previously implicated in HIV-mediated epithelial damage and mucosal dysfunction (32, 56, 63). CASP3, a central mediator of apoptosis, was also elevated in INR. This aligns with previous evidence that HIV-1 envelope proteins can trigger CASP3 dependent apoptosis in CD4⁺ T cells through CD4 receptor engagement (32, 56). Additional upregulated proteins included EIF1, RWDD1, and PI4KA, associated with translational control and intracellular signaling (45, 47, 64). Conversely, MSI2, a key regulator of RNA metabolism and autophagy (65), and POSTN, essential for extracellular matrix remodeling and tissue repair (48), were significantly downregulated, suggesting impaired tissue homeostasis and repair capacity.

The differential expression of structural and mitochondrial proteins, such as SPTBN1 and TIMM8A, further supports the presence of cellular stress and compensatory remodeling in INR mucosa (54, 55). Together, these findings reflect widespread disruption of mucosal immune regulation, apoptosis control, antigen processing, and cellular maintenance in INR, despite ART-mediated viral suppression.

B cell–related signaling emerged as one of the most consistently downregulated pathways in colon of INR compared to IR. Genes involved in antigen binding, the immunoglobulin production, and B cell activation were significantly suppressed. This included multiple immunoglobulin variable genes and key cytokine receptors.

Interestingly, immunoglobulin gene expression in INR was not significantly different from healthy controls, suggesting that the defect is not due to baseline normalization but rather to a failure of ART-mediated upregulation. This aligns with prior reports of deficient B cell responses in INR (33, 34) and indicates underlying dysfunction in mucosal B cell differentiation.

RNA deconvolution analysis revealed a reduction in plasmablast frequencies in INR, while memory B cell levels remained comparable. This suggests a specific impairment in B cell activation processes, particularly in proliferation and differentiation into plasmablasts during the immune response to HIV at the mucosal level.

Despite suppression of adaptive immune pathways, the colonic mucosa of INR exhibited increased expression of genes linked to inflammation, including NK cell–associated markers and epithelial stress mediators. REG3α, a known marker of epithelial inflammation, was significantly upregulated in the INR colon compared to HC in our dataset. We have previously reported elevated levels of REG3α in INR (16), and its persistent increase here likely serum reflects ongoing epithelial stress and persistent inflammation as a potential important phenotype in INR in the absence of effective adaptive immunity.

Proteomic profiling supported our transcriptomic findings, revealing dysregulation of proteins involved in translation, apoptosis, and mitochondrial function. The upregulated proteins EIF1, RWDD1, and PI4KA play roles in translation initiation and phosphoinositide signaling (45, 47, 64), and may represent stress-adaptive mechanisms.

In contrast, POSTN, involved in extracellular matrix remodeling and hematopoietic regulation, was downregulated, suggesting impaired tissue repair in INR (50, 66). Decreased expression of the immunoglobulin variable gene IGKV18 suggests impaired local antibody production. Increased levels of CASP3 and ACE2 further point to increased epithelial susceptibility and apoptotic signaling (32, 56). Indeed, in addition to being the receptor for SARS-CoV-2 entry, ACE2 plays a crucial role in modulating gut inflammation and gut microbiome and has also been implicated in the pathogenesis of simian immunodeficiency virus infection (63). Additional upregulation of mitochondrial and structural proteins (SPTBN1, TIMM8A) may reflect epithelial adaptation to sustained cellular stress (54, 55).

In contrast to the colon, the terminal ileum exhibited minimal molecular disturbance, with no significant differentially expressed genes. At the proteomic level, lower levels of MT-ATP8, HMGA1, SST, and ZWINT, proteins involved in mitochondrial function and cellular proliferation, were observed. These findings support a localized rather than systemic mucosal dysfunction in INR, focused on the colon.

Integration of transcriptomic and proteomic datasets highlighted a small group of altered markers, EIF1, RWDD1, PI4KA, and POSTN, that distinguished INR from IR and HC. The observed discordance between mRNA and protein levels for PI4KA and POSTN highlights the complexity of gene regulation and suggests that post-transcriptional mechanisms are involved. PI4KA upregulation at the protein level despite transcript downregulation suggests enhanced translation efficiency or reduced degradation, which may influence phosphoinositide signaling, vesicle trafficking, or viral persistence (61). Although POSTN mRNA levels were elevated in INR colon tissue, its protein levels were suppressed, suggesting a role for post-transcriptional regulation (60). N6-Methyladenosine (m6A) methylation of POSTN mRNA, particularly in the 3′ UTR, could mediate enhanced mRNA decay or impaired translation (60). These molecules may serve as candidate biomarkers for persistent mucosal immune dysregulation and provide mechanistic insights into failed immune reconstitution.

Interestingly, these findings exhibit several features that might relate to Inflammatory Bowel Disease (IBD), particularly through shared mechanisms of immune dysregulation and impaired gut barrier function. Both B cell dysfunction and altered immune responses are hallmark features of IBD, characterized by dysregulated innate and adaptive immune responses [reviewed in (67)]. Moreover, the hypothesis that impaired mucosal barrier function contributes to incomplete immune recovery in INR aligns with the pathogenic mechanisms observed in IBD, where increased intestinal permeability can lead to translocation of luminal antigens and microorganisms, contributing to intestinal inflammation [reviewed in (68)].

Notably, a recent study has shown that REG3α expression is absent in normal colonic tissue but is significantly induced in inflamed colonic biopsies, suggesting its potential role as a marker of inflammation in colonic conditions (69). Immunohistochemistry analyses indicated that REG3α was absent in colon from healthy controls but markedly upregulated in inflamed colonic tissue, suggesting its potential as a biomarker for colonic inflammation (69). We observed increased inflammation and immune activation in INR (16), which are common features in IBD, marked by chronic intestinal inflammation that triggers immune responses through similar pathways (70). Additionally, dysregulation in the adaptive immune response, cytokine signaling, and cellular processes related to tissue repair and growth are observed in IBD, where immune signalling and metabolic pathways are frequently altered, resulting in tissue damage and impaired repair mechanisms (67).

Furthermore, dysregulation of the adaptive immune response, cytokine signaling, and cellular mechanisms associated with tissue repair and growth has been noted in IBD (67). Within this framework, immune signaling and metabolic pathways frequently undergo alterations, resulting in tissue damage and hindered repair processes (66, 69).

While similarities can be identified between INR and IBD concerning disrupted immune and repair pathways, it is essential to acknowledge that persistent immune activation is also evident in IR, representing a shared characteristic among people living with HIV (70, 71). Consequently, instead of implying a direct disease overlap, upcoming research should investigate whether specific pathways or markers identified in IBD may have relevance for HIV-associated immune dysregulation, including INR (68, 70).

It is also important to emphasize that gut mucosal dysfunction should not be considered the sole driver of inflammation in PWH. Other contributing factors—including coinfections, immune senescence, and residual HIV viral replication—play significant roles in sustaining chronic immune activation (13, 15, 16).

Given these shared features, future research may benefit from exploring potential overlaps in pathways and biomarkers between IBD and INR, which could reveal therapeutic targets applicable to both. Specifically, dysregulation in B cell-mediated immunity, NK-cell interactions, and immune signaling pathways involved in tissue repair present viable avenues for targeted therapies. However, while such overlaps offer promising leads, these strategies must be carefully tailored to the distinct immunological context of INR (29, 70). A deeper understanding of the molecular mechanisms driving immune dysregulation in INR is essential before translating IBD-derived treatments, with careful consideration of safety, efficacy, and potential interactions with HIV-specific management strategies.

The strength of our study is the well characterized study groups matched on ethnicity, age, and sex for all groups and nadir CD4 T⁺ count for INR and IR. The biopsies were collected in the same hospital by the same gastroenterologist, from the same parts of the gut, using a well-defined protocol avoiding anatomical bias. Another important aspect of this study is that we investigated both the transcriptomic and proteomic regulation, which provide new insight to the pathogenesis of INR and a resource for developing novel adjuvant therapy. This study also has limitations such as low number of individuals and limited sample material per individual. Only one biopsy per individual was analyzed for RNA or protein regulation. As biopsies only sample a limited area of the mucosa, this can result in sampling bias. Finally, whereas we found markedly different regulation of several transcripts and proteins between INR and IR, the molecular mechanisms for these findings are not clear and the observed associations may not reflect direct causal relationships. Thus, functional validation and mechanistic investigation of these findings, such as through in vitro HIV infection models, B cell co-culture systems, or organoid-based approaches, are necessary to establish direct mechanistic roles, and should be prioritized in future studies.