Intestinal resident effector–memory CD4 T cells on the adaptive-innate spectrum comprise IL-18 reactivity and adaptive CMV specificity Authors: Tobias Kammann, Isabel Voigt, Nicole Wild, Efthymia Kokkinou, Curtis Cai, Takuya Sekine, Vera Nilsén, Whitney Weigel, Chris Tibbitt, Christopher Stamper, Anne Marchalot, John Bassett, Jyotsana Kaushal, Sabrina Ferreira, Elli Mouchtaridi, Elisa J. M. Raineri, Thomas R. Müller, Olga Rivera Ballesteros, Yu Gao, Akhirunnesa Mily, Sarah Adamo, Christian Constantz, Inga Rødahl, Demi Brownlie, Jakob Michaelsson, Jenny Mjösberg, Markus Birk, Marcus Buggert, Nicole Marquardt, Carl Jorns, Johan K. Sandberg Published: 2026 ABSTRACT The human tissue–resident T cell pool consists of a diverse array of conventional adaptive as well as unconventional invariant T cells that populate barrier sites and carefully balance immune defense in the context of microbial exposure. Here, we investigated the tissue-resident CD4 T cell compartment across donor-matched lymphoid and nonlymphoid tissues and uncovered a population enriched primarily in the ileum and liver displaying a rich polyfunctional profile with diverse T helper 1/17/22 characteristics. These CD4 T cells were highly responsive, susceptible to innate cytokine–driven activation, and identified by CD161 and CD56 coexpression. Their transcriptional and protein expression profile included innate-like effector features including NKp80, NKG2D, NKG7, as well as granzyme K, and exhibited characteristics of tissue residence, persistence, and barrier repair, and they were depleted in the colon of patients with inflammatory bowel disease. Notably, the CD161+ CD56+ CD4 T cell pool was expanded in cytomegalovirus (CMV)–seropositive individuals and enriched in CMV-specific T cell receptors and responses. Together, we identify a potent effector-memory CD4 T cell subset resident in the gastrointestinal tract and liver, defined by enhanced innateness and enriched in CMV-specificity. INTRODUCTION The human intestinal T cell compartment comprises a diverse array of specialized subsets that carefully balance tolerance and immune activation in the context of continuous exposure to microbiota, as well as bacterial and viral pathogens (1, 2). At mucosal barriers, the family of unconventional T cells contribute to rapid and targeted immune responses specific for evolutionarily conserved antigens while maintaining sensitivity to proinflammatory mediators in the tissue microenvironment (3, 4). These unconventional T cells often express the C-type lectin CD161, sometimes paired with the archetypical marker of human natural killer (NK) cells, CD56, a heavily glycosylated receptor that plays an enigmatic role in immunity besides its well-known function in mediating neural adhesion (3, 4). Beyond the unconventional T cells, CD161 and CD56 are expressed by substantial fractions of conventional adaptive αβT cells (5). Tissue-resident memory CD8 T (TRM) cells have emerged as key players in maintaining tissue homeostasis. CD8 TRM cells provide sentinel immunity through specialized effector programs (6–9). Some CD8 TRM populations express CD161 or CD56 and may share the capacity for innate cytokine–mediated activation with the unconventional T cells (10, 11). Effector-memory CD4 T cell populations are similarly distributed across human tissues, yet their roles are less well defined (12–14). In the human intestinal epithelium, CD4 T cells often dominate the resident lymphocyte population, with CD161 expression linked to the interleukin-17 (IL-17) producing helper T cell (TH17) phenotype (13, 15). In this study, we explore the characteristics of human adaptive CD4 αβT cell populations that express factors associated with innateness, in blood and donor-matched lymphoid and nonlymphoid barrier tissues. Our findings reveal a polyclonal effector-memory CD4 T cell population that is notably enriched and resident in the human gastrointestinal tract and liver, with much lower representation in lung, skin, and lymph nodes. This CD4 T cell population is identified by coexpression of CD161 and CD56, display enhanced transcriptional and phenotypic innateness, and a highly responsive polyfunctional profile. Furthermore, these cells are preferentially expanded in individuals seropositive for cytomegalovirus (CMV) and enriched in CMV-specific TCRs and CMV-specific responses. This population shares key phenotypic characteristics with the recently described innate-like CD4 T cells in human colon enriched in reactivity to Escherichia coli, Staphylococcus aureus, and Candida albicans (16). Collectively, these findings support a model where persistent antigens of both microbial and viral origin drive expansion and maintenance of this CD4 T cell population, which is positioned on the edge of the adaptive-innate spectrum and adapted toward high innate and adaptive responsiveness and barrier tissue residence. RESULTS Coexpression of CD161 and CD56 marks a conventional CD4 T cell population enriched in the liver and small intestine To investigate the composition of the CD4 T cell compartment in human tissues, we collected multiple tissues from human multiorgan donor transplantation surgeries (n = 23) (table S1). Lymphocytes isolated from fresh tissue were analyzed by flow cytometry for identification and phenotypic characterization of CD4 T cells, excluding the MR1-reactive Vα7.2+ MAIT cell population (Fig. 1A). CD4 T cells were highly abundant in lymph nodes, as expected, whereas lower percentages were observed in the liver, ileum, and colon (Fig. 1B). We next analyzed the phenotypic characteristics of the CD4 T cell pool using k-means clustering and projection on a minimum spanning tree. This unsupervised analysis revealed clustering by tissue origin (Fig. 1C and fig. S1A). Branch 1 was strongly enriched in hepatic and intestinal CD4 T cells (Fig. 1C and fig. S1A), and detailed analysis of branch 1 revealed cluster-specific expression patterns (Fig. 1D), with cluster 9 and cluster 15 being enriched in expression of the C-type lectin-like receptor CD161, the adhesion receptor CD56, and IL-7Rα (CD127) (Fig. 1D). Next, using biaxial gating on CD161 versus CD56, we analyzed the frequency of CD161+CD56+, CD161+CD56−, CD161−CD56+, and CD161−CD56− CD4 T cells among non-MAIT CD4 T cells (Fig. 1E). CD161+ CD56+ CD4 T cells were detectable in blood, as well as in all donor-matched tissues, and were enriched in the gastrointestinal tract and liver (Fig. 1E). In the ileum, CD161+ CD56+ CD4 T cells comprised 3 ± 3% of the non-MAIT T cell compartment and accounted for 11 ± 11% of the CD4 T cell pool (Fig. 1F). CD4 T cells defined by CD161 and CD56 coexpression enriched in the liver and gastrointestinal tract. (A) Biaxial gating strategy for the identification of human CD4 T cells by flow cytometry. (B) Box plot showing abundance of CD4 expression within conventional (non-MR1–restricted Vα7.2−) T cells isolated from human organ donors (n = 23). (C) Minimum spanning tree analysis of CD4 T cell k-means clusters. Nodes are colored by tissue origin of CD4 T cells (left) as in (B) or tree branch dependence (right). (D) Relative protein expression density of branch 1 clusters. (E) Gating strategy for the Boolean expression of CD161 and CD56 on CD4 T cells and quantification of CD161+ CD56+ CD4 T cells across major human body sites. (F) Average abundance of CD4 T cell populations with expression of CD161, CD56, or both. (G) Heatmap of Spearman correlations between the frequency of CD4 T cells coexpressing CD161 and CD56 and other subsets for each tissue. (H) Heatmap of Spearman correlations of frequencies of CD4 T cells coexpressing CD161 and CD56 between ileum, caecum, colon, and mesenteric lymph nodes (mLN). In boxplots, median is indicated by line, arithmetic average by square. In heatmap, color indicates Spearman coefficient, *P < 0.05, **P < 0.01, ***P < 0.001. Considering that CD161 and CD56 are both associated with innate-like functionality and are frequently expressed by tissue-resident unconventional T cells such as invariant NK T (iNKT) cells and γδT cells, we evaluated the composition of the CD161+ CD56+ CD4 T cell population in blood and tissues in a subset of donors (n = 10) (fig. S1B). Conventional T cells represented the vast majority of CD161+ CD56+ CD4 T cells, while MAIT cells, defined as Vα7.2+ and human MR1-5-OP-RU tetramer+ were also represented, as expected. In contrast, the Vα24+ Jα18+ iNKT cells and γδΤ cells were almost undetectable within this population, in particular in the intestines (fig. S1, C and D). These results indicate that the CD161+ CD56+ CD4 T cell population, once MAIT cells are gated out, can be considered as a conventional adaptive CD4 αβT cell population. We next analyzed the coexpression patterns of CD161 and CD56 in conventional CD4 and CD8 T cell subsets. Among CD4 T cells, CD56 expression was largely coupled to CD161 coexpression (fig. S1E) and largely restricted to the gastrointestinal tract, with CD161− CD56+ CD4 T cells being only marginally detectable in any tissue. In contrast, CD8 T cells contained a large proportion of cells expressing either CD56 or CD161 in most tissues (fig. S1F). In the gastrointestinal tract, the abundance of CD161+ CD56+ CD4 T cells correlated with the frequency of CD4 T cells expressing only CD161 and inversely correlated with the levels of CD161− CD56− CD4 T cells (Fig. 1G). Moreover, frequencies of CD4 T cells coexpressing CD161 and CD56 were correlated between the ileum, caecum, colon, and mesenteric lymph nodes, suggesting coordinated expansion of this subset across intestinal sites (Fig. 1H). Together, the CD4 T cell population coexpressing CD161 and CD56 is preferentially found enriched in the gastrointestinal tract and liver. Intestinal CD161+ CD56+ CD4 T cells have a tissue-resident profile adapted to the tissue site We next characterized the total CD4 T cell population from matched human donor livers and intestinal tissue sites. Principal components analysis (PCA) partly separated hepatic CD4 T cells from intestinal CD4 T cells along the PC2 axis, driven by higher expression of CXCR3 and GzmB and lower levels of CD127 and CD103 in resting hepatic cells (Fig. 2, A and B). Intestinal CD4 T cells clustered together, indicating overlapping phenotypes between the intestinal tissue sites (Fig. 2A). CD4 T cells were separated along PC1 by coexpression pattern of CD161 and CD56, forming an innateness gradient with CD4 T cells coexpressing CD161 and CD56 being the most distinct from their CD161− CD56− counterparts (Fig. 2C). CD4 T cells coexpressing CD161 and CD56 are phenotypically distinct with chemokine receptor expression similar to innate-like T cells. (A) PCA of CD4 T cell surface features measured by flow cytometry. Colors indicate origin of CD4 T cell populations from the liver, ileum, caecum, and colon. (B) Arrows indicate contribution of phenotypic markers to the first two principal components. (C) Color indicates CD4 T cell subset as defined by coexpression of CD161 and CD56. (D) Flow cytometry measurement of tissue residency markers expressed by intestinal CD4 T cell subsets defined by CD161 and CD56 and average CD103 protein expression. (E) Flow cytometry of the costimulatory receptor CD27 and the ATP-degrading enzyme CD39. (F) Heatmap of chemokine receptor density expressed on CD4 T cell subsets defined by CD161 and CD56 as well as innate-like T cell populations in intestinal tissues, or (G) peripheral blood. In boxplots, median is indicated by line, arithmetic average by square. Statistical comparison between CD4 T cell subsets is made by Wilcoxon signed-rank test and Benjamini-Hochberg correction for multiple comparison with *P < 0.05, **P < 0.01, ***P < 0.001. Intestinal CD4 T cells coexpressing CD161 and CD56 frequently expressed integrin-αE (CD103) and CD69, consistent with an intraepithelial location in the intestines (Fig. 2D and fig. S2A). Among CD103+ CD69+ tissue-resident T cells, CD161+ CD56+ CD4 T cells were found in similar frequency as γδΤ cells and were more abundant than iNKT or MAIT cells (fig. S2B). Intestinal CD161+ CD56+ CD4 T cells frequently expressed CD39, an immunoregulatory extracellular ATP-convertase (Fig. 2E), were low in CD27 (fig. S2C), and high surface density of CD127 indicating sensitivity toward IL-7–mediated regulation, a phenotype similar to intestinal MAIT cells (fig. S2D). In accordance with a tissue-resident effector-memory T cell phenotype, CD161+ CD56+ CD4 T cells expressed only low levels of markers associated with naïve or follicular TH cells, such as CD62L or CXCR5 (fig. S2 E). In line with a tissue-resident effector-memory phenotype, intestinal CD161+ CD56+ CD4 T cells presented a chemokine receptor pattern highly similar to innate-like T cell subsets with a TH17-like program, namely MAIT cells and γδT cells, with high expression of CCR5 and CCR6, mediating migration toward sites of inflammation, as well as CXCR6, mediating tissue infiltration, and CCR9, responsible for homing to the gut (Fig. 2F). Notably, the chemokine receptor profile of the CD161+ CD56+ CD4 T cell population was largely retained in peripheral blood (Fig. 2G and fig. S2F), and high expression of CCR2 and CXCR3 further supported the notion of a high migratory potential similar to that of innate-like unconventional T cell populations. CD161+ CD56+ CD4 T cells represent a highly functional intestinal TH1/17/22 population We next investigated the functionality of CD161+ CD56+ CD4 T cells in the human intestines. Cells isolated from donor-matched human intestinal tissues were stimulated with phorbol 12-myristate 13-acetate (PMA) and ionomycin for 5 hours before analysis by intracellular flow cytometry. CD4 T cell populations coexpressing CD161 and CD56 showed the strongest response to stimulation, indicated by PC1 (Fig. 3A). Intestinal CD161+ CD56+ CD4 T cells were characterized by a higher expression of tumor necrosis factor (TNF) and interferon-γ (IFN-γ) (Fig. 3B), as well as IL-17A and IL-22 (Fig. 3C), compared to their CD161+ CD56− and CD161− CD56− counterparts. Notably, intestinal CD161+ CD56+ CD4 T cells were more likely to mount bi- and trifunctional responses (Fig. 3D). CD161+ CD56+ CD4 T cells frequently coexpressed IL-17A and IL-22, or IFN-γ and IL-22, highlighting a complex effector response including mucosal barrier protection (Fig. 3E). CD161+ CD56+ CD4 T cells are highly polyfunctional effector–memory T cells with a barrier-protective profile. (A) PCA of CD4 T cell effector molecules measured by flow cytometry. Color indicates origin of CD4 T cell population from the ileum, caecum, and colon. Arrows indicate contribution of effector molecules to the first two principal components. Color indicates CD4 T cell subset as defined by coexpression of CD161 and CD56. (B) Flow cytometry and average production of TNF and IFN-γ or (C) IL-17A and IL-22 by CD4 T cell subset. (D) Polyfunctionality analysis of each CD4 T cell subset defined by CD161 and CD56 expression using SPICE. Pie chart segment color indicates the number of coexpressed effector molecules. Arch color represents average coexpression patterns of individual effector molecules. (E) Coexpression of IFN-γ and IL-22 (left) or IL-17A and IL-22 (right) by intestinal CD4 T cells coexpressing CD161 and CD56. (F) Flow cytometry of CD161 and CD56 expression on CD4 T cells isolated from large intestinal tissue obtained from healthy parts of tissue removed during colorectal cancer surgery (H, n = 7), or from surgically removed large intestinal tissue from patients with Crohn’s disease (IBD, n = 7). Population size of CD4 T cell subsets defined by CD161 and CD56 expression as frequency of conventional (non–MR1-reactive Vα7.2−) CD4+ T cells. Δ indicates average difference between healthy volunteers and patients with IBD for each CD4 T cell subset. Statistical comparison between CD4 T cell subsets is made by Wilcoxon signed-rank test and Benjamini-Hochberg correction for multiple comparison. Statistical comparison between cohorts is made by Wilcoxon rank-sum test with **P < 0.01 and ***P < 0.001. Given the relevance of lymphocytes producing these cytokines in inflammatory bowel disease (IBD), we next quantified the CD4 T cell population dynamics in samples obtained from surgically removed large intestines of patients with Crohn’s disease and from non-IBD controls (Fig. 3F and fig. S3). Strikingly, the relative population size of colonic CD161+ CD56+ CD4 T cells was reduced by 78% in IBD, whereas the CD161+ CD56− CD4 T cell population was reduced by 50% (Fig. 3F). This indicates that CD56-expressing CD4 T cells are dominantly affected by IBD conditions in the colon. Together, coexpression of CD161 and CD56 marks a highly polyfunctional intestinal CD4 T cell population with a mixed TH1/17/22 effector program, which is depleted in the colon tissue of patients with IBD. Transcriptional profiling indicates long-lived LYAR+TIAM1−TSHZ2− innate–like effector characteristics To gain insight into the transcriptional program of CD161+ CD56+ CD4 T cells, we analyzed single cell (sc) CITE-seq and scTCR-sequencing (scTCR-seq) data from CD4 T cells from four healthy blood donors (Fig. 4A), followed by reclustering of memory CD4 T cells expressing high levels of CD45RO (Fig. 4, B and C). Notably, the CD161+ CD56+ CD4 T cells clustered separately from their CD161+ CD56− counterparts. Coexpression of CD161 and CD56 was restricted to a single cluster on the Uniform Manifold Approximation and Projection (UMAP) (Fig. 4D), with high expression of IL7RA, CXCR3, reflecting the observations made on the protein level, as well as of the transcription factors TBX21 and EOMES, commonly associated with a cytotoxic effector-memory phenotype, and TYROBP encoding an adaptor molecule required for expression of some innate activating receptors (Fig. 4, E and G, and fig. S4, A and B). In line with this transcription factor program, the CD161+ CD56+ CD4 T cells were enriched in transcripts encoding members of the granzyme family like GZMK and GZMA, and gene set enrichment analysis (GSEA) and overrepresentation analysis (ORA) against the GO database confirmed an increased propensity for cytotoxic functionality (Fig. 4, E and F, and fig. S4C). Furthermore, these cells were high for the transcript encoding the NK cell granule protein 7 (NKG7), which is an indicator for cytotoxic functionality (17). CD161+ CD56+ CD4 T cells have an effector-memory transcriptome with innate-like characteristics. (A) UMAP of the total T cell compartment isolated from four healthy blood donors of which (B) memory (CD45RA− CD45RO+) and nonregulatory (FOXP3−) CD4 T cells were isolated for further phenotyping by transcriptomic analysis into (C) five memory CD4+ T cell clusters, highlighted by color. (D) Density plot of CD161 and CD56 protein coexpression reveals UMAP cluster specific expression patterns. (E) Differential gene expression analysis using MAST with donor origin as covariate between CD161+ CD56+ CD4 T cells and all other memory CD4 T cells. (F) GSEA for biological processes enriched in CD161+ CD56+ CD4 T cells compared to all other memory CD4 T cells. (G) Density plots for the expression of transcripts associated with innate-like functionality. (H) Flow cytometry of IL-18 receptor (IL-18R) expression on peripheral blood CD4+ T cells. Color indicates CD4 T cell subset defined by CD161 and CD56 coexpression. (I) Average expression of IL-22, or IFN-γ, by CD4 T cells following 24 hours of stimulation with IL-12 and IL-18. (J) Average expression of the activating receptors NKp80 and NKG2D by CD4 T cells at resting conditions. Statistical comparison between CD4 T cell subsets is made by Wilcoxon signed-rank test and Benjamini-Hochberg correction for multiple comparison with *P < 0.05 and **P < 0.01. In comparison to the total memory CD4 T cell compartment, resting CD161+ CD56+ CD4 T cells expressed higher levels of proinflammatory CCL5, encoding the chemokine CCL5/RANTES, as well as IL32 with expression commonly seen in regulatory T cells and controlled by oxygen sensing systems (Fig. 4E) (18). Accordingly, GSEA against the Reactome database indicated altered “IL-10 signaling” and “selenoamino acid metabolism” in CD161+ CD56+ CD4 T cells (Fig. 4F and fig. S4D). Conversely, CD161+ CD56+ CD4 T cells expressed lower levels of TSHZ2, a transcription factor stimulating proliferation and increasing with age in naïve CD4 T cells, and were largely devoid of TIAM1 expression, a Wnt-signaling–associated gene driving proliferation (Fig. 4E and fig. S4B) (19). The transcription of cell growth–regulating nucleolar protein LYAR was strongly up-regulated in CD161+ CD56+ CD4 T cells (Fig. 4E). LYAR has recently been shown to suppress type I IFN signaling (20), and ORA congruently highlighted a substantially altered “response to virus” pathway (fig. S4C). CD4 T cells coexpressing CD161 and CD56 were the dominant population expressing the transcriptional regulators TBX21 and EOMES, encoding T-bet and eomesodermin, respectively (Fig. 4G). Given that CD161 and CD56 are receptors associated with innate-like functionality of T cells (11, 21), we investigated the sensitivity of these cells to a proinflammatory cytokine environment as it could occur during viral infection. CD161+ CD56+ CD4 T cells expressed high levels of IL-18 receptor on the protein and transcriptomic level, overlapping with the high TBX21 and EOMES expression (Fig. 4, G to H). Following stimulation of peripheral blood mononuclear cells (PBMCs) with IL-12 and IL-18 for 24 hours, coexpression of CD161 and CD56 robustly identified CD4 T cells producing IL-22 and IFN-γ, whereas CD161+ CD56− CD4 T cells were insensitive to this stimulation (Fig. 4I). Last, we assessed the expression of NK cell–associated markers of antiviral immunity, such as NKp80 and NKG2D, which are both activating receptors surveying for viral infection. The expression of NKp80 and NKG2D was strongly associated with CD161 and CD56 coexpression (Fig. 4J). The innate-like CD161+ CD56+ CD4 T cells have a diverse TCR repertoire and are frequently CMV reactive The CD161+ CD56+ CD4 T cell population identified here shares key characteristics with the recently reported major histocompatibility complex (MHC) class II–restricted, innate-like, commensal-reactive CD4 T cell (Tmic) subset described in human colon (16, 22), which was found to be enriched in reactivity to E. coli, S. aureus, and C. albicans (16). We therefore next studied our scTCR-seq dataset and found that the CD161+ CD56+ CD4 T cells showed a diverse TCR repertoire and clonal distribution similar to the other memory CD4 T cell clusters, indicating that these cells were not clonally restricted like members of the unconventional T cell family (Fig. 5A and fig. S5, A to B). Clones among CD161+ CD56+ CD4 T cells, defined by the TRAV-TRAJ-TRAC (TRA) sequences, were partly shared with other CD4 T cell clusters and overlapped less with CD161+ CD56− CD4 T cells than with TOX+ TIGIT+ CD4 T cells or memory CD4 T cells (Fig. 5, B and C). Circulating CD161+ CD56+ CD4 T cells have a diverse TCR repertoire skewed toward reactivity against persistent viruses. (A) Relative abundance of TCRα chains (defined by TRA genes) within each memory CD4 T cell subset and (B) overlap in number of TRA clones between subsets. (C) Overlap of the top five largest TRA clones between memory CD4 T cell subsets. (D) Individual memory CD4 T cell clones with predicted reactivity against viral (black) or nonviral (red) antigen highlighted on UMAP. Blue outline and shade indicate CD161+ CD56+ CD4 T cell cluster. (E) Abundance of TRA clones with antigen specificity identified by reverse-epitope prediction against database included in Trex algorithm within all memory CD4 T cells. (F) Relative fraction of TRA clones within the CD161+ CD56+ CD4 T cell cluster or all other memory CD4 T cells with reactivity against various viruses or (G) specifically CMV. (H) T cell activation–induced marker assay after 12 hours of stimulation of PBMCs with CMV peptide pool and flow cytometry of TNF and CD69 (n = 6). Boxplots show responses of CD4 T cell populations defined by CD161 and CD56 to the CMV peptide pool. (I) Frequency of CD161+ CD56+ CD4 T cells as fraction of total T cells present in PBMCs, separated by CMV-IgG serostatus. Data point shape indicates gender of blood and tissue donors. (J) Spearman correlation of CD161+ CD56+ CD4 T cell frequency in CMV-seropositive individuals between peripheral blood and ileum tissue. (K) CD161+ CD56+ CD4 T cell frequency in CMV-seropositive donor-matched blood and intestinal tissue, separated by gender. (L) Spearman correlation of CD161+ CD56+ CD4 T cell frequency against donor age. Statistical comparison between CD4 T cell subsets is made by Wilcoxon signed-rank test and Benjamini-Hochberg correction for multiple comparisons with *P < 0.05 and **P < 0.01. We next compared the obtained TRA clones against the Trex database for reverse epitope discovery (23). Epitope-matching clones were found in all memory CD4 T cell clusters and were primarily of viral specificity (Fig. 5D and fig. S5C). Notably, the CD161+ CD56+ CD4 T cell TCR repertoire harbored an expanded clone (TRAV17-TRAJ15) with specificity for the CMV protein UL29 (Fig. 5C), and CMV-specific clones comprised most of the identified specificities (Fig. 5E). On the basis of the reverse epitope discovery results, the CD161+ CD56+ CD4 T cell cluster comprised a larger fraction of clones with predicted reactivity against viruses and CMV than the remaining memory CD4 T cell clusters (Fig. 5, F and G, and fig. S5D). To experimentally test the possibility that this cell population harbored an accumulation of CMV-specific cells, we stimulated PBMCs with a peptide pool of known CMV epitopes. In line with the prediction, there was an enrichment of CMV-specific responses within the CD161+ CD56+ CD4 T cell population as these cells up-regulated the activation marker CD69 and produced increased levels of TNF, whereas their CD161+CD56− and CD161− CD56− counterparts responded much less to CMV peptide stimulation (Fig. 5H). To study this in more detail, we next quantified the frequency of CD161+ CD56+ CD4 T cells in blood and intestinal biopsy tissue samples from individuals undergoing colonoscopy for preventive screening or diagnostic purposes (table S2). CMV–immunoglobulin G (IgG)–seropositive donors had increased levels of CD161+ CD56+ CD4 T cells in blood (Fig. 5I), which positively correlated with the population size in the ileum (Fig. 5J). We observed increased frequencies of CD161+ CD56+ CD4 T cells in the intestines of CMV-IgG+ males in comparison to females, whereas frequencies in blood were similar in both sexes (Fig. 5K). The CD161+ CD56+ CD4 T cell population in the ileum expanded with age (Fig. 5L), and as CMV prevalence generally increases with age, we compared the abundance of CD161+ CD56+ CD4 T cells in the ileum across age, gender, and CMV-serostatus. Notably, the population size increase was age-dependent in males, with older males more likely to be CMV-IgG+ and associated with CMV infection history in females (Fig. 5L). Hepatic CD161+ CD56+ CD4 T cells adopt Tmic1 and Tmic17 programs and are enriched in CMV-specific TCRs To explore the characteristics of CD161+ CD56+ CD4 T cells at a tissue site where these cells are highly enriched, we analyzed this population in a previously published CITE-seq and scTCR-seq dataset from CD45+ leukocytes isolated from human liver sinusoidal perfusates (24). Notably, CITE-seq is required for accurate identification of CD56 as detection of NCAM1 is notoriously difficult and unrelated to protein expression. A total of 14,005 CD4 T cells were identified in the liver sinusoidal perfusates with clear detection of CD161+ and CD56+ cells (Fig. 6, A and B, and fig. S6A). Clustering revealed major hepatic CD4 T cell populations, including several CD4 T cell helper cell subsets, cytotoxic CD4 T cells, regulatory CD4 T cells, as well as two populations that exhibited transcript and protein features similar to those we had identified in peripheral blood CD161+ CD56+ CD4 T cells (Figs. 5 and 6C). Notably, the liver-resident CD161+ CD56+ CD4 T cells showed mutually exclusive transcription factor expression of TBX21 and RORC, resulting in clusters favouring either a TH1-like program (Tmic1) or a TH17-like program (Tmic17), respectively (Fig. 6, B to D). Both CD4 Tmic populations were marked by the innate-like characteristics we had observed in blood, such as high expression of IL18R1 and GZMK (Fig. 6D and fig. S6B). Notably, the hepatic Tmic transcriptomic signature reflected the low propensity for proliferation evident in blood CD161+ CD56+ CD4 T cells, namely a very high DUSP2 expression paired with very low levels of TSHZ2 and TIAM1 (Fig. 6D). The hepatic CD4 Tmic1 and CD4 Tmic17 cells showed distinct usage of gene expression programs calculated by T-CellAnnoTator (TCAT) (25), and associated with TH1 and TH17 cells, respectively (Fig. 6, E and F). CD4 Tmic1 and CD4 Tmic17 cells were highly enriched in expression of a module score calculated on the 100-most differentially expressed genes in blood-circulating CD161+ CD56+ CD4 T cells. Similarly, reverse epitope prediction of TCRα clones by the Trex algorithm highlighted specificity for antigens of viral origin in the CD161+ CD56+ CD4 T cell population, including cells of both Tmic1 and Tmic17 characteristics (Fig. 6, G to I). Thus, these findings indicate that hepatic CD161+ CD56+ CD4 T cells largely replicate the enhanced innate characteristics and enrichment of CMV-specificity seen in cells from blood, with additional diversification as they segregate into Tmic1 and Tmic17 clusters. Innate-acting CD161+ CD56+ CD4 Tmic cells adopt TH1 and TH17 programs in the human liver. (A) UMAP visualization of CD45+ leukocytes isolated from human liver sinusoidal perfusates. T cells were identified by canonical markers, and CD4 protein detection was used to identify CD4 T cells. (B) Density plot of CD161 and CD56 protein coexpression overlayed on the UMAP to identify CD161+ CD56+ CD4 Tmic cells. (C) UMAP of CD4 T cells in human liver perfusions identifies CD4 Tmic1 and CD4 Tmic17 clusters based on expression of canonical genes. (D) Density plot of individual transcripts in hepatic CD161+ CD56+ CD4 Tmic cells. (E) Density plot of gene expression programs (GEP) calculated by TCAT. (F) Volcano plot showing log2-fold change average gene expression difference between liver CD4 Tmic17 cells or CD4 Tmic1 cells and all other hepatic CD4 T cell populations. (G) UMAP of epitopes for TCRα chains as reversely predicted by Trex (Levenshtein distance = 1). Clones were defined by gene and CDR3α nucleotide sequence. Cells with predicted reactivity against antigens of viral origin were combined as “virus-reactive,” and CMV reactivity was highlighted separately in the right panel. (H) Proportion of predicted TCR reactivity against identified epitopes of infectious agents. (I) Boxplot comparison of predicted virus reactivity or CMV reactivity in CD4 Tmic17 cells and all other memory CD4 T cells. In boxplots, median is indicated by line, arithmetic average by square. Significance of predicted reactivity was assessed by t test, *P < 0.05. Differential gene expression was examined by MAST test using donor origin as covariate. DISCUSSION Understanding tissue-specific adaptation in barrier tissue T cells and their relationship with persistent antigens has been a longstanding challenge in immunology. CD4 T cells expressing CD161 comprise a significant fraction of the human immune compartment in both blood and intestinal tissue (26, 27). Here, using donor-matched paired tissue specimens, we mapped the coexpression pattern of CD161 and CD56 on conventional αβ CD4 T cells across several human lymphoid and mucosal barrier sites. CD161+ CD56+ CD4 T cells showed preferential enrichment in the gastrointestinal tract and the liver, with a population size similar to that of γδT cells in the ileum. CD56 was nearly exclusively coexpressed with CD161 in CD4 T cells, a pattern distinct from CD8 T cells, indicating pathway differences in the ontogeny of these populations. Expression of CD161 in CD4 T cells is widely associated with a TH17 phenotype (13, 15, 22, 26). In contrast, CD56 expression has been linked to high cytolytic capacity in conventional CD8 T cells and in MAIT cells (5, 21, 28). The CD161+ CD56+ CD4 T cell population exhibited a hybrid highly polyfunctional TH1 and TH17 effector potential, displaying features characteristic of both lineages. The intestinal CD161+ CD56+ CD4 T cells expressed high levels of CD103, CCR6, and CD39, consistent with a resident TH17-like phenotype (29), in combination with high levels of CXCR6. Coexpression of these markers suggests a capacity for homing to tissue and interaction with CXCL16-expressing epithelial cells or dendritic cells (30). However, TCR clonality showed a limited overlap with the CD161+ CD56− CD4 T cell population, suggesting that the CD161+ CD56+ CD4 T cell population is distinct from the conventional TH17 pool. This interpretation is further supported by distinct surface expression of innate activating receptors and the segregation of these populations in single-cell clustering. In murine models, CXCR6+ CD4 T cells express reduced levels of TCF1 (31, 32), which is associated with stemness and proliferative capacity, suggesting reduced proliferative capacity. In our human datasets, CD161+ CD56+ CD4 T cells expressed less TSHZ2 and TIAM1 in both blood and liver transcripts associated with inhibitory effect on cell proliferation (19, 33). In line with this, CD103 expression was previously associated with enhanced persistence among CD4 T cells resident in transplanted human duodenum (34), and we identified CD103 expression elevated in mucosal CD161+ CD56+ CD4 T cells. Thus, coexpression of CD161 and CD56 may mark persistent and long-lived effector-memory CD4 T cells in the human intestine. CXCR6 expression is linked to tissue residency and tumor infiltration in CD8 T cells (35) and IL-22 production in innate lymphoid cells (36) but has also been described to mark an IFN-γ–producing pathogenic TH17 population in models of autoimmunity (31). However, IL-17A is critical for promoting mucosal barrier integrity, and enteric IL-17A–producing microbiota-reactive CD4 T cells have been identified as beneficial in maintaining intestinal homeostasis (22). We found CD161+ CD56+ CD4 T cells to be depleted in the colon of patients with IBD. Previous studies observed that a similar population was decreased in the lamina propria of patients living with Crohn’s disease (37), but not of those with ulcerative colitis (UC) (16). Notably, CD161+ CD56− CD4 T cells were previously described to be significantly increased in UC tissue, whereas CD161+ CD56+ CD4 T cells showed an opposite trend (16). Our findings highlight the importance to consider sex differences and the infection history of patients in such studies, given the tendency of higher levels of intestinal CD161+ CD56+ CD4 T cells in males. Further, our results suggest that the quantity of CD161+ CD56+ CD4 T cells circulating in the blood may serve as a surrogate biomarker for estimating their presence in the ileum, which was severely reduced in patients with IBD. Together, these findings warrant further investigation into the role of tissue-resident CD4 T cells coexpressing CD161 and CD56 in IBD pathology. Hackstein et al. (16) recently found that a similar CD4 T cell population, named Tmic cells, exhibit MHC II restriction, exist in the human colon, have increased innate-like characteristics, and are enriched in reactivity against some commensal microbes. Our donor-matched organ donor tissue data show that the CD161+ CD56+ CD4 T cell population is not limited to the human colon but is also distributed across the distal ileum and caecum, proportional to their representation in mesenteric lymph nodes, and prominently represented in the liver. Our findings indicate that CD161+ CD56+ CD4 T cells in both blood and liver encompass a highly reactive polyfunctional effector population enriched in reactivity toward epitopes of viral origin. Notably, the hepatic Tmic cells segregate in Tmic1 and Tmic17 clusters, suggesting yet another layer of tissue-dependent complexity. In light of the reported reactivity of this population against commensal microbes in the intestines (16), it is tempting to speculate that CMV in this regard may be viewed by the immune system as a commensal-like virus. CMV persists in the epithelium after infection, and while it is controlled by the host immune system with the cost of ongoing immune activation, CMV-reactive CD8 T cells gradually expand with age and adopt a cytotoxic effector–memory profile, driving so-called “memory inflation” (38, 39). Inflated memory T cells show increased levels of NK cell–associated receptors as well as cytotoxicity pathways (38, 39). In line with this, we find that CD161+ CD56+ CD4 T cells feature high levels of NKp80 and NKG2D, as well as granzyme expression. In mice, Rakebrandt et al. (40) recently described “innate-acting” CD4 T cells with high levels of CXCR6 and NKG2D that were sensitive to bystander activation and protective against secondary bacterial challenge. Thus, human CD161+ CD56+ CD4 T cells may fulfill a role akin to members of the innate-like T cell family by providing rapid amplification of stress signals (TNF and IFN-γ), while simultaneously mediating cytotoxicity (GZMK, NKG7, NKp80, and NKG2D) (17), chemotaxis (IL-17A), and tissue repair (IL-22). It is also interesting to note the high expression of LYAR, which may allow these cells to persist in the context of high IFN expression (20). The enrichment of these cells in the human gastrointestinal tract supports the hypothesis that they are reactive against microbial commensals and extends the spectrum of potential antigen targets to include persisting viral infections such as CMV. Furthermore, it is tempting to speculate that the loss of intestinal CD161+ CD56+ T cells in IBD may be important in the context of CMV colitis in immunocompromised patients and in patients with IBD (41) and the association of CMV with poor outcome in this patient group (42). In summary, we characterize the human tissue–resident CD4 T cell compartment across major donor-matched human body sites and identify coexpression of CD161 and CD56 as a hallmark of highly responsive and polyfunctional effector–memory CD4 T cells with tissue-resident characteristics enriched in the intestines and the liver. This population exhibits enhanced innateness characteristics on the adaptive-innate spectrum of conventional αβ T lymphocytes, shows enrichment of CMV-specific TCRs, and may play an important role in the maintenance of barrier integrity. MATERIALS AND METHODS Study design In this study, we characterized tissue resident CD4 T cells in matched lymphoid and nonlymphoid tissues from human organ donors, with the aim to better understand the role of resident CD4 T cells in tissue immune defense. In paired tissue specimens, we found tissue site-adapted phenotypes by flow cytometry. CD4 T cell phenotypes were then investigated in depth using flow cytometry, scCITE-seq and scTCR-seq. These studies were complemented by flow cytometry analysis of colonoscopy tissue biopsies from healthy donors and patients with IBD. Human organ donor cohort characteristics Human organ material was obtained through the Immunology Human Organ Donor Programme from deceased individuals. Criteria for inclusion were according to established national guidelines for organ donation and transplantation, and tissues not used for transplantation were sampled for research. All donor serums were tested for CMV, hepatitis B surface and core antigen, HIV-1, and syphilis according to clinical routine. Information on the organ donor characteristics is provided in table S1. The study was approved by the Swedish Ethical Review Authority (ID: 2019-05-016). Human tissue and blood processing Human tissues were processed immediately after transplantation surgery. Tissues were rinsed with phosphate-buffered saline (PBS) and processed in parallel in sterile petri dishes containing RPMI 1640 cell culture media (Gibco) supplemented with heat-inactivated fetal calf serum (10% FCS, MilliporeSigma), l-glutamine (1 mM, Invitrogen), penicillin (100 U/ml), and streptomycin (50 μg/ml) containing deoxyribonuclease I (DNase I, 1 nM, Sigma-Aldrich) (R10). Lung-associated lymph nodes were isolated from parenchymal lung and dissected separately. Lymph nodes were isolated from the mesenterium, and human ileum, caecum, and colon were processed independently. Gut specimens were washed with Hanks’ balanced salt solution buffer without calcium or magnesium (Thermo Fisher Scientific) containing gentamycin (10 mg/ml, Thermo Fisher Scientific), penicillin, and streptomycin (Cytiva, fungizone, 250 μg/ml). Adipose and muscle tissue were removed and tissues manually dissected before enzymatic digestion in RPMI 1640 supplemented with DNase I (1 nM, Sigma-Aldrich) and purified collagenase (80 μg/ml, Worthington, for gut specimens only) for 45 to 60 min under constant stirring at 37°C. Skin tissue was repeatedly punctured with an 8-mm skin biopsy punch needle (Observe Medical Nordic) instead of manual dissection. After incubation, the tissues were passed through a cell strainer and resuspended in R10 cell culture media at the start of the experiment. For spleen and lung tissues, strained cell suspensions were incubated in red cell lysis buffer (Invitrogen) before suspension in cell culture media. Organ donor whole blood was collected before organ perfusion. PBMCs were isolated by density-gradient centrifugation using Ficoll-Pacque Plus (Sigma-Aldrich) and resuspended in R10 cell culture media. Additional peripheral blood samples were obtained from adult healthy donors after informed consent at the blood transfusion clinic located at Karolinska University Hospital, as approved by the Swedish Ethical Review Authority (ID: 2020-02-604), and similarly processed. Mononuclear cells of each tissue were either unstimulated or treated with PMA (50 ng/ml) and ionomycin (1 μg/ml) for 5 hours with secretory pathway blocking with GolgiPlug and GolgiSTOP (BD BioSciences) according to the manufacturer’s instructions before flow cytometry. Alternatively, PBMCs were stimulated with a CMV peptide pool including 42 CMV peptides of which 14 are MHC-II restricted (Mabtech). Intestinal biopsy collection and preparation Intestinal pinch biopsies were collected from the terminal ileum, caecum, and sigmoid colon of donors undergoing colonoscopy screening at the Capio Gastro Center Stockholm (table S2), as approved by the Swedish Ethical Review Authority (ID: 2023-00840-01). Biopsies were collected in cold R10 medium and, upon arrival in the laboratory, transferred to medium lacking FCS but supplemented with DNase I (0.25 mg/ml, Sigma-Aldrich) and purified collagenase (240 U/ml, Worthington). The samples were incubated for 45 min under constant stirring at 37°C. After incubation, the tissues were passed through a 40-μm cell strainer and rinsed with R10 medium to stop the enzymatic digestion. Cell suspensions were pelleted before being resuspended in fluorescence-activated cell sorting buffer (PBS supplemented with 2% FCS and 2 mM EDTA) for flow cytometry staining and analysis. Presence of CD161+ CD56+ CD4 T cells in large intestinal tissues from seven adult patients with CD or UC colitis undergoing surgery and seven non-IBD controls undergoing surgery for colorectal cancer was determined on the basis of analysis of previously reported flow cytometry data (43). Analysis of T cells by flow cytometry PBMCs were seeded at a density of 2 × 106 cells per well of a 96-well U-bottom plate. Degranulation was assessed by the addition of BUV395-conjugated anti-CD107a (H4A3, BD Biosciences) antibody before treatment with recombinant IL-12 (10 ng/ml, Peprotech) and recombinant IL-18 (100 ng/ml) (Medical and Biological Laboratories). Secretion of effector molecules was inhibited during the last 6 hours of stimulation using GolgiPlug and GolgiSTOP (BD) before subjection of the samples to flow cytometric analysis. Cells were stained for 30 min at room temperature with phycoerythrin- or allophycocyanin-conjugated 5-OP-RU-human-MR1-tetramers provided by the National Institutes of Health (NIH) tetramer core facility, for the identification of MAIT cells, before washing with PBS supplemented with 2% FCS and 2 mM EDTA and subsequent cell surface staining using an antibody cocktail diluted in brilliant stain buffer (BD) containing dead cell marker (Live/Dead Fixable Aqua Dead Cell Stain Kit for 405-nm excitation, Invitrogen, or Live/Dead Fixable Near-IR Dead Cell Stain Kit, for 633- or 635-nm excitation). Surface staining was performed for 30 min at 4°C protected from light before washing and incubating with the Fixation/Permeabilization Kit (BD Biosciences) according to the manufacturer’s instructions. Intracellular proteins were stained for 30 min at 4°C before flow cytometry on a BD Symphony A5 (BD Biosciences) equipped with 355-, 405-, 488-, 561-, and 639-nm lasers. The list of antibodies used is provided in table S3. Flow cytometry analysis Data were deconvoluted and analyzed using FlowJo software (v10.8, BD) and single-stained compensation beads (BD Biosciences) after calculation of a compensation matrix with the autospill algorithm. For k-means clustering and minimum spanning tree analysis, CD4 T cell populations cells from each donor and tissue were randomly sampled to 3000 cells, if possible, and a maximum of 10,000 cells was clustered using the CytoTree package (v.1.03) including all measured phenotypic markers that were not used for cell type identification. A minimum number of 20 cells was required as threshold for phenotyping. For polyfunctionality analysis, Boolean combinations of effector molecules were created in FlowJo, formatted in R, and visualized using SPICE (44). CITE-seq and TCR-seq analysis We recently analyzed the total peripheral blood T cell compartment from four healthy individuals with scRNA sequencing (scRNA-seq) and scTCR-seq analysis including oligonucleotide-conjugated antibodies against CD4, CD8, CD45RA, CD45RO, Vα7.2, CD161, and CD56, among others (21). This dataset previously used for characterization of MAIT cells we here explored to investigate the CD4 T cell population sampled from the same donors and run in the same assay. In brief, MAIT cells were purified by 5-OP-RU-hMR1-tetramer and marked with a hashtag. The non-MAIT population was marked with a second distinct hashtag, and both populations were stained with anti-human CD3–fluorescein isothiocyanate (clone SK7, BioLegend), CD45-BV711 (clone HI30), CD14-V500 (clone M5E2, BD), and CD19-V500 (clone HIB19, BD). A total of 7500 non–5-OP-RU-hMR1–reactive CD45+ lymphocytes were sorted from each donor, and 5′ single-cell libraries were created following the 10X Genomics’ Next GEM Reagent Kit with Dual Index and Feature Barcode Technology according to the manufacturer’s protocol. Sequencing of pooled libraries was performed at the National Genomics Infrastructure (Science for Life Laboratory, Stockholm, Sweden). Sequences were aligned using CellRanger (v.7.1) on Uppmax supercomputing facilities (Uppsala, Sweden), and single cells were identified by HTODemux and scDblFinder (v.1.14.0) (45, 46) before controlling for quality metrics (nFeature_RNA > 200, nCount_RNA > 1000, nCount_RNA < 8000, percent_mito<10, percent_ribo>10). CD4 T cells were isolated from the total dataset following classification by canonical RNA transcript and surface markers. The CD4 T cell data were then rescaled, reintegrated, and reclustered. Memory CD4 T cells were identified by canonical memory markers and memory CD4 T cells clustered and visualized by UMAP following dimensionality reduction by PCA. Differentially expressed gene analysis was performed by the MAST test using the donor origin as covariate between memory CD4 T cell clusters. GSEA or ORA was performed on the background of all genes expressed in memory CD4 T cells using clusterProfiler (v.4.8.2), biomaRt (v.2.6), and ReactomePA (v.1.88). Selected gene sets are presented. Single-cell TCR sequences were identified by Cellranger. Donor- and CD4 T cell population–specific TCR repertoires were further analyzed using the scRepertoire package (v.2.0). Reverse epitope prediction of TRA was performed against public databases included in the Trex package (v.0.99) (47). TCR sequences predicted to react against a specific virus were collectively grouped as virus specific. The scRNA-seq count matrices are deposited at Zenodo: DOI: 10.5281/zenodo.14851876. CITE-seq and TCR-seq analysis of hepatic immune cells We applied our investigation of tissue-resident CD4 T cells to CITE-seq and TCR-seq data of liver sinusoidal mononuclear cells isolated from human liver perfusions (24). Processed data were obtained from GEO (GSE: GSE200173). The 33,627 cells that passed quality control criteria (nFeature_RNA > 200, nFeature_RNA < 2500, nCount_RNA > 1000, nCount_RNA < 8000, percent_mito<10) were demultiplexed (HTODemux) and controlled for doublets (scDblFinder). RNA transcript counts were then normalized and scaled before integration, clustering, and UMAP visualization. Barcode-conjugated antibody counts were centered log ratio normalized before scaling. CD4 antibody levels were used for identification of 14,005 CD4 T cells. Clusters were annotated by canonical markers after reintegration, clustering, and visualization. Assessment of induced gene expression programs was performed by applying the TCAT algorithm (25). The gene signature of blood CD161+ CD56+ CD4 T cells was scored by Seurat’s AddModuleScore function. Statistical analysis Flow cytometry data were exported from FlowJo and analyzed in R (v.4.4) using the tidyverse package (v.2) and visualized with the ggplot2 package (v.4.0.0). Comparisons of matched data, such as expression of phenotypic markers of CD4 T cells divided into CD56+/− and CD161+/− populations were made using the Wilcoxon signed-rank test implemented in the rstatix package (v.0.7.2). Comparisons of two unpaired groups were made by Wilcoxon rank-sum test, whereas more than two unpaired groups were analyzed using Kruskal-Wallis’ and Dunn’s post hoc test unless otherwise indicated, with significance levels *P < 0.05, **P < 0.01, and ***P < 0.001. Spearman correlations were calculated using the hmisc and corrplot package (v.5.12, v.0.92).