CD151 identifies a cytotoxic CD4 T cell population enriched in people with HIV that later develop cancer

Abstract

People with HIV (PWH) exhibit persistent immune activation despite effective antiretroviral therapy, contributing to risk of non-AIDS-associated comorbidities such as cancer. Cell populations reflecting immune remodeling trajectories preceding malignancy are poorly characterized. Using peripheral blood mononuclear cells from an adult cohort (25–65 yr), we quantified the tetraspanin CD151 on T cells in people without HIV (PWOH), PWH without documented cancer during follow-up, and PWH who subsequently developed a non-AIDS-defining cancer (PWHc), with samples collected a median of 5 yr prior to cancer diagnosis. In PWOH, CD4⁺CD151⁺ T cell frequencies increased with age, consistent with physiologic immune aging. In contrast, elevated CD4⁺CD151⁺ frequencies were observed at younger ages in both HIV-positive groups, with the typical age-associated increase attenuated. Notably, frequencies were highest in PWHc. CD151 expression was not associated with increased CD25 or CD69, indicating that expansion was not explained by generalized T cell activation. Instead, CD4⁺CD151⁺ T cells were enriched for granzyme B and localized predominantly to CD28⁻ effector memory (CD45RA⁻CCR7⁻) compartments, consistent with a cytotoxic CD4⁺ (cCD4) phenotype. Single-cell RNA sequencing of 1 participant per group identified a cCD4⁺ transcriptional cluster enriched in the PWHc sample. Here, we report that CD151 identifies a cCD4 T cell lineage that accumulates with age in PWOH but appears prematurely expanded in virally suppressed PWH, particularly in PWHc. These findings support CD151⁺cCD4 T cell expansion as a feature of altered immune remodeling detectable years before cancer diagnosis, suggesting a potential role in mechanisms linking chronic immune dysregulation to malignancy risk in PWH.

Introduction

Despite effective antiretroviral therapy (ART), people with HIV (PWH) exhibit persistent immune activation and are at elevated risk for non-AIDS-defining comorbidities such as metabolic syndrome, diabetes, cardiovascular disease, neurocognitive impairment, and cancer.^1–18 These conditions are associated with chronic inflammation and immune dysregulation; however, specific immune cell populations that drive these outcomes remain incompletely defined. Identifying functionally altered T cell subsets that underlie this persistent immune perturbation is therefore essential for understanding mechanisms of disease progression and for improving long-term risk stratification in virally suppressed PWH.

Among the immune alterations observed in chronic infection, the expansion of cytotoxic CD4⁺ (cCD4) T cells has emerged as a feature of sustained immune activation and immune senescence. These cells are typically rare in young individuals but increase with advancing immune age and in response to persistent viral infections, including but not limited to, cytomegalovirus (CMV), HIV, Epstein-Barr virus, influenza, dengue, and hepatitis viruses.^19–35 cCD4⁺ T cells are characterized by the expression of effector molecules such as granzyme B (GZMB) and perforin, transcriptional regulators including TBX21 and ZNF683, and loss of costimulatory receptors such as CD28. Accumulation of these cells has been associated with adverse clinical outcomes in both PWH and people without HIV (PWOH) populations,^22,23,27 yet their contribution to non-AIDS-defining comorbidities and their relationship to defined differentiation markers remain poorly characterized.

Aging is accompanied by progressive remodeling of the immune system that promotes a proinflammatory state, a process known as inflammaging. Inflammaging is characterized by persistent activation of the innate and adaptive immune systems, elevated levels of proinflammatory cytokines, and the accumulation of senescent cells.^36–40 In PWH, features of immune aging occur prematurely, characterized by chronic immune hyperactivation that reduces responses to pathogens and vaccines and increases susceptibility to inflammation driven, age-related comorbidities. As the PWH population ages, the growing burden of these conditions represents a major clinical challenge and highlights the urgent need to define mechanisms of immune dysregulation and identify measurable indicators or therapeutic targets.^{1–11,41–43}

Chronic immune hyperactivation in PWH is associated with an increase of T cells expressing activation markers (e.g. CD25, CD38, CD69, HLA-DR) in the peripheral blood, as well as elevated levels of soluble markers of immune activation in serum, including both lymphocyte-associated (e.g. CD26, CD30) and systemic (e.g. C-reactive protein, IL-6, soluble tumor necrosis factor receptor 1 and 2, and D-dimer) inflammatory markers.^44–55 These features are thought to arise from multiple, non-mutually exclusive processes, including insufficient CD4⁺ T cell recovery, functional impairment of immune cells,^46,56–62 and persistent antigenic stimulation such as microbial translocation.^63,64 However, these measures largely capture global immune activation and inflammation and do not resolve the specific immune cell populations that may drive disease progression or comorbidity risk. This represents a key gap in our understanding of how chronic immune dysregulation contributes to long-term outcomes in PWH.

One candidate population that may bridge activation and differentiation states is defined by expression of the tetraspanin CD151. We previously demonstrated that CD151 marks a subpopulation of hyperresponsive, proinflammatory T cells and is upregulated following activation independently of canonical activation markers such as CD25, CD38, and CD69.^65,66 In general, tetraspanins are involved in the regulation of cell morphology, motility, proliferation, and apoptosis.^67–72 They laterally coordinate cell surface molecules including other tetraspanins, integrins, CD2, CD4, CD8, CD5, CD19, CD21, Fc receptors, and major histocompatibility complex classes I and II,⁷⁰ which results in the formation of highly specialized structures known as tetraspanin-enriched microdomains.^73–75 Tetraspanins influence a spectrum of kinase activities (PKC, AKT, ERK/MAPK),^67,76–81 and tetraspanins on T cells (CD9, CD37, CD81, CD151) affect T cell activation in ways similar to costimulation.^69,82–86 CD151 specifically was shown to stabilize the immunological synapse following T cell activation.^87,88 Increased CD151 expression has been associated with enhanced proliferation and invasiveness across multiple tumor types, suggesting a broader role in regulating effector potential and cellular persistence.^89–97

Here, we examined the expansion and functional characteristics of CD151⁺ cCD4⁺ T cells in PWH with particular attention to individuals who subsequently developed non-AIDS-defining malignancies. Using multiparameter flow cytometry and single-cell RNA sequencing (scRNA-seq), we identify a CD4⁺ population coexpressing CD151 and markers of terminal differentiation, cytotoxic function, and effector memory phenotype. We demonstrate that this population is enriched in PWH relative to similarly aged individuals without HIV and is further elevated in PWH who later develop cancer, indicating that chronic infection is associated with altered immune-remodeling trajectories that are not solely explained by chronological age. These findings provide insight into the cellular architecture of immune dysfunction in virally suppressed PWH and support the concept that CD151⁺ cCD4⁺ T cells represent a lineage-defined feature of sustained immune activation and premature immune remodeling with a potential relevance to comorbidity risk.

Methods

Cohort characteristics

Frozen peripheral blood mononuclear cell (PBMC) samples were obtained from the 1917 Clinic cohort at the University of Alabama at Birmingham. Cohort (25-65 yr of age) included PWOH (n = 16), people with HIV without documented cancer during follow-up (PWHnc) (n = 16) and people with HIV who were diagnosed with a non-AIDS-defining malignancy after sample collection (PWHc) (n = 18). All PWH participants were on ART and with undetectable viral loads (<100 copies/mL). Cancer diagnoses were identified using International Classification of Diseases–Tenth Revision diagnostic codes within the institutional electronic medical record system and confirmed by manual chart review. Participants were screened for cancer diagnoses from the date of blood draw through the most recent available medical record at the time of data extraction. No minimum follow-up duration was imposed. All data and samples were collected in accordance with the University of Alabama at Birmingham Institutional Review Board, and all participants provided written informed consent for all biologic specimens and clinical data used in this study. A summary of patient demographic and clinical status data is shown in Table 1. Functional cytotoxicity assays included 5 additional PWOH participants lacking complete clinical metadata. These participants were included exclusively in functional GZMB analyses and were excluded from age-stratified, regression, and correlation analyses.

Table 1.

Characteristics of the cohort.

	PWOH (n = 16)	PWHnc (n = 16)	PWHc (n = 16)a
Age, yr	41 (25–62)	45 (37–62)	49 (33–58)
Sex
Female	11 (69)	5 (31)	2 (11)
Male	5 (31)	11 (69)	16 (89)
Race
White	10 (63)	5 (24)	10 (58)
Black or African American	5 (31)	15 (71)	11 (52)
Other	1 (6)	1 (5)	0 (0)
Years prior to cancer diagnosis	—	—	5 (0.4–10)
Cancer type distribution
Virally induced	—	—	6 (33)
Nonvirally induced	—	—	12 (67)
ART regimen
NRTI	—	13 (81)	11 (61)
NNRTI	—	1 (6)	3 (16)
PI	—	0	2 (11)
INSTI	—	2 (13)	1 (6)
EC	—	0	1 (6)
CD4
Count	—	655 (279–947)	659 (322–1,039)
Nadir	—	243 (27–733)	196 (45–659)

Values are median (range) or n (%).

People without HIV (PWOH), People with HIV without documented cancer (PWHnc), People with HIV who were diagnosed with a non-AIDS-defining malignancy (PWHc), Nucleoside reverse transcriptase inhibitor (NRTI), Non-nucleoside reverse transcriptase inhibitor (NNRTI), Protease inhibitor (PI), Integrase strand transfer inhibitor (INSTI), Elite Controller (EC).

^aPeople with HIV who were diagnosed with non-AIDS-defining malignancy varied in type of cancer diagnosis, and samples were taken up to 10 years prior to cancer diagnosis.

Cell culture

For experiments involving cell culture, cells were maintained in RPMI 1640 supplemented with 2 mM L-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, and 10% heat-inactivated fetal bovine serum (Cytiva HyClone; characterized fetal bovine serum). In all cases, cells were rested in medium overnight prior to stimulation.

Flow cytometric analysis

LIVE/DEAD Fixable Blue Dead Cell Stain (Molecular Probes; Invitrogen) was used to exclude dead cells from analysis. T cell phenotyping was performed by staining with monoclonal antibodies: CD3-APC efluor780 (SK7; Invitrogen), CD4-BV786 (SK3; BD Biosciences), CD8-V500 (RPA-T8; BD Biosciences), PD-1− PE/Cyanine7 (EH12.2H7; BioLegend), CD38-BUV395 (HB7; BD Biosciences), CD28-BV711 (CD28.2; BD Biosciences), CD151-PE (14A2.H1; BD Biosciences), and CD69-FITC (FN50; BD Biosciences). Antibodies recognizing CD45RA-APC (HI100; BD Biosciences) and CCR7-BV650 (2-L1-A; BD Biosciences) were used to define naïve T cells (T_N) (CD45RA⁺CCR7⁺), central memory T cells (T_CM) (CD45RA^– CCR7⁺), effector memory T cells (T_EM) (CD45RA^– CCR7^–) and T effector memory RA-positive cells (T_EMRA) (CD45RA⁺ CCR7^–), as previously reported.^65,66 Intracellular cytokine secretion experiments were stained with monoclonal antibodies: CD3-APC efluor780 (SK7; Invitrogen), CD4-BV786 (SK3; BD Biosciences), CD8-V500 (RPA-T8; BD Biosciences), CD14-FITC (M5E2; BD Biosciences), CD19-FITC (HIB19; BD Biosciences), CD151-PE (14A2.H1; BD Biosciences), CD38-BUV395 (HB7; BD Biosciences), CD28-BV711 (CD28.2; BD Biosciences), CD45RA-APC (HI100; BD Biosciences), and CCR7-BV650 (2-L1-A; BD Biosciences). For the detection of GZMB and IFN-γ, after thawing, cells were stimulated with SEB at 1 µg per milliliter for a total of 5 h. At 2 h, the protein transport inhibitors monensin and brefeldin A were added to block the transport of new protein to the cell surface (BD Biosciences).^65,66 Following stimulation, cells were stained with the surface antibodies stated previously. Cells were then washed in phosphate-buffered saline (PBS) with 2% fetal bovine serum, the cells were permeabilization with the Cytofix/Cytoperm reagent (BD Biosciences) for 20 min at 4 °C in the dark. Per the manufacturer’s instructions, the cells were then washed twice with PermWash and stained for Granzyme B-V450 (GB11; BD Biosciences) and IFN-γ-PE-Cy7 (B27; BD Biosciences). All experiments were analyzed on a BD Symphony flow cytometer in the University of Alabama at Birmingham Comprehensive Flow Cytometry Core Facilities. Flow cytometry results were analyzed using FlowJo software (v.10.10.0; BD Biosciences). For the gating strategy, please refer to Figure S1. CD151 positivity thresholds were established using fluorescence-minus-one controls. CD151 expression demonstrated a bimodal distribution in both CD4⁺ and CD8⁺ T cells, supporting discrete CD151⁺/CD151⁻ subset classification (Fig. S2).

Sample preparation for single-cell RNA sequencing

All cells were prepared in the University of Alabama at Birmingham Flow Cytometry and Single Cell Analysis Core. Frozen PBMC samples were thawed, counted, and assessed for viability. After incubation at 37 °C for 24 h, the PBMCs were sorted for CD4 expression (CD4-FITC [RPA-T4; BD Biosciences]). The sorted CD4⁺ T cells were prepared for single-cell library generation using Chromium Single Cell 3′ v3 Reagent Kits and captured using a 10x Chromium controller (10x Genomics). All steps were performed according to the manufacturer’s instructions.⁹⁸

Single-cell RNA sequencing was performed on CD4⁺ T cells isolated from 1 representative participant per cohort (PWOH, PWHnc, and PWHc: 53, 27, and 42 yr of age, respectively). Libraries were generated using the 10x Genomics Chromium Single Cell 3′ v3 platform.

Sequencing yielded a median of 52,372, 59,908, and 27,580 reads per cell for PWOH, PWHnc, and PWHc samples, respectively, with median gene detection ranging from 2,386 to 2,948 genes/cell.

scRNA-seq data processing

The scRNA-seq dataset was processed, explored, and visualized using Parse Bioscience Trailmaker. Trailmaker enables biologists to analyze their scRNA-seq datasets without a need for bioinformatics coding skills and allows every interested researcher to reproduce our analysis.

Quality control filtering included removal of cells below dataset-specific gene count thresholds, mitochondrial content filtering (≤6%), and doublet probability filtering. Data normalization, integration, clustering, and dimensionality reduction were performed using Trailmaker with Harmony integration, log normalization, Louvain clustering (resolution 0.3), and Uniform Manifold Approximation and Projection embedding. Full configuration parameters and classifier thresholds are provided in Supplemental Methods to enable reproducibility.

Statistical analysis

Statistical analyses were performed using GraphPad Prism software 10.0.02 for Mac (GraphPad Software). Nonparametric tests were used unless otherwise specified. Paired comparisons were evaluated using Wilcoxon signed rank test, while unpaired comparisons used Mann-Whitney U or Kruskal-Wallis tests followed by Dunn’s multiple comparisons correction. Associations between age and CD151 frequencies were assessed using Spearman rank correlation. Pearson correlations were also calculated, and differences between correlation coefficients were evaluated using Fisher’s r-to-z transformation. Permutation testing was used as a sensitivity analysis where appropriate. Multivariable linear regression models included age, ART class, and CD4 nadir to account for historical immune depletion and treatment-associated effects. To quantify the total frequency of CD151⁺ cells within distinct memory T cell populations (T_N, T_CM, T_EM, T_EMRA), we employed a hierarchical calculation. First, the frequency of the memory population was determined. Then, this value was multiplied by the frequency of CD151⁺ cells within the respective memory subset. This method ensured that the total frequency of CD151⁺ cells within each memory subset reflected both the proportional size of the memory population and the prevalence of CD151⁺ cells within that subset. Differences were considered to be significant on the basis of 95% confidence intervals (P < 0.05).

Results

CD4⁺CD151⁺ T cell frequencies exhibit age associated increases in PWOH but diminished age dependence in PWH

Aging is associated with a gradual shift toward a proinflammatory immune environment, a process often referred to as inflammaging.^36–39 We previously demonstrated that the tetraspanin CD151 marked T cells that exhibit an activated, proinflammatory phenotype.^65,66 We therefore investigated whether CD151⁺ T cells accumulate with age in PWOH, as their expansion could reflect age associated immune remodeling. We further hypothesized that elevated CD151⁺ T cell frequencies would be seen younger in life within PWH. Given their proinflammatory nature, the increased presence of CD151⁺ T cells in younger PWH would be consistent with altered immune remodeling rather than strictly age-dependent changes.^14,99–105

To evaluate whether CD151⁺ T cell frequencies differ by HIV status and clinical trajectory, we examined CD151 expression across 3 groups of individuals 25 to 65 yr of age: PWOH, PWHnc, and PWHc. Cancer was selected as a clinically relevant comorbidity given its well-established links to immune dysfunction and chronic inflammation.^37,106–113 Cancer status was ascertained through International Classification of Diseases–Tenth Revision diagnostic codes and clinical record review spanning the period from sample collection to the most recent available medical record. Importantly, samples from the PWHc group were collected a median of 5 yr prior to cancer diagnosis (range 0.4–10 yr), facilitating the evaluation of immune phenotypes preceding clinical disease. The interval between sample collection and cancer diagnosis varied across participants, reflecting the retrospective nature of cohort identification. All PWHnc and PWHc had sustained virologic suppression on ART, and CD4 counts were ≥250 cells/mm³ (range 279–1,039 cells/mm³) (Table 1). There was no indication that cancer etiology (virus associated versus non–virus associated) affected CD151⁺ T cell frequencies (Fig. S3).

By using flow cytometric analysis, we observed a significant age-related increase in CD4⁺CD151⁺ T cells in PWOH (Spearman ρ = 0.56, P = 0.02; Pearson r = 0.58, P = 0.02) (Fig. 1A). CD4⁺CD151⁺ T cell frequencies increased from approximately 5% to 37% as age increased from 25 to 65 yr, consistent with a gradual age-associated shift in CD4 T cell phenotype in otherwise healthy individuals.

Figure 1.

CD151⁺CD4⁺ T cell frequencies increase with age in PWOH but are prematurely elevated in PWHc. PBMCs from PWOH (n = 16), PWHnc (n = 16), and PWHc (n = 18) were analyzed using multiparameter flow cytometry. Scatter plots demonstrate the relationship between age and the frequency of CD4⁺CD151⁺ T cells in PWOH (A), PWHnc (B), and PWHc (C). Spearman and Pearson correlation analysis was performed for all scatter plots, and coefficients are reported. (D) Comparison of CD4⁺CD151⁺ T cells among groups. Horizontal bars indicate medians. Statistical comparisons were performed using Kruskal-Wallis (KW) testing with post hoc Dunn’s correction. P values are indicated above comparisons.

In contrast, this age associated relationship was not observed in PWHnc (Spearman ρ= −0.15, P = 0.58; Pearson r= −0.07, P = 0.79) (Fig. 1B) or in PWHc (Spearman ρ= −0.03, P = 0.91; Pearson r= −0.23, P = 0.37) (Fig. 1C). Instead, many individuals in both PWH groups exhibited elevated CD4⁺CD151⁺ T cell frequencies at younger ages, resulting in loss of the positive age correlation seen in PWOH. Direct comparison of Pearson correlation coefficients using Fisher’s r-to-z transformation demonstrated that the age association of CD4⁺CD151⁺ T cell frequencies in PWOH differed significantly from PWHc and trended different from PWHnc (Table S1). Within PWH, exploratory multivariable models adjusting for ART class, viral load at sampling, and CD4 nadir did not alter the observed group differences, although statistical power was limited (Table S2).

To directly compare CD4⁺CD151⁺ T cell frequencies independent of age-correlation effects, we next performed a cross-sectional comparison across the cohort. CD4⁺CD151⁺ frequencies were significantly higher in PWHc compared with PWOH (median 21.4% vs 11.7%, P = 0.02) (Fig. 1D), while PWHnc exhibited intermediate values that did not significantly differ from PWOH. In contrast, no significant age-associated relationships or overall group differences were detected for CD8⁺CD151⁺ T cells (Fig. S4).

These findings indicate that, rather than representing a simple acceleration of chronological aging, CD151⁺ T cell expansion in PWH reflects an altered immune remodeling trajectory characterized by elevated baseline frequencies and diminished age dependence.

CD151⁺ T cells exhibit a distinct activation and cytotoxic profile

To characterize the functional properties of the expanded CD151⁺ T cell population, we evaluated classical activation markers and cytotoxic mediators within CD4⁺ and CD8⁺ T cell subsets. Although CD151 has been associated with recent T cell activation,^65,66 its expression is regulated independently of canonical activation pathways and may instead reflect a differentiated effector phenotype. We therefore compared CD151⁺ and CD151⁻ T cells within each participant to determine whether CD151 expression defines a functionally specialized subset.

Classical activation markers such as CD69 and CD25 did not differ significantly across CD4⁺ T cell populations among PWOH, PWHnc, and PWHc (Fig. 2A, B), consistent with prior reports in which comorbidity associated immune changes were linked primarily to myeloid rather than T cell activation marker alterations.¹¹⁴ These findings indicate that CD151 expansion is not simply attributable to broad CD4⁺ T cell activation and suggest that CD151 identifies a phenotypically distinct subset regulated independently of classical activation pathways. Comparable patterns were observed within the CD8⁺ T cell compartment (Fig. S5).

Figure 2.

CD151⁺CD4⁺ T cells exhibit cytotoxic enrichment independent of classical activation markers. PBMCs from PWOH (n = 7), PWHnc (n = 6), and PWHc (n = 7) were subjected to flow cytometric analysis to determine classical activation markers and baseline cytotoxic mediator expression. (A, B) Frequencies of CD4⁺CD69⁺ and CD4⁺CD25⁺ T cells across groups. (C) Within-participant paired comparisons of GZMB expression frequency between CD4⁺CD151⁺ and CD4⁺CD151⁻ subsets in the 3 experimental groups. Wilcoxon signed rank tests were used. (D) Between-group comparison of CD4⁺CD151+GZMB⁺ T cell frequencies. Medians are indicated by horizontal bars. Statistical testing was performed using Kruskal-Wallis with post hoc Dunn’s correction.

We next assessed cytotoxic effector molecule expression as a function of CD151 status. We previously reported that in PWOH, both CD4⁺CD151⁺ and CD8⁺CD151⁺ T cells were more likely to exhibit a proinflammatory phenotype than their CD151⁻ counterparts and express preformed GZMB. Because GZMB contributes to chronic inflammatory conditions and immune remodeling,^115–119 early accumulation of CD151⁺ T cells in PWH could promote a sustained proinflammatory immune environment. Therefore, we analyzed CD4⁺ and CD8⁺ T cells from PWOH, PWHnc, and PWHc for preformed GZMB expression as a function of CD151 status.

Within all groups, CD4⁺CD151⁺ T cells consistently exhibited higher frequencies of GZMB⁺ cells compared with paired CD4⁺CD151⁻ subsets (paired Wilcoxon tests) (Fig. 2C). In PWOH, median CD4⁺GZMB⁺ frequencies increased from 0.70% in CD151⁻ T cells to 4.76% in CD151⁺ T cells (P = 0.02). Similar within-group increases were observed in PWHnc (0.71% to 9.48%; P = 0.03) and PWHc (0.54% to 10.3%; P = 0.02). However, overall CD4⁺CD151⁺GZMB⁺ frequencies did not significantly differ among PWOH, PWHnc, and PWHc (Fig. 2D), indicating that cytotoxic enrichment associated with CD151 expression is preserved across groups rather than selectively amplified.

To distinguish between cytotoxic cell frequency and per-cell effector intensity, we next evaluated GZMB median fluorescence intensity (MFI) specifically within the GZMB⁺ population (Fig. S6). Among CD4⁺GZMB⁺ T cells, CD151⁺ subsets exhibited significantly higher GZMB MFI than paired CD151⁻ counterparts in PWOH and PWHnc, whereas this difference was not observed in PWHc. Despite these within participant differences, overall CD4⁺CD151⁺GZMB⁺ MFI did not significantly differ across groups by Kruskal-Wallis testing, indicating that between-group differences arise primarily from expansion of cytotoxic CD4⁺ T cell frequencies, rather than increased granzyme content per cell.

A similar pattern was observed within the CD8⁺ T cell compartment, in which CD8⁺CD151⁺GZMB⁺ cells demonstrated higher granzyme intensity relative to paired CD151⁻ subsets in PWHc and trended in PWHnc but did not reach significance in PWOH (Fig. S6). Between-group comparisons of CD8⁺GZMB⁺ MFI were likewise not significant. Representative distribution plots supporting CD4 and CD8 CD151^+/− GZMB⁺ gating are provided in Fig. S7. To further assess whether CD151 expression level scaled with cytotoxic effector intensity, we examined correlations between CD151 MFI or CD151⁺ cell frequency and GZMB MFI within CD151⁺GZMB⁺CD4⁺ and CD8⁺ T cells. No significant correlations were detected (Fig. S8).

Collectively, these findings demonstrate that CD151 marks a functionally distinct T cell subset characterized by cytotoxic and proinflammatory potential that is maintained across HIV status and is not explained by generalized T cell activation. Importantly, the enrichment observed in PWHc appears to reflect subset expansion rather than heightened per-cell effector intensity, supporting a model in which CD151 identifies lineage identity and differentiation state, while group-level differences are driven primarily by increased abundance of cytotoxic CD4⁺ T cells, rather than amplification of cytotoxic strength per cell.

CD151 enrichment is compartment specific, preferentially localizing to senescent and effector memory CD4⁺ T cell subsets

Having established that CD151⁺ T cells retain a cytotoxic phenotype across groups, we next examined whether CD151 expansion preferentially occurred within T cell populations exhibiting features of terminal differentiation. Loss of CD28 expression is widely used as an indicator of replicative history and reduced costimulatory capacity in human T cells and reflects cumulative immune activation rather than chronological age alone.¹²⁰

Across the cohort, CD28 loss was more pronounced in CD8⁺ than CD4⁺ T cells and was generally elevated in PWH relative to PWOH. Kruskal-Wallis testing revealed a significant difference in CD4⁺CD28⁻ T cell frequencies among groups (P = 0.02), with post hoc Dunn’s testing indicating higher frequencies in PWHc compared with PWOH, whereas PWHnc did not significantly differ from PWOH (Fig. 3A). Comparable patterns were observed within CD8⁺ populations, although these differences did not reach statistical significance at the global level (Fig. S9).

Figure 3.

CD151⁺CD4⁺ T cells are selectively expanded within CD28⁻ senescent and effector memory compartments in PWHc. Samples from PWOH (n = 16), PWHnc (n = 16), and PWHc (n = 18) were subjected to flow cytometric analysis to evaluate CD151 expression as a function of T cell differentiation and memory subset composition. (A) Percentages of CD28⁻ T cells in CD4⁺ T cell populations of PWOH, PWHnc, and PWHc. (B) Percentages of CD151⁺ T cells in the functional CD4⁺CD28⁺ T cell populations of PWOH, PWHnc, and PWHc. (C) Percentages of CD151⁺ T cells in the senescent CD4⁺CD28⁻ T cell populations of PWOH, PWHnc, and PWHc. (D) Percentages of T_EM in PWOH, PWHnc, and PWHc. (E) Percentages of CD151⁺ cells within the CD4⁺ T_EM compartment in PWOH, PWHnc, and PWHc. Median values are indicated by horizontal bars. Statistical comparisons were performed using Kruskal-Wallis (KW) test with post hoc Dunn’s correction.

To determine whether CD151 expression preferentially localized to functionally distinct CD28 compartments, we next evaluated CD151 frequencies within CD28⁺ and CD28⁻ CD4⁺ subsets. CD151⁺ cells were detected within both functional (CD28⁺) and senescent (CD28⁻) populations; however, enrichment was most apparent within the CD28⁻ compartment. Within the CD4⁺CD28⁺ subset, overall group differences trended toward significance by Kruskal-Wallis testing (P = 0.06). Post hoc analysis demonstrated higher median CD151 frequencies in PWHc (18%) relative to PWOH (12%; Dunn’s P = 0.04), whereas differences between PWHnc (17%) did not significantly differ from PWOH (Fig. 3B). A similar pattern was observed within the CD4⁺CD28⁻ compartment, in which global differences also trended toward significance (Kruskal-Wallis P = 0.07). Median CD151⁺ frequencies were elevated in PWHc (51%) compared with PWOH (20%; Dunn’s P = 0.05), while PWHnc (31%) did not significantly differ from PWOH (Fig. 3C). This consistent directionality and magnitude of CD151 enrichment within both functional and senescent defined compartments support a model in which CD151 expansion preferentially aligns with terminal differentiation rather than generalized activation. Parallel CD8⁺ analyses did not reveal significant associations and are provided in Fig. S9.

We next asked whether this compartment-specific enrichment reflected redistribution across defined memory subsets. Consistent with prior reports, PWH exhibited reduced proportions of naïve CD4⁺ T cells with corresponding increases in differentiated memory subsets, a pattern described in both physiological aging and chronic immune activation.^{101,121–125} However, following Kruskal-Wallis testing, overall differences in total T_N, T_CM, T_EM, and T_EMRA frequencies did not reach statistical significance across groups (Fig. S10). Relative to PWOH (11.7%), CD4⁺ T_EM frequencies were comparable in PWHnc (11.0%) and modestly, nonsignificantly elevated in PWHc (17.0%) (Fig. 3D), indicating that expansion of the total effector memory pool alone does not fully account for cohort differences.

Therefore, we examined the proportions of CD151⁺ cells within each memory subset. Across most compartments, CD151⁺ frequencies were similar between groups or demonstrated only nonsignificant directional changes (Fig. S10). In contrast, a significant overall cohort effect was observed in the proportion of CD151⁺ cells within the CD4⁺ compartment (Kruskal-Wallis P = 0.048) (Fig. 3E). Although post hoc pairwise comparisons did not survive correction, CD151⁺ T_EM frequencies demonstrated a consistent directional increase in PWHc (6.46%) relative to PWOH (3.81%) (Dunn’s P = 0.06), whereas PWHnc values more closely paralleled those of PWOH. CD151 enrichment was not observed within T_N, T_CM, or T_EMRA compartments, and analogous analyses of CD8⁺ memory subsets did not reveal reproducible group-level differences. These findings indicate that CD151 expansion in PWHc does not arise from generalized memory redistribution but instead reflects preferential enrichment within the CD4⁺ effector memory pool.

Together, these data support a model in which CD151 marks a phenotypically distinct CD4⁺ effector subset characterized by cytotoxic and terminal differentiation features. Rather than broad activation across all T cell lineages, enrichment appears to be compartment specific and detectable years prior to cancer diagnosis. Given prior associations of CD151 expression with recent or repeated antigen exposure, this pattern is consistent with sustained immune stimulation within a defined CD4⁺ memory subset, rather than uniform activation across the entire T cell compartment.

Increased CD4⁺CD151⁺ T cell frequencies in PWHc are driven by an accumulation of cytotoxic CD4⁺ T cells

To obtain more detailed information on the composition and functional nature of the CD4⁺CD151⁺ T_EM cell population in PWHc (Fig. 3E), we performed scRNA-seq analysis of sorted CD4⁺ T cell populations using 1 participant per group (PWOH, PWHnc, PWHc), with sequencing depths ranging from ∼27,000 to 60,000 reads per cell following quality-control filtering. Unsupervised Uniform Manifold Approximation and Projection analysis identified 8 transcriptionally distinct clusters (Fig. 4A). Assessment of participant contribution to each cluster revealed that cluster 6 was disproportionately composed of cells from the PWHc participant (PWOH: 24%; PWHnc: 16%; PWHc: 60%) (Fig. 4B). When normalized to each participant’s total CD4⁺ T cell population, cluster 6 represented 2.96% of PWOH, 1.98% of PWHnc, and 7.22% of PWHc CD4⁺ T cells (Fig. 4C), indicating selective enrichment in the PWHc sample rather than uniform expansion across groups.

Figure 4.

scRNA-seq reveals the accumulation of cytotoxic CD4⁺ T cells in PWHc. Samples from 1 PWOH, 1 PWHnc, and 1 PWHc participant were subjected to scRNA-seq of CD4⁺ T cells to define transcriptionally distinct populations. (A) Uniform Manifold Approximation and Projection (UMAP) visualization of integrated CD4⁺ T cells with clusters annotated as T_N, T_CM, T_EM, regulatory T (T_REG), cCD4⁺ T cells are shown. (B) Proportional contribution of each participant to cluster 6 (cytotoxic CD4⁺ T cells). (C) Proportional contribution of each cluster to the overall T cell composition of each participant. Numbers indicate the relative percentage of cytotoxic CD4⁺ T cells/cluster 6 cells to the overall T cell population. (D) Dot plot analysis showing expression of lineage-defining genes associated with cytotoxic CD4⁺ T cells across clusters. (E) Numerical representation of the dot plot visualization of lineage-defining genes demonstrating enrichment of cytotoxic markers in panel D. FC, fold change.

Transcriptional profiling of cluster 6 cells revealed that, consistent with an assignment to a T_EM-like population,¹²⁶CD62L (SELL) and CCR7 expression in cluster 6 was low, whereas TBX21 (Tbet) and PRDM1 (Blimp1) expression was elevated (Fig. 4D, E). Cluster 6 cells expressed high levels of ID2 and ZEB2, both transcription factors associated with advanced memory states. Beyond conventional T_EM features, cluster 6 cells expressed ZNF683 and EOMES, two transcription factors that are considered markers of cCD4⁺ T cells.²⁷ The determination that cluster 6 cells were cCD4⁺ T cells was further supported by the presence of high baseline expression levels of cytotoxic genes including but not limited to GNLY, NKG7, GZMB, GZMH, and GZMA, as well as the expression of FASLG and CCL5 (RANTES) (Fig. 4D, E). Cells in the cCD4⁺ T cell cluster expressed CD151 but importantly had lost CD28 expression indicative of replicative senescence in PWH (Fig. 4D, E).¹²⁷ Similar to the accumulation of CD8⁺ T_EMRA cells, the accumulation of cytotoxic CD4⁺ T cells has been described in the setting of chronic viral infections.^30,128–133

Based on our scRNA-seq data, cCD4⁺ T cells would be defined as CD4⁺CD45RA^–CCR7^–CD28^–CD151⁺ T cells in our flow cytometric analysis. To validate that this marker combination also applied to the protein level, we performed a second set of flow cytometric analyses to confirm that CD4⁺CD45RA^–CCR7^–CD28^–CD151⁺ T cells express preformed GZMB. Indeed, >75% of CD4⁺CD45RA^–CCR7^–CD28^–CD151⁺ T cells in PWHnc and PWHc expressed preformed GZMB (Fig. 5A). Having confirmed the presence of preformed GZMB in CD4⁺CD45RA^–CCR7^–CD28^–CD151⁺ T cells, we determined the frequency of cCD4⁺ T cells in the 3 study groups. As expected, cCD4⁺ T cell frequencies were very low in PWOH (median 0.2%), and modestly higher in PWHnc (0.6%), with the highest median frequencies observed in PWHc (1.9%). Kruskal-Wallis testing demonstrated a significant overall group difference (P = 0.008). Dunn’s multiple comparisons testing identified a significant difference between PWOH and PWHc (P = 0.006), whereas the comparison between PWHnc and PWHc did not reach statistical significance (Fig. 5B).

Figure 5.

Flow cytometric analysis confirms expansion of cytotoxic CD4⁺CD151⁺ T cells in PWHc. (A) Intracellular staining of GZMB within CD4⁺CD45RA⁻CCR7⁻CD28⁻CD151⁺ T cells (marked by rectangles) in PWHnc (n = 6) and PWHc (n = 7). (B) Frequencies of cytotoxic CD4⁺ T cells identified as CD45RA^–CCR7^–CD28^–CD151⁺ T cells in PWOH, PWHnc, and PWHc. Horizontal bars represent medians. Statistical comparisons were performed using Kruskal-Wallis (KW) testing with post hoc Dunn’s correction.

These findings indicate that the elevated CD4⁺CD151⁺ T cell frequencies observed in PWHc are driven, at least in part, by selective enrichment of a cytotoxic, effector memory–like CD4⁺ T cell population rather than uniform activation across all CD4⁺ T cells. While single-participant scRNA-seq analysis limits statistical generalization, the convergence of transcriptional profiling and protein-level validation supports the presence of a phenotypically distinct cCD4⁺ subset preferentially expanded in PWHc. Notably, this phenotype is consistent with prior reports demonstrating that cCD4⁺ T cells are largely absent in younger individuals and instead accumulate with advanced immune age, in which they are considered a hallmark of immune senescence and chronic antigen exposure.^134,135

Discussion

Aging is associated with a progressive reshaping of the immune landscape, including the accumulation of terminally differentiated memory T cells and persistent, low-grade inflammation features commonly referred to as inflammaging.¹³⁶ This process is characterized by persistent immune activation, elevated proinflammatory cytokine levels, loss of naïve T cells, and the accumulation of senescent and cytotoxic T cell subsets.⁴⁰ In PWH, these immunologic shifts occur prematurely, even in the context of viral suppression via ART.^9,41,137 Here, we demonstrate that chronic HIV infection is associated with altered immune remodeling in virally suppressed PWH and identify a disproportionate expansion of CD151⁺ cCD4⁺ T cells, particularly in individuals who later develop cancer (PWHc).

Although studies of comorbid complications in PWH have emphasized myeloid dysregulation, including monocyte subset shifts linked to cardiovascular disease,^60,138–141 fewer studies have connected defined T cell subsets to comorbidity risk.^{114,142–144} Given the evidence of premature immune aging in PWH, including the loss of CD28 in CD8⁺ T cells and acquisition of NKG2D expression on CD4⁺ T cells,¹³⁴ we hypothesized that cytotoxic CD4⁺ subsets analogous to those accumulating during physiologic aging,⁴⁰ may emerge at earlier ages in PWH. Building on prior work identifying CD151 as a marker of recently activated T cells with increased proinflammatory potential,^65,66 we evaluated whether CD151 delineates such a subset.

In PWOH, CD4⁺CD151⁺ T cell frequencies increased steadily with age, consistent with physiologic immune aging and the gradual accumulation of effector memory populations. In contrast, PWH exhibited elevated CD4⁺CD151⁺ T cell frequencies at younger ages, with PWHc demonstrating the most pronounced expansion. This premature accumulation was accompanied by contraction of the naïve CD4⁺ T cell pool and expansion of central and effector memory subsets, supporting a model in which chronic HIV associated antigen stimulation alters the timing of immune remodeling rather than simply accelerating chronological aging. Consistent with this, CD151 expression in PWH appeared partially decoupled from age, suggesting that its expansion reflects disease associated immune remodeling.

The age range of our cohort (25–65 yr) spans a period in which cytotoxic and differentiated T cell populations remain dynamic, enabling detection of deviations from normative immune aging. Within this range, CD151⁺ T cell frequencies increased with age in PWOH, consistent with physiologic immune remodeling. In contrast, elevated CD151⁺ T cell frequencies were observed at younger ages in PWH, particularly in PWHc, indicating a shift toward earlier accumulation of cytotoxic and differentiated T cell states. Importantly, the interpretive value of CD151 expression appears context dependent. At older ages, in which cytotoxic and activated T cell populations accumulate in individuals with no comorbidities, as seen in supercentenarians,¹³⁵ the relative specificity of CD151 enrichment may diminish. Together, these findings highlight a context-dependent window in which CD151 most effectively reflects dysregulated immune remodeling.

A central observation of this study is the strong association between CD151 expression and cytotoxic CD4⁺ T cell differentiation. CD151⁺CD4⁺ T cells were enriched within effector memory and CD28⁻ compartments, expressed transcription factors associated with terminal differentiation (TBX21, PRDM1, ZEB2, ID2, EOMES, ZNF683), and demonstrated elevated baseline expression of cytolytic mediators including GZMB, GZMA, GZMH, GNLY, NKG7, and FASLG. These transcriptional and protein level features indicate that CD151 does not merely mark transient activation but instead identifies a stable, cytotoxic CD4⁺ lineage characterized by replicative history and sustained effector potential. As cytotoxic CD4⁺ T cells are uncommon in younger individuals and accumulate with immune aging,^134,135 their enrichment in PWHc supports CD151 as a biologically meaningful marker of immune remodeling rather than nonspecific activation.

Importantly, CD151 expression did not correlate with GZMB protein intensity at the single-cell level, indicating that CD151 enrichment reflects expansion of a differentiated cytotoxic lineage, rather than upregulation of cytolytic capacity on a per-cell basis. This distinction suggests that the immune risk conferred by CD151⁺ cCD4⁺ T cell accumulation in PWHc may derive from the cumulative burden of this subset within the circulating T cell pool, rather than from heightened per-cell effector function.

The mechanisms driving this immune remodeling remain incompletely defined. Persistent low-level HIV transcription within tissue reservoirs, despite effective viral suppression, may provide continuous antigenic stimulation sufficient to sustain cytotoxic T cell differentiation. Notably, emerging data suggest that cCD4⁺ T cells can exhibit features consistent with long-lived, antigen-experienced populations that are known to harbor HIV proviral DNA, raising the possibility that HIV either preferentially infects these cells or actively promotes their differentiation through chronic antigen exposure.²⁴ In parallel, microbial antigen-specific CD4⁺ T cells have been implicated as potential contributors to the HIV reservoir and have been observed to exhibit cytotoxic features, further linking antigen specificity, reservoir persistence, and effector differentiation.^24,145,146

Viral co-infections, particularly CMV, likely represent an additional source of sustained immune stimulation.¹⁴⁷ CMV seropositivity is highly prevalent in PWH, with a recent study finding more than 90% co-infected,¹⁴⁸ and has been associated with chronic immune activation, and expansion of terminally differentiated T cell populations. Both HIV and CMV may also contribute to disruption of epithelial barrier integrity,^149,150 consistent with early depletion of gut-associated lymphoid tissue and incomplete mucosal immune reconstitution following ART. This aligns with the “gut leak” hypothesis proposed by Brenchley and Douek,^151,152 which posits that microbial translocation from compromised gut epithelium sustains systemic immune activation during chronic HIV infection. Circulating microbial products such as lipopolysaccharide and soluble CD14 may further amplify this process and promote expansion of proinflammatory CD151⁺ T cell subsets across tissues. Consistent with the broader role for cCD4⁺ T cell expansion in chronic disease, cytotoxic CD4⁺CD28⁻ T cells has been associated with increased cardiovascular mortality and adverse outcomes across chronic inflammatory conditions,¹³¹ raising the possibility that CD151⁺ cCD4⁺ T cell expansion reflects broader immune vulnerability rather than oncologic risk alone.

These observations raise the possibility that therapeutic strategies aimed at attenuating CD151⁺ cCD4⁺ T cell accumulation, either directly or through modulation of upstream drivers such as CMV infection or microbial translocation, may have clinical relevance. Short-term CMV suppression with valganciclovir reduces immune activation in virally suppressed PWH,¹⁵³ and the CMV DNA terminase inhibitor letermovir, which has a more favorable safety profile, may provide a tractable approach to test whether CMV suppression attenuates accumulation of CD151⁺ cCD4⁺ T cells. In parallel, immunomodulatory interventions may also influence these processes. The Randomized Trial to Prevent Vascular Events in HIV (REPRIEVE) trial demonstrated that statin therapy reduces cardiovascular events in PWH,^154–156 and given their anti-inflammatory and antiviral effects,^157–175 it is plausible that part of their cardioprotective benefit may derive from dampening chronic immune activation and limiting accumulation of CD151⁺ cCD4⁺ T cells. Whether statins directly modulate CD151 expression or cytotoxic CD4⁺ differentiation remains to be determined, but this represents an intriguing avenue for future investigation.

Several limitations warrant consideration. The interval between sample collection and subsequent cancer diagnosis varied among participants, reflecting the retrospective design and limiting causal inference. In addition, the relatively small cohort size limits statistical power, and these findings should therefore be considered exploratory and hypothesis generating. Our stratification of PWH by subsequent cancer diagnosis was not intended to predict malignancy, but rather was intended to explore whether distinct immune phenotypes could be detected in individuals who would later develop a comorbidity. Although CD4 nadir contributed modestly to CD151 expansion, the persistence of group differences suggests broader immune remodeling rather than solely historical immune depletion. Similarly, ART class and plasma viral load did not significantly explain variation in CD151 frequencies within this virally suppressed cohort, supporting the interpretation that CD151 expansion reflects chronic immune activation patterns, rather than ART class alone. The single-participant scRNA-seq analysis limits statistical generalization; however, the concordance between transcriptional profiling and protein-level validation strengthens confidence in the biological relevance of the CD151⁺ cCD4⁺ phenotype.

In summary, our data support a model in which chronic immune hyperactivation in virally suppressed PWH is associated with expansion of a CD151⁺ cCD4⁺ T cell population exhibiting features of terminal differentiation, senescence, and sustained effector potential. While CD151 accumulation occurs during physiologic aging, its age-independent expansion in PWH, particularly PWHc, highlight its value as an indicator of dysregulated immune remodeling rather than chronological age alone. These findings provide new insight into the architecture of immune dysfunction in virally suppressed PWH and underscore the importance of examining lineage-specific T cell differentiation states, rather than global activation markers alone, in understanding long-term comorbidity risk.

Author contributions

M.P., A.S., T.F., C.T., D.M., and O.K. performed and analyzed the flow cytometric analysis experiments. M.P., H.H., S.S., N.E., and O.K. designed the experiments. E.C., B.D.G., K.E.L., F.Z., and O.K. performed and analyzed the single cell analysis experiments. S.O., L.P., J.K., S.H., N.E., and P.A.G. obtained clinical samples and data. O.K. and N.E. oversaw project administration. O.K., N.E., and M.P. wrote the manuscript with input from all authors.

M.P. (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Investigation [Lead], Methodology [Lead], Validation [Lead], Visualization [Lead], Writing—original draft [Lead], Writing—review & editing [Lead]), F.Z. (Data curation [Supporting], Formal analysis [Supporting], Methodology [Supporting], Validation [Supporting], Writing—review & editing [Supporting]), E.C. (Data curation [Supporting], Writing—review & editing [Supporting]), B.D.G. (Data curation [Supporting], Writing—review & editing [Supporting]), A.S. (Data curation [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), K.E.L. (Data curation [Supporting], Formal analysis [Supporting], Writing—review & editing [Supporting]), S.O. (Data curation [Supporting], Formal analysis [Supporting], Resources [Supporting], Writing—review & editing [Supporting]), T.F. (Data curation [Supporting], Writing—review & editing [Supporting]), C.T. (Data curation [Supporting], Writing—review & editing [Supporting]), A.D. (Validation [Supporting], Writing—review & editing [Supporting]), F.W. (Validation [Supporting], Writing—review & editing [Supporting]), D.M. (Data curation [Supporting], Writing—review & editing [Supporting]), L.P. (Resources [Supporting], Writing—review & editing [Supporting]), J.K. (Resources [Supporting], Writing—review & editing [Supporting]), S.H. (Resources [Supporting], Writing—review & editing [Supporting]), P.A.G. (Resources [Supporting], Writing—review & editing [Supporting]), H.H. (Conceptualization [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), S.S. (Conceptualization [Supporting], Investigation [Supporting], Methodology [Supporting], Validation [Supporting], Writing—review & editing [Supporting]), O.K. (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Funding acquisition [Lead], Investigation [Lead], Methodology [Lead], Project administration [Lead], Resources [Lead], Supervision [Lead], Validation [Lead], Visualization [Lead], Writing—original draft [Lead], Writing—review & editing [Lead]), and N.B.E. (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Investigation [Lead], Methodology [Lead], Project administration [Lead], Resources [Lead], Supervision [Lead], Validation [Lead], Visualization [Lead], Writing—original draft [Lead], Writing—review & editing [Lead])

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

This work was funded in part by National Institutes of Health grants R21-DA055490, R01-AI122842, R33-AI116188, and R33-AI133679 to O.K. Parts of the work were performed in the University of Alabama at Birmingham CFAR BSL2+ facilities and by the University of Alabama at Birmingham CFAR Advanced Technology Core, which are funded by National Institutes of Health/National Institute of Allergy and Infectious Diseases P30-AI027767.

Conflicts of interest

The authors declare no competing interests.

Data availability

Data available on request. RNA-seq datasets generated for this study will be shared on reasonable request to the corresponding author, O.K. (okutsch@uab.edu).