CD4+ T cells are key contributors in the induction of adaptive immune responses against pathogens. Even though CD4+ T cells are primarily classified as noncytotoxic helper T cells, it has become appreciated that a subset of CD4+ T cells is cytotoxic. However, tools to identify these cytotoxic CD4+ T cells are lacking. We recently showed that CD29 (integrin β1, ITGB1) expression on human CD8+ T cells enriches for the most potent cytotoxic T cells. In this study, we questioned whether CD29 expression also associates with cytotoxic CD4+ T cells. We show that human peripheral blood–derived CD29hiCD4+ T cells display a cytotoxic gene expression profile, which closely resembles that of CD29hi cytotoxic CD8+ T cells. This CD29hi cytotoxic phenotype was observed ex vivo and was maintained in in vitro cultures. CD29 expression enriched for CD4+ T cells, which effectively produced the proinflammatory cytokines IFN-γ, IL-2, and TNF-α, and cytotoxic molecules. Lastly, CD29-expressing CD4+ T cells transduced with a MART1-specific TCR showed target cell killing in vitro. In conclusion, we demonstrate in this study that CD29 can be employed to enrich for cytotoxic human CD4+ T cells.
Tcells are pivotal in mediating the clearance of pathogens, and in providing long-lasting immunity to recurring infections. Classically, T cells have been divided into cytotoxic CD8+ T cells and noncytotoxic helper CD4+ T cells. Several studies have, however, demonstrated that the labor distribution of CD4+ and CD8+ T cells may be more nuanced. For instance, we and others demonstrated that human CD8+ T cells contain noncytotoxic T cells (1–4). Furthermore, cytotoxicity has been readily demonstrated in the CD4+ T cell population (5–10).
Cytotoxic CD4+ T cells can contribute to immune responses against viruses, such as CMV, Swine flu, EBV, and HIV infections (11–15). They were also suggested to contribute to diseases such as IgG4-related disease (16) and have been associated with ulcerative colitis (17) and Graves’ disease (18). In addition, not only CD8+ T cells have been implicated in antitumor responses. Also cytotoxic CD4+ T cell responses were shown to substantially contribute to cancer clearance in mouse models (19, 20) and in various human cancers (6, 10, 21). Therefore, cytotoxic CD4+ T cells could possibly be exploited for therapeutic purposes, such as cancer cellular immunotherapies (21, 22). However, CD4+ T cells come in many flavors, and cytotoxic CD4+ T cells constitute only a subset of the CD4+ T cell pool. Furthermore, other CD4+ lineages such as regulatory T cells can hamper the antitumor potential (23). Therefore, preselection may be key for such approaches. Yet, the tools to do so are to date lacking.
We have recently shown that the expression of CD29 (integrin β1; ITGB1) on human CD8+ T cells identifies cells that produce high amounts of cytotoxic molecules (i.e., granulysin, granzyme A, granzyme B, and perforin), proinflammatory cytokines (i.e., IFN-γ and TNF-α), and proinflammatory chemokines (i.e., CCL3, CCL4, and CCL5) (1). Notably, the presence of CD29hiCD8+ T cells within tumors associated with better overall survival of melanoma patients, a feature that was also displayed in an increased in vitro killing capacity of tumor cells by CD29hiCD8+ T cells when compared with CD29lo-expressing CD8+ T cells (1). Whether the expression of CD29 is also linked to the cytotoxic function in human CD4+ T cells is to date not known.
In this study, we show that human blood–derived CD29hiCD4+ T cells also display a cytotoxic gene expression profile, which closely resembles that of the CD29hiCD8+ T cell counterpart. CD29 expression enriched for CD4+ T cells that produced cytotoxic molecules such as granulysin and granzymes, and the proinflammatory cytokines IFN-γ, IL-2, and TNF-α. This CD29hi cytotoxic phenotype of CD4+ T cells was observed ex vivo in blood-derived CD4+ T cells of healthy and CMV-infected individuals, as well as in CD4+ T cells infiltrating lung and liver tumors. This phenotype was also maintained upon T cell expansion in vitro. Lastly, we found that CD29-expressing CD4+ T cells were most potent in killing target cells. Therefore, we conclude that CD29 enriches for cytotoxic CD4+ T cells.
Materials and Methods
Cell isolation and culture
PBMCs from de-identified, fully anonymized healthy volunteers of 18–70 y old, were isolated by Lymphoprep (density gradient separation; StemCell) and stored in liquid nitrogen until further use. The study was performed according to the Declaration of Helsinki (seventh revision, 2013). Written informed consent was obtained (Sanquin, Amsterdam, the Netherlands).
Cryo-preserved PBMCs were thawed and cultured in T cell medium (IMDM [Lonza] supplemented with 8% FCS, 100 U/ml penicillin, 100 µg/ml streptomycin, and 2 mM l-glutamine). Cells were either directly used for flow cytometry analysis or activated in T cell medium as previously described (24). Briefly, nontissue culture–treated 24-well plates (Corning) were precoated overnight at 4°C with 4 μg/ml rat α-mouse IgG2a (MW1483, Sanquin) in PBS. Plates were washed with PBS and coated for >2 h with 1 μg/ml αCD3 (clone Hit3a, eBioscience) at 37°C. Then 0.8 × 106 PBMCs/well were seeded with 1 µg/ml soluble αCD28 (clone CD28.2, eBioscience) in 1 mL T cell medium. After 48 h of incubation at 37°C, 5% CO2, cells were harvested, washed, and further cultured in standing T25/75 tissue culture flasks (Thermo Scientific) at a density of 0.8 × 106/ml, supplemented with 10 ng/ml human recombinant IL-15 (Peprotech). Medium was refreshed every 5–7 d.
Flow cytometry analysis
The cytokine production profile was determined by intracellular cytokine staining (ICCS) after T cell activation with 10 ng/ml PMA and 1 µM ionomycin (Sigma-Aldrich) for 4 h in the presence of 2 µM monensin (eBioscience). Cells were stained in PBS + 1% FCS for live-dead marker (Invitrogen) and Abs against CD4 (RPA-T4, OKT4), CD8 (RPA-T8, SK1), CD29 (Mar4), CD38 (HIT2), CD27 (CLB-27), and CD45RA (HI100) for 30 min at 4°C in the dark. Cells were prepared with CytoFIX-CytoPerm kit following the manufacturer’s protocol, and stained with Abs against IFN-γ (4S.B3), IL-2 (MQ1-17H12), TNF-α (MAb11), granulysin (DH2), granzyme A (CB9), granzyme B (GB11), and perforin (dG49). Cells were acquired on LSR II, Fortessa, or Symphony (all BD) using FACS Diva v8.0.1 (BD Biosciences). Data were analyzed with FlowJo VX (Tree Star).
Single-cell RNA-sequencing analysis
Single-cell RNA-sequencing (scRNA-seq) datasets were collected and reanalyzed from Guo et al. (25), Zheng et al. (26), and Zhang et al. (27). Count matrix was filtered for “PTH” and “PTY” to select peripheral blood CD4+ T cells expressing low and intermediate CD25, respectively, or “TTH,” “TTY,” and “TTR” for CD4+ T cells found in the tumor of references (25, 26). In addition, we analyzed previously published datasets of CMV-infected individuals by Zhang et al. (28). The Seurat R package (version 3.1.5) (29) was used for scRNA-seq data analysis and batch-effect correction. Raw counts were used for further analysis. Differential expression analysis was performed, and significant genes were filtered for log2 fold change > 0.25 and adjusted p value < 0.05. ITGB1 (CD29) grouping was determined based on the “double peak” expression distribution of ITGB1 in total CD4+ T cells (see Supplemental Fig. 1A, 1D). To determine T cell differentiation subsets, unbiased clustering was used. To attribute cell identities to clusters, differential gene expression analysis was performed on each cluster and compared with the rest of the CD4+ T cells (log2 fold change > 1 and adjusted p value < 0.05). Naive-like clusters were characterized by high CD27, CCR7, SELL, LEF1, and TCF7 expression. Central memory T cell (Tcm)-like cells were characterized by low amounts of cytotoxic molecules, and high expression of CD27, CCR7, and SELL. Effector memory T cells (Tem) and effector T cells (Teff) were characterized by increasing expression level of cytotoxic molecules GNLY, GZMB, GZMA, PRF1, SLAMF7, and CX3CR1. T cell clusters were then used to assess ITGB1, SELL, CD27, and CX3CR1 expression. For the CMV-specific analysis, we used the cluster identity as published in the original paper (28).
Generation of MART1 TCR-expressing T cells
PBMCs from individual donors were activated for 48 h with αCD3/αCD28 as described above, harvested, and retrovirally transduced with the MART1 TCR, as previously described (1, 30). Briefly, nontissue culture–treated 24-well plates were coated with 50 μg/ml RetroNectin (Takara Bio) overnight and washed once with 1 mL/well PBS. Then 300 µL/well viral supernatant was added to the plate, and plates were centrifuged for 1 h at 4°C at 2820 × g. Then 0.5 × 106 T cells were added per well, centrifuged for 10 min at 180 × g, and incubated overnight at 37°C. The following day, cells were harvested and cultured in T25 flasks at a concentration of 0.8 × 106 cells/ml for 6–8 d in the presence of 10 ng/ml rhIL-15.
Functional assays with MART1 TCR-expressing T cells
The cytokine production of MART1 TCR-transduced CD4+ T cells was determined by ICCS as described above. A total of 100,000 TCR-transduced T cells (∼80% TCR+) were cocultured for 6 h with 100,000 HLA-A2+ MART1hi Mel 526 (MART1+) or HLA-A2− MART1lo Mel 888 (MART1−) tumor cell lines. Cells were subsequently measured by flow cytometry, and populations of interest were gated and analyzed.
For killing assays, CD4+ MART1 TCR+ T cells were FACS-sorted for total CD4+, CD29+ (CD29hi) or CD29− CD38+ (CD29lo) expression, as previously described (1). Briefly, T cells were stained with Abs against CD4, CD8, CD29, CD38, and murine TCRβ (H57-597) and for live-dead marker in PBS + 1% FCS for 30 min at 4°C. Cells were washed once with culture medium and sorted on a precooled FACS Aria III (BD Biosciences) sorter washed with 70% ethanol. Sorted T cells were collected in culture medium, washed, and rested overnight in the incubator in medium at 37°C. Tumor cells were labeled with 1.5 μM CFSE for 10 min at 37°C in FCS-free medium and washed three times with warm culture medium. Then 15 × 103 tumor cells were cocultured with MART1 TCR+ FACS-sorted T cells for 20 h, in a 3:1 or in a 30:1 ratio. Adherent and nonadherent cells were harvested from the wells to collect all remaining tumor cells. Dead tumor cells were quantified by flow cytometry with Near-IR live-dead marker on gated CFSE+ tumor cells.
Statistical analysis and data visualization
Data generated with flow cytometry were analyzed with paired and ratio paired t test as indicated using GraphPad PRISM version 8. Differences were considered significant with a p value < 0.05. Plots were generated with Seurat (ggplot2 version 3.0) and with GraphPad.
CD29 identifies human CD4+ T cells with cytotoxic gene expression
We first determined whether CD29 expression can identify cytotoxic CD4+ T cells, as we recently described for CD8+ T cells (1). To define the gene expression profile of CD29hiCD4+ T cells, we reanalyzed previously published scRNA-seq datasets of ex vivo blood-derived human CD4+ T cells (25–27). To exclude CD25hi-expressing regulatory T cells from the analysis, we selected CD4+ T cells index-sorted for low and intermediate expression of CD25 (see Material and Methods), which yielded 3243 CD4+ T cells. When we analyzed the ITGB1 (CD29) gene expression profile of these CD4+ T cells, we observed a heterogenous distribution of ITGB1 gene expression (Fig. 1A). Nevertheless, we could define a clear cut-off of T cells with high (ITGB1hi) and low (ITGB1lo) ITGB1 gene expression (Supplemental Fig. 1A). We confirmed the dichotomous expression of CD29 by flow cytometry on peripheral blood–derived CD4+ T cells (Supplemental Fig. 1B). Differential gene expression analysis revealed that ITGB1hiCD4+ T cells preferentially expressed cytotoxic genes, including GNLY (granulysin), GZMA, GZMB, GZMH (granzymes A, B, and H), PRF1 (perforin), FGFBP2, CCL5 (RANTES), and the cell surface markers HLA-DR, NKG7, and CX3CR1 (Fig. 1B). This gene expression signature closely resembled that of previous studies on cytotoxic CD4+ T cells (7, 31, 32). In contrast, ITGB1loCD4+ T cells showed higher gene expression of naive and central memory–like markers such as CCR7, CD27, and SELL (CD62L) and transcription factors LEF1 and SATB1.
Peripheral blood–derived CD4+ T cells are composed of T cells from distinct differentiation statuses: that is, naive T cells (Tn), Tcm, Tem, and Teff. We therefore asked how ITGB1 gene expression was distributed over the different CD4+ T cell subsets. To this end, we used unbiased clustering and identified four differentiation clusters (Fig. 1C, Supplemental Fig. 1C; see Materials and Methods). ITGB1 gene expression was very limited on Tn (Fig. 1D). Conversely, all nonnaive CD4+ T cell subsets contain T cells with high ITGB1 gene expression, albeit to a different extent (Fig. 1D). Measurements of CD29 protein expression on the different T cell subsets from blood-derived CD4+ T cells by flow cytometry corroborated these findings (Supplemental Fig. 1D). To exclude that the cytotoxic gene expression profile of ITGB1hiCD4+ T cells was influenced by an overrepresentation of naive (primarily ITGB1lo) T cells, we repeated the differential gene expression analysis on ITGB1hi and ITGB1lo nonnaive CD4+ T cells (Supplemental Fig. 1D, 1F). This analysis revealed a very similar gene expression profile of ITGB1hi and ITGB1loCD4+ T cells, with a high gene expression of GZMA, GZMB, GZMH, GNLY, FGFBP2, and PRF1 on ITGB1hiCD4+ T cells, and on ITGB1loCD4+ T cells a relatively high gene expression of SATB1, and of CCR7 and CD27 (Fig. 1E, Supplemental Fig. 1F, Supplemental Table I).
We next questioned how the cytotoxic gene expression profile of ITGB1hiCD4+ T cells related to that of ITGB1hiCD8+ T cells (1). We therefore compared the fold enrichment of differentially expressed genes of ITGB1hi and ITGB1loCD8+ T cells we previously described (1) with that of ITGB1hi and ITGB1loCD4+ T cells. Remarkably, the cytotoxic gene expression features of ITGB1hiCD8+ T cells was to a great extent shared with that of ITGB1hiCD4+ T cells (Fig. 1F). This included the transcription factors Hobit (ZFN683) (1, 31), HOPX (33), and ZEB2 (7, 34). We also found a set of genes that were commonly expressed by ITGB1lo T cells, such as CCR7, SATB1, TCF7, CD55, the TCR-signaling molecule TXK, and the antisense long noncoding RNA CD27-AS1. Combined, our findings reveal that ITGB1hiCD4+ T cells have a cytotoxic gene expression signature that is shared with that of ITGB1hiCD8+ T cells.
CD29hiCD4+ T cells comprise cells that produce cytotoxic molecules and proinflammatory cytokines
We next questioned how the differential gene expression profile of ITGB1hi and ITGB1loCD4+ T cells translated into their protein expression profile. To test this, we first measured the protein expression of the cytotoxic molecules granzyme A, granzyme B, granulysin, and perforin in nonactivated peripheral blood–derived CD4+ T cells by ICCS. In line with the gene expression profile, CD29hi (ITGB1hi) CD4+ T cells were the prime producers of all four cytotoxic molecules (Fig. 2A, 2B).
Not only the cytotoxic molecules but also the proinflammatory key cytokines IFN-γ, TNF-α, and IL-2 are critical to the CD8+ T cells’ immune response (35–37). In CD8+ T cells, IFN-γ, TNF-α, and their coexpression of IL-2 strongly correlated with CD29 protein expression (1). To test whether CD29hiCD4+ T cells also display the preferential production of these three key cytokines, we stimulated blood-derived CD4+ T cells with PMA-ionomycin for 4 h, and measured the production of IFN-γ, IL-2, and TNF-α protein by ICCS. As with CD29hi CD8+ T cells, the percentage of CD29hiCD4+ T cells producing these three cytokines was substantially higher than that of CD29loCD4+ T cells (Fig. 2C, 2D). Notably, a substantial proportion of CD29hiCD4+ T cells comprised double and triple cytokine producers, a feature that is associated with the most potent T cell responses (38, 39) (Fig. 2E). Thus, CD29hiCD4+ T cells include polyfunctional T cells that produce cytotoxic molecules and proinflammatory cytokines.
ITGB1hiCD4+ T cells increase in CMV-infected individuals
We next sought to determine the occurrence of CD29hiCD4+ T cells in the context of CMV infection. To this end, we reanalyzed recently published scRNA-seq dataset of CD4+ T cells from four healthy and four CMV-infected individuals (28). Unbiased clustering of the 18,815 cells revealed a distinct subset of cytotoxic CD4+ T cells found in CMV-infected individuals, termed “Tcmv” or “Tdiff” in the original publication (28) (Fig. 3A). Tcmv cells expressed high levels of ITGB1 (Fig. 3A, 3B). Tcmv displayed a cytotoxic gene expression profile, as characterized by high GNLY, FGFBP2, PRF1, GZMA, and GZMB expression (Fig. 3B). We next selected for CD4+ T cells with high (ITGB1hi) or with low (ITGB1lo) ITGB1 expression (Supplemental Fig. 2A). We noted a clear increase of ITGB1hiCD4+ T cells in CMV-infected individuals compared with that of healthy controls (Fig. 3C). This was in line with the natural frequency of CD29hi cells in peripheral blood–derived CD4+ T cells of healthy donors (Supplemental Fig. 1B). We next performed differential gene expression analysis between ITGB1lo and ITGB1hiCD4+ T cells (Fig. 3D, Supplemental Table I). We observed a strong cytotoxic gene signature that was marked by the expression of GNLY, FGFBP2, PRF1, GZMA, and GZMB expression in ITGB1hiCD4+ T cells (Fig. 3D). Of note, we also found that the transcription factors HOPX, ZNF683, and ZEB2 were increased in expression in ITGB1hiCD4+ T cells from CMV-infected donors (Fig. 3D), similar to what we found in blood-derived ITGB1hiCD4+ T cells (Fig. 1B, 1F). Thus, cytotoxic CD4+ T cells in CMV-infected patients expressed high ITGB1 levels and displayed a strong cytotoxic gene expression profile.
ITGB1hiCD4+ T cells are associated with cytotoxic gene signature in lung and liver tumors
Cytotoxic CD4+ T cells have recently been described to infiltrate several tumor types (10). This prompted us to investigate whether ITGB1 expression related to cytotoxic CD4+ T cells in tumors. We reanalyzed the scRNA-seq datasets of FACS-sorted CD4+ T cells from non–small cell lung tumors (25) and from liver tumors (26). Using unbiased clustering, we isolated “conventional” CD4+ T cells (lung = 2316 cells; liver = 754 cells) thereby excluding regulatory and follicular helper CD4+ T cells. We examined the expression of ITGB1 and cytotoxic markers in both tumor types (Fig. 3E, 3F). We observed in lung and in liver tumors a cluster of cells with high gene expression levels of GNLY, FGFBP2, PRF1, GZMA, and GZMB that also expressed high levels of ITGB1 (Fig. 3E, 3F). Notably, the expression of FGFBP2 and GNLY was restricted to the ITGB1hi cell cluster (Fig. 3E, 3F), similar to what we observed by flow cytometry in blood-derived CD4+ T cells (Fig. 2A, 2B). We next divided cells based on low (ITGB1lo) or high (ITGB1hi) ITGB1 gene expression (Supplemental Fig. 2B, 2C), and we performed differential gene expression analysis (Fig. 3G, 3H, Supplemental Table I). ITGB1hiCD4+ T cells from lung and liver tumors alike had a cytotoxic gene signature that was marked by the expression of GNLY and FGFBP2, as well as GZMB, and PRF1 in ITGB1hiCD4+ T cells of lung tumors. We therefore conclude that also in the context of lung and liver tumors, ITGB1 expression enriches for cytotoxic CD4+ T cells.
In vitro cultured CD29hiCD4+ T cells maintain their production profile
To determine whether the increased production of cytotoxic molecules and inflammatory cytokines by CD29hiCD4+ T cells was maintained upon T cell culture and expansion, we stimulated PBMCs with αCD3-αCD28 for 2 d, removed them from the stimulus, and expanded the cells for an additional 5 d in the presence of human recombinant IL-15. To better separate CD29hiCD4+ T cells from CD29loCD4+ T cells, we used the same gating strategy as previously employed for in vitro activated CD8+ T cells, and we included CD38 for separation. Specifically we selected for CD29hiCD38lo (CD29hi)-expressing T cells, and CD29loCD38hi (CD29lo)-expressing T cells for further analysis (Supplemental Fig. 3A, 3B) (1). This gating strategy improved the distinction of the two T cell populations (Fig. 4A, left panel).
We next measured the cytokine production of the cultured CD4+ T cells by ICCS after T cell activation with PMA-ionomycin for 4 h. Also upon T cell expansion in vitro, the percentage of IFN-γ producing T cells was significantly enriched in CD29hiCD4+ T cells compared with the total CD4+ T cell population, or with CD29loCD4+ T cells (Fig. 4A, 4B). Even though the percentage of IL-2– or TNF-α–producing cells was similar between CD29lo and CD29hiCD4+ T cells, double or triple cytokine producers were enriched in CD29hiCD4+ T cells (Fig. 4C), a feature that we also observed in CD29hiCD8+ T cells (1). In addition, although the percentage of TNF-α producing T cells was equally produced by CD29lo and CD29hiCD4+ T cells (Fig. 4B), the amount of cytokine produced per cell was higher in CD29hiCD4+ T cells as measured by the geometric mean fluorescence intensity (geoMFI; Supplemental Fig. 3C).
When we analyzed the cytotoxic protein expression profile of nonactivated in vitro cultured CD4+ T cells, we did not observe differences in the percentage of granzyme A–expressing CD29hiCD4+ T cells compared with CD29loCD4+ T cells (Fig. 4D, 4E). Yet, CD29hiCD4+ T cells showed higher granzyme A protein expression per cell, as determined by the geoMFI (Supplemental Fig. 3D). For granzyme B, perforin, and granulysin, however, we found substantial differences in nonactivated CD4+ T cells (Fig. 4D, 4E). In particular, perforin and granulysin were almost exclusively produced by nonactivated CD29hiCD4+ T cells (Fig. 4D, 4E). We thus conclude that CD29 expression enriches for CD4+ T cells with a cytotoxic phenotype also upon in vitro culture.
CD29hiCD4+ T cells are superior IFN-γ and TNF-α producers in response to TCR activation
We next investigated the cytokine production profile of CD29hiCD4+ T cells in response to cognate Ag exposure. To this end, we retrovirally transduced CD4+ T cells with the codon-optimized MART1 TCR that recognizes the HLA-A*0201 restricted 26-35 epitope of MART1 (30, 40). The MART1 TCR is MHC class I restricted. It can nonetheless elicit Ag-specific responses in MART1 TCR-expressing CD4+ T cells (30, 40, 41), even though the MHC class I–restricted TCR requires higher affinity for full activation (42) and may thus limit CD4+ T cell activation as these require higher affinity for full activation.
The MART1 TCR transduction efficiency with ∼80% was comparable between CD29hi and CD29loCD4+ T cells, and the TCR expression levels similar but slightly lower in CD29hiCD4+ T cells (Supplemental Fig. 3E). To determine the cytokine expression profile, MART1 TCR-engineered CD4+ T cells were cocultured in a 1:1 ratio for 6 h with HLA-A201+ MART1hi-expressing melanoma tumor cell line (MART1+), or with HLA-A201− MART1lo-expressing tumor cell line (MART1−) (43, 44). The cytokine production was measured by flow cytometry in CD29hi and CD29loCD4+ T cells (Fig. 5A). The percentage of MART1 TCR-engineered CD29hiCD4+ T cells contained substantially more IFN-γ– and TNF-α–producing T cells in response to MART1+ tumor cells than the CD29loCD4+ T cells (Fig. 5A, 5B). The IL-2 production of CD29hiCD4+ T cells was similar to that of CD29loCD4+ T cells. Notably, CD29loCD4+ T cells comprised mostly IL-2 and TNF-α single producers, whereas most CD29hiCD4+ T cells produced IFN-γ and TNF-α and were enriched in IFN-γ+ TNF-α+ IL-2+ polyfunctional cells (Fig. 5C). We detected granulysin expression almost exclusively in resting MART1 TCR-expressing CD29hiCD4+ T cells (Fig. 5D). Thus, MART1 TCR-expressing CD29hiCD4+ T cells maintain their cytokine production profile also when exposed to target cells.
CD29 enriches for cytotoxic CD4+ T cells
Lastly, we determined the cytotoxic capacity of MART1 TCR-engineered CD29hiCD4+ T cells in response to tumor cells. CFSE-labeled MART1+ and MART1− tumor cells were cocultured at an E:T ratio of 3:1 with FACS-sorted MART1 TCR-engineered total CD4+ T cells, or with CD29lo or CD29hi FACS-sorted TCR-engineered CD4+ T cells. After 20 h, tumor cell killing was determined by live-dead marker labeling of all (adherent and nonadherent) tumor cells present in the coculture.
At an E:T ratio of 3:1, the overall killing capacity of MART1 TCR-engineered total CD4+ T cells was low, but significant. MART1+ tumor cells, but not MART1− tumor cells, showed increased cell death in the presence of MART1 TCR-engineered T cells (Fig. 6A, left panel; (Fig. 6B). Notably, in line with their cytotoxic gene and protein expression profile, FACS-sorted CD29hiCD4+ T cells from the same donors showed superior killing capacity when compared with CD29loCD4+ T cells, or with total CD4+ T cells (Fig. 6A, 6B). The superior killing capacity of CD29hiCD4+ T cells was also apparent at an E:T ratio of 30:1 (Fig. 6C). At this higher E:T ratio, we observed that nonsorted CD4+ T cells (containing CD29hi cells) were superior in killing when compared with CD29lo FACS-sorted CD4+ T cells (Fig. 6C). We thus conclude that the preselection of CD29hiCD4+ T cells enriches for cytotoxic CD4+ T cells.
We also defined how the overall CD29 expression levels of MART1 TCR-engineered CD4+ T cells related to their killing capacity. We therefore tested the tumor cell killing of MART1 TCR-engineered, FACS-sorted total CD4+ T cells from 34 individual donors. Importantly, we observed that the percentage of CD29-expressing CD4+ T cells correlated with the killing capacity of the total CD4+ T cell population (p < 0.0001; (Fig. 6D). In conclusion, CD29-expressing CD4+ T cells comprise cytotoxic CD4+ T cells that display a polyfunctional cytokine production profile and that can effectively kill target cells.
In this study, we show that CD29-expressing CD4+ T cells contain polyfunctional cytotoxic cells, a feature that is shared between CD4+ and CD8+ T cells. Although CD29 expression does not specifically mark cytotoxic CD4+ T cells, it enriches for these. Of note, if and how CD29-expressing CD4+ T cells employ receptor-mediated killing, such as through Fas ligand, is not yet known. Our findings are concordant with another recent report on human cytotoxic T cells (32). Interestingly, the shared gene expression of the transcription factors ZNF683 (Hobit), HOPX, and ZEB2 in CD29hiCD4+ and CD29hiCD8+ T cells suggests that cytotoxic CD4+ and CD8+ T cells share a transcriptional regulation network. In humans, Hobit is primarily found in effector CD8+ T cells, cytotoxic CD4+ T cells, and in resident memory T cells (31, 45, 46). Hobit is also known to repress CCR7, a gene we found associated with ITGB1loCD4+ T cells (47). HOPX is involved in the regulation of granzyme B production in CD4+ T cells (33), and in the maintenance of effector memory within the type 1 helper T cell subset (48). Finally, the transcription factor ZEB2 works in coordination with T-bet to mediate the acquisition of cytotoxicity by CD8+ T cells (34). Differential gene expression of another transcription factor that was recently correlated with cytotoxic CD4+ T cells, ThPOK (33), was not found in ITGB1hiCD4+ T cells. Nevertheless, with our data combined, it is tempting to speculate that a transcription factor network involving ZNF683, ZEB2, HOPX, and possibly ThPOK are involved in the development or maintenance of CD29hiCD4+ and CD8+ T cells. Whether these transcription factors are involved in the expression of ITGB1 is to date unresolved.
Also how CD29hi cytotoxic CD4+ T cells are generated is to date not known. Cytotoxic CD4+ and their progenitor were recently described to be CD127lo and CD127hi, respectively (7). Even though our scRNA-seq analysis did not reveal the CD127 (IL7R) gene to be differentially expressed in ITGB1hi or ITGB1loCD4+ T cells, CD29hi T cells could very well arise from CD127hi progenitors, and CD127 gene expression may redistribute upon differentiation. In addition, the signals that are required to induce CD29 expression in T cells are not known. In CD8+ T cells, the cytokines IL-2, IL-6, IL-9, IL-11, IL-15, and IL-21 failed to do so in in vitro cultures (1). It is also possible that CD29 expression requires some signal or a combination thereof that cannot be mimicked in in vitro settings. Future research should address these questions.
Whether integrin β1 has a role in the cytotoxic features or is a “bystander” is yet to be determined. Integrin β1 can form heterodimers with 12 different α integrins (49). The compositions of the α-β integrin dimers confer ligand specificity to the integrin complex. In T cells, the integrin complexes are important for cell adhesion, cell migration, motility, and signaling (50). For example, α4-β1 can act as a costimulatory molecule (51) and as a receptor for CX3CL1 (fractalkine) (52) and can skew CD4+ T cells toward a type 1 helper phenotype (53). It is also possible that CD29 expression contributes to T cell signaling and helps trigger cytotoxic functions, as observed in murine CD8+ T cell clones (54). Altogether, it is tempting to speculate that CD29 expression on cytotoxic CD4+ T cells is not merely a marker, but may also exert a function, for instance by providing additional signaling or by stabilizing the cell–cell interaction.
Even though CD8+ T cells have been in the spotlight for therapies, recent findings suggest that cytotoxic CD4+ T cells are also involved in immune responses against tumors and viruses. One could speculate that a broad spectrum of cytotoxic cells is desirable, as they can attack infected cells and tumor cells from different angles. Indeed, the expression of MHC class I (required for CD8+ T cells) and MHC class II (required for CD4+ T cells) is subject to changes, and so is the loss of MHC molecules (required for NK cell killing). Having different flavors of cytotoxic cells should thus provide protection against a diversity of pathogens and tumors. In fact, the expansion of cytotoxic CD4+ T cells and their polyfunctionality is associated with healthy aging in supercentenarians (32). In addition, cytotoxic CD4+ T cells are shown to contribute to antiviral responses (11–14). In this study, we show that cytotoxic CD4+ T cells expanded in CMV-infected individuals (5, 28) express high levels of ITGB1. In addition, ITGB1hiCD4+ T cells include cytotoxic CD4+ T cells in lung and liver tumors. These finding are in line with a recent study that described the presence of cytotoxic CD4+ T cells in several cancers (10). Furthermore, cytotoxic CD4+ T cells may not be solely bystanders, but may substantially contribute to antitumor responses, as observed in bladder cancer patients (6). Whether the functional contribution of cytotoxic CD4+ T cells is also applicable to other cancer types is not yet known and should be addressed. In the context of cellular immunotherapies such as chimeric Ag receptor T cells and tumor infiltrating lymphocyte therapy, the selection of cytotoxic CD29-expressing CD4+ T cells could be of use to potentiate the efficacy of these therapies. Of note, the presence of cytotoxic CD4+ T cells may be beneficial in many but not all cases. In fact, in SARS-Cov-2 infected individuals, a higher number of cytotoxic CD4+ T cells was observed in severe compared with mild infections (55). This finding suggests that cytotoxic CD4+ T cells contribute to the pathology in SARS-Cov-2 infections, similar to what was previously observed in the IgG4-related disease (16).
In summary, we show in this study that cytotoxic CD4+ T cells can be enriched for by the surface marker CD29, a feature that is shared with cytotoxic CD8+ T cells. Thereby, CD29 provides a robust marker to study human cytotoxic T cell responses.
We thank the blood donors for donation; E. Mul, M. Hoogenboezem, and S. Tol for FACS sorting; and R. Gomez-Eerland and T.N.M. Schumacher for providing the MART1 TCR system. We are grateful to B. Popovic, A. Jurgens, N. Zandhuis, and S. Castenmiller for critical reading of this manuscript.
This work was supported by Dutch Cancer Society (KWF Kankerbestrijding) Grant 10132 (to M.C.W.) and by the Oncode Institute.
B.P.N. and M.C.W. designed experiments and wrote the manuscript; B.P.N. and A.G. performed experiments and analyzed data; M.C.W. directed the study.
The online version of this article contains supplemental material.
Abbreviations used in this article
intracellular cytokine staining
central memory T cell
effector T cell
effector memory T cell
naive T cell
The authors have no financial conflicts of interest.