The human immunodeficiency virus-1 (HIV) is a lymphotropic virus that mainly infects and depletes CD4+ T cells, leading to chronic immune activation, immune dysfunction, and, ultimately, immunodeficiency (1–4). Although highly active antiretroviral therapy (HAART) potently suppresses viral replication, no cure is available to date (5, 6).

Lately, the role of unconventional T cells for HIV pathogenesis and HIV cure approaches has come more into focus (7–11). Among these T-cell populations, γδ T cells seem to have important immunomodulatory properties relevant for the disease (4). γδ T cells express a T-cell receptor with gamma and a delta chain (12–17). They are “innate-like” T cells that make up 1-15% of circulating leukocytes and exert a direct cytotoxic activity independently of MHC presentation (12–22). About 30% of γδ T-cells express a CD8+ T cell receptor, less than 1% a CD4+ T-cell receptor and 70% none of the conventional T-cell receptors (23).

γδ T cells recognize stress-induced molecules, non-peptide, and phosphoantigens, self- or MHC-related molecules, and lipids associated with different kinds of pathogens (20, 21, 24–29). γδ T cells are generally seen to be pro-inflammatory and involved in the initiation and propagation of immune responses, but lately, it has been shown that they also act as immunomodulators and can inhibit T- and B-cell responses (20, 21, 30–32). They produce the pro-inflammatory cytokines TNF-α, IL-17, and IFN-γ as well as the anti-inflammatory cytokines IL-10 and TGF-β and IL-2, IL-22, IL-21, IL-4, IL-5, IL-13 (31, 33–38). γδ T cells can alleviate or maximize inflammation in the blood and different tissues and might be used to directly target HIV-infected cells but have also been described as potential targets of HIV (18, 22, 39, 40).

The two main subsets of γδ T cells, Vδ1 and Vδ2 are present in different anatomic compartments (23, 41–44). While Vδ1 cells can be found in the intraepithelial layer of mucosal surfaces, the Vδ2 population is mostly present in the blood and secondary lymphoid tissues of healthy adults (ratio in the peripheral blood Vδ1:Vδ2 3:10) (23, 41–49).

Early during primary HIV infection, an inversion of the Vδ1:Vδ2 ratio can be observed in the blood whereas the frequency of total γδ T cells remains relatively stable (18, 22). Vδ2 cells are depleted while Vδ1 cells expand (4, 18, 50–52). It has been shown that Vδ2 cells express high levels of the HIV co-receptors CCR5 and α4β7, which possibly contributes to their preferential depletion in HIV infection (53–56). Their number, but also functionality remains below that of healthy controls even after successful implementation of ART and restoration of the CD4+ T-cell compartment (51, 52). It has been implied that Vδ1 cells are involved in antiviral immunity and their expansion may be an indirect consequence of viral infection and reflects an increased translocation of stimulatory bacterial products across the gut epithelium in non-human primate studies (57–59). In contrast, direct cytotoxicity towards HIV-infected cells has been demonstrated to be largely restricted to Vδ2 cell clones, and the frequency of cervical Vδ2 cells correlates with SIV viral load (60–62). Interestingly, elite controllers exhibit higher frequencies of Vδ2 cells than untreated or antiretroviral treated HIV progressors (63, 64).

The ectonucleotidases CD39 and CD73, members of the adenosine pathway that are expressed on several lymphocyte populations, convert extracellular pro-inflammatory ATP and ADP to anti-inflammatory Adenosine (ADO) (65–67). In healthy individuals, the level of extracellular ADO is low but can increase 100-1000-fold in situations of strong inflammation and tissue injury (56). ADO strengthens epithelial barrier functions and inhibits leukocyte extravasation by binding to four different receptors: A1R, A2AR, A2BR, and A3R (68, 69). ADO binding to A2AR increases intracellular cAMP and inhibits the production of cytokines and T-cell proliferation (70–74). Importantly, both CD39 and CD73 can work in cis (interaction on the same cell), trans (interaction with enzymes expressed on different cells), and as soluble forms, and detailed knowledge about the respective microenvironment is essential (67, 75).

Over the last decade, multiple roles of CD39 and CD73 in the regulation of inflammation and immune responses have been revealed (75–82). It has also been demonstrated that mutations in the purine system can cause severe primary immunodeficiency diseases (78, 83–85). In HIV-infected untreated individuals, an over-expression of CD39 on and an increased hydrolysis of ATP by lymphocytes has been observed. Also, a variant of the CD39 gene associated with low CD39 expression and a slower progression to AIDS has been described in lymphocytes (86–88). Furthermore, the CD39/CD73/adenosine axis has been linked to inhibition of HIV-1 replication as well as immune suppression by CD39+ regulatory T cells (Tregs) (66, 88, 89).

In viremic HIV patients, the frequency of CD73+ cells in different T-cell subsets, especially Tregs and CD8+ T cells, is markedly reduced, and the function of CD73+ CD8+ T cells is impaired (82, 90). On B cells, low CD73 and CD39 expression are associated with low CD4+ T-cell counts (77). While both CD39 and CD73 are co-expressed on murine Tregs, only a small fraction of human peripheral Tregs expresses CD73 (65, 66, 82, 91).

Higher CD73 levels have been associated with immunosuppression and poor prognosis in e.g. breast or ovarian cancer (76, 92–95). In mice, suppressive activities of CD73+ γδ T cells via adenosine were shown (96).

Only recently it has been shown that γδ T cells can also act in an immunosuppressive manner and that they can infiltrate tumors and suppress dendritic cells and T cells (30, 97, 98). Liang et al. demonstrated that the regulation of γδ T cells in autoimmunity is associated with ADO (96). Hu et al. described CD39⁺ γδ T cells as capable of suppressing T cells via the adenosine-mediated pathway but independent of IL-10 and TGF-beta expression (98). In contrast, Otsuka et al. have reported a potential role of CD39+ γδ T cells with a regulatory phenotype mediated by IL-10 secretion in mice (98, 99).

In HIV infection, the plasma concentration of IL-10 increases over time and limits specific T-cell responses (100). CD39+ NK cells secreting IL-10 also contribute to this increase: Dierks et al. demonstrated that elevated levels of CD39+ NK cells in viremic patients correlated directly with viral load and activation, and negatively correlated with CD4+ T-cell count (101). IL-10 secretion was associated with the expression of CD39 (99, 101–103).

We and others have previously shown that CD39 expression of Tregs correlates with the progression of HIV infection and that Tregs of HIV elite controllers show the lowest levels of CD39 (79, 104). We have also recently identified CD39+ γδ T cells with an immunosuppressive phenotype in the gut (81). Bhatnagar et al. suggested a suppressive activity especially of Vδ2 via TGF-β, which is dysregulated in progressed HIV infection (35).

In HIV infection, a comprehensive assessment of the expression of CD39 and CD73 on different γδ subsets including Vδ1 and Vδ2 γδ T cells has never been performed. Therefore, we sought to characterize CD39+ and CD73+ expression on γδ T cells in relation to phenotype and function in a large cohort of healthy individuals and people living with HIV with different disease statuses including HIV elite controllers and long-term non-progressors.

The ectonucleotidases CD39 and CD73 have been described as important immunoregulatory molecules on Tregs and T effector cells (67, 75–79, 81, 82, 92, 93, 98, 99, 101, 104, 111). In mice, CD39+ γδ T cells with a regulatory phenotype have already been described (99, 112). In humans, CD39+ immunosuppressive γδ T cells have been described in the context of colon cancer (98). Little is known about the expression of these two molecules on γδ T cells in healthy humans and the context of viral infections. In this study, we aimed at the detailed assessment of the CD39 and CD73 expression pattern on peripheral γδ T cells in healthy and HIV-infected individuals with respect to their differentiation, activation, and exhaustion status and their immunomodulatory properties in terms of their cytokine profiles.

In line with previously published data, we found that the percentage of total γδ T cells was stable during HIV infection regardless of the stage of HIV infection while the ratio between the subsets Vδ1 and Vδ2 was inversed ( Supplementary Figure 1 ; see Supplementary Figures 2 and 3 for the gating strategy) (4, 18, 22, 40). In healthy individuals, the percentage of Vδ2 (68,5%) was significantly higher than that of Vδ1 γδ T cells (31,4%, p=0,0027), whilst it was significantly lower in viremic (Vδ1: 81,6%, Vδ2: 18,4%, p<0,0001) and patients on ART (Vδ1: 73,3%, Vδ2: 26,8%, p<0,0001). Interestingly, in long-term non-progressors, the proportions of Vδ1 (50,34%) and Vδ2 γδ T cells (49,7%) were similar. In elite controllers, we observed a difference between Vδ1 (69,7%) and Vδ2 (30,3%) γδ T-cell proportion that did not reach statistical significance, most likely due to the small patient number.

γδ T cells are part of the first-line defense against pathogens since they exert direct cytotoxic functions independent of MHC proteins (15, 17), but also immunomodulatory γδ T cells have been described in different immunological settings such as cancer and inflammatory bowel disease (121, 122). Alterations of CD39 and CD73 expression on several lymphocyte populations have been described and are important for HIV pathogenesis (82, 88, 90, 101, 123). In the current study, a detailed phenotypical and functional characterization of CD39 and CD73 expression on different γδ T-cell populations from healthy individuals and HIV patients with different disease courses was carried out.

Our results show that the expression pattern of these ectoenzymes is associated with distinct functional states and can be used as a marker to identify activated cells. We find significant differences in CD39 and CD73 expression on total γδ T cells, as well as on Vδ1 and Vδ2 cells between healthy and HIV-infected individuals depending on the clinical status. Importantly, we define a small population of γδ T cells co-expressing CD39 and CD73 that produce IL-10 after in vitro stimulation in healthy individuals and HIV patients.

In chronic HIV infection, IL-10 concentrations in the blood plasma were reported to increase over time, mediated by different lymphocyte populations (100, 124). The level of IL-10 production correlates with disease progression and causes reversible T-cell dysfunction to enable a balance between protective responses and immunopathology. IL-10 expression is associated with the expression of CD39 and the frequency of CD39+ cells secreting IL-10 has been correlated with viral load and immune activation (99, 101–103). Furthermore, IL-2 production is inhibited via the CD39/ADO pathway in CD39+ Tregs (125, 126).

We and others demonstrate that CD39, PD-1, and IL-10 were increased on γδ T cells in viremic HIV infection and provide an immunosuppressive environment in which the immune system is unable to clear the HI virus. Interestingly, our data show that neither IL-10 nor PD-1 nor CD39 increase strongly in EC, who can control the infection spontaneously.

One way to (partially) revert the function of virus-specific effector T cells (e.g. Vδ1 cells) is by combined checkpoint inhibitor blockade and blockade of adenosine signaling. Interestingly, CD39+ T cells CD8+ often also express PD-1 and other markers of cellular exhaustion. Li et al. demonstrated a reversion of CD8 exhaustion by concomitant blockade of PD-1 and adenosine pathways (127).

(IL-10+CD39+) γδ T cells could be reactivated by blockade of IL-10 or PD-1, CD39, or combinations thereof (80). Restored CD4 T cell function was previously achieved by immune checkpoint blockade of PD-1 and IL-10 in HIV-infected patients and Tang et al. demonstrated an improved function of MAIT cells during HIV/tuberculosis infection (128,129).

Overall, a strong correlation between the frequency of CD39+ and CD73+ γδ T cells and immune activation as well as disease progression (viral load and CD4+ T-cell count) could be determined. Thus, we propose that CD39 expression as well as the down-regulation of CD73 on γδ T cells can be seen as markers of activation, which has also been proposed in other immunological and disease contexts (99, 112, 130). The shifts of CD39 and CD73 have important implications regarding homing, functionality, nucleotide metabolism (that can occur in cis and/or trans), as well as interaction with other lymphocyte populations (131).

In this study, only HIV elite controllers showed an expression of both enzymes on γδ T cells comparable to healthy controls, while the expression pattern of CD39 and CD73 was altered in viremic and not fully normalized in individuals on ART. Viremic HIV patients had the highest CD39 expression on their γδ T cells – and these CD39+ γδ T cells produced the most IL-10 after in vitro stimulation. CD39 has lately been defined as a potential marker of immunomodulatory cells like Treg and NK cells, and CD39+ Vδ2 T cells might have a peculiar immunomodulatory role in HIV infection (64, 101, 104). One might interpret the CD39+CD73+IL-10+ γδ T cells as a counter-reaction to viremia to abrogate excessive inflammation, and speculate how this IL-10 production can inhibit HIV-specific immune responses and therefore act detrimentally in HIV pathogenesis (100, 124). Our data from elite controllers, who maintain low levels of CD39 and produce considerably less IL-10 than viremic HIV-infected patients, fit this hypothesis.

To see whether the observed alterations were specific for HIV infection, we also examined samples from patients with other viral infections, i.e. HCV or HBV. There, no significant increase of CD39+ γδ T cells compared to healthy controls could be measured. Similar to HIV infection, there is a loss of peripheral Vδ2 cells, an expansion of peripheral Vδ1 cells, and strong immune activation in (chronically) HBV‐infected subjects (132, 133). The transcriptional pathways and factors (e.g. cytokines) that regulate CD39 and CD73 expression need to be better defined for γδ T cells (see also below), but also other lymphocyte populations. Several cytokines that have also been shown to be altered in HIV infection regulate CD39 on lymphocytes (134). For example, IL-6 and TGF-ß are likely to lead to an up-regulation of CD39 on lymphocytes (134). In vitro, TCR engagement and IL-2 increased CD39 expression. In mice, IL-27 signaling triggers CD39 expression in Tregs by a STAT-1-dependent mechanism (134). Another factor that influences CD39 expression on human T cells includes genetic variations (single nucleotide polymorphisms) (134).

Overall, the role of γδ T cells in HIV remains elusive and seems double-edged (41). Pan Vδ2 γδ T cells have been associated with a protective role in HIV: a study in non-human primates identified a relationship between cervical Vδ2 frequency and SIV viral load, and Vδ2 γδ T cells expressing CD16 are capable of mediating potent ADCC (62,135). Another study reported that elite controllers maintain significantly higher frequencies of Vδ2 T cells than untreated patients or those on ART (63, 64). We also find a higher frequency of Vδ2 γδ T cells in samples from long-term non-progressors compared to HIV progressors.

On the other hand, Soriano-Sarabia et al. report that replication-competent HIV could be recovered from purified Vδ2 γδ T cells in 14 of 18 long-term ART recipients and thus concluded that these cells form part of the viral reservoir (55). It has been demonstrated that Vδ2 cells express high levels of the HIV co-receptors CCR5 and α4β7 which contribute to their infectibility (53–56). CD39+ cells are more activated than CD39- cells, thereby inducing transient expression of CD4 on Vδ2 γδ T cells in vivo and promoting infection. Follow-up studies should test the hypothesis that CD39+ Vδ2 γδ T cells form part of the viral reservoir, as has been shown for CD39+ Tregs (55, 79). In naïve Tregs, a correlation between HIV DNA and frequency of CD39+ naïve Tregs was demonstrated by Song et al. (79)

We observed a loss of polyfunctionality, defined as cells capable of producing three or more cytokines after in vitro stimulation, within the CD39+ Vδ2 γδ T-cell population of viremic HIV patients that was not fully regained in γδ T cells of HIV-infected individuals under ART. These results are in line with results from Casetti et al., who also measured a reduction of polyfunctionality (cytokine/chemokine production and cytotoxicity) in Vδ2 γδ T cells from ART-treated patients (136, 137).

This first study on the CD39 and CD73 expression pattern and functionality of γδ cells in HIV patients has some limitations. A first one is given by the limited number of parameters that could be measured in a respective panel by flow cytometry analysis. In future studies, the expression of CD16, CD56, and NKG2D, as well as the transcriptional profile [FOXP3, HIF-1, and AhR (138–141)], should be included in the flow cytometry experiments or assessed, e.g. by use of single-cell transcriptional RNA expression analysis (142). NKG2D can activate γδ T cells in an innate TCR-independent manner and is expressed by the vast majority of Vδ2 T cells (29, 43, 143, 144).

We found that the frequency of circulating γδ T cells expressing both CD39 and CD73 is particularly low. However, CD39+CD73+ γδ T cells had the highest frequencies of IL-10-producing cells after in vitro stimulation. One could hypothesize that this small sub-population of CD39+CD73+ γδ T cells that secretes an anti-inflammatory cytokine stays relatively unaffected from HIV infection. By contrast, the majority of γδ T cells show shifts of the CD39/CD73 expression ratio comparable to the changes observed in the effector cell compartment, most likely due to generalized immune activation in HIV (86, 145). It will be interesting to further investigate this scarce population of cells alongside the other γδ T-cell populations, especially regarding their suppressive capacities. Since the frequency of peripheral CD39+, CD73+ γδ T cells is too low for live cell sorting and subsequent co-culture with activated T cells, transcriptional analyses such as single-cell sequencing with regard to the transcriptome will have to be used to understand the capabilities of this and other γδ T-cell populations. Alternatively, γδ T-cell subpopulations could be expanded in vitro before life-sorting, co-culture, and flow-based read-out, with the disadvantage that this expansion may alter the phenotype and the function of the γδ T cells.

Previous studies point towards a regulatory role of CD39 in the inflammatory microenvironment of the gut which is caused i.a. by microbial ATP (109). Our group has previously demonstrated that the frequency of mucosa-derived CD39+ γδ T cells is decreased in patients with inflammatory bowel diseases compared to healthy controls (81). Upon stimulation, these cells produced less IL-17 and more IL-10 than CD39- γδ T cells, also pointing towards a regulatory phenotype of these cells. At the same time, the number of Tregs in the mucosal compartment was increased, which could serve as a compensatory mechanism for the loss of the CD39+ γδ T cells (81). The gut is one of the major sites for virus dissemination and formation of the viral reservoir (146–148). It will therefore be worthwhile to examine alterations in the number and function of CD39+ γδ T cells in the gut-associated lymphoid tissue (GALT) from HIV-infected individuals compared to healthy controls.

It will also be interesting to study patients with primary HIV infection. Bhatnagar et al. have demonstrated that Vδ2 γδ T cells transform from an anti-inflammatory phenotype in primary infection into a pro-inflammatory cytokine profile in chronic infection (35).

Taken together, the CD39/CD73 expression ratio on γδ T cells in untreated HIV is inversed and is associated with immune activation and disease progression. We find altered functionality and higher levels of IL-10 production in viremic HIV patients. Also, we defined a small population of CD39+CD73+ γδ T cells producing IL-10 at high frequencies after in vitro stimulation.

We hypothesize an immunomodulatory role of CD39+ and CD73+ γδ T cells in the pathogenesis of chronic HIV infection potentially mediated by IL-10 secretion. Similar to the deleterious role of suppressive cells in the microenvironment of tumors, the frequency of CD39+ γδ T cells was correlated with HIV disease progression in this study. This is further supported by our findings in elite controllers, who maintain stable frequencies of (IL-10-producing) CD39+ and CD73+ γδ T cells compared to healthy controls. Also, double-positive CD39+CD73+ produced significantly more IL-10 than γδ T cells expressing only one ectonucleotidase. Finally, a link between CD39 and IL-10 expression and disease progression has already been established in NK cells (90).

Future studies have to understand the role of adenine metabolism for γδ T cell function and elucidate the effects of the alterations of CD39 and CD73 expression on γδ T cells in HIV in more detail.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Review Board, Ärztekammer Hamburg, Hamburg, Germany. The patients/participants provided their written informed consent to participate in this study.

KK and MW have contributed equally to this work and share the first authorship. KK and JS designed the initial study design. MW and JS wrote the first draft of the manuscript together with KK. JS gave funding and was in charge of the overall research project. KK and MW conducted most of the experiments. A-DH, OD, H-JS, and JS recruited the patients and collected patient data. KK and MW analyzed the data under the supervision of JS, PH, and FH. MW and KK prepared the figures and got input from FH, JS and all other authors. PH and FH aided in interpreting the results. All authors discussed the results and critically reviewed the manuscript. All authors contributed to the article and approved the submitted version.

FH, and JS get funding from the DFG (SFB1328 A12), JS and MW get funding by the H.W. & J. Hector Stiftung (project M2101). KK, MW, and JS get funding from the DZIF (TTU 04.816), JS and CA are also funded by the European Union Horizon 2020 program (European HIV Vaccine Alliance 681032).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.