HIV Infection Functionally Impairs Mycobacterium tuberculosis-Specific CD4 and CD8 T-Cell Responses

Mycobacterium tuberculosis and human immunodeficiency virus (HIV) infections are coendemic in several regions of the world, and M. tuberculosis/HIV-coinfected individuals are more susceptible to progression to tuberculosis disease. We therefore hypothesized that HIV infection would potentially impair M. tuberculosis-specific protective immunity in individuals suffering from latent tuberculosis infection (LTBI) or active pulmonary tuberculosis (PTB). In this study, we demonstrated that M. tuberculosis/HIV-coinfected individuals have fewer circulating M. tuberculosis-specific CD4 T cells and that those that remained were functionally impaired in both LTBI and PTB settings. In addition, we showed that HIV infection significantly interferes with M. tuberculosis-induced systemic proinflammatory cytokine/chemokine responses. Taken together, these data suggest that HIV infection impairs functionally favorable M. tuberculosis-specific immunity.

HIV infection influences M. tuberculosis-specific CD4 T-cell frequencies and cytokine profiles. The frequencies and functional profiles of M. tuberculosis-specific CD4 T-cell responses were assessed by intracellular cytokine staining (ICS) according to the gating strategy shown in Fig. 1A. In particular, the ability of antigen-specific CD4 T cells to produce IFN-␥, TNF-␣, IL-2, IL-4, IL-5, and/or IL-13 in response to M. tuberculosis (ESAT-6 and CFP-10 peptide pools) or HAd5 (hexon-derived overlapping peptide pool) antigen-specific stimulation was assessed by multiparametric flow cytometry in 20 LTBI and 67 PTB individuals and compared to that in 15 HIV/LTBI-and 8 HIV/PTB-coinfected individuals. Of note, Th2 cytokines, i.e., IL-4, IL-5, and IL-13, were all assessed in the same flow cytometry fluorescence channel, which allowed the assessment of total Th2 cytokine production but prevented direct identification of individual IL-4, IL-5, or IL-13 antigen-specific CD4 T-cell responses.
We next analyzed the cytokine profile of M. tuberculosis-specific memory CD4 T cells of individuals with LTBI, LTBI/HIV, PTB, and HIV/PTB (Fig. 2B, pie charts). The cytokine profiles of M. tuberculosis-specific memory CD4 T cells of individuals with LTBI or PTB were significantly different from those of individuals with HIV/LTBI or HIV/PTB, respectively (P Ͻ 0.05) (Fig. 2B, pie charts). Again, no significant differences were observed between individuals with LTBI and PTB or between individuals with HIV/LTBI and HIV/PTB (P Ͼ 0.05) (Fig. 2B, pie charts).
Taken together, our data indicate that HIV infection strongly influences M. tuberculosis-specific memory CD4 T-cell frequencies and cytokine profile, among which IL-2, IL-17A/F, and IL-21 production/secretion capacity appeared to be the most impacted, independently of M. tuberculosis disease status.
The combined data showed that the percentages of memory CD4 T cells expressing Gata-3 or ROR␥t were significantly reduced in individuals with HIV/LTBI or HIV/PTB compared to those in individuals with LTBI or PTB (Gata-3, 2.4% and 2% versus 6.7% and 6.4%, respectively [P Ͻ 0.05]; ROR␥t, 1.1% and 0.8% versus 2% and 1.9%, respectively [P Ͻ 0.05]) ( Fig. 3A and B). In contrast, the percentage of memory CD4 T cells expressing high levels of T-bet (T-bet high ) was significantly increased in individuals with HIV/LTBI or HIV/PTB compared to that in individuals with LTBI or PTB (13% and 17% versus 0.9% and 3.4%, respectively [P Ͻ 0.05]) (Fig. 3C). Notably, the frequencies of memory CD4 T cells expressing T-bet high were significantly higher in individuals with PTB than in individuals with LTBI (3.4% versus 0.9% [P Ͻ 0.05]) (Fig. 3C). Interestingly, the percentage of T-bet high memory CD4 T cells was inversely correlated with the percentage of memory CD4 T cells expressing Gata-3 (r ϭ Ϫ0.6685; P Ͻ 0.0001) (Fig.  3D), supporting previous observations (25).
We then determined whether the M. tuberculosis-specific CD4 T-cell cytokine profiles observed in individuals with LTBI or PTB coinfected or not with HIV were associated with T-bet, Gata-3, or ROR␥t expression profiles. To address this issue, we plotted the percentage of M. tuberculosis-specific CD4 T cells producing IFN-␥ or IL-4/5/13 against the percentage of memory CD4 T cells expressing T-bet high or Gata-3 from the same patients ( Fig. 3E and F) or the levels of IL-17A/F detected in M. tuberculosis-stimulated cell culture supernatants against the percentage of memory CD4 T cells expressing ROR␥t from the same patients (Fig. 3G). The cumulative data demonstrate that the percentage of IFN-␥-producing M. tuberculosis-specific CD4 T cells directly correlated with the percentage of T-bet high memory CD4 T cells (r ϭ 0.4599; P ϭ 0.0002) (Fig. 3E) and the percentage of IL-4/5/13-producing M. tuberculosis-specific CD4 T cells directly . Undetectable values were arbitrarily defined as 0.1 pg/ml. Individuals were color coded (A to C); Individuals with LTBI, blue; individuals with HIV/LTBI, red; individuals with PTB, green and individuals with HIV/PTB, orange. Red asterisks indicate statistical significance. Statistical significance (P Ͻ 0.05) was calculated using one-way ANOVA (Kruskal-Wallis test) followed by a Mann-Whitney test (A and C). Statistical analyses of the global cytokine profiles (pie charts) (B) were performed by partial permutation tests using the SPICE software as described previously (54). correlated with the percentage of memory CD4 T cells expressing Gata-3 (r ϭ 0.4085; P ϭ 0.0011) (Fig. 3F). In addition, the levels of IL-17A/F detected in M. tuberculosisstimulated cell culture supernatant directly correlated with the percentage of memory CD4 T cells expressing ROR␥t from the same patients (r ϭ 0.3349; P ϭ 0.0187) (Fig. 3G). Notably, the percentages of Gata-3 and T-bet high memory CD4 T cells negatively correlated with the percentage of M. tuberculosis-specific CD4 T cells producing IFN-␥ and IL-4/5/13, respectively (Gata-3 versus IFN-␥, r ϭ Ϫ0.3707 and P Ͻ 0.05; T-bet high versus IL-4/5/13, r ϭ Ϫ0.3476 and P Ͻ 0.05) ( Fig. 3H and I).
Taken together, these data indicate that HIV infection strongly influenced Gata-3, ROR␥t, and T-bet high T-cell lineage transcription factor expression profiles. In particular, HIV coinfection resulted in a significant shift of the M. tuberculosis-specific cytokine profile from a mixed Th1/Th2/Th17 cytokine profile associated with increased Gata-3 and ROR␥t and reduced T-bet high expression observed in individuals with LTBI or PTB to a Th1-restricted cytokine profile associated with increased T-bet high and reduced Gata-3 or ROR␥t expression observed in individuals with HIV/LTBI or HIV/PTB.

HIV infection influences PD-1 expression on M. tuberculosis-specific CD4 T cells.
Our data showed that HIV infection significantly reduced IL-2 production/secretion from M. tuberculosis-specific CD4 T cells. Since IL-2 production/secretion capacity might be reduced in effector memory cells (EM; CD45RA Ϫ CCR7 Ϫ ) and/or by coinhibitory molecule expression (26,27), we therefore assessed whether HIV infection might have influenced CCR7 and/or PD-1 surface expression on M. tuberculosis-specific CD4 T cells.
The percentage of M. tuberculosis-specific memory CD4 T cells expressing PD-1 was significantly increased in individuals with HIV/LTBI and HIV/PTB compared to individuals with LTBI and PTB, respectively (51% and 38% versus 17% and 24%; P Ͻ 0.05) ( Fig. 4A and B). In contrast, the percentages of M. tuberculosis-specific memory CD4 T cells expressing CCR7 did not differ between individuals with HIV/LTBI or HIV/PTB and individuals with LTBI or PTB (P Ͻ 0.05) ( Fig. 4A and C).
We then investigated whether the M. tuberculosis-specific CD4 T-cell cytokine profiles observed in individuals with LTBI or PTB coinfected or not with HIV were associated with PD-1 expression. To address this issue, we plotted the proportion of M. tuberculosis-specific CD4 T cells producing IL-2 (either total IL-2-producing M. tuberculosis-specific memory CD4 T cells or IFN-␥/IL-2/TNF-␣-producing M. tuberculosis-specific memory CD4 T cells) against the percentage of M. tuberculosis-specific memory CD4 T cells expressing PD-1 from the same patients ( Fig. 4D and E). The cumulative data showed that the percentage of M. tuberculosis-specific CD4 T cells expressing PD-1 inversely correlated with the proportion of IL-2-producing M. tuberculosis-specific memory CD4 T cells (r ϭ Ϫ0.5781; P Ͻ 0.0001) (Fig.  4D) or the proportion of IFN-␥/IL-2/TNF-␣-producing M. tuberculosis-specific memory CD4 T cells (r ϭ Ϫ0.4798; P ϭ 0.0007) (Fig. 4E).
These data indicate that PD-1 expression on M. tuberculosis-specific CD4 T cells is induced by HIV infection and associated with reduced IL-2 production/secretion by M. tuberculosis-specific CD4 T cells.
HIV infection did not influence HAd5-specific CD4 T-cell frequencies, cytokine profiles, or PD-1 expression. In order to determine whether HIV infection specifically influenced M. tuberculosis-specific CD4 T-cell responses in the present study, HAd5specific CD4 T-cell responses were also assessed in LTBI with or without HIV coinfection. The cumulative data indicated that the frequencies and cytokine profiles of HAd5specific CD4 T cells were not significantly influenced by HIV infection (P Ͼ 0.05) (Fig. 5A  and B). In addition, the percentages of HAd5-specific CD4 T cells expressing PD-1 and/or CCR7 did not significantly differ between individuals with LTBI and individuals with HIV/LTBI (P Ͼ 0.05) (Fig. 5C and D). Taken together, these data indicated that in contrast to the case with M. tuberculosis-specific CD4 T cells, HIV infection influenced neither HAd5-specific CD4 T-cell frequencies nor their cytokine profiles or PD-1 expression.
HIV   between individuals with HIV/LTBI or HIV/PTB and individuals with LTBI or PTB, respectively (7% versus 0% and 80% versus 46%; P Ͼ 0.05) (Fig. 6A). Due to the limited number of LTBI and HIV/LTBI subjects with detectable M. tuberculosis-specific CD8 T cells (Fig. 6A), M. tuberculosis-specific CD8 T-cell frequencies, cytokine profile and phenotype were restricted to individuals with PTB (Fig. 6B). The cumulative data indicated that the frequencies of IL-2-producing M. tuberculosis-specific CD8 T cells were significantly reduced in individuals with HIV/PTB compared to individuals with PTB (P Ͻ 0.05), while frequencies of IFN-␥-or TNF-␣-producing M. tuberculosis-specific CD8 T cells did not significantly differ between individuals with PTB and individuals with HIV/PTB (P ϭ 0.7202 and P ϭ 0.4585, respectively) (Fig. 6B). The cytokine profiles of M. tuberculosis-specific CD8 T-cell responses in PTB versus HIV/PTB differed significantly (P Ͻ 0.05) (Fig. 6C, pie charts). A significant loss of polyfunctional IFN-␥ ϩ IL-2 ϩ TNF-␣ ϩ -producing M. tuberculosis-specific CD8 T cells in individuals with HIV/PTB compared to individuals with PTB (0% versus 23%; P Ͻ 0.05) (Fig. 6C) was observed.
Next we measured the PD-1 expression level of M. tuberculosis-specific CD8 T cells. The cumulative data indicate that PD-1 expression was significantly increased on M. tuberculosis-specific memory CD8 T cells in HIV/PTB compared to PTB (P Ͻ 0.05) (Fig. 6D), indicating that HIV infection strongly influenced PD-1 expression on M. tuberculosis-specific CD8 T cells and was associated with reduced IL-2 production capacity. (18,19). In addition, PD-1 expression level directly correlated with HIV viral load, suggesting that excessive and continuous antigen stimulation functionally impaired HIV-specific T cells (17,30). However, the impact of ongoing HIV replication on M. tuberculosis-specific CD4 T cells remains unclear. We then determined whether M. tuberculosis-specific CD4 T-cell cytokine profiles observed in M. tuberculosis/HIV-coinfected individuals were associated with HIV viral load. To address this issue, we plotted the proportion of M. tuberculosisspecific CD4 T cells producing IL-2 (either total IL-2-producing M. tuberculosis-specific memory CD4 T cells or IFN-␥/IL-2/TNF-␣-producing M. tuberculosis-specific memory CD4 T cells) or expressing PD-1 against HIV viral load (Fig. 7). Our results show that the proportion of M. tuberculosis-specific CD4 T cells producing IL-2 or the percentage of M. tuberculosis-specific CD4 T cells expressing PD-1 did not correlate with HIV viral load (r ϭ Ϫ0.0051, r ϭ Ϫ0.0735, and r ϭ Ϫ0.0017; P Ͼ 0.05; respectively) ( Fig. 7), suggesting that the degree of M. tuberculosis-specific CD4 T-cell impairment was independent of ongoing HIV replication.

HIV infection dampens the level of systemic inflammation markers in PTB/HIVcoinfected individuals.
One objective of the present study was to determine whether HIV infection may influence systemic inflammation markers in the Tanzanian cohort studied. Hence, the serum levels of IL-1␣, IL-6, C-reactive protein (CRP), IL-23, and IP-10 were assessed in individuals with LTBI or PTB coinfected or not with HIV using a multiplex bead assay (Fig. 8). Notably, the serum levels of IL-6, CRP, and IP-10 were significantly increased in individuals with PTB compared to individuals with LTBI or HIV/LTBI (P Ͻ 0.05), supporting previous observations (31) (Fig. 8). Interestingly, the serum levels of IL-1␣, IL-6, CRP, IL-23, and IP-10 were significantly reduced in individuals with HIV/PTB compared to individuals with PTB (P Ͻ 0.05) (Fig. 8).
Taken together, these data suggest that HIV infection significantly interferes with M. tuberculosis-induced systemic proinflammatory cytokine/chemokine responses.

FIG 5 Legend (Continued)
Percentages of HAd5-specific CD4 T cells isolated from individuals with LTBI (n ϭ 9) or HIV/LTBI (n ϭ 10) expressing PD-1 (C) and/or CCR7 (D). Statistical significance (*; P Ͻ 0.05) was calculated using the Mann-Whitney test (A, C, and D). Statistical analyses of the global cytokine profiles (pie charts [B]) were performed by partial permutation tests using the SPICE software as described previously (54).

DISCUSSION
Previous studies indicate that the cytokine profile and the frequencies of M. tuberculosis-specific CD4 T cells could be impacted by HIV coinfection (14,15,32) and that these responses were not fully restored under cART (12,13). In the present study, we hypothesized that progression from LTBI to PTB might not be due to CD4 T-cell depletion only but might also be driven by M. tuberculosis-specific CD4 and CD8 T-cell functional impairment. To test this hypothesis, we assessed the cytokine profile, the transcription factor expression profile, and the phenotype of M. tuberculosis-specific T cells of untreated Tanzanian individuals suffering from LTBI or PTB and compared them to those of untreated M. tuberculosis/HIV-coinfected individuals suffering from LTBI or PTB.
We showed that HIV-infected individuals harbored reduced M. tuberculosis-specific CD4 T-cell frequencies associated with significant changes in M. tuberculosis-specific CD4 T-cell cytokine profiles. Interestingly, HIV infection did not influence all cytokines to the same extent. In particular, we showed that HIV infection significantly reduced the proportion of IL-2-or Th2 (IL-4/IL-5/IL-13)-producing M. tuberculosis-specific CD4 T cells in LTBI or PTB. These data, however, contrast with a recent study that did not report significant changes in IL-2 production although changes in M. tuberculosis-specific CD4 T-cell cytokine profiles were observed (15). These differences might be attributed at least in part to the cohort studied, i.e., South African versus Tanzanian individuals. Indeed, we recently observed that M. tuberculosis-specific CD4 T-cell cytokine profile of PTB patients from Tanzania were primarily composed of polyfunctional Th1 and Th2 cells, while M. tuberculosis-specific CD4 T-cell cytokine profiles of TB patients from South Africa were dominated by single IFN-␥ and dual IFN-␥/TNF-␣ (25).
In addition, we showed that M. tuberculosis/HIV-coinfected individuals harbored reduced capacity to secrete IL-17A, IL-17F, and IL-21 cytokines following in vitro M. tuberculosis antigen stimulations, supporting previous observations for LTBI subjects (33) and suggesting that Th17 cells may contribute to control of M. tuberculosis replication.
Interpretation of cytokine production from stimulated cells remains challenging, especially within individuals with disease. Notably, the reduced cytokine levels observed in some individuals might be associated with reduced M. tuberculosis-specific T-cell frequencies and/or functional impairment. In addition, various parameters such as transcription factor expression (22,23), memory subsets (26), and/or the expression of coinhibitory molecules (17)(18)(19)27) could influence the cytokine profile of antigenspecific CD4 T cells. Here we demonstrate that changes of M. tuberculosis-specific CD4 T-cell cytokine profiles in M. tuberculosis/HIV-coinfected individuals was associated with an increase of T-bet and PD-1 expression accompanied by a reduction of Gata-3 and ROR␥t expression in the memory CD4 T cells. The proportion of IL-4/5/13-producing M. tuberculosis-specific CD4 T cells was inversely correlated with the percentage of memory CD4 T cells expressing T-bet high , and the proportion of IL-2-producing M. tuberculosis-specific CD4 T cells was inversely correlated with the percentage of PD-1expressing M. tuberculosis-specific CD4 T cells. Consistent with previous studies, we did not observe any influence of HIV infection on the differentiation profile of M. tuberculosis-specific CD4 T cells (15,34), indicating that HIV infection might influence the M. tuberculosis-specific CD4 T-cell cytokine profile by influencing the transcription factor profile and PD-1 expression. Several clinical studies performed in multidrugresistant TB cases showed an increased rate of Mycobacterium clearance associated with radiological lung improvement when recombinant IL-2 was applied (35).
To determine whether HIV infection might specifically impact M. tuberculosis-specific CD4 T-cell responses, we also assessed HAd5-specific T-cell responses in LTBI with or without HIV coinfection in the same volunteers. HAd5 is an intracellular pathogen that usually causes upper and lower respiratory tract infections (36), controlled by efficient polyfunctional CD4 and CD8 T-cell responses (37)(38)(39), leading to persistent sub-clinical infections in most of immunocompetent individuals (40,41). In contrast, in immunocompromised individuals, reactivation of HAd5 replication can lead to fatal disease progression (40). Interestingly, HAd5-specific CD4 T-cell frequencies, cytokine profiles, and phenotypes did not differ between HIV-uninfected and HIV-infected individuals, suggesting that HIV infection might specifically influence M. tuberculosis-specific CD4 T-cell responses. Notably, we cannot exclude that distinct scenario or set of circumstances were associated with these observations. However One potential mechanism by which HIV infection may influence TB disease progression and therefore M. tuberculosis-specific T-cell cytokine profile and phenotype might be through granuloma disorganization. Indeed, M. tuberculosis replication is partially contained by organized granulomas formed by macrophages and CD4 T cells in the lung tissues (1), which HIV infection may disrupt through CD4 T-cell depletion, favoring M. tuberculosis dissemination and extrapulmonary disease (43,44).
Finally, during PTB, M. tuberculosis-exposed/infected lung macrophages produce large quantities of proinflammatory cytokines and chemokines involved in the chemoattraction of monocytes and T cells to the sites of infection (45,46). These cytokines/ chemokines are detectable in the sera of individuals with PTB, and the serum cytokine profile may be indicative of the level of systemic inflammation, antigen load, TB disease severity, and hospitalization duration (47). Therefore, in the present study, we were interested in how and if HIV infection may interfere with M. tuberculosis-induced cytokine/chemokine production. We compared the serum levels of proinflammatory cytokines and chemokines of individuals with LTBI and PTB coinfected or not with HIV. We found that the serum levels of IL-1␣, IL-6, CRP, IL-23, and IP-10 were significantly reduced in HIV/PTB-coinfected individuals compared to the levels in individuals with PTB only, suggesting that HIV infection significantly dampens M. tuberculosis-induced systemic proinflammatory cytokine/chemokine response.
The mechanism by which HIV suppresses the M. tuberculosis-induced systemic proinflammatory cytokine response in PTB requires further investigation. However, one potential mechanism might be through M. tuberculosis-specific CD4 T-cell depletion. Indeed, M. tuberculosis-induced systemic proinflammatory cytokine response in PTB is mainly mediated by macrophages, epithelial cells, and fibroblasts in response to proinflammatory cytokines released by M. tuberculosis-specific CD4 T cells (48)(49)(50), which might be reduced in functionally impaired M. tuberculosis-specific CD4 T cells.
In conclusion, the present study provides evidence that M. tuberculosis-specific CD4 and CD8 T-cell responses are impacted by HIV coinfection, resulting in pronounced variations in the qualitative and quantitative profile of M. tuberculosis-specific T cells in human populations. The precise mechanism by which HIV infection interferes with M. tuberculosis-specific protective immunity still remains to be determined and probably involves both the depletion and the functional impairment of M. tuberculosis-specific T cells.

MATERIALS AND METHODS
Study group and cell isolation. In total, 112 subjects participated to this study and were recruited at the Mwananyamala hospital, Dar es Salaam, and at the TB clinic of Bagamoyo (TZ) ( Table 1). Pulmonary TB patients were selected based on sputum smear microscopy confirmed by GeneXpert assay, and HIV infection was defined based on a rapid serological test (Alere Determine HIV-1/2 test). Individuals with