HIV infection impairs Th1 and Th17 Mycobacterium tuberculosis-specific T cell responses

Abstract

Background

HIV-infected individuals have a higher risk of developing active tuberculosis than HIV-uninfected individuals, but the mechanisms underpinning this are unclear. We hypothesized that depletion of specific components of Mycobacterium tuberculosis (M.tb)-specific CD4+ and CD8+ T cell responses contributed to this increased risk.

Methods

M.tb-specific T cell responses in 147 HIV-infected and 44 HIV-uninfected control subjects in a TB-endemic setting in Bloemfontein, South Africa were evaluated. Using a whole-blood flow cytometry assay, we measured expression of IFNγ, TNFα, IL-2 and IL-17 in CD4+ and CD8+ T cells in response to M.tb antigens (PPD, ESAT-6/CFP-10 (EC) and DosR regulon-encoded α-crystallin (Rv2031c)).

Results

Fewer HIV-infected individuals had detectable CD4+ and CD8+ T cell responses to PPD and Rv2031c than HIV-uninfected subjects. M.tb-specific T cells showed distinct patterns of cytokine expression comprising both Th1 (CD4 and CD8) and Th17 (CD4) cytokines, the latter at highest frequency for Rv2031c. Th17 antigen-specific responses to all antigens tested were specifically impaired in HIV-infected individuals.

Conclusions

HIV-associated impairment of CD4+ and CD8+ M.tb-specific T cell responses is antigen-specific, particularly impacting responses to PPD and Rv2031c. Preferential depletion of Th17 cytokine-expressing CD4+ T cells suggests this T cell subset may be key to TB susceptibility in HIV-infected individuals.

Introduction

Infection with Human Immunodeficiency Virus (HIV) predisposes individuals to active tuberculosis (TB), resulting in a global HIV/TB co-infection epidemic that disproportionately affects sub-Saharan Africa [1–3]. The key components of protective mycobacterial immunity remain incompletely defined. Mycobacterium tuberculosis (M.tb)-specific T cells play an important role in protection against TB, with the Th1 CD4+ T cell response believed to be critical [4,5]. HIV co-infection is associated with early and preferential depletion of Th1 M.tb-specific CD4+ T cells, possibly explaining the increased TB susceptibility before the onset of profound CD4+ T cell loss [6,7]. In addition to quantitative M.tb-specific CD4+ T cell loss, HIV infection induces functional changes to the Th1 T cell response, including the suppression of T cell proliferation and cytokine secretion, and reduced polyfunctionality in both peripheral blood and the lung [8–12].

M.tb-specific Th1 CD4+ T cell responses to the best described TB antigens (M.tb purified protein derivative (PPD), early secreted antigenic target 6 kDa (ESAT-6, Rv3875) and culture filtrate protein 10kDa (CFP-10, Rv3874) are not sufficient for protection and are poor correlates of TB risk [13–15]. There is therefore a need to explore other M.tb-specific antigens, such as those of the DosR regulon (e.g. α-crystallin 16kDa (Rv2031c)[16]) expressed during in vitro conditions of hypoxia, low pH and high NO intermediates believed to exist in M.tb granulomata during latent M.tb infection (LTBI) [17,18].

M.tb-specific Th17 CD4+ T cells also contribute to the M.tb-specific CD4+ T cell response, although the effect of HIV on this subset is not well understood [19]. Animal studies suggest a role for M.tb-specific CD8+ T cell responses[20–22], although their importance in humans is controversial, but with some evidence for a preserved if poorly functional response [23][24]. Evidence from mouse models suggests that loss of IFNγ from CD4+ T cells may negatively impact CD8+ T cell responses, although this is yet to be shown in humans [25].

Here, we explore the effect of HIV infection on M.tb-specific Th1 and Th17 CD4+ and CD8+ T cell cytokine responses in a well-characterised South African cohort. Responses to cytomegalovirus (CMV) are used as a comparator.

Materials and Methods

Study site and participants

Participants were recruited from antiretroviral therapy (ART) initiation sites in Bloemfontein, Free State Province, South Africa at Heidedal, MUCCP and National Hospital clinics. HIV-infected individuals were ART-naïve and recruitment was based on CD4+ T cell counts to reflect varied levels of disease progression. Exclusion criteria included active TB, current AIDS-defining illness, pregnancy, previous ART, current use of immunosuppressive medication and age <18 years. HIV-uninfected individuals were recruited from the same healthcare facilities. The study was approved by the University of the Free State Ethics Committee (ETOVS 171/08). Participants were aged 18 years or older, were informed in their own language and provided written consent.

Collection and processing of blood samples

Within two hours of venepuncture, 1ml of heparinized whole blood was incubated with antigen at 37°C, with anti-CD28 and anti-CD49d antibodies (Becton Dickinson, UK) [26]. Stimulatory antigens were: positive control (Staphylococcal enterotoxin B, SEB; 5 μg/mL; Sigma-Aldrich, UK), negative control (co-stimulatory antibodies alone), M.tb PPD (PPD; 20 μg/mL; Statens Serum Institut, Denmark), ESAT-6 and CFP-10 recombinant proteins in a single combined stimulation designated “EC” (EC; ESAT-6 10 μg/mL, CFP-10 10 μg/mL; Lionex, Germany), Rv2031c (Alpha crystallin 16kDa) recombinant protein (Rv2031c; 10μg/mL; Lionex, Germany), cytomegalovirus lysate (CMV; 5μg/mL, Virusys, USA). The M.tb recombinant proteins were expressed and purified from E.coli, had purity of >95% and endotoxin contents below 25 IU/mg (Limulus Amebocyte Lysate assay). Brefeldin A (5 μg/mL; Sigma-Aldrich, UK) was added after six hours and further incubated for five hours. Red blood cells were lysed and white cells fixed before long-term storage in liquid nitrogen.

Flow cytometry and intracellular cytokine staining (ICS)

Cryopreserved white cells were stained as previously described [26]. The following conjugated monoclonal antibodies directed at cell-surface markers and intracellular cytokines were used: CD3-eFluor 450 (clone UCHT1), IFNγ-PE/Cy7 (clone 4S.B3) and TNFα-PerCP/Cy5.5 (clone MAb11) (eBioscience, UK); CD4-APC (clone RPA-T4), CD14-APC/Cy7 (clone HCD14), CD19-APC-Cy7 (clone HIB19) and IL-17-Alexa Fluor 488 (clone BL168) (BioLegend, UK), CD8-VioGreen (clone BW135/80) and IL-2-PE (clone N7.48A) (Miltenyi Biotec, Germany).

Stained cells were acquired on a MACSQuant flow cytometer (Miltenyi Biotec, Germany) and all events were captured. At least 50,000 events were captured for each sample included in the analysis. Data were analysed using FlowJo vX (Tree Star, USA). The gating strategy is shown in Supplementary Figure 1.

Boolean cytokine combination gates were created using FlowJo vX and analysed with PESTLE (v1.7; National Institutes of Health (NIH), USA) and SPICE software (v5.35; NIH, USA). A positive T cell cytokine response was defined as a frequency of total cytokine-positive CD4+ and/or CD8+ T cells at least 0.1% of total CD4+/CD8+ T cells after background subtraction and at least twice the frequency of total cytokine-positive CD4+/CD8+ T cells in the negative control. As opposed to M.tb, the genome of attenuated M. bovis BCG lacks the RD1-region encoding the genes for the proteins ESAT-6 and CFP-10. In addition, Rv2031c and components of PPD can be found in mycobacteria other than M.tb. Therefore, as per accepted convention, we classified individuals with LTBI only if demonstrating a cytokine-positive T cell response to ESAT-6 and CFP-10 as used in combination in our assay.

Statistical Analyses

Standard statistical approaches for parametric and non-parametric data were used as appropriate. Specific tests are detailed in the results and include Fisher-exact T tests, Kruskal-Wallis tests and Mann-Whitney tests (Prism 5 for Mac OS X v5.0a; GraphPad, USA). Partial permutation analyses were calculated using SPICE (v5.35; NIH, USA). Multivariate logistic regression was performed using SPSS (v22; IBM, USA).

Results

Participant characteristics

44 HIV-uninfected and 147 HIV-infected individuals were studied (Table 1). Of the latter, CD4+ T cell counts ranged from ≤200 cells/μL (n=51), 201-350 (n=48) and >350 (n=48). 14 HIV-infected individuals had a history of active TB; there were no previous TB cases in the HIV-uninfected group. All were believed to have received BCG vaccination in childhood, although only 20 (45%) HIV-uninfected and 89 (61%) HIV-infected individuals had a BCG vaccination scar. HIV-infected subjects were stratified by CD4+ T cell count.

Table 1. Demographics of Study Participants.

	HIV-uninfected	HIV-infected
	n= 44	Total n= 147	CD4 >350 n= 48	CD4 201 – 350 n= 48	CD4 ≤ 200 n= 51
Gender
Female – n (%)	20 (45)	96 (65)	37 (77)	34 (71)	25 (49)
Male – n (%)	24 (55)	51 (35)	11 (23)	14 (29)	26 (51)
Age (years) *	24 (20–32)	36 (30–44)	36 (31–44)	37 (29–46)	36 (31–44)
CD4 count (cells/μL)*	879 (686–1054)	254 (178–379)	473 (380–596)	260 (231–295)	155 (120–183)
CD8 count (cells/μL) *	586 (409–923)	946 (673–1542)	1234 (806–1986)	925 (715–1447)	821 (492–1206)
CD4:CD8 Ratio*	1.63 (1.03–2.23)	0.27 (0.17–0.42)	0.51 (0.30–0.77)	0.27 (0.18–0.39)	0.17 (0.12–0.26)
Viral load (RNA copies/ml) *	N/A	48173 (11490–161087)	15306 (6914–64855)	58963 (16423–181851)	139402 (34425–456048)
BMI (kg/m2) *	24.3 (22.7–27.8)	24.9 (22.3–28.7)	26.5 (22.7–31.5)	24.9 (21.0–28.1)	24.9 (21.0–28.2)
Prev TB – n (%)	0 (0)	14 (10)	6 (13)	5 (10)	3 (6)
BCG scar – n (%)	20 (45)	89 (61)	31 (65)	25 (52)	33 (65)

Definition of abbreviations: BMI = Body mass index; TB = tuberculosis; BCG = Bacillus Calmette-Guerin, N/A = not applicable.

^*Data expressed as median (IQR)

Impairment of PPD and Rv2031c-specific CD4+ T cell responses in HIV-infected individuals

As expected, median total CD4+ T cell counts were significantly lower in HIV-infected than uninfected individuals, whilst CD8+ T cell counts were higher (P<0.001 for all, Mann-Whitney) (Table 1). The median CD4:CD8 ratio of 1.63 (IQR 1.02 – 2.23) in the HIV-uninfected group was as expected for healthy African individuals [27,28], but was significantly reduced in HIV-infected individuals (Table 1).

We examined the effect of HIV infection on M.tb (EC, PPD and Rv2031c) and CMV-specific CD4+ T cell cytokine responses. In the HIV-uninfected control group, a cytokine-positive response to EC was present in 21/44 (47.7%) of participants, reflecting the proportion with LTBI, and similar to other sub-Saharan African populations (Table 2) [29,30]. This was not significantly different to the proportion of HIV-infected participants responding to EC (58/146 (39.7%) (P=0.385)) (Table 2). However, the proportions of individuals with CD4+ T cell responses to PPD and Rv2031c were significantly lower in the HIV-infected group compared with healthy controls (66.4% vs 86.4% (P=0.013) and 38.6% vs 56.8% (P=0.038), respectively) (Table 2). Similar proportions of HIV-infected and uninfected participants responded to CMV lysate (91.8% vs 97.6% (P=0.304)) (Table 2).

Table 2. Comparison of cytokine-positive CD4+ responders between HIV-infected and HIV-uninfected subjects.

Antigen	CD4 Responders			CD8 Responders
	HIV - Number (%)	HIV + Number (%)	P value	HIV – Number (%)	HIV+ Number (%)	P value
EC	21/44 (47.7)	58/146 (39.7)	0.385	10/44 (22.7)	22/146 (15.1)	0.250
PPD	38/44 (86.4)	97/146 (66.4)	0.013	17/44 (38.6)	21/146 (14.4)	0.001
Rv2031c	25/44 (56.8)	56/145 (38.6)	0.038	22/44 (50.0)	46/145 (31.7)	0.032
CMV	41/42 (97.6)	135/147 (91.8)	0.304	26/42 (61.9)	110/147 (74.8)	0.120

Definition of abbreviations: EC= ESAT-6 and CFP-10, PPD = Purified protein derivative, Rv2031c = alpha crystallin 16kDa, CMV = Cytomegalovirus lysate. P values are indicated for Fisher’s exact tests.

We next compared the frequencies of cytokine-positive CD4+ T cells responding to M.tb and CMV antigens in individuals across different CD4+ T cell strata (≤200, 200-350 and >350)(Figure 1). The frequencies of total CD4+ T cells responding to PPD and Rv2031c were significantly lower in HIV-infected individuals compared with uninfected controls (P<0.001 for both) (Figure 1B and 1C). CD4+ T cell responses to PPD were reduced in participants with CD4+ T cell counts >200 cells/μL, including those with CD4+ T cell counts above 350 cells/μL, indicating specific depletion of PPD-specific CD4+ T cells despite relatively preserved total CD4+ T cell counts (Figure 1B). Reduced frequencies of Rv2031c-specific CD4+ T cells between HIV-uninfected and HIV-infected subjects were more evident at lower CD4+ T cell counts, with relative preservation above 350 cells/μL (Figure 1C). Interestingly, the lack of significant differences in the proportions of HIV-infected and HIV-uninfected subjects responding to CMV and EC (Table 2) was also reflected in similar frequencies of CD4+ T cell responses to these antigens in these groups (Figure 1A and D).

Figure 1. Reduced PPD- and Rv2031c-specific CD4+ T cell frequencies in HIV-infected individuals.

Antigen-specific CD4 T cell frequencies were compared between the HIV-uninfected and HIV-infected subjects within different CD4 T cell count strata. The data are shown as total cytokine-positive CD4 T cell frequencies as measured using flow cytometry and therefore include cells expressing any of the 4 cytokines measured singly or in combination. The horizontal bars indicate the median and inter-quartile range. P values on the top right of each panel indicate Kruskal-Wallis P values. Significant differences between individual groups were tested using Dunn’s Multiple Comparisons test. * P <0.05, ** P <0.01, *** P <0.001

M.tb and CMV-specific T cell responses exhibit distinct patterns of cytokine expression

As the loss of antigen-specific CD4+ T cell responses following HIV infection has been associated with different patterns of cytokine expression, we characterised and compared the cytokine expression patterns of each antigen-specific CD4+ T cell population in our cohort[31]. To set a comparator group, we first measured the relative contributions of IFNγ, TNFα, IL-2, and IL-17 to the total CD4+ T cell response within the HIV-uninfected control group.

Th1 cytokines (IFNγ, TNFα and IL-2) dominated the CD4+ T cell cytokine responses to both EC and PPD, with all three expressed by more than 50% of cytokine-expressing antigen-specific cells in responding participants (Figure 2A). Rv2031c also induced Th1 cytokines, dominated by IFNγ, but with increased production of IL-17 (30% of cytokine-expressing cells), which was also seen for EC or PPD, but to a much lesser extent (7% and 5%, respectively). The CMV-specific CD4+ T cell responses showed a similar hierarchy as EC and PPD, but with decreased IL-2 expression and very low IL-17 expression (Figure 2A). For all antigen responses there was very little co-expression of Th1 cytokines and IL-17 (Figure 2B). Interestingly, despite EC and PPD-specific CD4+ T cell responses showing markedly different patterns of depletion in HIV-infected subjects, their cytokine expression profiles were similar in HIV-uninfected control individuals (P = 0.235) (Figure 2B).

Figure 2. Mtb- and CMV-specific CD4+ T cells exhibit distinct cytokine expression profiles in HIV-uninfected controls.

Expression of IFNγ, TNFα, IL-2, and IL-17 in HIV-uninfected antigen-responders was assessed using flow cytometry (EC, n = 21; PPD, n = 38; Rv2031c, n = 25; CMV, n = 41). (A) The fraction of total cytokine-positive CD4+ T cells expressing each cytokine is shown. (B) Values shown are the fraction of total cytokine-positive CD4+ T cells responding to each antigen (coloured as per the upper legend) producing the given combination of cytokines (upper panel). Filled bars represent median values and whiskers show the inter-quartile range. The coloured bars under the graph represent the number of cytokines produced by each cell (crimson, 4; orange, 3; blue, 2; yellow, 1). The same colours are used in the pie charts (lower panel) summarise the proportion of cells expressing 1, 2, 3 or 4 cytokines simultaneously. The arcs surrounding the pie charts indicate the proportion of the responses contributed by each of the single cytokines. P values shown in the table at the bottom compare pie charts and are calculated using a partial permutation analysis[50].

M.tb and CMV-specific Th17 CD4+ T cell responses are significantly depleted in HIV infected individuals

We then characterised CD4+ T cell cytokine responses in the HIV-infected participants to determine if the observed HIV-associated loss of antigen-specific responses would be reflected in changes to particular cytokine-expressing CD4+ T cell sub-populations. Specifically, we were interested in the different patterns of Th1 and Th17 cytokine expression. As before, Th1 CD4+ T cell responses were defined by the expression of IFNγ, TNFα and IL-2 alone or in combination, whilst Th17 CD4+ responses were defined by expression of IL-17 alone.

Amongst EC-specific CD4+ T cells, only those cells expressing IL-2 and IL-17 were reduced compared with uninfected controls (P=0.036 and P=0.037, respectively), whereas those expressing IFNγ and TNFα were preserved (Figure 3A). For PPD, which showed the greatest proportionate decrease in CD4+ T cell frequencies between HIV-infected and uninfected participants, there was significant reduction of CD4+ T cells expressing all cytokines representing both Th1 and Th17 responses (P≤0.001 for all comparisons tested) (Figure 3B). Interestingly, the impact of HIV infection on the CD4+ T cell response to Rv2031c was manifested in a significant loss of both IFNγ and IL-17-expressing CD4+ T cells, the two most dominant components of the response in HIV-uninfected controls (Figure 3C).

Figure 3. HIV-infected individuals have impaired Th17 CD4+ T cell responses to all Mtb and CMV antigens tested, in addition to impairment of certain antigen-specific Th1 CD4+ T cell responses.

The frequencies of antigen-specific CD4+ T cells expressing each cytokine were compared between HIV-infected and HIV-uninfected individuals. The Th1 CD4+ T cell response included CD4+ T cells expressing IFNγ, TNFα or IL-2 alone or in combination whilst the Th17 response was defined by IL-17 expression. Horizontal bars indicate median values and inter-quartile ranges. P values were generated using Mann-Whitney tests.

The CMV-specific response was also impacted by HIV infection, with HIV-infected participants making statistically significantly larger TNFα responses (Figure 3D). In contrast, the median frequency of IL-17-positive CMV-specific CD4+ T cells was reduced (P = 0.005) (Figure 3D). In summary, the Th17 components of all antigen-specific CD4+ T cell responses were significantly depleted in HIV-infected subjects which was not uniformly the case for antigen-specific Th1 CD4+ T cell responses, suggesting increased susceptibility of antigen-specific Th17 CD4+ T cells to depletion during HIV infection.

Chronic HIV infection is not associated with changes in polyfunctional M.tb-specific CD4+ T cell subsets

Polyfunctional antigen-specific CD4+ and CD8+ T cell responses to viral antigens have been shown to be protective in humans and animals, but the association is less clear for anti-M.tb immunity[32]. Having shown the differential impact of HIV infection on Th1 and Th17 CD4+ T cell responses, we wished to determine how HIV infection might influence polyfunctionality. We compared the proportion of single (1+), dual (2+) and triple (3+) cytokine-expressing Th1 CD4+ cells responding to EC, PPD, Rv2031c and CMV (including the positive control, SEB) between HIV-infected and uninfected participants (Figure 4).

Figure 4. The effect HIV infection on CD4+ T cell polyfunctionality is variable and antigen-specific.

The effect of HIV infection on polyfunctionality of antigen-specific CD4+ T cell populations was assessed through a comparison of proportions of single (1+), dual (2+) and triple (3+) cytokine-expressing CD4+ T cells contributing to the total cytokine-positive antigen-specific response, between HIV-infected and HIV-uninfected individuals. Only individuals with a demonstrated cytokine-positive CD4+ T cell response were included in the analysis (SEB: HIV-/HIV+, n = 44/147; EC: HIV-/HIV+, n = 21/58; PPD: HIV-/HIV+, n = 38/97; Rv2031c: HIV-/HIV+, n =25/56; CMV: HIV-/HIV+, n = 41/135). Each pie chart summarises the proportion of cells expressing 1, 2 or 3 cytokines alone or in combination. The arcs surrounding the pie charts indicate the proportion of the responses contributed by each of the single cytokines. P values shown between pie charts give an overview of different cytokine expression profiles between HIV-uninfected and HIV-infected individuals for each antigen and are calculated using a partial permutation analysis[50]. P values shown next to the pie charts compare the proportion of 1+, 2+ or 3+ CD4+ T cells in HIV-uninfected and HIV-infected individuals and are calculated using Mann-Whitney tests. Data in the right-hand panel are shown as relative distributions of the different cytokine-expressing CD4+ T cell populations within the total response to each antigen. P values were calculated using Student’s t tests and are denoted “+” if p<0.05.

HIV-infected subjects had a significantly smaller proportion of 3+ SEB-stimulated CD4+ T cells and a greater proportion of 1+ CD4+ T cells. Similarly, there was a reduction in the polyfunctionality of the CMV-specific response, with HIV-infected subjects having relatively smaller proportion of 3+ and larger proportion of 2+ CD4+ T cells, compared with controls. However, there were no significant differences between HIV-infected and uninfected individuals in the proportions of 2+ or 3+ CD4+ T cells responding to any of the M.tb antigens (Figure 4).

PPD and Rv2031c-specific CD8+ T cell frequencies are reduced in HIV-infected individuals

We next characterised the antigen-specific CD8+ T cell cytokine response. The proportions of HIV-infected individuals with detectable CD8+ T cell cytokine responses to PPD and Rv2031c were significantly lower than HIV-uninfected controls (P=0.001 and P=0.032 respectively) (Table 2). In a similar pattern to the CD4+ T cell responses, there was no difference in the proportions of EC or CMV CD8+ T cell responders between HIV-infected and HIV-uninfected individuals (P=0.250 and P=0.120 respectively) (Table 2).

A comparison of the frequencies of cytokine-positive CD8+ T cells responding to EC between HIV-uninfected individuals and HIV-infected individuals within different CD4+ T cell count strata did not show any significant differences (P=0.432) (Figure 5A). However, consistent with a decrease in the proportion of individuals making a PPD-specific CD8+ response, there was a reduction in PPD-specific cytokine-positive CD8+ T cell frequencies in individuals across all CD4+ T cell strata compared with the HIV-uninfected control group (P<0.001) (Figure 5B). For Rv2031c-specific CD8+ T cell responses, there was evidence for a global decrease in CD8+ T cell frequencies in HIV-infected individuals compared with controls (P=0.043), but this was not attributable to differences between any particular CD4+ T cell subset (Figure 5C). There was no evidence of an impact of HIV infection on the CMV-specific CD8+ cytokine response, even at CD4+ T cell counts below ≤200 cells/μL (Figure 5D).

Figure 5. Impaired PPD- and Rv2031c-specific CD8+ T cell frequencies in HIV-infected individuals.

Antigen-specific CD8+ T cell frequencies were compared between the HIV-uninfected and HIV-infected subjects within different CD4+ T cell count strata. The data are shown as total cytokine-positive CD8+ T cell frequencies and therefore include cells expressing any of the 4 cytokines measured singly or in combination. The horizontal bars indicate the median and inter-quartile range. P values on the top right of each panel indicate Kruskal-Wallis P values. Significant differences between individual groups were tested using Dunn’s Multiple Comparisons test. * P <0.05, ** P <0.01, *** P <0.001

Loss of antigen-specific CD8+ T cell cytokine responses to EC and PPD are associated with loss of antigen-specific CD4+ T cell responses

As CD8+ T cells are not frequently infected by HIV we hypothesized that the impact of HIV infection on the antigen-specific CD8+ T cell cytokine response was secondary to the depletion of CD4+ T cell help, as has been described in animal models of TB and in human CMV studies[25,33]. We therefore examined the effect of different variables on the odds of having a detectable CD8+ T cell cytokine response to each of the antigens tested using multivariate logistic regression (Supplementary Table 1). Individuals included in the analysis had a demonstrated T cell response to the antigen of interest and variables included total CD4+ T cell count, total CD8+ T cell count, HIV plasma viral load and antigen-specific CD4+ T cell count. A CD8+ T cell cytokine response to both EC and PPD was positively associated with antigen-specific CD4+ T cell magnitudes (OR [95% CI] 1.7 [1.2-2.6]; P=0.008 and 1.3 [1.1-1.5]; P<0.001 respectively) but not with total CD4+ T cell count, total CD8+ T cell count or plasma HIV viral load (Supplementary Table 1). Making a Rv2031c-specific CD8+ T cell response was, however, not associated with a Rv2031c-specific CD4+ T cell response, although was positively associated with HIV plasma viral load (OR [95% CI] 2.7 [1.3-5.8]; P=0.008). No factors assessed (including the magnitude of the CMV-specific CD4+ response) significantly affected the likelihood of making a CMV-specific CD8+ response.

Discussion

In this study we have analysed M.tb-specific CD4+ and CD8+ T cell responses to PPD, ESAT-6/CFP-10, and the M.tb DosR regulon antigen Rv2031c, as well as CMV-specific responses in HIV-infected and HIV-uninfected individuals.

We detected cytokine-positive CD4+ and CD8+ T cell responses to all the antigens tested in both HIV-uninfected controls and HIV-infected individuals. The high proportion of CD4+ and CD8+ T cell responders to Rv2031c was striking as the frequency of responders has been lower in other studies [34,35]. This discrepancy may be related to differences in endemic rates of LTBI and exposure to non-tuberculous and environmental mycobacteria between the study populations. In addition, different assay techniques were used to measure Rv2031c-specific CD4+ T cell responses. Arlehamn et al used cryopreserved PBMCs in an IFNγ ELISPOT assay[34], whilst Sutherland et al employed a whole blood IFNγ ELISA[35]. The addition of co-stimulatory antibodies during antigen incubation as well as the measurement of TNFα, IL-2 and IL-17 in addition to IFNγ by flow cytometry in the whole blood ICS assay used in our study, may have contributed to the increased sensitivity we observed [36].

Different rates of CD4+ T cell decline have previously been shown for M.tb and CMV-specific CD4+ T cell responses[31]. We confirmed this finding and also showed that different M.tb-specific CD4+ T cell populations are depleted to different extents in chronic HIV infection. PPD and Rv2031c-specific CD4+ T cells were depleted in HIV-infected individuals whilst EC-specific CD4+ T cells were maintained. This relative sparing of EC-specific CD4+ T cells has been shown before using IFNγ release assays (IGRAs) [37–39]. Despite this, EC-based IGRA’s have shown suboptimal accuracy for diagnosing LTBI in HIV-infected individuals in some studies and could be enhanced through the inclusion of latency-associated antigens [40]. In contrast, PPD-specific CD4+ T cell responses measured by IGRA, flow cytometry and TST have consistently been shown to be reduced in HIV infection[10,41,42]. The differential impact of HIV infection on responses to these two antigens is intriguing, particularly considering that the ESAT-6 and CFP-10 antigens are likely to be part of the antigen mix of PPD, although at much lower concentrations. Depletion of PPD-specific CD4+ T cells has been associated with increased IL-2 expression, a less mature phenotype and reduced production of MIP-1β[31]. We found no differences however, in the pattern of cytokine expression between PPD- and EC-specific CD4+ T cells.

We also found a specific loss of CD4+ responses in HIV-infected individuals to Rv2031c, a M.tb DosR regulon-encoded antigen. HIV-infected individuals with LTBI also show impairment of responses to heparin-binding haemagglutinin (HBHA), another antigen associated with LTBI, and supporting a role for responses to antigens that are preferentially expressed during conditions of M.tb latency[43,44].

Analysis of cytokine profiles in HIV-uninfected controls provided insight into antigen-specific responses prior to HIV infection. Most notable was the increased IL-17 expression of Rv2031c-specific CD4+ T cell responses and the relatively decreased IL-2 expression of CMV-specific T cell responses in comparison with EC and PPD-specific responses. The importance of Th17 CD4 T cell responses to antigens associated with LTBI is supported by a study by Dreesman et al which investigated responses to HBHA in children exposed to TB[45]. We also showed that the IL-17 expressing components of all the antigen-specific CD4+ T cell responses were preferentially depleted in HIV-infected individuals. Although Th17 CD4+ T cells are known to be depleted in HIV infection, this has not been confirmed for M.tb-specific Th17 CD4+ T cell responses [46,47]. This preferential depletion has been linked to the increased expression of CD4, CXCR4 and CCR6, resulting in enhanced permissiveness to HIV infection [48]. The increased proportion of IL-17-expressing CD4+ T cells within the Rv2031c CD4+ T cell response may therefore contribute to preferential depletion of this antigen-specific CD4+ T cell population. Further evidence from a population of HIV-uninfected adolescents with LTBI in South Africa correlated suppression of Th17 responses with progression to active TB, highlighting the importance of Th17 CD4 T cell responses in LTBI control [49].

Although the importance of polyfunctional CD4+ T cells in control of viral infections has been described, this has not been confirmed in TB. Whilst we show that polyfunctional M.tb-specific CD4+ T cell responses are present in our study participants, we found no evidence to suggest that this was impacted by HIV infection.

The effect of HIV infection on M.tb-specific CD8+ T cell responses has not been examined in detail. Sutherland et al. showed that CD8+ T cell responses were maintained in advanced HIV infection but lost polyfunctionality [24]. We show that CD8+ T cell cytokine responses to PPD and Rv2031c are significantly impaired in HIV-infected individuals, possibly indicating that loss of M.tb-specific CD8+ T cell cytokine responses may contribute to susceptibility to TB disease in HIV infection. We further determined that impairments of antigen-specific CD8+ T cell responses to some M.tb antigens were associated with loss of antigen-specific CD4+ T cell help. Whilst effective stimulation of CD8+ T cell cytokine responses was achieved in this study, the full breadth of these responses may not have been identified due to the use of protein antigens.

Our study is limited by it cross-sectional nature. Ideally a longitudinal analysis of M.tb-specific T cell perturbations in acute and chronic HIV infection would provide further clarity. In conclusion, our data show that HIV infection is associated with a complex pattern of M.tb-specific T cell impairment affecting both CD4+ and CD8+ T cell populations, and suggest that depletion of antigen-specific Th1 and Th17 CD4+ T cells may directly lead to TB disease susceptibility compounded by loss of corresponding CD8+ T cell function.

Summary.

HIV infection is associated with a complex pattern of M. tuberculosis-specific T cell impairment. In South African participants, we report an HIV-associated depletion of TB-specific Th1 and Th17 CD4+ T cells compounded by loss of corresponding CD8+ T cell function.