Th22 cells are a major contributor to the mycobacterial CD4+ T cell response and are depleted during HIV infection

Abstract

HIV-1 infection substantially increases the risk of developing tuberculosis (TB). Mechanisms such as defects in the Th1 response to Mycobacterium tuberculosis (M.tb) in HIV-infected persons have been widely reported. However, Th1-independent mechanisms also contribute to protection against TB. To identify a broader spectrum of defects in TB immunity during HIV infection, we examined IL-17A and IL-22 production in response to mycobacterial antigens in peripheral blood of persons with latent TB infection (LTBI) and HIV co-infection. Upon stimulating with mycobacterial antigens, we observed a distinct CD4+ T helper lineage producing IL-22 in the absence of IL-17A and IFN-γ. Mycobacteria-specific Th22 cells were present at high frequencies in blood and contributed up to 50% to the CD4+ T cell response to mycobacteria, comparable in magnitude to the IFN-γ Th1 response (median 0.91% and 0.55%, respectively). Phenotypic characterization of Th22 cells revealed that their memory differentiation was similar to M.tb-specific Th1 cells (i.e. predominantly early-differentiated CD45RO+CD27+ phenotype). Moreover, CCR6 and CXCR3 expression profiles of Th22 cells were similar to Th17 cells, while their CCR4 and CCR10 expression patterns displayed an intermediate phenotype between Th1 and Th17 cells. Strikingly, mycobacterial IL-22 responses were three-fold lower in HIV-infected persons compared to uninfected persons, and the magnitude of responses correlated inversely with HIV viral load. These data provide important insights into mycobacteria-specific T helper subsets in humans and suggest a potential role for IL-22 in protection against TB during HIV infection. Further studies are needed to fully elucidate the role of IL-22 in protective TB immunity.

INTRODUCTION

Tuberculosis (TB) is the leading cause of death from an infectious agent, claiming 1.4 million lives in 2019, with 10 million new TB cases that year (1). This considerable burden of disease, along with a host of challenges in diagnosing, treating and managing TB, emphasize its significance as a global health threat. Although TB is curable and successful treatment outcomes are typically >80%, cure is achieved less frequently with drug resistant TB (56%), and outcomes during HIV co-infection are worse (2). Importantly, cure does not lead to protection from re-infection or disease reactivation. HIV-infected persons are particularly vulnerable to developing TB, with an estimated increase in risk of 20–30 fold (3). The widespread introduction of ART has coincided with only a modest decline in TB in regions most affected by HIV (4), as TB risk still remains elevated in HIV-infected persons compared to HIV-uninfected persons, despite immune reconstitution (5).

The development of an effective TB vaccine is hampered by a lack of understanding of correlates of immune protection (6), particularly the functional and phenotypic characteristics of effector T cells that mediate control of Mycobacterium tuberculosis (M.tb), and how this immune response might be balanced by immunoregulatory T cell populations to limit inflammation and avoid pathology. The recent demonstration of the first candidate TB vaccine capable of protecting adults from pulmonary TB with an efficacy of 54% (7), provides the field with an opportunity to define correlates of vaccine protection, and has the potential to uncover unique insights into immunological control of TB.

TB and HIV co-infection presents us with a further prospect to improve our understanding of the mechanisms of immune control of M.tb, by identifying how HIV renders the immune response to M.tb defective, leading to increased risk of TB disease. CD4+ T cells and specifically the Th1/IFN-γ response to M.tb are critical for protective immunity to TB (8). Most studies of HIV-TB co-infection focus on Th1 immunity, and have demonstrated depletion or dysfunction of M.tb-specific Th1 responses in both blood (9–12) and the airways (13–15) during HIV infection.

However, there is evidence of a role for IFN-γ-independent mechanisms in immune control of TB (16) that may also contribute to, or synergize with, Th1 responses to TB. Recently, we characterized the profile of Th subsets specific for M.tb using lineage-defining transcription factors, revealing the broad spectrum of Th subsets involved in mycobacterial immunity, demonstrating that the inflammatory environment associated with HIV infection skewed these profiles (17). Th17 cells form part of this spectrum of M.tb-specific Th responses, and are believed to play an important role in immune protection from TB (18). Suppression of Th17-related genes was recently shown to be associated with progression to TB disease in M.tb-infected adolescents (19). In line with this, M.tb-specific IL-17A-producing CD4+ T cells were significantly depleted in HIV-infected individuals from a TB-endemic area, compared to HIV-uninfected individuals (20).

Whilst IL-17 responses in M.tb immunity have been relatively well-studied (21–26), IL-22 responses have been overlooked in part due to their classification as a Th17 cytokine from studies in mice (27). In humans, however, IL-22 is produced by a distinct subset of CD4+ T cells (28–30), termed “Th22 cells”. IL-22 is a member of the IL-10 family of cytokines, and functions mainly to protect tissues from inflammation and infection, through stimulating proliferation and repair, and the production of antimicrobial peptides (31). Until recently, IL-22 was thought to be dispensable for control of M.tb, since deficiency or neutralization of IL-22 in mice had no effect control of M.tb using lab strains H37Rv and Erdman (32–35). However, the recent observation that IL-22 deficient mice infected with a clinical strain of M.tb (HN878) had an impaired ability to control M.tb, leading to increased bacterial burden and greater dissemination of infection (36), has triggered renewed interest in IL-22 and its role in TB control.

Given the paucity of data on M.tb-specific IL-22 CD4+ responses, and the knowledge that HIV infection results in the preferential targeting and depletion of Th22 cells (37), we sought to characterize HIV-induced defects in adaptive immunity to M.tb, with a focus on Th22 cells. Our findings highlight the large contribution IL-22 makes to the human CD4+ T cell response to TB (equivalent in magnitude to the IFN-γ response), with M.tb-specific Th22 cells being entirely distinct from Th1 and Th17 cells. Moreover, we show for the first time that M.tb-specific Th22 cells are depleted during HIV co-infection to a similar extent as Th1 responses. These findings emphasize the potential importance of this understudied CD4+ Th subset in TB immunity, and suggest that the loss of M.tb-specific Th22 cells may contribute to the increased risk of TB during HIV infection.

MATERIALS AND METHODS

Study Participants

Volunteers were recruited from Cape Town, South Africa, and fell within the following groups: ART naive HIV-seropositive persons with CD4 counts >400 cells/mm³ (n=25; median age 31; 96% female) and HIV-seronegative persons (n=25; median age 23; 60% female). HIV RNA levels were determined using an Abbott m2000 RealTime HIV-1 assay and blood CD4 counts by the Flow-CARE^™ PLG CD4 test. All volunteers were TB sensitized based on a positive IFN-γ release assay (IGRA; Quantiferon, Cellestis), and active TB was excluded, based on symptoms and radiological evidence.

Healthy donors were recruited from the University of Cape Town, South Africa. Participants were >18 years of age, weighed >55 kg, did not have any chronic disease, did not use immunosuppressive medication and were not pregnant or lactating. These studies were approved by the Research Ethics Committee of the University of Cape Town (158/2010, 279/2012). All participants provided written, informed consent.

Whole blood stimulation assays

Venous blood was collected and processed within 4 hours. Whole blood stimulation was performed as previously described (38) with the following antigens: Bacillus Calmette-Guerin (BCG; MOI of 4; SSI), Purified Protein Derivative (PPD) of M. tuberculosis (20μg/ml; Statens Serum Institute), ESAT-6 and CFP-10 peptide pools (4μg/ml), M. tuberculosis whole cell lysate (10μg/ml; BEI Resources) or PMA and Ionomycin (0.01μg/ml and 1μg/ml, respectively, Sigma), in the presence of anti-CD28 and anti-CD49d (1μg /ml each). Optimal antigen concentrations were determined experimentally. Unstimulated cells were incubated with co-stimulatory antibodies only. Brefeldin A (BFA, 10μg/ml; Sigma) was added 7 hours after the onset of stimulation, and 5 hours after BFA addition, cells were either stained immediately, or red blood cells were lysed, the cell pellet stained with a violet viability dye, ViViD (Molecular Probes), fixed with FACS Lyse (BD Biosciences) and cryopreserved in 10% DMSO in FCS. For peripheral blood mononuclear cells (PBMC) and whole blood co-culture experiments, PBMC were isolated by density gradient centrifugation. Freshly isolated PBMC were stained with the fluorescent dye Oregon Green (0.5 mg/ml; Invitrogen) for 4 min at room temperature. After washing in PBS, 2×10⁶ cells were combined with 500μl blood from the same donor and stimulated as described above.

Antibody Staining and Flow Cytometry

Cryopreserved or freshly stimulated whole blood was stained as previously described (15). For intracellular markers, cells were permeabilized with Perm/Wash buffer (BD Biosciences) and then stained intracellularly. The following antibodies were used for characterizing cytokine expression and memory phenotype: CD3 APC-H7 (SK7; BD Biosciences), CD4 PE-Cy5.5 (S3.5), CD8 QDot705 (3B5; both Invitrogen), CD45RO ECD (UCLH1), CD27 PE-Cy5 (1A4CD27; both Beckman Coulter), IFN-γ Alexa700 (B27), IL-17A Alexa488 (N49–653; both BD Biosciences) and IL-22 PE (22URTI; e-Bioscience). To characterise chemokine receptor expression on Th1, Th17 and Th22 cells, whole blood was stained directly after stimulation, without red blood cell lysis, with CD14-Pacific blue, CCR4-BV510 (L291H4; Biolegend), CCR6-BV605 (11A9); CCR10-PE (1B5) and CXCR3-PECy7 (1C6/CXCR3; all BD Biosciences). The cells were then permeabilized and stained intracellularly with CD3-APC-H7 (SK7), IFN-γ-Alexafluor 700 (B27; both BD Biosciences), IL-17A-FITC (BL-168; Biolegend) and IL-22-eFluor450 (22URTI; e-Bioscience). An additional panel included KLRG1 PE-vio770 (REA261; Miltenyi Biotec) and CD26 FITC (M-A261; BD Biosciences). Cells were acquired on a BD Fortessa using FACSDiva software and data analysed using FlowJo (TreeStar) and Pestle and Spice (39). Fluorescent minus one (FMO) controls were run to optimize each antibody panel and ensure that there was no spillover from other channels and to set gates for positive populations (Supplemental Figure S1A). The gating strategy is indicated in Supplemental Figure S1B. Cells were gated first on time, followed by singlets, lymphocytes, live CD3+, total CD4+ IFN-γ+ cells and then CD4+ CD8- cells. After that, cytokine and memory gates were drawn. A similar gating strategy was employed for the analysis of chemokine receptor expression on cytokine positive populations. A positive cytokine response was defined as twice background (unstimulated sample) and a net response >0.025%, and all data are reported after background subtraction. The backgrounds were low overall, namely for IFN-γ: median 0.007% (IQR, 0.002–0.01%), IL-22: median 0.02% (IQR, 0.014–0.036%) and IL-17A: median 0.012% (IQR, 0.007–0.027%). A minimum of 30 cytokine-positive events was required for memory or chemokine receptor phenotyping.

Plasma cytokine measurement

Whole blood stimulation was performed as described above (but without the addition of BFA), and plasma was harvested 12 and 24 hours after stimulation with M.tb lysate. Plasma was assayed using a commercial sandwich ELISA (Invitrogen) to quantify soluble IFN-γ, IL-22 and IL-17A, according to manufacturer’s instructions. The sensitivity of the kits ranged from 4 pg/ml (IFN-γ and IL-17A) to 8 pg/ml (IL-22).

Statistical Analysis

Statistical analyses were performed using Prism 7 (GraphPad). Non-parametric tests (Mann-Whitney U test, Wilcoxon matched pairs test and Spearman Rank test) were used for all comparisons. Kruskal-Wallis and Friedman tests with Dunn’s post-test was used for multiple comparisons. A p value of <0.05 was considered significant.

RESULTS

IL-22 responses are a major component of the CD4+ mycobacterial response

We examined CD4+ T cell cytokine profiles in response to a range of mycobacterial antigens in 25 healthy, HIV-uninfected persons sensitized by M. tuberculosis (M.tb IGRA+; Table 1). Figure 1A shows representative flow cytometry plots of IFN-γ, IL-22 and IL-17A CD4+ responses to M. bovis BCG, M.tb PPD and a pool of ESAT-6 and CFP-10 peptides from M.tb. As expected, CD4+ T cell IFN-γ responses to BCG were detected in all donors (median 0.55%, IQR: 0.28–1.46%; Figure 1B). Remarkably, IL-22 accounted for the greatest proportion of the CD4+ response to BCG (median 0.91%, IQR: 0.52–1.24%). In fact, the frequency of IL-22+ cells was greater than IFN-γ in 75% of participants. IL-17A CD4+ responses to BCG were significantly lower (median 0.11%, IQR: 0.06–1.66%) than both IFN-γ (p=0.007) and IL-22 (p=0.0008). Stimulation with M.tb PPD led to the detection of a similar IFN-γ response as BCG (median 0.74%), with comparatively lower frequencies of PPD-specific IL-22+ CD4+ T cells (median 0.21%; p=0.02) and IL-17A+ cells (median 0%; p<0.0001) (Figure 1B). The ESAT-6/CFP-10 response was dominated by IFN-γ (median 0.07%), with low to undetectable IL-17A and IL-22 responses (medians of 0%). Taken together, these data demonstrate that different mycobacterial antigen preparations result in detection of different CD4+ T cell cytokine profiles. Of note, IL-22 made a substantial contribution to the anti-mycobacterial CD4+ response, with responses equivalent to or greater than the IFN-γ response to BCG.

Table 1:

Characteristics of study participants

HIV-uninfected (n=25)			HIV-infected (n=25)
PID	CD4 count (cells/mm3)		PID	CD4 count (cells/mm3)	Viral Load (RNA copies/ml)
1032	ND		1086	1449	4250
1035	1459		1151	988	1848
1052	1412		1075	965	5922
1031	1169		1150	894	311
1023	1120		1006	802	4141
1070	1028		1039	790	4521
1024	939		1080	774	14100
1025	915		1152	749	<40
1057	871		1154	714	2954
1058	866		1018	681	4614
1054	832		1143	656	618
1094	827		1073	632	12274
1028	814		1084	619	9192
1033	813		1134	599	6383
1011	801		1079	591	32485
1095	760		1153	571	9697
1038	743		1137	560	18797
1072	741		1141	552	9826
1047	680		1076	543	908
1061	674		1045	522	59125
1015	659		1129	510	4559
1049	655		1074	478	10093
1001	631		1142	441	544849
1066	621		1126	433	32994
1010	580		1020	406	31145
Median	813			619	6383
IQR	675.5–933			532.5–782	3548–16449

Figure 1: CD4+ T cell cytokine responses to mycobacterial antigens in latent TB infection.

(A) Representative flow cytometry plots of the production of IFN-γ, IL-22 and IL-17A from CD4+ T cells after stimulation with M. bovis BCG, M.tb PPD and ESAT-6/CFP-10 peptides, in one study participant. UNS corresponds to the unstimulated control. The frequency of cytokine-producing cells is shown as a percentage of the total CD4+ T cell population, after gating on live, CD3+ lymphocytes. (B) Individual IFN-γ (blue), IL-22 (red) or IL-17A (green) responses to BCG, PPD or ESAT-6/CFP-10 in individuals with latent TB infection (LTBI; n=25). The frequency of cytokine-producing cells is shown as a percentage of the total CD4+ T cell population, after gating on live, CD3+ lymphocytes. (C) The relationship between the frequency of CD4+ T cells producing IFN-γ and IL-22 in response to BCG (n=24). (D) Populations of CD4+ T cells producing different combinations of IFN-γ, IL-22 and IL-17A in response to BCG. The pie charts indicate the proportion of cytokine combinations that makes up the BCG response. Each slice of the pie represents a specific subset of cells, defined by a combination of cytokines shown by the color at the bottom of the graphs. Data are shown as box and whisker (interquartile range) plots and horizontal bars represent the median. Each dot represents one individual. Statistical comparisons were performed using Friedman’s test with Dunn’s multiple comparison, and a non-parametric Spearman rank test for the correlation. **p≤0.01, ***p≤0.001, ****p≤0.0001.

We next focused on the high magnitude IL-22 response detected to BCG, to further characterize IL-22 CD4+ responses and their relationship with IFN-γ and IL-17A. There was a highly significant positive correlation between IFN-γ and IL-22 responses to BCG (p<0.0001, r=0.830; Figure 1C). The frequency of IL-17A+ CD4+ T cells also correlated with both IFN-γ and IL-22 production (p=0.039, r=0.424 and p=0.005, r=0.559, respectively; data not shown). Given these associations between IFN-γ, IL-22 and IL-17A, we examined the co-expression patterns of the cytokines following BCG stimulation (Figure 1D). The majority of BCG-responding CD4+ T cells produced only IL-22 (median 47%; IQR: 36.6–59.6), whilst CD4+ cells secreting IFN-γ-alone made up a median of 37% (IQR: 27.1–47.4). There was minimal co-expression of IL-22 with both IL-17A (median 0.5%) and with IFN-γ (median 6.4%). When examining all CD4+ T cells producing IL-22, a median of 78% produced IL-22 alone (IQR: 71.1–89.2%), while 14% and 1.5% co-expressed IFN-γ or IL-17A, respectively (data not shown). We also investigated IL-22 production in combination with other cytokines and found low or negligible co-expression with TNF-α, IL-2 and IL-21 (medians 0.3%, 0.5% and 3%, respectively, data not shown). Our data reveal that the large proportion of BCG-specific IL-22 was produced predominantly by CD4+ T cells secreting IL-22 in the absence of either IL-17A or IFN-γ, consistent with being a distinct ‘Th22’ lineage (28–30).

We next sought to determine whether we could detect soluble IL-22 in response to mycobacterial stimulation. For these studies, we stimulated whole blood with M.tb whole cell lysate. To ensure that we were measuring similar cytokine responses to BCG, we compared IFN-γ, IL-22 and IL-17A induced by each antigen and found highly comparable frequencies of CD4+ T cell responses in the same donors, using optimal antigen concentrations determined by titration (Supplemental Figure S2). Then, using a direct ELISA, we quantified soluble cytokine secretion in plasma of whole blood stimulated with M.tb lysate for 12 and 24 hours in 8 healthy donors with LTBI (Figure 2). IFN-γ was readily detectable in stimulated plasma at 12 hours (median 722pg/ml; range: 542–1640pg/ml). Of note, soluble IL-22 was also abundantly detectable in plasma, at a median of 3232pg/ml (range 2600–7489pg/ml), five-fold higher than IFN-γ concentrations. Levels of both IFN-γ and IL-22 more than doubled at 24 hours compared to 12 hours (p=0.004). In contrast, low concentrations of IL-17 were detectable at 12 hours and decreased significantly by 24 hours of stimulation (p=0.02). These data support the cytokine production demonstrated by flow cytometry, indicating robust secretion of IL-22 in response to mycobacteria.

Figure 2: Cytokine secretion in M.tb-stimulated whole blood.

The concentration of soluble IFN-γ (white), IL-22 (black) and IL-17A (gray) was measured by ELISA (pg/ml) in plasma from whole blood stimulated with M.tb lysate in individuals with LTBI (n=9). Plasma was harvested at 12 and 24 hours after stimulation. Horizontal bars represent the median. Statistical comparisons were performed using a Wilcoxon test. *p≤0.05, **p≤0.01.

IL-22 detection differs between whole blood and PBMC

The finding of high frequencies of Th22 cells specific for mycobacteria in LTBI contributing so substantially to the overall T cell response that we report here was surprising, given that Th1/IFN-γ responses are widely reported to be the dominant response to mycobacteria. PBMC are most often used in studies analyzing human cytokine expression, so we compared our assay format, direct stimulation of whole blood with mycobacterial antigens, with cytokine responses from stimulated PBMC. Figure 3A (bottom left panel) shows representative flow cytometry plots of the CD4+ T cell IFN-γ and IL-22 responses to M.tb in whole blood and PBMC. We noted that whilst IFN-γ production by CD4+ T cells was readily detected in both assays, IL-22 was barely detectable, and this was confirmed in 8 healthy donors with LTBI (Figure 3B). Based on these poor CD4+ IL-22 responses we observed in PBMC, we investigated whether CD4+ T cells from PBMC retained the ability to produce IL-22 under optimal conditions. To determine whether the CD4+ IL-22 response from PBMC could be recovered, a mixed reaction of autologous PBMC added to whole blood was stimulated (Figure 3A). Strikingly, the IL-22 response was detectable in PBMC from the mixed reaction to such an extent that it was similar to IL-22 from both whole blood and whole blood from the mixed culture (medians 0.65%, 0.73% and 0.98%, respectively). The PBMC IFN-γ response was also higher in the mixed reaction compared to PBMC alone, and comparable to whole blood from the co-culture (medians 0.72% and 0.98%). Our results demonstrate that CD4+ T cells from PBMC retain the ability to produce IL-22 and this response can be restored under specific culture conditions.

Figure 3: Detection of the CD4+ T cell IL-22 responses in PBMC.

(A) Schematic of mixed stimulation assay with representative flow cytometry data. PBMC were stained with a fluorescent dye, Oregon Green (Or Green), and added to autologous whole blood, and subsequently stimulated with M.tb lysate. Representative flow cytometry plots show the production of IFN-γ and IL-22 from CD4+ T cells in fresh PBMC (PBMC+ M.tb) alone, whole blood (WB + M.tb) alone, and the mixed stimulation assay (WB in mix, PBMC in mix), after stimulation with M.tb lysate in one participant. (B) Comparison of the frequency of CD4+ T cells producing IL-22 and IFN-γ in each experimental condition in the same individuals (n=8). Horizontal bars represent the median. Statistical comparisons were performed using Friedman’s test with Dunn’s multiple comparison. **p≤0.01, ***p≤0.001.

Phenotypic characteristics of mycobacteria-specific IL-22-producing CD4+ T cells

In order to characterize the Th22 subset in more detail, we determined the memory differentiation profile of mycobacteria-specific Th22 cells (i.e. those producing IL-22 alone) compared to cells producing only IFN-γ or IL-17A. Figure 4A shows representative flow cytometry plots of CD45RO and CD27 expression on total CD4+ cells with overlays of BCG-specific cytokine-producing CD4+ T cells (IFN-γ, IL-22 or IL-17 alone). The memory profile of BCG-specific CD4+ T cells was comparable, regardless of their cytokine secretion profile, with approximately 79% having an early differentiated phenotype (ED: CD45RO+CD27+, comprising central and transitional memory cells). Of the remaining cells, a median of ~17% were late differentiated (LD: CD45RO+CD27-, comprising effector memory and intermediate cells), with few terminally differentiated (TD: CD45RO-CD27-; ~0.3%) or naïve-like (CD45RO-CD27+; ~2 %) cells (Figure 4B). Thus, CD4+ T cells producing IFN-γ, IL-22 or IL-17A shared a similar memory differentiation phenotype.

Figure 4: Memory profiles of CD4+ T cells producing IFN-γ, IL-22 or IL-17A in response to BCG.

(A) Representative flow cytometry plots of total CD4+ memory subset distribution in one individual based on CD45RO and CD27 staining. Naïve: CD45RO-CD27+, early differentiated (ED: CD45RO+CD27+), late differentiated (LD: CD45RO+CD27-) and terminally differentiated (TD: CD45RO-CD27-). The overlays indicate the antigen specific CD4+ T cells producing IFN-γ (blue), IL-22 (red) or IL-17A (green). The frequencies of each subset are indicated. (B) The memory distribution of cells producing IFN-γ (blue), IL-22 (red) or IL-17A (green) in response to BCG (n=20, 25 and 7, respectively). Only individuals with a positive cytokine response and more than 30 cytokine events were included in the phenotyping. Each dot represents one individual. Data are shown as box and whisker (interquartile range) plots and horizontal bars represent the median. Statistical comparisons were performed using Kruskal-Wallis and Dunn’s multiple comparison test.

To further characterize the phenotype of the different cytokine-producing subsets, we examined chemokine receptor expression profiles on CD4+ cells producing IFN-γ, IL-22 or IL-17A. For these studies, we stimulated whole blood with M.tb whole cell lysate. Figure 5A shows representative flow cytometry plots of M.tb-specific CD4+ T cell production of IFN-γ, IL-22 and IL-17A overlaid onto chemokine receptor expression profiles (CXCR3, CCR6, CCR4 and CCR10). Whilst a majority of IFN-γ-producing cells expressed CXCR3 (median 61.5%), a sizable fraction also expressed CCR6 (median 54.5%), with a low proportion expressing CCR4 (median 4.1%) and negligible CCR10 (median 0.5%; Figure 5B). In contrast, CD4+ cells producing IL-22 were almost all CCR6 positive (median 94.7%), and compared to cells producing IFN-γ, significantly fewer expressed CXCR3 (median 27.7%), and significantly more expressed CCR4 and CCR10 (median 23.3% and 2.7%, respectively). Th17 cells (IL-17A+) shared comparable expression profiles for CCR6 and CXCR3 (medians 96.1% and 17.2%, respectively), but a higher proportion expressed CCR4 (median 50.8%) and CCR10 (median 5.8%) compared to Th22 cells. Of note, cells co-producing IFN-γ and IL-22 had a similarly high expression of CCR6 as Th22 and Th17 cells, but were otherwise intermediate between IFN-γ+ and Th22 for the remaining chemokine receptors (data not shown). These findings demonstrate distinct patterns of chemokine receptor expression on different cytokine-producing subsets. These data are consistent with previous descriptions (28, 30), but also serve to highlight the substantial overlap in chemokine receptor expression between T helper subsets producing distinct cytokines.

Figure 5: Chemokine receptor expression of CD4+ T cells producing IFN-γ, IL-22 or IL-17A in response to M.tb whole cell lysate.

(A) Representative flow cytometry plots of the expression of CCR6, CCR4, CXCR3 and CCR10 on total CD4+ T cells in one individual. The overlays indicate the antigen specific CD4+ T cells producing IFN-γ (blue), IL-22 (red) or IL-17A (green). The frequencies of each subset are indicated. (B) The chemokine receptor distribution of cells producing IFN-γ (blue), IL-22 (red) or IL-17A (green) in response to M.tb lysate (n=19, 19 and 11, respectively). Only individuals with a positive cytokine response and more than 30 cytokine events were included in the phenotyping. Each dot represents one individual. Data are shown as box and whisker (interquartile range) plots and horizontal bars represent the median. Statistical comparisons were performed using Kruskal-Wallis and Dunn’s multiple comparison test. *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001

We also investigated the homing potential of M.tb-specific CD4+ Th subsets using KLRG1 and CD26. Killer cell lectin-like receptor G1 (KLRG1)-expressing cells appear to be retained within lung blood vasculature, while KLRG1⁻ cells migrate to the lung parenchyma (40). Dipeptidyl peptidase IV (CD26) is involved in enzymatic chemokine modification that enhances T cell migration (41, 42). Expression of these markers was significantly different between Th subsets (Supplementary Figure S3A). Th22 cells were characterized by a near absence of KLRG1 expression compared to Th1 and Th17 cells. In contrast, 50% of Th22 cells expressed CD26, compared to a median of 34% of Th1 cells and 11% of Th17 cells (Supplemental Figure S3B). These data suggest that M.tb-specific Th22 are endowed with a distinct homing potential compare to Th1 and Th17 cells.

The effect of HIV infection on the Th22 response to mycobacteria

Th1 responses to M.tb are impaired or reduced during HIV infection (3). However, little is known about the effect of HIV co-infection on the Th22 response to M.tb. Hence, we examined IFN-γ, IL-22 and IL-17A responses to BCG and PPD in 25 HIV-infected individuals with a median CD4 count of 619 cells/mm³ (IQR: 532.5–782) and a median plasma viral load of 6.38 ×10³ copies/ml (IQR: 3.55–16.45 ×10³; Table 1). Consistent with previous reports, the frequency of M.tb-specific CD4+ T cells producing IFN-γ was significantly lower in HIV-infected participants compared to uninfected participants in response to BCG (p=0.0004, medians 0.12% and 0.55%, respectively; Figure 6A). Notably, the IL-22 response to BCG was also lower in HIV-infected individuals, to a similar degree as the IFN-γ response (p=0.0005; medians 0.28% and 0.91%, respectively). Additionally, IL-17A responses were also significantly lower in HIV-infected individuals in response to BCG compared to the HIV-uninfected group (p<0.0001, medians 0% and 0.11%, respectively). After adjusting for CD4 count, these differences became even more evident (Figure 6B), despite the relatively well-preserved CD4+ T cell numbers in our HIV-infected cohort. HIV-infected participants had 8-fold (p<0.0001) and 3-fold (p=0.0003) fewer CD4+ T cells producing IFN-γ or IL-22, respectively, compared to uninfected participants. There were also fewer cells producing IL-17A in HIV-infected individuals (median 0; p<0.0001). Similar results were obtained for IFN-γ and IL-22 in response to PPD (Supplemental Figure S4A and B). Overall, HIV-infected participants had lower M.tb-specific IFN-γ, IL-22 and IL-17A responses. Whilst the decrease in M.tb-specific IFN-γ and IL-17A responses during HIV infection has been reported, we report here a striking loss of M.tb-specific CD4+ T cells producing IL-22.

Figure 6: CD4+ T cell responses to BCG in HIV-infected and uninfected individuals and their relationship with CD4 count and HIV viral load.

(A) The individual IFN-γ, IL-22 or IL-17A responses in HIV uninfected or infected individuals in response to BCG (n=24 in each group). (B) The cytokine frequency adjusted for CD4 count in HIV-infected and HIV-uninfected individuals in response to BCG. (C) The median fluorescent intensity (MFI) of IFN-γ, (n=24 and n=15 for HIV-uninfected and infected, respectively), IL-22 (n=23 and n=20 for HIV-uninfected and infected, respectively) and IL-17A (n=22 and n=8 for HIV-uninfected and infected, respectively) in response to BCG. For each cytokine, MFI was only graphed for individuals with positive cytokine responses. HIV-uninfected participants are shown with open circles and HIV-uninfected individuals with closed circles. Each dot represents one individual. Horizontal bars represent the median. (D) The association between IFN-γ (white), IL-22 (black) or IL-17A (gray) responses to BCG and CD4 count or (E) viral load (n=24). The dotted line indicates linear regression for statistically significant correlations. Statistical comparisons were performed using a non-parametric Mann Whitney test, and a non-parametric Spearman rank test for the correlations. *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001.

To further investigate the impact of HIV on BCG-specific Th22 responses, we measured the amount of IL-22 produced per cell, using median fluorescent intensity (MFI). The MFI of IL-22 was significantly lower in HIV-infected individuals compared to uninfected individuals (p<0.0001; medians 4169 and 6215, respectively; Figure 6C), whereas no differences in the MFI of IFN-γ and IL-17A was observed. This suggests that HIV may have a unique effect on Th22 cells in response to BCG. However, we found no differences in the MFI of any cytokines produced in response to PPD (Supplemental Figure S4C).

To investigate whether the lower cytokine responses to mycobacterial antigens in HIV-infected individuals related to clinical parameters, the association between IFN-γ, IL-22 and IL-17A responses and CD4 count or plasma viral load was examined. We observed a significant positive correlation between both the IFN-γ and IL-22 response to BCG and CD4 count (p=0.04, r=0.43; and p=0.004, r=0.57, respectively; Figure 6D). Likewise, in response to PPD, IFN-γ (p=0.03, r=0.45) and IL-22 (p=0.004, r=0.57) correlated directly with CD4 count (data not shown). This suggests that the decrease in these responses could be a consequence of overall CD4+ T cell depletion, despite the relatively narrow CD4 count range and modest CD4 decreases in our study (84% of participants had CD4 counts >500 cells/mm³). No association between the frequency of IL-17A and CD4 count was observed for either BCG (Figure 6D) or PPD (data not shown). Finally, there were no significant associations between plasma viral load and IFN-γ or Th17 responses to BCG; Figure 6E) or any cytokine in response to PPD (data not shown). However, the frequency of Th22 cells responding to BCG was significantly inversely correlated with plasma viral load (p=0.006, r=−0.54, Figure 6E, middle panel).

Overall, we demonstrate the detrimental effect of HIV infection on CD4+ T helper subsets in response to mycobacteria. In particular, the Th22 subset exhibited both a decrease in the magnitude of the response to mycobacteria, and a defect in IL-22 production on a per cell basis. Furthermore, unlike the other cytokine-producing subsets examined (Th1 and Th17), the frequency of Th22 cells correlated inversely with HIV viral load, suggesting a direct relationship between HIV infection and the loss of Th22 cells specific for mycobacteria.

DISCUSSION

Th1/IFN-γ responses are needed for an effective response to TB (8), however a range of immune mechanisms beyond Th1 immunity may also contribute to protection from TB (6). Since HIV-infected individuals are considerably more susceptible to TB disease (3), key components required for effective immune control of M.tb are likely to be defective in these individuals, and we sought to identify these. In addition to IFN-γ/Th1 immunity, this study examined IL-17A and IL-22 responses to mycobacteria in M.tb-sensitized, HIV-infected and uninfected individuals. Consistent with previous studies, we identified distinct populations of CD4+ T cells expressing IFN-γ, IL-17A or IL-22 in response to mycobacterial antigens (43, 44). The IL-22 response was unexpectedly abundant, contributing up to 50% of the mycobacterial response measured using these three cytokines, and the source was a distinct subset of CD4+ T cells producing IL-22 alone. Importantly, IL-22 response was impaired in HIV-infected individuals in both magnitude and function, suggesting that depletion of this subset may contribute to TB risk.

IL-22 has classically been characterized as a Th17-related cytokine, since in mice it is co-secreted with IL-17A and has overlapping functions with IL-17A (27). However, IL-22 is a member of the IL-10 family (45), and in humans, IL-22 is not co-expressed with IL-17A (28–30). Consequently, ‘Th22’ cells were proposed as a novel CD4+ T helper cell lineage in humans, with shared but distinct features and functions compared to Th17 cells.

The role for IL-17 and Th17 responses in TB is well appreciated in mouse models (21, 24–26, 46, 47), with accumulating evidence from non-human primates (48, 49). In humans, genetic mutations and polymorphisms that suppress Th17 function were associated with the development of TB disease in humans (50, 51). Consequently, lower Th17 responses were found in individuals with active TB and in those who went on to develop active disease (19, 43). Importantly, IL-17 producing tissue-resident cells in the human lung enhanced immune control of M.tb (52). However, other studies report associations between IL-17 responses and TB pathogenesis (18, 53–55). More studies are clearly required to determine the conditions in which Th17 cells confer protection. Here, we found that IL-17A responses made only a modest contribution to the total mycobacterial response in M.tb-exposed individuals, consistent with previous reports (20, 43). In contrast, we detected ample mycobacteria-specific IL-22 production from CD4+ T cells in the absence of IL-17A (and IFN-γ), consistent with a distinct Th22 subset and in agreement with earlier observations in LTBI and TB disease (43, 56). Of the CD4+ T cells producing IL-22, we found only a small proportion (10%) also expressed IL-17. Thus, our data, along with previous reports, emphasize that Th22 cells should be considered a distinct contributor to the human mycobacterial response in their own right.

Phenotypic profiling demonstrated that whilst the memory differentiation phenotype of Th22 cells was similar to that of Th1 and Th17 cells, the bulk of Th22 cells expressed CCR6, with expression frequencies of CXCR3, CCR4 and CCR10 intermediate between Th1 and Th17 cells, somewhat consistent with published reports (28, 30). M.tb-specific Th22 cells were also characterized by higher CD26 and absent KLRG1 expression compared to both Th1 and Th17 cells. Altogether, these characteristics emphasize the shared and unique features of mycobacteria-specific Th22 cells relative to Th1 and Th17 cells, which may relate to distinct homing capabilities.

The previously unappreciated, sizeable contribution Th22 cells make to the mycobacterial response prompts the question of whether Th22 responses play a role in protective immunity against M.tb. Previous studies demonstrated that deficiency or neutralization of IL-22 in mice did not affect control of laboratory strains of M.tb (H37Rv and Erdman) (32–35). However, renewed interest in IL-22 has been garnered since the observation that IL-22 deficient mice infected with a hypervirulent clinical strain of M.tb (HN878) have an impaired ability to control the infection, resulting in both increased bacterial burden and greater dissemination of infection (36). Additional evidence from a range of models suggest that IL-22 may indeed participate in TB immunity. IL-22 has been found at sites of TB disease; soluble IL-22 and IL-22 transcripts were elevated in the airways, lung tissue, granuloma, and in pleural and pericardial effusions during TB disease (43, 56–60). Along with IFN-γ, IL-22 was one of the strongest genes upregulated in bovine TB (61), and gene expression signatures revealed that IFN-γ and IL-22 were the dominant correlates of protection from bovine TB in blood in BCG-vaccinated cattle (62). Human genetic studies demonstrated the association between increased susceptibility to TB and a single nucleotide polymorphism in the IL-22 promoter that decreased IL-22 expression (60).

If IL-22 is involved in TB immunity, how might it mediate a protective function? IL-22 functions as a key regulator of tissue-specific antimicrobial immunity (31). The receptor for IL-22 is a heterodimer consisting of the IL-10R2 and the IL-22R, and expression is primarily restricted to non-hematopoietic cells, particularly epithelial cells in the skin, digestive tract and respiratory tract (31). IL-22 has been shown to be essential for mediating protective immunity to a range of extracellular and intracellular bacteria, such as Klebsiella and Chlamydia in the lung and Citrobacter in the intestine (63–66). Neutralization of IL-22 led to bacterial dissemination, exacerbated pathology, and lower Th1 and Th17 responses in the lung (65). The protective role at barrier sites appears to be mediated by three distinct functions, namely; maintenance of barrier integrity by promotion of epithelial homeostasis, stimulating epithelial proliferation and preventing apoptosis, as well as enhancing mucin production and tight junction formation; inducing antimicrobial peptides such as β-defensins; and regulating chemokine secretion from epithelial cells to co-ordinate recruitment of immune cells, such as neutrophils, to inflamed tissue (27, 29, 65, 67). Indeed, Treerat and colleagues demonstrated that the TB-protective role of IL-22 resulted from the secretion of S100 and Reg3γ from epithelial cells, and induction of CCL2 that mediated macrophage recruitment to the infected lung (36). It is worth noting that several studies have independently documented IL-22R expression on M.tb-infected monocyte-derived macrophages (MDMs), as well as macrophages in TB granulomas in humans and non-human primates (36, 68, 69). Consistent with these findings, IL-22 from CD4+ T cells and innate cells, as well as recombinant IL-22, reduced mycobacterial replication in MDMs by improving phagolysosome fusion (36, 68–70). These data suggest that direct effector function for IL-22 in limiting mycobacterial growth cannot be ruled out.

HIV-infected individuals remain one of the most vulnerable populations at risk of TB (3). The early depletion of M.tb-specific Th1 responses, considered fundamental to TB immunity, has been reported during HIV infection (9, 10). Here, we investigated the relative effect of HIV on Th22 and Th17 responses to mycobacteria compared to Th1 responses. An important and novel finding from our study was that the mycobacteria-specific Th22 response was depleted during HIV infection, to a similar extent as Th1/IFN-γ responses. Mycobacterial Th17 responses, albeit low in magnitude, were also significantly lower during HIV infection. Several studies have described a global and preferential loss of Th22 and Th17 cells during HIV/SIV infection, leading to mucosal gut damage and systemic immune activation, driving HIV disease progression (37, 71–74). The CCR6+CD4+ T cell subset (within which all Th22 and Th17 cells reside) is more permissive to HIV infection and replication, and is enriched for HIV DNA (75–77). Elevated expression of HIV co-receptors CCR5 and CXCR4 has been reported on CCR6+CD4+ T cells, which could facilitate HIV entry (78). In addition, post-entry mechanisms appear to create a more permissive cellular environment for HIV replication in CCR6-expressing cells, demonstrated by specific transcriptional signatures favoring HIV replication (79–81). We report here that higher HIV plasma viral load correlates with lower frequencies of Th22 cells specific for mycobacteria, consistent with a mechanism of direct, preferential infection of Th22 cells by HIV. Overall, multiple mechanisms may contribute to the loss of Th22, Th17 and Th1 subsets specific for M.tb (82, 83), and their combined depletion may contribute to TB risk during HIV infection. Although ART leads to expansion of the CD4+ T cell compartment, mycobacteria-specific T cells may not recover proportionally (84–86). Indeed, the incomplete reconstitution of M.tb-specific Th1 cells in individuals on ART has been described by our group, with the extent of reconstitution dependent on the memory differentiation phenotype (87). Thus, given that the memory phenotype of Th1, Th17 and Th22 cells were similar in the current study, we may speculate that ART would lead to a similar, partial recovery of Th17 and Th22 subsets. Studies examining recovery of these subsets in individuals on ART are ongoing.

Another important observation from our study is that while IL-22 is readily detectable from stimulated whole blood, it is not easily detected in a traditional PBMC assay, which may partly account for its unappreciated contribution to the mycobacterial T cell response. Interestingly, the IL-22 response in PBMC could be recovered when these cells were cultured with whole blood followed by stimulation with M.tb, suggesting that CD4+ T cells capable of producing IL-22 are not lost in PBMC, but that there are factors in whole blood that are required for IL-22 production. These could be cell types responsible for antigen presentation and co-stimulation, such as granulocytes, or soluble factors in plasma. A range of innate cytokines have been shown to enhance IL-22 expression (27, 28, 83). Experiments to understand the conditions necessary for IL-22 secretion are underway.

In conclusion, our new findings add to a growing body of evidence in support of a role for IL-22 in protective immunity to TB. However, a number of questions remain unanswered. Does IL-22 contribute to protective immunity to TB, or only during infection with specific clinical strains, or during HIV infection, when multiple immunological defects manifest? Does IL-22 assume a direct effector or indirect regulatory role in immunity to TB, or both? Does the inflammatory context dictate whether IL-22 might be beneficial to the host or pathological (88)? Ultimately, will it be necessary to induce Th22 responses for a TB vaccine to be effective? Notwithstanding these gaps in our knowledge, our study highlights the substantial contribution that Th22 cells make to mycobacterial immunity, and the importance of further elucidating the role of IL-22 in the control of M.tb infection and disease.

KEY POINTS.

1. IL-22 responses to mycobacterial antigens contributed ~50% of the total response.

2. Mycobacterial Th22 responses were significantly lower in HIV-infected individuals.