Peripheral blood mononuclear cells (PBMC) and plasma were obtained from adults with or without HIV enrolled in the UCSF SCOPE cohort (27, 32). The participants selected for the current study were tested for sensitization to Mtb by tuberculin skin test (TST). All participants gave written informed consent using protocols approved by the UCSF institutional review board. Blood was collected into ethylenediaminetetraacetic acid–containing tubes for cell counts and into Acid Citrate Dextrose–containing tubes for purification of plasma and PBMC. Processed plasma was stored at −80°C, while PBMC were cryopreserved in liquid nitrogen. Demographic characteristics of the participants are in Tables 1 , 2 . We did not have samples from all the participants in the longitudinal study (cohort 2/ Table 2 ) at the start of isoniazid (INH) therapy.

Tryptophan and kynurenine concentrations were measured by liquid chromatography–mass spectrometry as previously described (27, 31).

Cryopreserved PBMC were thawed in a 37°C water bath and transferred into pre-warmed R10 media (RPMI 1640 containing L-glutamine with 10% FBS, 1% PenStrep, and 1% Hepes). Cells were centrifuged at 900g for 5 min at room temperature, supernatants discarded, and cells resuspended before adding 5 ml of warm R10 and transferred to a 6-well culture plate and rested overnight in 37°C/5% CO₂ incubator. After resting, 1 × 10⁶ live cells resuspended in 200 µL of R10 were transferred to each well of a 96-well round bottom plate, and duplicate wells were stimulated with Mtb peptide mega pool (Mtb300) (33) (2 µg/ml), cytomegalovirus (CMV) pp65 peptide (1 µg/ml) or staphylococcal enterotoxin B (SEB) positive control (1 µg/ml) in the presence of costimulating antibodies anti-CD28 (1 µg/ml) and anti-CD49d (1 µg/ml) (BD). Mtb peptide megapool represents commonly recognized epitopes from 90 Mtb antigens recognized by HLA class II–restricted CD4 T cells in diverse populations and across species (3, 34–42). CMV pp65 (PepTivator^®CMV pp65 – Milteny Biotec, San Jose, California, USA) is a peptide pool covering the complete sequence of the pp65 protein of CMV. Negative controls received no stimulation. After 2h, GolgiStop and GolgiPlug (BD Biosciences, Franklin Lakes, New Jersey, USA) were added to one well of each of the stimulation conditions, and the cells were incubated for an additional 18h.

After 20h total stimulation, cells were washed, stained with Live/Dead Fixable Blue Dead Cell stain kit (Invitrogen, Waltham, Massachusetts, USA), washed again, then surface antibody cocktail (αCD3 BV510 clone UCHT1, 1:80; αCD8 BV570 clone RPA-T8, 1:80; αCCR7 BV785 clone G043H7, 1:20; αCD95 Alexa Fluor700 clone DX2, 1:80; αCCR6 FITC clone G034E3, 1:80; αCXCR3 BV 605 clone G025H7, 1:40; αCD161 APC-Fire750 clone HP-3G10, 1:80; αCD69 BV650 clone FN50, 1:20; αCD137 BV711 clone 4B4-1, 1:20, αOX40 PE-Cyanine7 clone Ber-ACT35, 1:40; and αTRAV1-2 (Vα7.2) BV421 clone 3C10, 1:40 (all from BioLegend, San Diego, California, USA), αCD4 BUV496 clone SK3, 1:80, αCD45RA BUV395 clone 5H9, 1:40, αCD27 BUV615 clone L128, 1:80, αCD25 BUV563 clone 2A3, 1:80, αCD39 BUV737 clone TU66, 1:80 and αCD26 BUV805 clone M-A261, 1:80 (all BD Biosciences, Franklin Lakes, New Jersey, USA) diluted in Brilliant Violet buffer (BD Biosciences, Franklin Lakes, New Jersey, USA) was added then kept in the dark for 20 min at room temperature. After washing, cells were fixed and permeabilized using eBioscience FOXP3/Transcription Factor staining kit (Invitrogen, Waltham, Massachusetts, USA). Cells were washed twice with 1X eBioscience Perm diluent, then resuspended in intracellular antibody mix (αRORγT Alexa Fluor 647 Clone Q21-559, 1:40, αIFNγ BB700 clone 2B7, 1:160 (both from BD Biosciences, Franklin Lakes, New Jersey, USA), αIL-17 PE clone BL168, 1:20, αT-bet PE-Dazzle 594 clone 4B10, 1:40 (both from BioLegend, San Diego, California, USA) and αFoxP3 PE-Cyanine5.5 clone PCH101, 1:40 (Thermo Fisher Scientific, Waltham, Massachusetts, USA) in eBiosciences perm diluent. Cells were then washed (centrifuged at 900g for 5 min at room temperature) twice with 1X eBioscience Perm diluent and fixed in 2% PFA. Data were acquired on a 5-Laser Aurora Spectral Flow cytometer (Cytek Biosciences, Fremont, California, USA).

Spectral flow fcs files were initially analyzed using SpectroFlo v3.0 (Cytek Biosciences, Fremont, California, USA) for unmixing and autofluorescence correction. Identification of T-cell subsets from unmixed fcs files was performed using Flowjo v10 (Flowjo LLC, Ashland, Oregon, USA). Antigen-specific cytokine levels are reported after subtraction of values of unstimulated cells for each participant. K/T ratio was determined from the concentrations of tryptophan and kynurenines after quality control checks. Non-parametric tests were used for comparison between two groups for paired (Wilcoxon test) or unpaired (Mann–Whitney test) and p< 0.05 was considered significant. Spearman correlation was used to test the association between parameters with a p< 0.05 considered significant. All statistical analyses used Prism 9.0 (GraphPad, Boston, Massachusetts, USA).

To study the effects of LTBI and ART-treated HIV (HIV-ART) on Th17 cells defined by different criteria, we designed a spectral flow cytometry panel encompassing cytokines, chemokine receptors, transcription factors, and cell surface markers ( Supplementary Figure S1 , Figures 1A, C ). We used the following nomenclature: (a) subset 1 is CD26⁺CD161⁺ (16, 17); (b) subset 2 is CCR6⁺CXCR3^- (12, 15); and (c) Th1* (Th1Th17) is CCR6⁺CXCR3⁺ (15, 43). We first gated out Vα7.2⁺CD4⁺T cells to exclude mucosal-associated invariant T (MAIT) cells (Supplementary Figure 2) that share certain markers with Th17 cells (44–49), and since MAITs can produce IL-17 (50).

We also stained for RORγT, the Th17 lineage defining transcription factor (51), and used unsupervised clustering and t-distributed stochastic neighbor embedding (tSNE) to determine whether subset 1 and subset 2 Th17 cells express RORγT and/or produce IL-17 upon stimulation with an Mtb peptide mega pool (Mtb300) ( Figure 1C ). To identify Th1* (also termed Th1Th17) cells that respond to Mtb antigens (15), we included T-bet, the Th1 lineage transcription factor (52). We found a positive and significant association (r= 0.4472, p= 0.0002) ( Figure 1B ) between subset 1 and subset 2 Th17 cells and found that cells in both populations express RORγT and produce IL-17 ( Figure 1C ). Therefore, subset 1 and subset 2 Th17 cells express other properties of bona fide Th17-cell populations. T-bet expressing cells were present within subset 1 and subset 2 Th17-cell populations, suggesting a Th1* population that expresses both CCR6 and CXCR3 and likely produce both IFNγ and IL-17 ( Figure 1C ). Thus, we identified the Th17 subsets based on chemokine receptors (CCR6⁺in the absence of CXCR3 expression) or surface markers (CD26 and CD161) co-expression. Since protein transport inhibitors alter the expression of some of the T cell markers (Supplementary Figure 3), our classification of Th17 subsets was performed on samples in which GolgiStop and GolgiPlug were not added.

As expected, circulating CD4 T cells were reduced in HIV-ART compared with HIV-uninfected participants, ( Figure 2A ). We first phenotyped CD4⁺T-cell populations based on CD45RA and CCR7 (3) to identify naïve, central memory, effector memory, and terminally differentiated T cells ( Supplementary Figure S4A , left). The predominant populations were naïve and central memory T cells and there was no difference in the frequency of cells in these memory populations in participants with or without HIV ( Figure 2B ). The expression of CD45RA and CCR7 was not affected by 20hour stimulation of PBMC with antigens (Supplementary Figure 5). CD4⁺CD45RA⁺CCR7⁺(naïve T cells) are a heterogeneous population, including cells that produce cytokines upon short-term stimulation (53), so we characterized the naïve T cells further based on co-expression of CD95 and CD27 to identify cells with a stem cell-like memory phenotype ( Supplementary Figure S4A , right). We found no difference in frequencies of CD95⁺CD27⁺stem cell–like memory cells or CD95⁻CD27⁺“true naïve” cells in those with or without HIV ( Supplementary Figure S4B ). Thus, whereas individuals with treated HIV have lower overall frequencies of circulating CD4⁺T cells than those without HIV ( Table 1 , Figure 2A ), their memory states do not differ. These observations are in accord with studies that antiretroviral therapy induces a burst in CD4 T-cell memory regeneration and general reconstitution of CD4 T-cell count relative to pre-treatment time points (54–56).

Next, we quantitated Th17 cells in those with LTBI without HIV or with HIV-ART and found significantly lower frequencies of subset 1 (CD4⁺Vα7.2⁻CD26⁺CD161⁺) and Th1* (CD4⁺Vα7.2⁻CCR6⁺CXCR3⁺) cells (expressed as % of CD4⁺ T cells) in HIV-ART than in those without HIV ( Figure 2C ). Although subset 2 cells (CD4⁺Vα7.2⁻CCR6⁺CXCR3⁻) and Th1 (CD4⁺Vα7.2⁻CCR6⁻CXCR3⁺) cells were also present at lower frequencies in HIV-ART, the difference was not significant ( Figure 2C ). These data demonstrate that CD26⁺CD161⁺(subset 1) and CCR6⁺CXCR3⁺ (Th1*) circulating Th17 cells are disproportionately depleted in HIV-ART, while CCR6⁺(subset 2) Th17 cells and Th1 (CCR6^-CXCR3⁺) cells are relatively preserved.

Th17 and regulatory T cells (Tregs) develop under related but distinct cytokine environments (57). Tregs contribute to immune homeostasis by maintaining unresponsiveness to self-antigens, suppressing exaggerated immune responses, and promoting epithelial tissue integrity (58). Human Tregs are identified by the expression of high levels of CD25 and the transcription factor FoxP3 (59). Since the kynurenine pathway has differential effects on Th17 and Treg differentiation, we characterized Treg cells ( Supplementary Figure S6A , left) and found that the frequency of circulating Treg was similar in HIV-ART and those without HIV ( Supplementary Figure S6B ). We also found that the frequency of activated (CD39⁺) Tregs did not differ in the two groups ( Supplementary Figure S6A , right), although there was considerable variation in CD39⁺ Treg, especially in HIV-ART ( Supplementary Figure S6C ). We also compared the ratio of Th17 subset 1 and Th17 subset 2 cells to Tregs between the groups and found a trend toward a lower Th17 subset1/Treg in HIV-ART than in those without HIV ( Supplementary Figure S6D ), but the difference was not significant, and Th17 subset 2/Treg ratios were similar in HIV-ART and those without HIV ( Supplementary Figure S6E ). Taken together, we found that CD26⁺CD161⁺ Th17-cell subset is disproportionately reduced in LTBI with treated HIV, suggesting that even effective treatment of HIV does not reconstitute all Th17-cell subsets equally. We did not find differences in Treg by HIV status.

Having established that subset 1 (CD26⁺CD161⁺) and subset 2 (CCR6⁺) Th17 cells are bona fide Th17 populations, we determined whether subset 1 and subset 2 cells differ in IL-17 production after Mtb antigen stimulation. HIV-ART participants with LTBI had significantly more IL-17⁺cells in subset 1 Th17 cells than in subset 2 Th17 cells upon stimulation with the Mtb300 peptide pool ( Figure 3A , left). All except one participant had IL-17⁺cells in subset 1 cells compared with 11 of 21 participants with IL-17⁺cells in subset 2 Th17 cells ( Figure 3A , left). We did not have enough cells from participants without HIV (cohort 1) for cytokine analysis. We confirmed this observation in cohort 2; regardless of the sampling time point, there were significantly more IL-17⁺cells in subset 1 Th17 cells than in subset 2 Th17 cells ( Figure 3A , right). We performed a similar analysis after stimulation with SEB and found an opposite pattern: IL-17 production was significantly higher in subset 2 than in subset 1 Th17 cells ( Figure 3B ). Additionally, CMV-peptide stimulation induced equivalent production of IL-17 by subset 1 and subset 2 Th17 cells, although there was a trend toward higher responses in subset 2 Th17 cells ( Supplementary Figure S7A ). These data reveal enrichment of IL-17 production to Mtb antigens by a subset of Th17 cells marked by expression of CD26 and CD161 (subset 1), which are distinct from IL-17–producing cells that are marked by expression of chemokine receptor CCR6 (subset 2) and those that produce IL-17 in response to SEB or CMV (Supplementary Table 1).

Next, we assessed IFNγ and IL-17 production after Mtb300 stimulation. In HIV-ART, only four of 21 (19%) had detectable IFNγ responses compared with 11 of 21 (52%) who had detectable IL-17 responses (p= 0.0516, Fisher’s exact test). Additionally, the median IL-17 response was higher than the median IFNγ response, although this did not reach significance ( Figure 3C , left). Similar IL-17 responses to Mtb antigen stimulation were observed in people with HIV before and after LTBI diagnosis and treatment. Treatment of LTBI did not change either IFNγ or IL-17 responses, although the response magnitudes varied by participant ( Figure 3C , right). Similarly, more participants had detectable IL-17 responses than detectable IFNγ responses at all three sampling times, prior to being TST positive, and at INH initiation and treatment completion: 2/11 versus 10/11, 3/11 versus 9/11, and 6/14 versus 12/14, respectively ( Figure 3C , right).

To determine whether the predominant production of IL-17 over IFNγ was specific to Mtb antigens, we used a CMV peptide pool to stimulate PBMC from participants ( Tables 1 , 2 ) who had sufficient cells ( Figure 3D ). This revealed a pattern distinct from that with Mtb300: participants with LTBI and HIV-ART (cohort 1) exhibited higher magnitude and higher prevalence of IFNγ than IL-17 responses [IFNγ: 11/16 (69%) vs. IL-17: 5/16 (31%)] ( Figure 3D , left) (p= 0.0756, Fisher’s exact test). Similarly, in Cohort 2, there was a higher magnitude of IFNγ responses than IL-17 responses at all time points, reaching statistical significance at baseline (prior to TST⁺) time point ( Figure 3D , right). As with CMV and in contrast to Mtb300, SEB induced higher magnitude IFNγ than IL-17 responses ( Supplementary Figure S7B ). Together, these data demonstrate that in LTBI and HIV-ART, IL-17 responses to Mtb are relatively preserved compared with IFNγ responses, and this is distinct compared to CMV and SEB responses. This is also contrary to the commonly held impression that IFNγ production is the predominant response to Mtb.

We quantitated tryptophan and kynurenine concentrations in plasma of participants with LTBI with or without HIV-ART and inferred IDO activity based on the K/T ratio. We found significantly lower plasma tryptophan concentrations in participants with HIV-ART compared with those without HIV ( Figure 4A ), and kynurenine concentrations were higher in HIV-ART but did not achieve statistical significance when compared with the HIV uninfected group ( Figure 4B ). Finally, there was a significantly higher K/T ratio in people with HIV-ART compared to those without HIV ( Figure 4C ). Although we cannot ascertain the contribution of LTBI to the observed IDO activity because we did not have participants who were not exposed to Mtb, these observations agree with prior reports showing that IDO-1 mediated tryptophan catabolism is elevated in PWH, including those initiated on early treatment and ART-suppressed viral loads (31).

Since kynurenines favor the development of Tregs while suppressing Th17 generation (27), we determined the correlation of plasma K/T and Th17 cells, Tregs, and Th17/Treg ratios in cohort 1 participants. This revealed a significant inverse relationship between K/T and circulating subset 1 and subset 2 ( Figures 4D, E ), while there was no correlation between K/T and Tregs ( Figure 4D , middle). We also found significant inverse correlations between K/T and subset 1/Treg and subset 2/Treg ( Figures 4D, E ). Importantly, we found no correlation between K/T and Th1*, Th1, or IL-17⁺cell frequencies ( Supplementary Figures S8A , S8B and data not shown). Although we confirmed high IDO activity ( Figure 4 ) and found a reduction in circulating Th17 (especially subset 1) cells in HIV-ART compared to HIV uninfected people ( Figure 2 ), Tregs and the Th17/Treg ratio were similar in the two groups ( Supplementary Figure S3 ). Thus, increased IDO activity is associated with a reduction in circulating Th17 cell subsets, but the frequency of IL-17–producing CD4 T cells does not correlate with IDO activity.