Adults with newly diagnosed TB were enrolled at referral TB clinics in Reynosa and Matamoros, Mexico. Exclusion included age older than 60 years, HIV-positive, reported alcohol/drug abuse, or type 1 diabetes. Enrollment procedures followed guidelines from the Institutional Review Boards in Mexico (110/2018/CEI) and the United States (HSC-SPH-19-0308; HSC-SPH-14-1007), and participants signed informed consent. TB diagnosis was based on isolation of Mtb or a positive smear for acid-fast bacilli and an abnormal chest x-ray. Body-mass index and waist:hip ratio (WHR) were documented as before (33). T2D was based on hyperglycemia (fasting glucose ≥126 mg/dL or random ≥200 mg/dL or HbA1c ≥ 6.5%). Estimates of insulin resistance and % beta-cell function used the homeostatic model assessments HOMA1-IR and HOMA2, respectively (33, 34), on participants with fasting blood glucose and no insulin use.

PBMC isolated from heparinized peripheral blood were cryopreserved in liquid nitrogen. Cells were thawed and transferred into R10 media (RPMI 1640 containing L-glutamine with 10% FBS, 1% PenStrep, and 1% Hepes). Cells were centrifuged at 900g for 5 minutes, resuspended in 5 ml R10, transferred to a 6-well plate, and rested overnight at 37°C/5% CO₂. Cells were counted and 1×10⁶ live cells in 200 μL of R10 were stimulated with Mtb peptide mega pool (Mtb300) (2μg/ml) or staphylococcal enterotoxin B (SEB) positive control (1μg/ml) in the presence of costimulating antibodies anti-CD28/CD49d (1μg/ml) (BD Biosciences). Cells with no antigen stimulation were included as negative controls. Mtb peptide megapool represents immunodominant epitopes from 90 Mtb antigens recognized by HLA class II-restricted CD4 T cells in diverse populations (35, 36). After 2h, GolgiStop and GolgiPlug (BD Biosciences) were added (0.5uL/200uL) to each well, and incubated for an additional 18h.

After stimulation, cells were washed, stained with Live/Dead viability dye 1:1000 (Invitrogen), washed then surface antibody cocktail (αCD3 BV510 1:80; αCD8 BV570 1:80; αCCR7 BV785 1:80; αCCR6 FITC 1:80; αCXCR3 BV605 1:40; αCD161 APC-Fire750 1:80; αCD69 BV650 1:20; αCD154 BV711 1:80; and αVα7.2 BV421 1:40; (all BioLegend), αCD4 BUV496 1:80; αCD45RA BUV395 1:80; αCD25 BUV563 1:80; αCD39 BUV737 1:80; and αCD26 BUV805 1:80; (all BD Biosciences), αCD153 Alexa Fluor 488 1:20; (R & D Systems) diluted in Brilliant Violet buffer (BD Biosciences) was added then kept in the dark for 20 min at room temperature. Next, cells were fixed and permeabilized using eBioscience FOXP3/Transcription Factor kit (Invitrogen) for 30 minutes on ice. Cells were washed twice with 1X eBioscience diluent, then resuspended in intracellular antibody mix (αRORγT Alexa Fluor 647 1:40; αIFNγ BB700 1:160; (both BD Biosciences), αIL 17 PE 1:20; αT-bet PE-Dazzle 594 1:40; (both BioLegend), αKi-67 eFluor450 1:40; and αFoxP3 PE-Cy5.5 1:40; (both Thermo Fisher Scientific) in eBiosciences perm diluent for 20 minutes at room temperature in the dark. Next, cells were washed twice and fixed in 2% PFA. Data was acquired on Aurora Spectral Flow cytometer (Cytek Biosciences).

Spectral fcs files were analyzed using SpectroFlo v3.0 (Cytek Biosciences) for unmixing and autofluorescence correction and T-cell subsets were identified using Flowjo v10 (Flowjo LLC). Antigen-responsive cytokine levels are reported after subtracting no antigen stimulation tests for each participant.

Statistical analysis was performed using SAS version 9.4 (SAS Institute Inc.). Chi-square or Fisher’s exact tests were used to compare categorical variables, and non-parametric tests were used to compare median values in study groups (Mann-Whitney U or Wilcoxon rank-sum tests) with Dunn’s post-hoc correction. To evaluate the independent contribution of T2D to different CD4 T cell subsets among TB patients, we performed GENMOD analysis to fit a generalized linear model to the data, using maximum likelihood estimation of the parameter vector β adjusted for sex, body mass index, and age. P-values were considered significant if ≤ 0.05 while p-values between 0.05 – 0.099 were considered borderline significant. Graphs were created using GraphPad Prism v9.0 (GraphPad Software).

We selected 23 newly diagnosed TB patients with T2D (n=12) or without T2D (n=11) matched by age and sex ( Table 1 ). Other sociodemographics were similar. 64% of TB-only patients had pre-T2D while TB-T2D patients had known their T2D diagnosis for a median of 8.8 years before the current TB episode. 11/12 of the T2D patients reported taking a glucose-lowering medication in the past month, mainly metformin (83%) ( Supplementary Table S1 ). The participants with TB-T2D had significantly higher measures of glucose or estimates of insulin resistance and lower estimates of % of beta-cell function ( Table 1 ). Complete blood counts were comparable between study groups except for a lower neutrophil: lymphocyte ratio in TB-T2D vs TB-no T2D ( Supplementary Table S2 ). All patients without T2D had taken antimycobacterial drugs for 1 to 10 days, compared to 5 (42%) with TB-T2D (p=0.005). Other TB symptoms were similar by T2D status ( Supplementary Table S3 ).

To identify alterations in circulating T cells in people with TB by T2D status, we designed a high-parameter spectral flow cytometry panel to measure T cell maturation states, proliferation, differentiation, and effector functions ( Supplementary Figure 1 ). We found no difference in the frequency of circulating bulk CD3, CD4, and CD8 T cells ( Figure 1A ). Further analysis was focused on the characterization of circulating CD4 T cells since this broad subset plays a prominent role in TB control (11, 12, 15, 37). First, we evaluated cells without antigen stimulation and found a significantly higher frequency of CD4 T cells expressing the Th1 master transcription factor T-bet in TB-T2D compared to TB alone. Accordingly, T2D patients had a borderline higher proportion of circulating Th1 T cells (CD4⁺Vα7.2^-CCR6^-CXCR3⁺; Table 2 , Figure 1B ). For regulatory T cells, there was no difference in the frequency of their master transcription factor (FoxP3), nor their cell surface markers (FoxP3⁺CD25⁺). However, activated (CD39⁺) regulatory T cells were borderline higher in T2D ( Figure 1C , Table 2 ). Together, these findings suggest that in TB-T2D patients versus TB alone, there are no differences in the overall frequencies of CD4 and CD8 T cells, but within the CD4 T cell subsets, there are increases of Th1 and activated regulatory T cells.

Emerging evidence suggests a beneficial role for Th17 and Th17-like cells in controlling Mtb in humans (23–25) and animal models (38, 39). To identify Th17 cell subsets, we used the gating strategy we described previously (29) that defines Th17 cells based on the expression of cell surface markers (CD26 and CD161) and chemokine receptors (CCR6 and CXCR3) (40, 41) among CD4 T cells (see extended data of 1xN plots ( Supplementary Data Sheet 1 )). We excluded mucosa-associated invariant T cells (MAIT) that also express CD26 and CD161 in addition to the canonical MAIT cell marker Vα7.2 (42). We identified three categories of Th17 cells (29); subset 1 (CD4⁺Vα7.2^-CD26⁺CD161⁺), subset 2 (CD4⁺Vα7.2^-CCR6⁺CXCR3^-), and Th1* (Th1/Th17), (CD4⁺Vα7.2^-CCR6⁺CXCR3⁺). We found a significant increase in circulating subset 1, subset 2, and Th1* cells in TB-T2D versus TB ( Figure 2A , Table 2 ), despite finding no difference in the expression of RoRγt, the master regulatory transcription factor for the Th17 lineage ( Table 2 ). We further stratified participants into no T2D, pre-T2D, and T2D to screen for the extent of dysglycemia when Th17 phenotypes change. Results suggested a gradual increase in the Th17 phenotypes through the No T2D < pre-T2D < T2D states, but significant increases were only detected for Th17 subsets 1 and 2 ( Figure 2B ). Characterization of the expression of CD26, CD161, and CCR6 on CD8⁺Vα7.2^- T cells also revealed an increase in all three circulating cytotoxic type 17 cell subsets in TB-T2D compared to TB-only ( Supplementary Figure 2a ). Thus, T2D is associated with an increase in three subsets of circulating Th17 cells in people with TB, and this shift may start at a pre-T2D stage.

To further characterize CD4 T cells during TB and with or without T2D, we analyzed their maturation states based on CCR7 and CD45RA expression, with classification into naïve (CCR7⁺CD45RA⁺), central memory (CM: CCR7⁺CD45RA^-), effector memory (EM: CCR7^-CD45RA^-), and terminally differentiated memory (TEMRA: CCR7^-CD45RA⁺) T cells (25, 26, 29). The frequency of circulating naïve CD4 T cells was significantly lower in TB-T2D than in those with TB alone ( Figure 3A , Table 2 ). This difference was accompanied by a trend towards more circulating mature CD4 T cells, particularly central memory cells ( Figure 3A , Table 2 ). Stratified by the extent of dysglycemia, the results suggested that as T2D was developing, the frequency of circulating naïve CD4 T cells was decreased while those with a central memory phenotype were increased ( Figure 3B ). We found no difference in CD8 T cell maturation states, except for a trend toward higher circulating memory CD8 T cells in TB-T2D than in TB alone ( Supplementary Figure 2b ). Together, T2D is associated with a shift from fewer naïve to more mature circulating CD4 T cell states.

To assess the capacity of T cells to proliferate in response to stimulation, we measured the intracellular expression of the proliferation marker, Ki67, on CD4 T cells after 20-hour stimulation with the Mtb300 antigen megapool. We found no difference in the proportion of Ki67⁺ cells in the bulk CD4 T cell population in response to antigen stimulation between TB-only and TB-T2D (data not shown). Given the differences by T2D status in CD4 T cell maturation states ( Figure 3A ), we evaluated whether Ki67 expression differed in Mtb300-stimulated CD4 T cells by maturation state. We found no difference in the proliferation of antigen-stimulated naïve CD4 T cells, but the proliferation of central memory and effector memory CD4 T cells was significantly reduced in TB-T2D compared to TB ( Figure 3C ). Further stratification of participants by extent of dysglycemia showed a lower frequency of Ki67⁺ T cells among the central memory or effector memory CD4 T cells in T2D, and not in the pre-T2D stage ( Figure 3D ). In summary, T2D is associated with decreased proliferation of central and effector memory CD4 T cells in people with TB.

To compare the effector responses of CD4 T cells to Mtb antigens in TB by T2D status, we stimulated PBMC with the Mtb300 antigen megapool and measured the expression of the activation markers CD69 and CD154 in combination with CD153. While we found no difference by T2D status in the expression of CD69, CD153, or CD154 as single markers (data not shown), the frequency of CD69⁺CD154⁺ T cells was higher in TB-T2D than in TB alone ( Table 2 ). We also measured the frequency of CD4 T cells expressing IFNγ, IL17, and TNFα after Mtb300 stimulation. We found no differences in the frequencies of any of the three cytokines individually or as combinations on bulk CD4 T cells between TB-T2D and TB-only individuals ( Table 2 ). When we combined the groups and assessed IL17 production by Th17 cell subsets 1 and 2, we found that Th17 subset 1 preferentially produced IL17 compared to Th17 subset 2 in response to Mtb300 peptide pool stimulation ( Supplementary Figure 3 ), in concurrence with our previous finding (29). Taken together, our analysis has revealed that T2D differentially alters distinct features of circulating T cells in people with TB, including maturation, proliferation, differentiation, and the activation of Mtb antigen-responsive CD4 T cells, despite similar cytokine responses of Mtb-responsive CD4 T cells in TB patients with or without T2D.

To determine the independent contribution of T2D to the CD4 T cell characteristics with significant or borderline significant associations by univariable analysis ( Figures 1 - 3 , Table 2 ), we performed a multivariable analysis with variable selection guided by host characteristics that differed by T2D status and were unrelated to glucose control ( Table 1 ). Potential confounders included sex, vascular diseases, and days of TB treatment. The estimated beta coefficients and adjusted p-values are shown in Table 3 . Among CD4 T cells without antigen stimulation, individuals with T2D had a higher proportion of cells committed to Th1 lineage (T-bet+), or Th1 and Th17 (subsets 1, 2, and Th1*), and activated regulatory T cell subsets. T2D also remained independently associated with a shift in maturation states with lower proportions of naïve T cells and higher central memory T cells. When PBMC were stimulated with the Mtb300 antigen megapool, T2D remained associated with an increase in the proportion of CD69⁺CD154⁺ activated T cells, but differences in proliferation were no longer significant.

We performed additional analyses focusing on the association between Th17 cell subsets and markers of TB severity. Our findings indicate an association between Th17 cell subsets and T2D comorbidity in TB patients. Since T2D has been associated with adverse TB outcomes (27, 43–47), we examined whether a higher proportion of Th17 cells could be a contributor, by evaluating associations with features of a more severe TB. Table 4 identified associations, after controlling for T2D (already known to have higher Th17 cells), as well as TB treatment initiation within 1–10 days of enrollment and age (associations by univariable analysis). Sex was not associated, and due to the small sample size, not controlled for in the multivariable models. We found correlations between the Th17 subset 1 and blood inflammatory biomarkers (platelets, neutrophils, and C-reactive protein) and duration of reported TB symptoms. Th17 subset 2 was associated with high blood pressure, higher platelet counts, and smear grade, and the Th1* subset was also associated with a higher smear grade. Together, these data support the independent contribution of T2D to alterations in CD4 T cell phenotypes in TB patients and suggest likely contribution of Th17 cells to poor outcomes of TB.