Th1, Th17, and Th1Th17 Lymphocytes during Tuberculosis: Th1 Lymphocytes Predominate and Appear as Low-Differentiated CXCR3⁺CCR6⁺ Cells in the Blood and Highly Differentiated CXCR3^+/−CCR6⁻ Cells in the Lungs

Host protection against Mycobacterium tuberculosis depends on the successful cooperation between cells of innate and acquired immunity. Understanding the cellular and molecular components underlying this cooperation is a prerequisite for developing better strategies for tuberculosis (TB) control and prevention.

IFN-γ–producing Th1 lymphocytes able to activate macrophages are thought to play a central role in TB protection. The concept is supported by the development of severe mycobacterial infections in humans and animals bearing defects in the CD4⁺ population or genes of the IFN-γ/IL-12 axis (1–4). However, the degree to which host protection depends on Th1/IFN-γ is currently debated. Several recent studies reported a lack of correlation between the extent of Th1/IFN-γ responses and the degree of protection against TB (5–12). Some investigators found that overproduction of IFN-γ by Th1 cells may be deleterious to the host (13, 14). Further, the involvement of immune cells other than “classical” Th1 lymphocytes in M. tuberculosis control has been suggested. Specifically, Arlehamn et al. (15) have associated protection against TB with the persistence of IFN-γ–producing CD4⁺ lymphocytes expressing the CXCR3 (X3)⁺CCR6 (R6)⁺ phenotype of “nonclassical” (also called ex-Th17) Th1 cells.

In addition to Th1 populations, the possible effect of Th17 lymphocytes in TB control is actively discussed. Because of their involvement in the induction of neutrophilic inflammation and the ability of neutrophils to mediate TB pathology, Th17 lymphocytes were first considered a potentially pathological population. However, experimental data have evidenced the implications for Th17 in TB defense. In mice, Th17 cells conferred protection during experimental infection induced by the highly virulent M. tuberculosis isolate HN878 (16) and were important mediators of vaccine-induced protection (17, 18). The underlying mechanisms included induction of Th1 lymphocytes, stimulation of their recruitment to the infectious site, and formation of ectopic lymphoid follicles and granulomas (16, 18, 19).

Data on the extent of Th17 responses during TB in humans are highly controversial. Some studies reported elevated levels of IL-17 and memory Th17 cells in subjects with latent TB infection (LTBI) compared with active TB patients, supporting the involvement of Th17 in TB protection (20, 21). The concept was further enforced by the demonstration of high mortality in TB patients having low levels of IL-17 in the serum (22). In contrast, several groups found reduced Th17 responses during LTBI and increased frequencies of IL-17–producing cells in patients with active and multidrug-resistant TB (23–25). Some investigators did not find significant differences between LTBI and TB with regard to the extent of Th17/IL-17 responses (26). Finally, it has been reported that M. tuberculosis–specific Th17 lymphocytes are present at extremely low frequencies during LTBI and TB, and a minimal role for Th17 in immunity to mycobacteria has been suggested (27). Thus, there is no clarity on whether Th17 lymphocytes play a protective or pathological role, or any significant role at all, in immunity to M. tuberculosis in humans.

Further complexity comes from the interplay between the Th1 and Th17 populations. Th17 lymphocytes do not inhibit and even favor the generation of Th1 lymphocytes (28). In contrast, IFN-γ and T-bet, the master regulator of Th1 cells, negatively regulate the expression of Th17-specific genes and inhibit Th17 differentiation (29). Yet, a population of Th1Th17 cells (also called Th17.1) coexpressing T-bet and RORγt/RORC and coproducing type 1 cytokines (IFN-γ, TNF-α) and IL-17 has been described (30). Th1Th17 lymphocytes play a pathological role during autoimmune diseases (31), but the role for this population during active TB remains undetermined.

Another unexplored aspect of TB immunity is to what degree Th1, Th17, and Th1Th17 contribute to local responses in the lungs. Existing data are scarce and, with regard to Th17 responses, are largely limited to the analysis of pleural fluid and bronchoalveolar lavage fluid (32). Examinations of Th17 and Th1Th17 populations at the site of pulmonary TB lesions in humans are missing.

Th1, Th17, and Th1Th17 cells can be identified on the basis of IFN-γ and/or IL-17 production or on the basis of chemokine receptor expression. Specifically, chemokine receptors X3 and R6 are considered surface markers for Th1 (phenotype, X3⁺R6⁻), Th17 (X3⁻R6⁺), Th1Th17 (X3⁺R6⁺), and nonclassical Th1 (also X3⁺R6⁺) cells (30, 33). During TB, Th1 lymphocytes are usually evaluated on the basis of IFN-γ production. Thus, the classical versus nonclassical nature of M. tuberculosis–specific Th1 cells and their relationship with Th17 and Th1Th17 lymphocytes remain poorly understood.

In this study, we have analyzed Th1, Th17, and Th1Th17 populations during pulmonary TB. We present data on the frequencies of M. tuberculosis–specific CD4⁺ lymphocytes producing IFN-γ, IL-17, and IFN-γ/IL-17 in the blood and lungs of TB patients, describe the patterns of X3/R6 expressed by blood and lung CD4⁺ lymphocytes, show that nonclassical X3⁺R6⁺, classical X3⁺R6⁻, and X3⁻R6⁻ CD4⁺ cells differ with regard to their degree of differentiation, and examine the relationships among these populations using an in vitro culture model. The results add to the general understanding of immune mechanisms operating during TB and are relevant for the development of new strategies to stimulate a protective antimycobacterial immune response.

The study was approved by the Institutional Review Board Number 1 of the Central Tuberculosis Research Institute (CTRI), was performed between 2015 and 2017, and was conducted in accordance with the principles expressed in the Helsinki Declaration. All participants gave written informed consent to participate in the study.

Forty-one patients diagnosed with pulmonary TB were recruited at CTRI (Table I). All patients were HIV seronegative. TB diagnosis was based on clinical and radiological evidence of TB and the identification of M. tuberculosis and/or M. tuberculosis DNA in the sputum (n = 31) or positive clinical and radiological responses to anti-TB therapy (assessed 2 mo following treatment by independent clinicians, n = 10). Blood cell analyses were performed in 29 patients who had recent TB and had not received antituberculosis therapy before their admission to CTRI (14 women, 15 men; median age 35 y; range 18–63 y). Of the 29 patients, 15 had TB infiltrate, 4 had tuberculoma, 3 had focal TB, 3 had cavitary TB, and 4 had disseminated TB (Table I). Immunological analyses were performed within 2 wk of admission to CTRI.

Twelve patients had chronic TB and underwent lung surgery because of poor response to anti-TB drug treatment (six women, six men; median age 29 y; range 25–62 y). Blood and lung tissue–derived cells from these patients were analyzed on the day of surgery. Of the 12 surgery patients, 1 had TB infiltrate, 5 had tuberculoma, 2 had cavitary TB, and 4 had cirrhotic TB (Table I).

For blood cell analysis, venous blood samples were collected in heparin tubes, and PBMCs were isolated within 1 h of sample collection by Ficoll density gradient centrifugation.

To isolate lung tissue cells, samples of surgically resected lung tissue were obtained within 30 min of tissue resection. Samples from patients with cavitary TB were taken in a such a way to avoid the inclusion of the grossly fibrotic wall of the cavity. Samples were immersed in ice-cold RPMI 1640 medium supplemented with 5% FCS, 20 U/ml heparin, and 50 μg/ml gentamicin (all from Thermo Fisher Scientific, Waltham, MA). Samples were rinsed to remove blood clots, carefully minced with a sterile surgical blade, chopped with scissors, and vigorously pipetted. The resulting suspension was passed through a 100-μm-mesh stainless steel sieve and washed by centrifugation. After additional washing, the cells were counted and treated with Abs to analyze surface phenotype or cultured to identify intracellular cytokines. Cell recoveries ranged from 1.0 to 4.2 × 10⁷ of total live cells per gram of lung tissue (median 1.8 × 10⁷ [interquartile range 1.2 × 10⁷–3.1 × 10⁷]) and from 0.1 to 3.2 × 10⁶ of CD4⁺ cells per gram of lung tissue (1.6 × 10⁶ [0.4 × 10⁶–2.9 × 10⁶]).

In some experiments, populations of blood CD4⁺ X3⁺R6⁺, X3⁺R6⁻, and X3⁻R6⁻ cells were sorted. For biosafety reasons, in these experiments, blood cells were obtained from TB contacts (n = 2) or from treated and sputum-converted TB patients (n = 2). TB contacts were CTRI employees working in close contact with TB patients for >2 y and having no clinical or radiographic evidence of TB (two women aged 54 and 28 y). TB patients (two men aged 35 and 30 y) had infiltrative TB, had been treated for 6 mo, and had converted sputum culture by the time of blood collection.

Surface expression of analyzed markers was assessed by staining 100 μl of blood or 1 × 10⁶ lung cells with PerCP–Cy-5.5–anti-CD4, allophycocyanin–anti-X3 (CD183), BV510–anti-R6 (CD196) plus PE–anti-CD161 or FITC–anti-CD27, PE–Cy7–anti-CD62L, and allophycocyanin–H7–anti-CD28 Abs (all from BD Biosciences). Cell viability was assessed by staining cells with a LIVE/DEAD Fixable Far-Red Dead Cell Stain Kit (Thermo Fisher Scientific), using the manufacturer’s recommendations.

For intracellular cytokine staining, 500 μl of whole blood or 2 × 10⁶ lung cells were incubated with purified protein derivate (PPD; 10 μg/ml, 3 h; Accurate Chemical & Scientific, Westbury, NY) and then cultured in the presence of Brefeldin A (GolgiPlug; BD Biosciences) for an additional 12 h. The cells were stained with PerCP–Cy-5.5–anti-CD4, BV421–anti-X3, BV510–anti-R6 plus PE–anti-CD161 or PE–anti-CD27, PE–Cy7–anti-CD62L, and allophycocyanin–H7–anti-CD28 Abs. The cells were treated with BD FACS Lysing solution, washed, treated with BD FACS Permeabilizing Solution 2, stained with FITC–anti–IFN-γ and allophycocyanin–anti–IL-17A Abs, washed, and fixed (all reagents were from BD Biosciences; allophycocyanin–anti–IL-17A was from eBioscience). During the analysis, whole samples were acquired on a flow cytometer. The stability of surface marker expression following intracellular cell staining was tested in a series of preliminary experiments (see 8Results and Supplemental Fig. 1).

For T-bet detection, the cells were cultured as described above, stained for surface markers, permeabilized using a human FOXP3 buffer set (BD Biosciences), and stained with FITC–anti–IFN-γ Ab (BD Biosciences) and allophycocyanin–anti–T-bet Ab (Sony Biotechnology, San Jose, CA).

Unstained, single-stained, and unstimulated cells were used as controls. T-bet expression was also verified by fluorescence-minus-one control. Cells were analyzed on a BD FACSCanto II flow cytometer equipped with 405-, 488-, and 647-nm lasers (BD Biosciences) using BD FACSDiva (BD Biosciences) and FlowJo (TreeStar) software.

PBMCs were isolated by Ficoll gradient centrifugation (Dia-M, Moscow, Russia). Approximately 2 × 10⁷ PBMCs were used to isolate CD14⁺ cells, which served as APCs. The cells were isolated using CD14 MicroBeads (Miltenyi Biotec, Auburn, CA; cell purity > 90%) and stained with CFSE (Thermo Fisher Scientific), according to the previously described procedure (34). The remaining PBMCs (∼5 × 10⁷ from each donor) were stained with PerCP–Cy5.5–anti-CD4, PE–anti-R6, and allophycocyanin–anti-X3 Abs or with PerCP–Cy5.5–anti-CD4, PE–anti-R6, allophycocyanin–anti-CD62L, and BV421–anti-X3 Abs. The cells were sorted using an SH800S Cell Sorter (Sony Biotechnology). Depending on the experiments, CD4⁺X3⁺R6⁺ and CD4⁺X3⁺R6⁻ populations or CD4⁺X3⁺R6⁻CD62L⁻ and CD4⁺X3⁻R6⁻CD62L⁻ populations were sorted. Sorted cells (3.5–5 × 10⁴ cells per well of a 96-well plate) were stimulated in vitro with immobilized anti-CD3 Ab (1 μg/ml) and anti-CD28 Ab (5 μg/ml) (BioLegend) in the presence of excess autologous APCs (7–10 × 10⁴ cells per well). On days 1 and 5, culture supernatants were collected, centrifuged to remove contaminating cells, and frozen at −70°C for subsequent multiplex analyses. Cultured cells were harvested, counted, stained with anti-CD4, anti-X3, and anti-R6 Abs, and analyzed on a BD FACSCanto II flow cytometer. Following the analysis, expression of X3 and R6 on CD4⁺CFSE⁻ cells was analyzed. CFSE⁺ cells were excluded from the analysis to ensure that CD4⁺ lymphocytes contaminating the APCs (<5%) did not affect the results. Concentrations of IFN-γ, TNF-α, IL-17A, and IL-17F in culture supernatants were quantified using a Bio-Plex Multiplex System (Bio-Rad, Hercules, CA), according to the manufacturer’s protocol.

Data are presented as medians and interquartile ranges. Differences between two groups were analyzed using a Mann–Whitney U test (R; www.r-project.org, Foundation for Open Access Statistics). Differences between several groups were analyzed using one-way ANOVA or Kruskal–Wallis test with the Dunn posttest (GraphPad and/or R software) and the false discovery rate (FDR) method with q = 0.05 (35). Relationships between T cell populations and the bacterial load in the sputum were determined using Spearman rank correlation.

Th1, Th17, and Th1Th17 populations differ by the production of IFN-γ and IL-17 and by the expression of chemokine receptors, particularly X3 and R6.

To characterize M. tuberculosis–specific Th1, Th17, and Th1Th17 populations, we first obtained blood samples from TB patients (Table I), incubated whole blood with PPD, and determined the frequencies of CD4⁺ cells producing only IFN-γ (IFN-γ⁺), only IL-17 (IL-17⁺), and both cytokines (IFN-γ⁺IL-17⁺). Compared with other populations, the frequency of IFN-γ⁺ cells was significantly higher (0.42% [0.23–0.75%], p < 0.0001), IL-17⁺ cells were much less frequent (0.018% [0.009–0.034%]), and IFN-γ⁺IL-17⁺ lymphocytes were almost undetectable (0.007% [0.003–0.011%]) (Fig. 1).

In humans, different lineages of Th cells are characterized by differential expression of chemokine receptors participating in cell homing to lymphoid and peripheral tissues. Specifically, classical Th1 lymphocytes express X3 in the absence of R6, Th17 lymphocytes express R6 in the absence of X3, and Th1Th17 lymphocytes coexpress X3 and R6. Recently, Th1 lymphocytes producing only IFN-γ but coexpressing X3 and R6 were described (30). During LTBI, X3⁺R6⁺ Th1 cells were shown to form the main population of M. tuberculosis–specific IFN-γ⁺ CD4⁺ lymphocytes (15). The X3/R6 phenotype of IFN-γ⁺ lymphocytes persisting during active TB remained uninvestigated. Thus, it was unclear whether coexpression of X3 and R6 by IFN-γ⁺ cells is a specific feature of LTBI or a general characteristic of M. tuberculosis–specific lymphocytes. Therefore, we analyzed the X3/R6 profile of IFN-γ⁺ lymphocytes in TB patients.

First, we examined X3/R6 expression by all CD4⁺ lymphocytes using a surface staining procedure. Based on X3/R6 coexpression, four populations were identified: X3⁺R6⁻, X3⁺R6⁺, X3⁻R6⁺, and X3⁻R6⁻ (Fig. 2A). In the blood, double-negative X3⁻R6⁻ lymphocytes were the most frequent (53.0% [43.8–62.0%]). Among the other populations, the X3⁺R6⁻ subset predominated (X3⁺R6⁻ 18.3% [14.7–24.7%], X3⁺R6⁺ 12.2% [9.1–17.2%], X3⁻R6⁺ 13.2% [10.4–16.5%], p < 0.005, Fig. 2B).

Next, we set out to analyze the relationships between the expression of X3/R6 and the production of IFN-γ/IL-17 by CD4⁺ lymphocytes from TB patients. In preliminary experiments, we demonstrated that procedures used to identify cytokine-producing cells (i.e., cell stimulation and permeabilization) did not alter X3/R6 expression by blood CD4⁺ cells (Supplemental Fig. 1A, 1B). Thus, we determined the frequencies of cytokine-producing cells within each of the four “X3-R6” populations. IL-17⁺ cells were most frequent within the X3⁺R6⁺ population (p ≤ 0.0001 versus the X3⁺R6⁻ and X3⁻R6⁻ populations, p < 0.05 versus the X3⁻R6⁺ population, Fig. 2C). IFN-γ⁺IL-17⁺ lymphocytes were found almost exclusively within the X3⁺R6⁺ population (p < 0.0001 versus the other subsets). IFN-γ⁺ cells were found within the X3⁺R6⁺ and X3⁺R6⁻ populations and were most frequent in the former (p < 0.0005 versus all other populations, Fig. 2C).

Phenotypic analysis of M. tuberculosis–specific cells showed that IL-17⁺ lymphocytes were primarily composed of X3⁺R6⁺ and X3⁻R6⁻ cells (67.5% [50.1–93.5%]); IFN-γ⁺IL-17⁺ lymphocytes were largely X3⁺R6⁺ (85% [73.3–90.9%], Fig. 2D, see figure legend for details of the analysis). Within IFN-γ⁺ lymphocytes, X3⁺R6⁺ lymphocytes also predominated (52.6% [39.7–66.4%]). Individual patients varied with regard to the frequencies of X3-R6 populations within IFN-γ⁺ (i.e., M. tuberculosis–specific) cells. However, this variability was not associated with sputum status of the patients (|r| < 0.3, p > 0.16). Of note, all IFN-γ⁺ cells were alive (Fig. 1B), suggesting that their expression of X3 and R6 could not be attributed to unspecific ligation of the corresponding Abs by dead cells. Also, cells belonging to all four X3-R6 populations were similarly viable (≥99.5%, Fig. 2E), indicating that low frequencies of IFN-γ⁺ cells within some of these populations could not be attributed to their lower viability.

On the basis of IFN-γ/IL-17 production and X3/R6 expression, 12 populations of M. tuberculosis–specific CD4⁺ lymphocytes could be identified. Among these populations, the IFN-γ⁺X3⁺R6⁺ population was predominant (Fig. 2F).

The X3⁺R6⁺ phenotype of M. tuberculosis–specific IFN-γ⁺ CD4⁺ blood cells suggested an affiliation with nonclassical Th1 lymphocytes. Another marker of nonclassical Th1 cells is CD161, a lectin-like receptor whose expression is indicative of current or past production of IL-17 (33, 36). Examination of CD161 showed that, in TB patients, 23.3% [17.9–30.4%] of CD4⁺ blood lymphocytes were CD161⁺ (Fig. 3A, 3B). The frequencies of CD161⁺ lymphocytes were higher in R6⁺ populations compared with R6⁻ populations (p < 0.0001, Fig. 3A, 3B, i.e., there was a link between the expression of the two markers of Th17/ex-Th17 lymphocytes [R6 and CD161]).

Evaluation of CD161 expression by IFN-γ/IL-17–producing lymphocytes showed that CD161⁺ cells formed the majority of IFN-γ⁺IL-17⁺ lymphocytes (53.3% [46.2–90.0%]). CD161⁺ cells were also present within IL-17⁺ and IFN-γ⁺ lymphocytes (28.6% [16.3–52.3%] and 17.9% [11.6–29.4%], respectively, Fig. 3C). Of note, procedures used to identify intracellular cytokines hampered CD161 expression (Supplemental Fig. 1C, 1D), yet CD161⁺IFN-γ⁺ cells could be identified. The percentages of CD161⁺ cells within IFN-γ–producing lymphocytes were close to numbers reported by other investigators (33). From these experiments, we concluded that at least some M. tuberculosis–specific IFN-γ⁺ lymphocytes were CD161⁺ and could be considered ex-Th17 cells (33).

On the basis of CD161, X3, and R6 coexpression, CD4⁺ lymphocytes could be divided into eight subsets. IFN-γ⁺ lymphocytes were most frequent within the X3⁺R6⁺CD161⁻ and X3⁺R6⁺CD161⁺ populations, whereas the frequencies of IFN-γ⁺ cells within the X3⁺R6⁻CD161⁻ population (expressing the phenotype of classical Th1 cells) were significantly lower (p < 0.002 and p < 0.02, respectively, Fig. 3D).

Thus, detailed phenotypic analysis suggested the nonclassical nature of a substantial portion of M. tuberculosis–specific IFN-γ⁺ CD4⁺ lymphocytes circulating in the blood during active TB.

Th1 lymphocytes are characterized by the expression of the transcriptional master regulator T-bet. We next analyzed whether IFN-γ⁺ lymphocytes expressing the X3⁺R6⁺ phenotype expressed T-bet. Blood cells were incubated with PPD and stained for surface CD4, X3, and R6, as well as intracellular T-bet and IFN-γ. T-bet^hi cells could be easily identified within CD4⁺ lymphocytes (Fig. 4A). Application of fluorescence-minus-one control resulted in the detection of cells expressing T-bet at intermediate levels. T-bet^hi cells were more frequent within the X3⁺R6⁻ and X3⁻R6⁻ populations than within the X3⁺R6⁺ and X3⁻R6⁺ populations (Fig. 4B). T-bet^int cells were present within all four X3-R6 populations. All IFN-γ⁺ lymphocytes, including those with the X3⁺R6⁺ phenotype, expressed T-bet (Fig. 4C). The results confirmed the Th1 lineage of IFN-γ⁺X3⁺R6⁺ lymphocytes and demonstrated higher expression of T-bet by CD4⁺ lymphocytes lacking R6.

During infections, Ag-specific lymphocytes preferentially accumulate at the infectious foci. Currently, little is known about the prevalence and phenotypic markers of Th1, Th17, and Th1Th17 lymphocytes located at the site of pulmonary TB. To characterize these populations, we obtained samples of lung tissue surgically resected from TB patients (n = 12) and prepared lung cell suspensions. Blood samples were also obtained from each patient on the day of surgery. Lung and blood cells were stained to determine surface expression of X3, R6, and differentiation markers by CD4⁺ lymphocytes and were incubated with PPD and stained for surface markers and intracellular IFN-γ and IL-17.

On X3-R6 dot plots, surface stained (SS) lung CD4⁺ lymphocytes appeared as a diffuse population expressing various levels of X3 and largely negative for R6 (Fig. 5A). Although some cells fell into “R6⁺” gates, they represented the extent of the R6⁻ population rather than a real R6⁺ subset (Fig. 5A, patients 1 and 2; all gates were placed based on unstained and single-stained lung cell controls). Only in one patient were R6⁺ cells identified (Fig. 5A, patient 3). Analysis showed fewer X3⁻R6⁺ cells (p < 0.005) and more X3⁺R6⁻ cells (p < 0.05) in the lungs compared with blood (Fig. 5B) (i.e., the phenotype of lung lymphocytes was biased toward a classical Th1 type).

The frequencies of IFN-γ⁺, IFN-γ⁺IL-17⁺, and IL-17⁺ lymphocytes in the lungs were significantly higher compared with blood (24-, 26-, and 16-fold higher, respectively, Fig. 5C, 5D). As in blood, IFN-γ⁺ lymphocytes in the lungs were the main cytokine-producing cells (11.4% [2.1–34.4] compared with 0.31% [0.11–0.73%] and 0.30% [0.20–0.83%] for IFN-γ⁺IL-17⁺ and IL-17⁺ lymphocytes, respectively). Of note, IFN-γ⁺IL-17⁺ lymphocytes (absent from blood) were readily detected in the lungs (Fig. 5C, 5D).

Examination of X3/R6 expression by lung IFN-γ⁺ cells showed that all were X3⁻R6⁻. However, a thorough comparison of X3/R6 expression by SS and intracellularly stained (ICS) lung cells showed that procedures used to identify cytokine-producing cells altered chemokine receptor expression (i.e., decreased the expression of X3 and resulted in a complete loss of R6, Supplemental Fig. 1E, 1F). Mechanisms underlying these alterations could be different [e.g., include receptor shedding or internalization (37)], but in any case, the X3/R6 phenotype of lung IFN-γ⁺ cells could not be determined reliably.

CD4⁺ lung lymphocytes are known to contain more effectors and be much more differentiated than lymphocytes circulating in blood (38, 39). We supposed that altered expression of R6 by SS CD4⁺ lung lymphocytes could be due to a higher degree of differentiation and loss of R6 expression during the differentiation process. To test this hypothesis, we examined the differentiation status of X3-R6 populations. Within each population, effector (CD62L⁻) lymphocytes were identified, and their coexpression of CD27 and CD28 was determined. This approach allowed us to identify early effectors (EEs; CD27⁺CD28⁺), late effectors (LEs; CD27⁻CD28⁺), and terminally differentiated effectors (TDEs; CD27⁻CD28⁻, Fig. 6A) (40–42). Analyses were performed on freshly isolated blood and lung cells using a surface staining procedure (to avoid alteration of X3/R6 expression by lung cells during the intracellular staining procedure).

In the blood, CD62L⁻ effectors were rare within the X3⁻R6⁻ population (median < 5%), but they constituted >25% of other X3-R6 populations (Fig. 6C). Effectors that expressed X3⁺R6⁺ and X3⁻R6⁺ phenotypes (i.e., R6 expressing) were primarily composed of EEs and LEs and did not contain TDEs (TDE percentages within X3⁺R6⁺ and X3⁻R6⁺ effectors were 0.3% [0.2–0.9%] and 0.3% [0.1–0.8%], respectively, Fig. 6B, 6D). In contrast, effector cells belonging to the X3⁺R6⁻ and X3⁻R6⁻ populations (i.e., R6⁻) contained fewer EEs and LEs and a substantial proportion of TDEs (27.3% [11.9–45.4%] and 58.8% [21.8–70.1%], respectively, p < 0.0001 versus R6⁺ populations). A comparison of X3⁺R6⁺ and X3⁺R6⁻ effectors showed that the former contained more EEs (p < 0.0001) and LEs (p < 0.005) but fewer TDEs (p < 0.0001) (i.e., were less differentiated). In turn, X3⁺R6⁻ effectors were less differentiated than X3⁻R6⁻ effectors: they contained more LEs (p < 0.0001) and fewer TDEs (p < 0.01). Flow cytometric analysis of the expression of X3/R6 by EEs, LEs, and TDEs showed that all TDEs lacked R6 and exhibited decreased expression of X3 (phenotype, X3^lo/−R6⁻, Fig. 6B).

In the lungs, all four X3-R6 “populations” (i.e., cells that formally fall into the corresponding gates) were much more differentiated compared with blood: the cells contained larger proportions of CD62L⁻ effectors (>70%, Fig. 6E) and lower percentages of EEs (<25% of effector CD62L⁻ cells, Fig. 6F). TDEs were found within all four X3-R6 populations of lung effector lymphocytes and formed a substantial part of these populations (Fig. 6F). There was no significant difference in the differentiation profile of X3-R6 populations in the lungs. This was in good agreement with our previous notion that CD4⁺ lung lymphocytes formed one homogeneous population.

Overall, our analysis revealed that R6⁻ CD4⁺ blood lymphocytes were significantly more differentiated than R6-expressing lymphocytes, that the degree of differentiation of blood X3-R6 populations increased in the order of X3⁺R6⁺ < X3⁺R6⁻ < X3⁻R6⁻, and that CD4⁺ lung lymphocytes were uniformly highly differentiated and reminiscent of blood X3^+/−R6⁻ populations.

A higher degree of differentiation of R6⁻ populations compared with R6⁺ populations suggested that R6⁻ populations could originate from R6⁺ precursors. To address this hypothesis, we sorted X3⁺R6⁺ and X3⁺R6⁻ populations of CD4⁺ blood lymphocytes (three donors, two independent experiments). Cells were stimulated with anti-CD3/anti-CD28 Abs in the presence of CFSE-labeled APCs or were left unstimulated. On days 1 and 5, the levels of type-1 and type-17 cytokines in culture supernatants were evaluated, and the percentages and numbers of cells belonging to the four X3-R6 populations were determined.

Sorted X3⁺R6⁻ cells produced IFN-γ and TNF-α but virtually no IL-17A/F. In contrast, sorted X3⁺R6⁺ lymphocytes produced IFN-γ, TNF-α, and a significant amount of IL-17A/F (Fig. 7A). The results were in accordance with what was expected for the sorted populations and documented the Th1 profile of X3⁺R6⁻ lymphocytes and the polyfunctional properties of X3⁺R6⁺ cells [also reported by other investigators (33, 43)].

In cultures started with X3⁺R6⁺CD4⁺ lymphocytes, the frequencies and numbers of X3⁺R6⁻CD4⁺ cells were initially low and remained low in unstimulated cultures, but they increased significantly in anti-CD3/anti-CD28–stimulated cultures (3–17-fold increase in percentages; 6–41-fold increase in numbers, Fig. 7B, 7C). There was also an increase in X3⁻R6⁻CD4⁺ cells in stimulated cultures (3–9-fold increase in percentages, 7–22-fold increase in numbers, Fig. 7C). Concomitantly with the accumulation of X3⁺R6⁻ and X3⁻R6⁻ cells, the percentages and numbers of X3⁺R6⁺ lymphocytes decreased (2–33-fold).

Because sorted X3⁺R6⁺ populations were initially contaminated with some amount of X3⁺R6⁻ cells, it could not be excluded that the latter may have expanded by day 5 and gave rise to the accumulating X3⁺R6⁻ population. However, in cultures started with X3⁺R6⁻ CD4⁺ lymphocytes, no increase in the numbers of X3⁺R6⁻ cells in response to anti-CD3/anti-CD28 stimulation was noted (Fig. 7C); X3⁻R6⁻ cells increased slightly in two of three cultures (Fig. 7C, culture 2). These results strongly suggested that the accumulation of X3⁺R6⁻ cells in cultures started with X3⁺R6⁺ CD4⁺ lymphocytes could not be attributed to a better survival or proliferation of X3⁺R6⁻ lymphocytes present in initial cultures; rather, it was due to X3⁺R6⁺ → X3⁺R6⁻ cell transition.

Our data demonstrated that effector CD4⁺ lymphocytes expressing the X3⁻R6⁻ phenotype were highly/terminally differentiated and could be derived from X3/R6-expressing precursors, at least in vitro. In contrast, IFN-γ⁺ lymphocytes were present within the X3⁺R6⁺ and X3⁺R6⁻ populations but were not detected within the X3⁻R6⁻ population. This raised the question of whether X3⁻R6⁻ lymphocytes were functionally active. One possibility was that the cells were exhausted; however, we noticed that the X3⁻R6⁻ population contained many fewer effector cells compared with the X3⁺R6⁺ and X3⁺R6⁻ populations (Fig. 6C). Therefore, we reasoned that the seeming absence of IFN-γ⁺ X3⁻R6⁻ cells could be due to a low proportion of effector cells within this subset. To test this hypothesis and analyze the functional properties of the X3⁻R6⁻ population in more detail, we compared the frequencies of IFN-γ⁺ cells specifically within effector (i.e., CD62L⁻) cells expressing the X3⁺R6⁺, X3⁺R6⁻, and X3⁻R6⁻ phenotypes. We found that IFN-γ⁺ cells were most frequent within X3⁺R6⁺ effectors (p < 0.003 versus X3⁺R6⁻ effectors, p = 0.02 versus X3⁻R6⁻ effectors). The frequencies of IFN-γ⁺ cells within X3⁻R6⁻ effectors were similar to those detected within X3⁺R6⁻ effectors (i.e., classical Th1 cells, Fig. 8A). Because the frequencies of cytokine-producing cells do not directly mirror the levels of cell activity, we next sorted X3⁻R6⁻ and X3⁺R6⁻ effector cell populations and analyzed their culture supernatants for the presence of type 1 cytokines IFN-γ and TNF-α. We found that X3⁻R6⁻ effectors produced type 1 cytokines even more efficiently than did X3⁺R6⁻ effectors (Fig. 8B). Thus, our initial data showing low frequencies of IFN-γ⁺ cells within the X3⁻R6⁻ population (Fig. 2C) likely resulted from a low content of effector cells within this population. To adjust for the different content of effector cells within the four X3-R6 populations, we calculated the absolute numbers of IFN-γ⁺ cells expressing different X3-R6 phenotypes in the blood (i.e., cells per microliter of blood). X3⁺R6⁺ IFN-γ⁺ cells appeared to be most prevalent (Fig. 8C), confirming our initial data.

In this study, we examined Th1, Th17, and Th1Th17 cells during TB in humans. Our specific aims were to evaluate the extent of their responses, examine them in pulmonary TB lesions, and characterize the classical versus nonclassical nature of the Th1 population specific to M. tuberculosis.

Th1 cells are known to mediate protection against M. tuberculosis infection. In contrast, there are contradictory data on the role, generation, and extent of Th17 responses during TB. Different studies reported elevated or decreased levels of Th17/IL-17 in TB patients or did not detect Th17/IL-17 during TB and LTBI (20–25, 27). Evaluation of Th17 (IL-17⁺) and Th1Th17 (IFN-γ⁺IL-17⁺) lymphocytes in this study showed that both populations were extremely rare in the blood of TB patients. In contrast, IFN-γ⁺ lymphocytes accounted for a large proportion of M. tuberculosis–specific CD4⁺ cells in blood. The results are in line with the data obtained by Perreau et al. (27), who reported that M. tuberculosis–specific Th17 lymphocytes were undetectable in peripheral blood of TB patients.

In contrast to blood, in the focus of pulmonary TB infection, IL-17⁺ and IFN-γ⁺IL-17⁺ lymphocytes could be easily identified; their frequencies were 16–24-fold higher compared with blood. To the best of our knowledge, this is the first description of Th1Th17 and Th17 populations in pulmonary lesions of TB patients. The data suggest the involvement of these populations in local immune responses; however, it is not clear whether and to what extent the cells contribute to local pathology or protection. Indeed, Th17 lymphocytes are known to recruit neutrophils, whose accumulation during TB has been associated with pathology and tissue destruction (44–46). In contrast, Th17 lymphocytes can confer protection against M. tuberculosis infection and contribute to the induction of Th1 responses (18, 28). Th1Th17 lymphocytes have been implicated in tissue pathology during various autoimmune diseases (31, 47), but their role during TB remains uncertain (48).

Our analysis of Th1 (IFN-γ⁺) lymphocytes in blood showed that a considerable portion expressed the X3⁺R6⁺ phenotype, which is characteristic of nonclassical (ex-Th17) Th1 cells (30, 33). Numerically, IFN-γ⁺X3⁺R6⁺ cells were also more prevalent than other X3-R6 populations of IFN-γ⁺ cells. This corresponds to the results by Arlehamn and colleagues (15, 49), who first described the X3⁺R6⁺ phenotype of M. tuberculosis–specific IFN-γ⁺ T cells. However, in the study by Arlehamn et al. (15), IFN-γ⁺X3⁺R6⁺ cells were identified during LTBI and were more numerous in LTBI subjects compared with healthy individuals. It was suggested that the cells can be considered protective and mediate LTBI control. In our study, X3⁺R6⁺ cells constituted more than half of M. tuberculosis–specific IFN-γ⁺ lymphocytes during active TB. Thus, it is unclear whether the X3⁺R6⁺ phenotype is a specific feature of protective Th1 cells that persist during LTBI or a general characteristic of M. tuberculosis–specific Th1 lymphocytes. Direct comparisons of Th1 phenotypes during LTBI and active TB are needed to clarify this question.

A similar comparison has recently been performed by Strickland et al. (50). They analyzed X3/R6/CCR4 expression by M. tuberculosis–specific CD4⁺ lymphocytes in TB patients and LTBI subjects and found that TB patients exhibited decreased X3⁺R6⁺ populations and expanded X3⁻R6⁻ populations. These data agree with some of our findings (discussed below); however, they seem to contradict our results on the prevalence of the X3⁺R6⁺ population of IFN-γ⁺ lymphocytes during TB. In this regard, we should note that we have not compared TB patients and LTBI subjects (the study is ongoing). Thus, the prevalence of the IFN-γ⁺X3⁺R6⁺ population reported in this study does not mean that this population was not reduced in TB patients compared with LTBI subjects. Additionally, there are several methodological differences between the two studies that can account for the discrepancy. Particularly, Strickland et al. (50) identified M. tuberculosis–specific lymphocytes using tetramer staining (i.e., detected cells irrespective of their ability to produce cytokines). Our analysis of Th1 cells was specifically focused on IFN-γ⁺IL-17⁻ lymphocytes. Strickland et al. (50) examined CD4⁺ lymphocytes specific to ESAT-6/CFP-10 Ags. In our study, cells responding to all Ags present in PPD were analyzed. Thus, it is likely that different populations of M. tuberculosis–specific CD4⁺ cells were analyzed for their expression of X3/R6 in the two studies.

Another finding that should be discussed is our demonstration that the X3⁻R6⁻ population contains functionally active lymphocytes producing type 1 cytokines. Th1 lymphocytes are usually considered X3⁺R6⁻ or X3⁺R6⁺. The possibility that Th1 cells may lack X3 expression is rarely discussed. Our results suggest that this is largely due to the low content of effector lymphocytes within the X3⁻R6⁻ population, which masks the presence of IFN-γ⁺X3⁻R6⁻ Th1 cells in the blood. To summarize, existing data are not plentiful and are not entirely consistent; however, they indicate that chemokine receptor expression may appear to be an important characteristic of M. tuberculosis–specific CD4⁺ cells and a potential indicator of M. tuberculosis infection activity. How and why the pattern of chemokine receptor expression is associated with M. tuberculosis infection activity are interesting topics to discuss.

We have found that X3-R6 populations circulating in blood differ by their degree of differentiation. Specifically, the degree of differentiation of blood CD4⁺ cells increased in the following order: X3⁺R6⁺ < X3⁺R6⁻ < X3⁻R6⁻. These results can be explained in several ways. One is that X3⁺R6⁺, X3⁺R6⁻, and X3⁻R6⁻ populations represent sequential stages of effector cell differentiation, with X3⁻R6⁻ cells being the most differentiated. Another explanation is that X3⁺R6⁺, X3⁺R6⁻, and X3⁻R6⁻ populations represent independent sublineages of CD4⁺ lymphocytes that, for some reason, have reached distinct differentiation stages (e.g., because they have differentiated in distinct microenvironments). It is also possible that the X3⁺R6⁺ and X3⁺R6⁻ populations develop independently but transit into the X3⁻R6⁻ state at late stages of their differentiation. Our culture experiments showed that sorted CD4⁺ X3⁺R6⁺ lymphocytes converted their phenotype to X3⁺R6⁻ and X3⁻R6⁻ in vitro, supporting the first hypothesis. However, changes in the cell surface phenotype observed in vitro do not necessarily mirror ongoing processes in vivo. Thus, more investigations are needed to gain insight into the relationships among the X3⁺R6⁺, X3⁺R6⁻, and X3⁻R6⁻ populations during TB.

A number of studies have demonstrated Th17 lymphocyte plasticity and their capacity to differentiate into X3⁺R6⁺ Th1Th17 and/or nonclassical (ex-Th17) Th1 lymphocytes (33, 51–54). Thus, if the capacity of X3⁺R6⁺ Th1 lymphocytes to convert into X3⁺R6⁻/X3⁻R6⁻ Th1 cells is confirmed in further studies, this will indicate that Th17 → Th1 differentiation is possible. Certainly, this hypothesis requires careful examination; however, transition of Th17 lymphocytes into Th1-like cells has been recently demonstrated directly in a mouse model of colitis (55).

Independently of the relationships among the X3⁺R6⁺, X3⁺R6⁻, and X3⁻R6⁻ populations, the significance of our data is that they have linked the pattern of X3/R6 expression to cell-differentiation status. Previously, several studies have associated the degree of differentiation of Ag-specific CD4⁺ cells with M. tuberculosis infection activity (41, 56). In the aggregate, this links X3/R6 expression with M. tuberculosis infection activity and suggests a mechanism to explain the reduction in the X3⁺R6⁺ population and the increase in the X3⁻R6⁻ population during active TB [observed in a recent study by Strickland et al. (50)].

Our study has some limitations that should be noted and that should provide opportunities for future research. First, as a result of the limited number of fluorochromes that we could detect simultaneously and our attempt to minimize artifacts and the usage of tandem dyes, we restricted our analysis to two chemokine receptors (X3 and R6, and we did not examine CCR4). However, CCR4 is a marker of the Th2 population, which we did not analyze in this study. In addition, identification of Th1Th17 populations in this study was based on X3/R6 coexpression, as well as by detecting intracellular IFN-γ and IL-17; thus, we find our approach appropriate. Another limitation is that, due to problematic sorting of live cytokine-producing cells, our hypothesis about the generation of Th1 lymphocytes from IL-17–producing precursors is based on indirect observations: X3⁺R6⁺ → X3⁺R6⁻ transition in vitro and the higher degree of differentiation of the X3⁺R6⁻ population compared with the X3⁺R6⁺ population. Further studies will be needed to directly address that hypothesis. Finally, only patients with active TB were included in the study. A comparison of Th1, Th1Th17, and Th17 populations in TB patients and healthy M. tuberculosis–exposed and unexposed individuals is a subject of ongoing investigations.

In aggregate, we have demonstrated the lack of evident Th17 and Th1Th17 populations in the blood of TB patients and their presence in pulmonary TB foci; documented the prevalence of IFN-γ⁺X3⁺R6⁺ lymphocytes in patients with recent TB; showed that Th1 lymphocytes may lack X3 expression (i.e., be X3⁻R6⁻); described the pattern of X3/R6 expression by CD4⁺ lymphocytes residing in pulmonary TB foci for the first time, to our knowledge; and documented an association between the pattern of X3/R6 expression and the degree of differentiation of Th1 lymphocytes. The results provide new clues for understanding the relationships between distinct Th cell populations during TB and add to our understanding of immune mechanisms operating during TB.

We thank Dr. Tatevik Bagdasarian and Dr. Irina Burmistrova for providing clinical information and Dr. Larisa Chernousova and Dr. Tatiana Smirnova for providing microbiological data. We are grateful to Nadezda Kolokolova and Polina Tugusheva for technical assistance.

The authors have no financial conflicts of interest.