Immunophenotypic characterization of peripheral T lymphocytes in Mycobacterium tuberculosis infection and disease

Abstract

The cellular immune response probably plays a pivotal role in determining the clinical outcome after exposure to Mycobacterium tuberculosis. We used multi-parameter flow-cytometry to evaluate the distribution of T-lymphocyte subsets during infection and disease caused by M. tuberculosis. Samples were obtained from 71 volunteers to identify the T CD4⁺ and CD8⁺ lymphocyte numbers, and the activation plus memory/naïve phenotypes, as defined by CD38, HLA-DR, CD45RA and CD27 markers. Subjects were divided into 18 healthy volunteers without detectable reaction to purified protein derivative (PPD⁻), 18 health care workers with a recent conversion to PPD, 20 patients with active pulmonary tuberculosis (TBC) and 15 patients with treated TBC at 6 months of therapy. By multiple-comparison analyses, the T CD4⁺ lymphocyte number of the TBC group was lower than the PPD^– group (P < 0·05). This difference was apparently lost after treatment. The higher and the lower number of naïve T CD4⁺ cells was observed in the PPD^– and TBC group, respectively. CD8⁺ T lymphocytes were also statistically different among the four groups (P = 0·0002), lower in the TBC group (P < 0·05). CD8⁺ T lymphocyte activation was evaluated by the CD38 and HLA-DR surface expression. The percentage distribution of these markers was statistically different between the four groups (P = 0·0055). TBC patients had a higher percentage of CD38⁺ cells and mean fluorescence index, suggesting an overall increase of cell activation. These results suggest that peripheral T lymphocytes reflect cellular activation during TBC, along with possible redistribution of naïve, memory/effector and late differentiated memory/effector phenotypes in the peripheral blood after infection and disease caused by M. tuberculosis.

INTRODUCTION

Tuberculosis is a major global public health problem and one of the leading causes of infectious disease-related deaths in the world. Around 12 million cases are reported every year, with an estimate of 30 million deaths between 1994 and 2004 [1]. Most cases are concentrated in developing countries, including India, China, Nigeria, Bangladesh, Pakistan, Philippines, Congo, Russia and Brazil.

A better understanding of the mechanisms governing the host reaction to the pathogen and control of disease is fundamental for the development of new and more immunogenic vaccines and adjunctive therapies. It has been recognized that the cellular immune response, particularly by T lymphocytes, plays a central role in controlling Mycobacterium tuberculosis replication [2]. T cell-deficient mice have demonstrated increasing susceptibility to disease [3], similar to the higher susceptibility to disease development observed in immunodeficient patients [4]. Lymphocyte cloning experiments [5] and passive T-cell transfer [6] have demonstrated the reactive capacity and protective ability of CD4⁺ cells. This T-cell activity appears to work via cytotoxicity against infected target cells [7]. The effect is at least partially related to T cell interferon gamma (IFNγ) production in response to M. tuberculosis antigens [8]. Also, the CD8⁺ T lymphocyte cytotoxic effect is strongly related to the host capacity to block the development of disease due to M. tuberculosis. β₂-microglobulin knock-out mice are highly susceptible to disease [9], and protection by CD8⁺ lymphocytes has been demonstrated in a BALB-c mouse passive transfer model [10]. Similar to CD4⁺ lymphocytes, IFNγ production seems to be a key marker for the ability of CD8⁺ lymphocytes to exert their cytotoxic activity in mice and humans [11,12]. Some reports have shown that CD1-restricted CD8⁺ T lymphocytes can recognize distinct M. tuberculosis antigens and play an important role against the pathogen [13]. Although still a matter of debate, CD4^–CD8^–TCRαβ⁺ T cells may also participate in the defense line against M. tuberculosis[14], probably through their recruitment to infected tissue [15], reactivity to bacterial proteins [16] and cytokine release in response to M. tuberculosis antigens [17].

The immune response after M. tuberculosis infection and disease may be captured by the measurement of T-lymphocyte phenotypes in the human peripheral blood. CD4⁺ and CD8⁺ T-cell numbers are decreased in patients with active tuberculosis [18–21]. These results were reproduced in children with tuberculosis when compared with healthy controls who had a positive delayed-type hypersensitivity (DTH) skin test response to purified protein derivative (PPD) [22]. Some reports have suggested a correlation between the CD4⁺ T-lymphocyte depletion and the severity of tuberculosis [21,23].

In that context, CD4⁺ and CD8⁺ T-lymphocyte counts were determined in healthy subjects who had a repeatedly negative PPD, recent converters, patients with active pulmonary tuberculosis and patients evaluated after 6 months of successful therapy against the disease. The immunophenotyping was expanded to determine the naïve and memory/effector subpopulations of CD4⁺ T lymphocytes and the activation status of CD8⁺ T lymphocytes.

MATERIALS AND METHODS

Volunteers

Written informed consent was obtained from all the participants according to the guidelines of the Brazilian Ministry of Health. Five groups of volunteers were enrolled. The first consisted of 18 healthy volunteers, who had two consecutive non-detectable delayed-type hypersensitivity (DTH) intradermal reactions to purified protein derivative (PPD). These were performed 6–12 months apart (PPD negative group). The second consisted of 18 health care workers who had a positive DTH response to PPD after a non-reactive response 6–12 months earlier, defining a probable recent infection by M. tuberculosis (PPD positive group). These subjects were identified in a prospective cohort of 2500 workers at the Federal University of São Paulo (Dr E. A. S. Medeiros). The third group consisted of 20 patients with recently diagnosed active, typical pulmonary tuberculosis (active TBC group). The fourth consisted of 15 patients with successfully treated, typical pulmonary tuberculosis at 6 months of therapy (treated TBC group). All the subjects were sero-negative for human immunodeficiency virus infection. The male/female distribution was 13/5 in the PPD negative group, 11/7 in the PPD positive group, 7/13 in the active TBC group and 6/9 in the treated TBC group (P = NS). The mean age in years (range) was 32(18–58), 34(20–58), 33(21–53) and 40(22–58) for the PPD negative, PPD positive, active TBC and treated TBC groups, respectively (P = NS).

DTH was performed by intradermal injection on the forearm of 5 TU (0·1 ml) of PPD (Tubersol, Connaugh Laboratories, Willowdale, Ontario, Canada). Induration was read 48–72 h after the injection, and was measured in millimetres at the greatest diameter. Those patients with DTH readings considered to be non-reactive (less than 5 mm) were submitted to a booster DTH within 14 days. If the booster DTH was again non-reactive, the test was considered negative. Subjects with a previous non-reactive DTH to the PPD performed 6–12 months earlier were screened for conversion, defined as an increase in induration of at least 10 mm.

Active pulmonary tuberculosis was defined by a medical history and clinical findings compatible with pulmonary tuberculosis, a thoracic roentgenogram showing lung involvement suggestive of tuberculosis, and the presence of at least twosputum smear positive for acid-fast bacilli or the growth of M. tuberculosis.

Monoclonal antibodies and sample preparation

HLA-DR fluorescein isothiocyanate (FITC, clone L243), CD27 phycoerythrin (PE, clone L128), CD45RA FITC (clone L48), CD8 peridin chlorophyll protein (PerCP, clone SK1), CD4 PerCP (clone SK3) and CD3 allophycocyanin (APC, clone UCHT1) monoclonal antibodies were obtained from Becton Dickinson Immunocytometry Systems (BDIS, San Jose, CA, USA), and CD38 FITC (clone HIT2) was obtained from PharMingen (San Diego, CA, USA). EDTA-treated blood (100 μl) was incubated at room temperature with a combination of monoclonal antibodies for 15 min in the dark, and then treated with haemolysis buffer for further a 10 min. Cells were washed and resuspended in phosphate saline buffer supplemented with 0·1% sodium azide for cytometric analysis.

For blood T CD4⁺ and CD8⁺ lymphocyte absolute counts, a TriTest and TrueCount reagent kit (BDIS) were used according to the manufacturer’s instructions. Briefly, TrueCount tubes received appropriate amounts of TriTest reagents, containing CD3/CD4/CD45 or CD3/CD8/CD45, and 50 μl of blood, and were incubated for 15 min at room temperature. Lysing solution (500 ml) was added and the tubes were incubated for 15 min before acquisition. Samples were acquired and analysed using Multiset software (BDIS).

Cell samples were analysed on a FACSCalibur flow cytometer (BDIS) equipped with an argon and a diode laser for four-colour detection. Acquisition and analyses were performed using CellQuest software (Becton Dickinson). Fluorescence voltages and compensation values were determined using singly fluorochrome-stained cells from a healthy volunteer. The subpopulation of naïve, effector and memory cells were analysed by the absolute number and percentage of cells expressing different combinations of CD45RA and CD27 in the CD4⁺ T lymphocytes. The activation of CD8⁺ T lymphocytes was determined by the expression of CD38 and HLA-DR molecules. The state of activation of CD8⁺ T lymphocytes was also analysed by a CD38 expression index, calculated by dividing the mean fluorescence intensity of the CD38 FITC-stained cells by the mean fluorescence intensity of the FITC-conjugated isotype control.

Statistical analysis

Statistical analysis used Statistica (StatSoft, Tulsa, OK, USA) and NCSS (Keysville, UT, USA) software. Group results were compared using Kruskal–Wallis one way anova on ranks test. For those variables identified as having a P < 0·05, multiple pair-wise group comparisons were further performed using as threshold for statistical significance a Z-value calculated with the Bonferroni correction.

RESULTS

From August 1999 to July 2000, 71 subjects were enrolled in the following four groups: PPD negative (n = 18), PPD positive (n = 18), active TBC (n = 20) and treated TBC (n = 15). Whereas the first two groups were composed mostly of women (72% and 64%), the latter two had a higher percentage of males (64% and 60%), without statistically significant difference (P = 0·081). Age was evenly distributed in the four groups.

CD4⁺ T lymphocytes were significantly different among the four groups (P = 0·0004). The PPD negative group was significantly higher than the active TBC group (P < 0·05; pair-wise comparison), suggesting an association between disease and low CD4 T lymphocyte counts (Fig. 1). This difference was lost after treatment.

Fig. 1.

Absolute peripheral blood numbers of CD4⁺ and CD8⁺ T lymphocytes. The box plot represents the median, 25/75 percentile, 10/90 percentile, 5/95 percentile and extreme values. Arrows above the plots represent P-values calculated by the Kruskal–Wallis one way anova on ranks test. Arrows inside the plot show statistically significant differences between groups using the Bonferroni correction.

Statistically significant differences were observed among the four groups in the analysis of the number of naïve (CD45RA⁺CD27⁺) (P = 0·005), memory/effector (CD45RA^–CD27⁺) (P = 0·002) and late differentiated memory/effector (CD27^–) (P = 0·002) subsets within the CD4⁺ T-lymphocyte population. Patients with active TBC had the lowest numbers of CD4⁺ T cells and their naïve, effector and late differentiated memory subsets. This was in contrast to the PPD negative subjects who had the highest counts. Pair-wise comparisons between these two groups were statistically significant for the total number of CD4⁺ T lymphocytes and the subsets of memory and naïve cells (Figs 1 and 2). No differences were noted when the percentage instead of the absolute counts of naïve, late differentiated memory and effector populations were analysed (data not shown).

Fig. 2.

Absolute peripheral blood numbers of CD4⁺ T-lymphocyte subpopulations. CD45RA⁺CD27⁺ denote naïve cells, CD27^– late differentiated memory/effector cells and CD45RA^–, CD27⁺ memory/effector cells. The box plot represents the median, 25/75 percentile, 10/90 percentile, 5/95 percentile and extreme values. Arrows above the plots represent P-values calculated by the Kruskal–Wallis one way anova on ranks test. Arrows inside the plot show statistically significant differences between groups using the Bonferroni correction.

The CD8⁺ T lymphocyte counts were also significantly different among the four groups (P = 0·0002), and highest in the PPD negative group when compared with the other groups (Fig. 1). CD8⁺ T-lymphocyte activation was evaluated by the surface expression of CD38 and HLADR. In comparison with the PPD negative group, the absolute numbers of CD8⁺CD38⁺ T cells were low in the other groups, but in percentage terms they were higher, particularly in the group of patients with active pulmonary disease, suggesting an overall increase in cellular activation in response to ongoing M. tuberculosis disease (Fig. 3). A statistically significant difference was noted in the CD38 expression index among the four groups (P = 0·0391), with higher values observed in the active pulmonary tuberculosis group (4·78 ± 1·77, 5·27 ± 4·17, 6·47 ± 3·07 and 4·10 ± 4·53 for the PPD negative, PPD positive, active TBC and treated TBC groups, respectively). Similarly to absolute numbers and percentage of HLADR⁺ in the CD8⁺ T-lymphocyte population, no statistically significant difference was observed in the HLADR expression index.

Fig. 3.

Percentage of peripheral blood CD8⁺ T lymphocytes expressing activation markers CD38 and HLA-DR. The box plot represents the median, 25/75 percentile, 10/90 percentile, 5/95 percentile and extreme values. Arrows above the plots represent P-values calculated by the Kruskal–Wallis one way anova on ranks test. Arrows inside the plot show statistically significant differences between groups using the Bonferroni correction.

DISCUSSION

In this study, a peripheral CD4⁺ T-lymphocyte depletion was observed in patients with tuberculosis and was restored to normal levels after successful chemotherapy, consistent with other observations 18-21,23,24. CD8⁺ T-lymphocyte counts were markedly decreased and also recovered after therapy. This phenomenon may reflect the immunodeficiency observed in tuberculosis patients, documented by the loss of a specific DTH response in severe cases [25], changes in cytokine production profile and decreased lymphocyte proliferative capacity [26]. However, the depletion of peripheral T cells was shown not to correlate with higher mortality [27].

Remarkably, a higher number of T cells was observed in the PPD negative group, suggesting that the infection by M. tuberculosis by itself may exert a significant impact in the immune response, resulting in changes in peripheral T-lymphocyte numbers, possibly reflecting cell homing and differentiation.

To address the phenotype of CD4⁺ T-cell depletion, the naïve, effector and memory subpopulations were analysed. A drop in all three phenotypic populations was noted in tuberculosis patients. Between PPD negative and PPD positive healthy volunteers, there was a significant difference only in the naïve subpopulation. However, samples from patients with tuberculosis were also associated with effector/memory CD4⁺ T-cell loss when compared with samples from PPD positive subjects. These observations may reflect a higher cell differentiation and tissue homing after infection and disease caused by M. tuberculosis, respectively. Lymphocyte homing has been described during tuberculosis in lymph nodes [28], lungs [29] and pleural spaces [30]. Alternatively, T-lymphocyte depletion may be explained by impaired thymus function [31] or increased apoptosis levels [32].

Due to the importance of cytolytic activity of CD8⁺ T cells in M. tuberculosis infection, the state of cellular activation of this population of lymphocytes was assessed by the detection of surface activation markers. CD38 is an ectoenzyme with an unknown function, but its presence is associated with cellular activation [33]. A higher percentage of cells expressing the CD38 molecule was observed in patients with active tuberculosis, which returned to control levels after therapy. However, no difference was noted in HLA-DR expression among groups. Higher CD38 expression on CD8⁺ T cells has already been described in other infectious diseases, such as HIV-1 [34] and diabetes mellitus type 1 [35], whereas higher HLA-DR expression was noted in Epstein-Barr virus [36], cytomegalovirus [37] and HTLV-I infections [38]. To our knowledge, this is the first time that this activation phenomenon has been described in HIV-negative patients with tuberculosis using CD38 marker.

This transitory state of activation of CD8⁺ T lymphocytes during active infectious diseases, including tuberculosis, possibly mirrors the cytotoxic activity against intracellular pathogens [10,39]. Studies in animal models have shown that CD8⁺ T-lymphocyte deficiency can result in susceptibility to tuberculosis [9], and that CD8⁺ T cells can contain the dissemination of M. tuberculosis[40]. In humans, antigen-specific IFNγ-producing CD8⁺ T cells have been identified, which have a potential cytotoxic effect [12]. However, protection of human hosts against the pathogen played by these cells in vivo still needs to be clearly demonstrated.

The results presented here reveal a marked effect in the T-lymphocyte number and distribution related to M. tuberculosis infection and disease, reflecting a shift from a naïve to a memory/effector pool of CD4⁺ T lymphocytes and a higher activation of CD8⁺ T lymphocytes. At least partially, the lymphocyte disturbances seem to return to baseline after successful treatment of the disease. The mechanisms involved in these effects need to be evaluated further, and may lead to an understanding of the pathogenesis of the disease, identification of prognostic markers and even development of adjunct immunomodulatory strategies to improve specific therapy.