T Cells from Programmed Death-1 Deficient Mice Respond Poorly to Mycobacterium tuberculosis Infection

Background Programmed Death-1 (PD-1; CD279) receptor molecule is widely believed to be a negative regulator predominantly expressed by exhausted/activated mouse T cells. Upon interaction with its ligands, PD-L1 and PD-L2, PD-1 inhibits activation of T cells and cytokine production, which has been documented in various viral and fungal infections as well as in vitro studies. Therefore, inhibition of T cell responses by PD-1 resulted in disease resistance in a variety of mouse infection models studied heretofore. Methodology/Principal Findings Here, we report that PD-1 deficient (PD-1−/−) mice infected with Mycobacterium tuberculosis (M. tb) H37Rv by the aerosol route have increased susceptibility as compared with their wild type littermates. Surprisingly, M. tb antigen-specific T cell proliferation was dramatically reduced in PD-1 deficient animals compared with wild-type littermates, and this was due to increased numbers of regulatory T cells (Tregs) and recruitment of mesenchymal stem cells. Furthermore, PD-1−/− mice exhibited decreases in the autophagy-induced LC3-B marker protein in macrophages. Conclusions/Significance Our findings suggest that PD-1 does not play an inhibitory role during M. tb infection and instead promotes mycobacterial clearance in mice.


Introduction
Mycobacterium tuberculosis (M. tb) infection is a global health problem with an estimated one third of the global population latently infected, and two million deaths annually [1,2]. Coinfection with human immunodeficiency virus (HIV) dramatically enhances the risk of mortality and poses the threat of a pandemic in this modern era [3,4]. Cellular immune responses, which are mediated by IFN-c, play a central role in protective immunity against M. tb [5,6]. Therefore, animals that are impaired in mounting T helper (Th) 1 responses or IFN-c signalling exhibit dramatic susceptibility to M. tb infection [7,8,9]. On the other hand, Th2 responses exacerbate M. tb infection [10,11]. Likewise, regulatory T cells (Tregs) have been shown to assist disease progression by inhibiting protective immune responses [12,13]. Recently, we have shown that M. tb recruits mesenchymal stem cells (MSCs) to granulomas to promote persistent infection and Treg expansion in infected mice [14]. The role of Th17 cells in M. tb is not entirely clear, however, some reports have suggested a role of IL-17 in host resistance against M. tb and granuloma formation [15]. Therefore, a delicate balance between Th1, Th2, Th17, and Treg cells determines the disease outcome in M. tb infections.
The divergent roles of CD4 + T cells (IFN-c-producing Th1 cells and IL-4-and IL-13-producing Th2 cells) in manipulating the macrophage's phagocytic activity to control M. tb infection by activating and repressing autophagy, respectively, have been investigated [11,32]. Historically, it has been known that M. tb resides in phagosomes and survives by interfering with phagolysosome biogenesis within macrophages or dendritic cells [33,34]. Autophagy has a diverse role in several pathologies to protect the host against infections, cancer and aging [35]. The identification of signals that regulate autophagy and genes that execute autophagy has assisted in the development of reagents to manipulate the autophagy pathway. Activation of mammalian autophagy factor LC3 during autophagy results in a non-soluble form called LC3-II (also called LC3-B or Atg8) that stably associates with the autophagosomal membrane [36]. Autophagy can be induced by rapamycin, which can result into colocalization of the mycobacterial phagosome with the autophagy factor LC3, and induction of autophagy and suppression of intracellular survival of bacteria [34,37,38].
Here we report that PD-1 deficient (PD-1 2/2 ) animals infected with M. tb exhibit higher susceptibility when compared to their wild type littermates. Consistently, the number of responding activated CD4 + memory T cells was dramatically lower in lung and spleen of PD-1 2/2 animals. Furthermore, production of T cell-derived cytokines was significantly higher in PD-1 2/2 animals. In addition to this, PD-1 2/2 mice recruited more MSCs and Tregs to the site of infection as compared with WT littermates. Furthermore, macrophages from these animals expressed lower levels of the autophagy marker LC3-B. Taken together, these observations suggest that, unlike most other infections, PD-1 plays an important role in culminating host resistance in M. tb infection.

Results
M. tb H37Rv augments PD-1 expression in WT mice and enhances bacterial burden in PD-1 2/2 mice It is well established that PD-1 is expressed on activated/ exhausted T cells and inhibits T cell responses. Interestingly, the amount of PD-1 expression and its engagement with ligands (PD-L1 and PD-L2) decides the threshold for T cell inactivation and the amount of cytokine production [39,40,41]. A study with human subjects indicated that substantial numbers of peripheral T cells express PD-1 in tuberculosis (TB) patients [31], and the inhibition of PD-1 resulted in higher M. tb antigen-specific proliferation in vitro. Therefore, we examined the expression of PD-1 on T cells from M. tb infected wild type C57BL/6 mice. It was found that both CD4 + and CD8 + T cells significantly upregulated expression of PD-1 when compared with uninfected littermates (Fig. 1A). Absolute numbers of CD4 + and CD8 + splenocytes in infected WT mice also showed significant differences in PD-1 expression compared with control mice (Fig. 1B). Therefore, to investigate the role of PD-1 in M. tb infection, we infected PD-1 2/2 mice and determined bacterial burden in different organs. Consistently, lung tissue histology revealed that PD-1 2/2 mice had very large and immature granulomas (Fig. 1C). Furthermore, PD-1 2/2 mice have deformed and perforated lungs (gross morphology) compared to WT after M. tb infections (Fig. 1D). Unexpectedly and paradoxically, infected PD-1 2/2 mice had higher bacterial load in spleen and lungs compared to infected wild type littermates (Fig. 1E). Therefore, unlike viral and fungal infections, the absence of PD-1 promotes susceptibility to M. tb infection.
M. tb infected PD-1 2/2 mice exhibit dramatically lower antigen-specific immune responses The higher bacterial burden in PD-1 2/2 mice prompted us to investigate the M. tb antigen-specific immune responses in these animals. We examined proliferation of splenocytes and lung cells induced by M. tb complete soluble antigen (CSA) at day 30 after infection. Surprisingly, cells from PD-1 2/2 animals exhibited notably reduced proliferation throughout disease progression in both lung and spleen ( Fig. 2A and B). However, control uninfected PD-1 2/2 mice did not exhibit any significant difference in T cell proliferation compared with wild type littermates in response to M. tb lysate protein stimulation (Fig. 2). This observation suggested that there is a defect in T cell activation in PD-1 2/2 mice. Next, we tested whether there is any defect in B cell activation in these animals. To do so, we performed B cell proliferation assays in the presence of the B cell mitogen LPS. Indeed, we observed that B cell proliferation was also abnormal in infected PD-1 2/2 animals compared to WT mice (Fig. 2C). An earlier time point at 15 days provided similar findings (data now shown). To provide insight into the mechanism of such unexpected findings, we examined the number of activated T cells in M. tb infected splenocytes. We observed that PD-1 2/2 animals contained significantly lower numbers of CD44-positive (infected PD-1 2/2 , 30.9619.6% [mean6STDEV] versus infected WT, 39.5623.3%; n = 9, p = 0.0003) and CD25-positive (14.666.9% versus 22.5611.0%; n = 9, p = 0.007) CD4 + T cells when compared with their wild type littermates in spleen (Fig. 3A, C). Similarly, in lungs, CD4 + T cells also showed down regulation of CD25 (5.962.1% versus 9.263.3%; n = 9, p = 0.05) expression in PD-1 2/2 mice. The expression of CD44 was reduced in lungs of PD-1 2/2 compared Figure 1. M. tb H37Rv augments PD-1 expression in WT mice and enhances bacterial burden in PD-1 2/2 mice compared to WT mice. Total splenocytes from WT mice infected with H37Rv were stained with anti-CD4, anti-CD8 and anti-PD-1 using fluorescently-labelled antibodies to determine the expression of PD-1 on T cell subsets by flow cytometry. A. PD-1 expression on CD4 and CD8 cells of infected and uninfected WT mice. Cells were gated on CD4 + and CD8 + T cells and analyzed for the % of cells expressing PD-1 on the gated population. The upper dot plots (top panels) show PD-1 expression for CD4 cells, and the lower dot plots (bottom panels) show PD-1 expression for CD8 cells. Data shown here is a representative experiment of three. B. Total PD-1 costimulatory molecule expression by CD4 + and CD8 + splenocytes (mean6STDEV) in uninfected and infected WT mice. PD-1 expression is significantly higher in both CD4 + (p = 0.01) and CD8 + (p = 0.03) splenocytes, as determined by two-tailed student's t test. Data shown here is a representative experiment of three. C. Gross pathology photographs of the lungs from infected WT and PD-1 2/2 mice. PD-1 2/2 mice have very distinct deformed and perforated lungs after M. tb infection. It appears that PD-1 2/2 mice develops cavities in the lungs, which is very unique to these mice compared to WT mice. D. Histology of the lung tissue sections after 30 days of infection in WT and PD-1 2/2 mice stained with both Haematoxylin & Eosin and Acid Fast dyes. WT mice show very well structured granulomas compared to PD-1 2/2 mice, whereas PD-1 2/2 mice had very large granulomas with perforated lungs (26larger in size as shown in the figure). PD-1 2/2 mice developed severe lesions in the lungs after 30 days of infection with H37Rv. E. Bacterial burden is significantly higher at day 75 in PD-1 2/2 mice compared to WT in both spleen (p = 0.0001) and lungs (p = 0.03). Data shown here is a representative of two independent experiments. Each CFU experiment has been carried out in triplicates (3 mice per experiment). doi:10.1371/journal.pone.0019864.g001 to WT animals, however no statistical significance was found (Fig. 3B). CD25 is a marker for activated T cells as well as regulatory CD4 + T cells (Tregs). Therefore, we determined expression of Foxp3, a Treg specific fork-head box transcription factor, among CD25 + T cells in M. tb infected PD-1 2/2 and wild type animals, using a Foxp3 reporter system. Tregs increased both in lungs and spleens of PD-1 2/2 mice (Fig. 4A, B). Therefore, the differences in the proliferation were likely due to the expansion of Tregs during the progression of disease. Similarly, CD25 expression was determined on CD8 + T cells, but we could not detect any difference (data not shown) between WT and PD-1 2/2 mice. Furthermore, we found that PD-1 2/2 mice recruited more MSCs to spleens compared to WT animals (Fig. 4C). Therefore, these data suggested that the numbers of T cells responding to M. tb antigens are reduced due to increased numbers of Tregs and MSCs in PD-1 2/2 mice. M. tb antigen-specific cytokine response in PD-1 2/2 animals In the previous section, we found that PD-1 2/2 animals induced lower M. tb antigen-specific T cell proliferative responses. Therefore, to examine whether antigen-specific responses exhibited a bias towards specific T helper cell subsets, we determined cytokines in the culture supernatant. We observed that PD-1 2/2 animals produced almost equal amount of cytokines irrespective of their association with the Th1, Th2, or Th17 phenotype after twenty four hours of antigen stimulation. However, when we determined the cytokines after 48 and 72 hours of antigen stimulation, we found an increase in cytokines in PD-1 2/2 mice   (Fig. 5A, B). Similarly, serum cytokines were also higher in PD-1 2/2 mice compared to WT mice (Fig. 5C). This observation suggested that M. tb antigen responding T cells in PD-1 2/2 mice acquired an anergic phenotype, but that activation of neighbouring T cells resulted in cytokine production that overcame their anergic status. Therefore, we tested the response of spleen cells in the presence of IL-2. Supplementation of IL-2 had no effect on the proliferative responses in these animals (data not shown). Thus, we tested cytokines by intracellular cytokine staining. Surprisingly, we found that cytokines were produced by non-proliferating cells in both lungs and spleens of PD-1 2/2 animals. Furthermore, intracellular cytokine staining revealed a significant increase in IL-17-producing cells in lungs whereas alterations in IL-4 and IFN-c were not significant ( Fig. 6A and B). In the case of spleen, there was a similar trend for IL-17 and IFN-c, and IL-4 was found to be significantly increased as well (Fig. 6A and B). This observation suggested that M. tb antigen-specific T cells from PD-1 2/2 animals may be defective in cellular proliferation. Furthermore, because of their inability to proliferate, we postulate that the capacity of these cells to consume cytokines is reduced, which results in the observed elevation of cytokines at late time points of antigen stimulation.

M. tb infected PD-1 2/2 macrophages do not undergo autophagy
In macrophages and DCs, it has been shown that autophagy and defence against intracellular pathogens via immunity-related GTPase Irgm1 (LRG-47) has a protective role against M. tb, resulting in clearance of the M. tb organisms from macrophages [37]. We have seen very minor differences in LC3-B expression in DCs of infected lungs of WT and PD-1 2/2 mice. However, in infected PD-1 2/2 mice CD11b + and CD11c + cells had strikingly lower LC3-B expression compared with WT mice (Fig. 7). PD-1 2/2 mice have very low autophagy and, therefore, we speculate that due to this defect these mice probably are unable to clear bacteria from the peritoneal macrophages. Nonetheless, autophagy plays a significant role in the response of PD-1 2/2 mice to M. tb infection.

Discussion
The PD-1/PD-L pathway has been extensively studied in various infection models. However, in the case of M. tb this pathway is still under exploration. Recently, it has been shown that PD-1 2/2 mice have higher mycobacterial burden compared to WT infected mice and have higher serum cytokines [42,43], and we have confirmed those findings here. When PD-1 binds to PD-L, T cells become exhausted and thereby inhibit T cell responses [22]. Various studies with viral and fungal infections reported that PD-1/PD-L interactions inhibit T and B cell proliferation [23,24,25,26,27], and inhibition of such interactions dramatically rescues T cell functions and acquired host resistance [28]. However, some studies revealed that PD-1/PD-L2 (B7-DC) interaction drives the proliferation of CD4 + and CD8 + T cells [29]. To delineate the mechanism of PD-1/PD-L action during M. tb infection, we have performed T and B cell proliferation assays and surprisingly, we found that PD-1 2/2 mice after M. tb infection exhibit defective proliferation of lymphocytes both in lungs and spleen. Studies investigating the role of PD-1 and its ligands PD-L1 and PD-L2 in patients infected with M. tb suggested that blockade of this pathway enhances IFN-c production by CD3 + T lymphocytes and additionaly showed that T cells also express PD-L2 [31]. Unlike these human studies, we did not observe any increase in IFN-c production in PD-1 2/2 mice compared to WT mice; however, we did observe reduced proliferation of T lymphocytes from PD-1 2/2 mice. It has been shown that, in the absence of PD-1 in the mouse model of TB infection, PD-L1 binds to B7-1 molecules, which inhibits T cell proliferation and cytokine regulation [44], a finding that further support our observations in PD-1 2/2 mice. Therefore, our data strengthen the idea of bidirectional inhibitory interactions between B7-1 and PD-L1 in the absence of PD-1 molecules.
It is well known that natural regulatory T cells help in disease progression of M. tb infection by inhibiting protective immune responses [12,13]. We have also found that PD-1 2/2 Foxp3 gfp mice have higher numbers of natural Tregs compared to WT mice. It appears that M. tb recruits Treg cells to the site of infection, which is further assisted by MSCs [14]. MSCs dampen the immune response by inhibiting T cell proliferation.
A recent report also suggested that CD4 + T cells promote rather than inhibit the bacterial burden in PD-1-deficient mice [42]. In agreement with this report, we suggest that MSCs have an important role in the susceptibility of PD-1 2/2 mice to M. tb infection. Furthermore, IFN-c, which plays a central role in protective immunity against M. tb [5,6], was found to be very similar in WT and PD-1 2/2 mice using intracellular staining, but our PD-1 2/2 serum data suggested that these mice have abundant amounts of accumulated IFN-c. Therefore, PD-1 2/2 animals, which are impaired in mounting Th1 responses or in producing IFN-c, are susceptible to M. tb infection. On the other hand, Th2 responses, which exacerbate M. tb infections [10,11], were found to be increased in PD-1 2/2 animals, suggesting that these mice exhibit a severe imbalance in cytokine regulatory mechanisms due to loss of PD-1. PD-1 2/2 mice also produce increased levels of inflammatory cytokines such IL-17, which should be protective, yet due to the concomitant increase in IL-4 production, these mice are unable to control the bacterial burden.
Our findings that macrophages from PD-1-deficient exhibit reduced autophagy suggested that these cells are unable to generate phagolysosomes to kill the engulfed bacteria [34]. Thus, overall this report suggests that PD-1-deficient mice have a defect in T cell proliferation and phagolysosomal formation. Therefore, our study provides a mechanistic link between autophagy and costimulatory pathways, which may be involved in higher M. tb bacterial growth and disease progression in PD-1-deficient animals.
In summary, our findings indicate that, in sharp contrast with other microorganisms that cause chronic infection, infection of PD-1-deficient mice with M. tb causes increased inflammation and bacterial load. Furthermore, we found an indirect role of reduced autophagy for the increased bacterial burden in PD-1 2/2 mice. These findings therefore suggest a protective role of the PD-1/PD-L pathway in host immunity against M. tb. Manipulation of this pathway in patients infected with M. tb or co-infected with M. tb and HIV could therefore lead to unexpected outcomes.

Antibodies and Reagents
We used the following antibodies: anti-CD4 (clone: GK1.

Flow cytometry: surface and intracellular staining
Spleens were isolated from PD-1 2/2 and WT mice, either infected or uninfected, and macerated by frosted slides in 10% RPMI 1640 (Gibco, Invitrogen, UK) and made into a single cell suspension. Red blood cells (RBCs) were lysed with RBC cell lysis buffer, incubated at room temperature for three to five minutes and washed with 10% RPMI 1460. The cells were counted and 1610 6 cells were used for surface staining. For intracellular staining 1610 6 cells were cultured per well in 24 well plates (Nunc, USA) and activated with 50 ng/ml phorbol 12-myristate 13acetate (PMA) and 750 ng/ml ionomycin (Sigma, USA) overnight, and 10 mg/ml brefeldin A (eBiosciences, USA) was added during the last 4 hours of culture. Cells were washed twice with PBS and stained with antibodies directed against surface markers. After staining, cells were washed again with PBS and cells were fixed with 100 ml fixation buffer (eBiociences, USA) for 30 minutes, then re-suspended in 200 ml permeabilization buffer (eBiosciences, USA) and stained with fluorescently labelled anti-cytokine antibodies. Fluorescence intensity of fluorochrome-labelled cells was measured by flow cytometry (FACS Canto TM II, BD Biosciences, USA). LC3-B staining was performed according to the manufacturer's protocol (Cell Signalling, USA) and cell viability dye (7-AAD) was added to the LC3-B stained cells 15 minutes before analyzing the cells by flow cytometry. FACS Diva was used for acquiring the cells and final data analysis was performed by Flow Jo (Tree star, USA).

T and B cell proliferation assay
Spleens were isolated from PD-1 2/2 and WT mice, either infected or uninfected, and macerated by frosted slides in 10% RPMI 1640 (Gibco, Invitrogen, UK) and made into a single cell suspension. Red blood cells (RBCs) were lysed with RBC cell lysis buffer and incubated at room temperature for three to five minutes and washed with 10% RPMI 1460. The total splenocytes were passed through nylon wool columns as described earlier to isolate T lymphocytes [45] and B cells were isolated using CD19 + magnetic beads and a B cell isolation kit using the manufacturer's instructions (Miltenyi Biotech, Germany). Similarly, lung T lymphocytes were isolated by nylon wool column method. T lymphocytes were counted and plated at 0.3610 6 cells/well in a 96-well plate together with equal numbers of mitomycin C (Sigma, USA) treated syngeneic splenocytes as APCs and stimulated with different concentrations of H37Rv bacterial protein lysate or M. tb complete soluble protein (CSA). B cells were activated with LPS (Sigma, USA) at various concentrations for B cell proliferation.-Cells were cultured for 48 hours and then pulsed with tritiated thymidine (3H-TdR, 1.0 mCi per well; Amersham Biosciences, Figure 6. Cytokine storm in M. tb infected PD-1 2/2 mice. Lung tissues and spleens were homogenised and passed through nylon wool columns. The isolated T lymphocytes were co-cultured in the presence of mitomycin C-treated syngeneic splenocytes and H37Rv protein lysate overnight in the presence of PMA and Ionomycin. The next day cells were treated with brefeldin A for four hours. Next, cells were surface stained with anti-CD4 and anti-CD8, and fixed cells were used for intracellular staining with anti-IL-4, anti-IFN-c and anti-IL-17A antibodies. Cells were acquired on a flow cytometer. A. Total IL-4, IFN-c and IL-17 production by lung or spleen mononuclear cells from WT and PD-1-deficient mice is shown by flow cytometry in dot plots. Data shown here is a representative experiment of three. B. Summary of the data for IL-4, IFN-c and IL-17 production. Summary statistics show mean6STDEV and Student's t-test was applied for estimating significance between two parameters. doi:10.1371/journal.pone.0019864.g006

M. tb aerosol infection and colony forming unit (CFU) estimation
M. tb strain H37Rv (ATCC 27294; American Type Culture Collection, Rockville, MD) [46] infections were performed by aerosol challenge. M. tb strain H37Rv was grown to mid-log phase (OD600 ,0.6) in Middlebrook 7H9 media (Difco TM , USA) with 0.1% Tween 80 (Sigma, USA), 0.2% glycerol and 10% Middlebrook albumin, dextrose and catalase (ADC) enrichment medium (Difco TM , USA). Bacteria were stored at 280uC in 20% glycerol stocks for further experiments. For aerosol infection, cultured stock was washed with phosphate buffer saline (PBS) twice and made into single cell suspensions. Mice were infected with ,125 CFU of M. tb H37Rv using an aerosol chamber. Mice were sacrificed at various time points and organs were harvested, homogenised in 0.2 mm filtered PBS containing 0.05% Tween 80 and plated onto 7H11 Middlebrook (Difco TM USA) plates containing 10% oleic acid, albumin, dextrose and catalase (OADC) (Difco TM USA). Undiluted, ten-fold diluted and one hundred-fold diluted lung and spleen cell homogenates were plated in doublet on the above 7H11 plates and incubated at 37uC for 21 days. Colonies were counted and CFU were estimated.

Histology staining
Lung tissues were stained with Acid Fast Bacilli (AFB) stain and Hematoxylin and Eosin (H&E) dyes as previously described [47].

Statistical analysis
All data were analyzed by Excel 2007. In all the figures mean values were calculated with standard deviation (STDEV) until unless stated otherwise. For all statistical analyses Student's t-test was performed to compare two groups, p,0.05 was considered significant.