Effect of Attenuation of Treg during BCG Immunization on Anti-Mycobacterial Th1 Responses and Protection against Mycobacterium tuberculosis

Background The functional equilibrium between natural regulatory T cells (Treg) and effector T cells can affect the issue of numerous infections. In unvaccinated mice, the influence of Treg in the control of primary infection with mycobacteria remains controversial. Methodology Here, we evaluated the role of Treg during prophylactic vaccination with Mycobacterium bovis BCG (Bacillus Calmette-Guérin) on the induction of T cell responses and on the protective effect against subsequent M. tuberculosis challenge in mice. Principal Findings We demonstrated that, subsequent to BCG injection, Treg were recruited to the draining lymph nodes and negatively control anti-mycobacterial CD4+ — but not CD8+ — T-cell responses. Treatment of BCG-immunized mice with an anti-CD25 mAb (PC61) induced an increase IFN-γ response against both subdominant and immunodominant regions of the protective immunogen TB10.4. In Treg-attenuated, BCG-immunized mice, which were then infected with M. tuberculosis, the lung mycobacterial load was significantly, albeit moderately, reduced compared to the control mice. Conclusions Our results provide the first demonstration that attenuation of Treg subset concomitant to BCG vaccination has a positive, yet limited, impact on the protective capacity of this vaccine against infection with M. tuberculosis. Thus, for rational design of improved BCG, it should be considered that, although the action of Treg does not represent the major cause of the limited efficiency of BCG, the impact of this cell population on the subsequent control of M. tuberculosis growth is significant and measurable.


Introduction
The only vaccine available against infection with Mycobacterium tuberculosis is the live attenuated M. bovis BCG (Bacillus Calmette-Guérin). With more than three billion of doses injected, BCG is the most used vaccine in the world, although its efficiency is highly variable. Even though, in non-endemic zones, BCG is efficient up to 80% against disseminated infections with M. tuberculosis, this vaccine is not able to protect efficiently against the adult pulmonary tuberculosis in endemic zones. Moreover, the resurgence of tuberculosis in immuno-compromised individuals and the rapid expansion of multi-drug resistant and extensively drug resistant tuberculosis [1] reinforce the need of fundamental understanding of the variability and the limited efficiency of BCG for a better rational design of new strategies of antituberculosis vaccines [2].
Variability in the efficiency of BCG against infection with M. tuberculosis can be explained by differences in vaccination protocols and in BCG strains used in different areas of the world and human genetic factors, predisposing individuals to develop an active tuberculosis [3,4]. Moreover, this variability can also be due, in endemic zones, to pre-exposure to environmental mycobacteria, such as M. avium, which could induce immune responses against antigens shared with BCG. The resulted immune cross-recognition may decrease the persistence of subsequently injected BCG and may lead to variable protective efficiency [5,6]. On the other hand, the limited efficiency of BCG against infection with M. tuberculosis can be explained by the following hypotheses. (i) Deletion of genes coding for protective immunogens in BCG and/ or over attenuation of BCG, due to the loss of genetic Regions of Differences (RD) [7,8] could explain the weak efficiency of BCG. This hypothesis has been confirmed by genetic reintroduction of RD1 into the BCG Pasteur strain [9,10]. The resulted BCG::RD1 is able to induce marked Th1 responses to M. tuberculosis-specific antigens, i.e., Early Secreted Antigenic Target, 6 kDa (ESAT-6 3 ) and Culture Filtrate Protein, 10 kDa. It displays an increased virulence and persistence, which are in direct correlation with the increased protective capacity of this vaccine candidate against M. tuberculosis [10,11] (ii) Despite the induction of anti-mycobacterial Th1 cells by BCG immunization, intracellular mycobacteria sequestered within granuloma, may be inaccessible to these effector T cells. (iii) Finally, subsequent to BCG immunization, concomitant induction of regulatory T cells (Treg) [12], in parallel to the induction of anti-mycobacterial T-cell responses, could reduce the efficiency of the latter in protection against M. tuberculosis.
During infection, like IL10 and TGF-b, the principal function of CD4 + CD25 + FoxP3 + Treg is the control of intense inflammation, induced at the site of infection, to avoid tissue damage. However, this immune regulation can counteract the function of CD4 + or CD8 + effector T cells and paradoxically dampen the control of infection. Subsequent to the pioneering work on the role of Treg in the persistence of Leishmania major in mice [13], it has been shown that the functional balance between natural Treg and effector T cells can influence the control of numerous viral, bacterial or parasitic infections [14].
With respect to the control of primary mycobacterial infections in unvaccinated mice, the role of Treg remains controversial in the few works addressing this question. Indeed, Quinn et al. showed that anti-CD25 mAb treatment in mice has no consequence on the control of mycobacterial growth [15]. In contrast, when effector T cells and Treg were adoptively co-transferred, at high Treg/ effector T cells ratio into recipients, Treg were able to significantly inhibit the action of effector T cells on the control of M. tuberculosis growth [16]. An improved control of M. tuberculosis infection was also observed under conditions of total depletion of FoxP3 + T cells, giving rise to a substantial polyclonal -but not mycobacteriaspecific -activation of T cells [17]. Besides these discrepancies, it is important to know whether the prophylactic anti-mycobacterial vaccination with BCG leads to induction of Treg, with the potential to influence the T-cell mediated control of subsequent infection with virulent M. tuberculosis. Here, by use of a standard mouse model of anti-tuberculosis vaccine assay [18], we studied the possible recruitment/expansion of Treg in mice immunized s.c. with BCG or infected with M. tuberculosis via aerosol route. By modulating the Treg compartment during BCG immunization, we evaluated the role of these cells on the set up of anti-mycobacterial CD4 + or CD8 + T-cell responses. Finally, in mice immunized with BCG and treated with anti-CD25 mAb (PC61), we investigated the T-cell responses at the site of infection and the outcome of M. tuberculosis infection. Our results show that, during immunization with BCG, attenuation of Treg function by anti-CD25 mAb (PC61) treatment, finely tunes anti-mycobacterial Th1 responses and slightly, yet significantly, improves the protective efficiency of BCG against infection with M. tuberculosis.

Recruitment/expansion of Treg in lymphoid organs of BCG-immunized mice
We first analyzed the possible recruitment/expansion of Treg in the inguinal draining lymph nodes and spleen of BALB/c mice immunized s.c. with 1610 6 CFU of BCG. At 3 weeks postimmunization, percentages of CD25 + FoxP3 + , and also of CD25 2 FoxP3 + , within the CD4 + T-cell compartment of the inguinal lymph nodes and of the spleen of BCG-immunized mice ( Figure 1A, top) were comparable to those of age-matched nonimmunized controls ( Figure 1A, bottom). However, after s.c. immunization with BCG, following a 4-week-kinetics study, we observed that, in the inguinal lymph nodes, despite the non variable percentage of CD25 + FoxP3 + within the CD4 + T-cell subset ( Figure 1B top, left), the absolute number of these cells statistically increased ( Figure 1B bottom, left), albeit in a comparable manner to the increase of the number of total lymph node cells ( Figure 1B top, right), or of the number of CD4 + T cells ( Figure 1B bottom, right). The increased Treg numbers in the draining lymph nodes may result from their recruitment, local proliferation/retention or conversion from CD4 + CD25 2 T cells. No change was detected in the number of CD4 + CD25 + FoxP3 + cells in the spleen (data not shown). Thus, the size of the Treg compartment grows proportionally to that of the total cells and of the CD4 + T-cell subset in the draining lymph nodes of BCGimmunized mice.
Treg-mediated negative regulation of anti-mycobacterial CD4 + T-cell responses in BCG-immunized mice To investigate the possible role of Treg on anti-mycobacterial T-cell responses following BCG immunization, we evaluated IFNc CD4 + T-cell responses of mice immunized by BCG and treated with anti-CD25 mAb (PC61). As shown in the Figure 2A, mice received i.p., at day 22 and day 11, PBS or 1 mg of a control rat IgG or PC61. At day 0, mice were immunized s.c. with 1610 6 CFU of BCG. At day 15 (not shown), as well as at day 21 ( Figure 2B), we observed a significant decrease (Anova, p,0.001) in the percentage of FoxP3 + within CD4 + T splenocytes of anti-CD25-treated mice, compared to their counterparts injected with PBS or with the control IgG ( Figure 2B). Compared to BCGvaccinated individuals, unvaccinated mice, treated with PC61 displayed the same range of reduction in percentages of FoxP3 + within the CD4 + cell subset (not shown). The partial depletion of CD4 + FoxP3 + T cells in anti-CD25 mAb-treated mice has been previously described [19][20][21]. It is noteworthy that such anti-CD25 mAb treatment with PC61 does allow efficient attenuation of the Treg function in numerous studies and in different models. Moreover, this anti-CD25 mAb treatment does not block the action of activated T cells, as it is reported in many cases, T-cell responses increased following this treatment [19,[22][23][24].
At day 21, we studied IFN-c CD4 + T-cell responses of these mice against TB10.3 (Rv3019c) and TB10.4 (Rv0288) mycobacterial immunogens. TB10.3 and TB10.4 are expressed by both BCG and M. tuberculosis and belong to the immunogenic family of ESAT-6 proteins [25]. Moreover, TB10.4 has been shown to be a protective immunogen [26,27]. The frequencies of IFN-cproducing CD4 + T splenocytes, specific to MHC-II-restricted immunodominant epitopes contained in TB10.3:74-88 or TB10.4:74-88 peptides ( Figure 2C), were significantly increased in anti-CD25-treated mice, compared to the control mice injected with PBS or with the control IgG. IFN-c responses to these peptides were only due to CD4 + T cells, as shown by CD4 + T-cell depletion that fully abolished these responses [26]. It is noteworthy that in anti-CD25 mAb-treated mice, the partial depletion of Treg corresponds to a loss of 2 to 2.5% of total splenocytes and thus cannot account for the observed increase in the frequency of IFNc-producing splenocytes. Stimulation with the negative control peptide PV1:103-116 did not reveal any IFN-c producing cells (not shown).
Based on the fact that responses to all T-cell epitopes are not affected equally by Treg, it has been suggested that these cells could modify the profile of T-cell responses to immunodominant or subdominant regions [28]. We thus sought to determine whether, under the conditions of Treg attenuation, the profile of Th1 responses to immunodominant and subdominant regions of a protective mycobacterial immunogen is modified. To this end, we evaluated, in individual BCG-immunized mice (n = 3), treated with the control IgG or anti-CD25 (PC61) mAb, as detailed in Figure 2A, IFN-c production by splenocytes stimulated with individual overlapping 15-mer peptides, covering the entire sequence of TB10.4 at a single amino acid step [26]. Unvaccinated control mice did not produce IFN-c after in vitro stimulation with these peptides (data not shown, [26]). We previously established that IFN-c responses detected in these conditions with TB10.4 15mer peptides are only due to the CD4 + T-cell subset [26]. As shown in Figure Figure 3A), but stimulate higher responses in PC61-treated mice ( Figure 3B). In this pepscan assay, no differences were detectable between the two experimental groups with respect to the immunogenicity of the C-terminal TB10.4:69-96 (peptides 69-82) region. This could be due to the high peptide concentration (20 mg/ml) used in the pepscan assay to detect responses against weak epitopes. Indeed, compared to their IgG-treated counterparts, in the PC61-treated, BCG-immunized mice, higher IFN-c responses to TB10.4:74-88 were detected by use of ELISPOT ( Figure 2C  Absence of Treg-mediated negative regulation of antimycobacterial CD8 + T-cell responses in BCG-immunized mice Based on the demonstration that Treg are able to down-regulate CD8 + T cell responses in immunization against Listeria monocytogenes in mouse model [29], we evaluated CD8 + T-cell responses of BCG-immunized BALB/c mice, injected with PBS, the control IgG or anti-CD25 mAb, as shown in the Figure 2A. At day 21, splenocytes from individual mice were stimulated with TB10.3/ 4:20-28 peptide, containing a MHC-I-restricted epitope. The frequency of specific T cells within CD8 + T-cell population was determined, as shown in Figure 4A, by use of a combination of anti-CD8, anti-CD44 mAbs and a H-2K d pentamer complexed with the homologous peptide [30,31]. We observed that, compared to the control groups, the anti-CD25 treatment of BCG-immunized mice, slightly but not significantly, increased the percentage of TB10.3/4:20-28-specific CD8 + T cells in the responding mice and did not improve the frequency of mice with detectable TB10.3/4:20-28-specific CD8 + T cells ( Figure 4B). Accordingly, in the numerous experiments that we performed to evaluate anti-mycobacterial CD8 + T-cell responses specific to Ag85A or TB10.3/4, using either the 51 Cr release cytotoxicity assay or the MHC-I pentamer FACS assay, we always detected only 30 to 50% of positive mice in BCG-immunized groups with large variability in percentages of anti-mycobacterial CD8 + T cells among individuals of the same experimental group [30, data not shown].

Recruitment/expansion of Treg in lung parenchyma of M. tuberculosis-infected mice
To investigate the recruitment/expansion of Treg subset at the site of M. tuberculosis infection, we infected BALB/c mice with <100 CFU of M. tuberculosis H37Rv by aerosol route. Similar to the draining lymph nodes of BCG-immunized mice, following a 4week-kinetics study of the lung parenchyma, despite the constant percentage of CD25 + FoxP3 + cells ( Figure 5A), and CD25 2 FoxP3 + within the CD4 + T-cell subset the absolute number of these cells statistically increased in the infected mice ( Figure 5B), in parallel to the well-characterized increase in the total number of lymphocytes or of CD4 + T cells recovered from the infected lungs [32].

Study of Th1 responses in anti-CD25-treated, BCGimmunized and M. tuberculosis-infected mice
We then wondered whether, in the conditions of BCG vaccination, Treg exert any inhibitory effect on the induction or on the action of T cell-responses, responsible of the partial control of subsequent M. tuberculosis challenge. To address this issue BALB/c mice received, at days 22, 12 and 26, i.p. injections of PBS, of control IgG or of anti-CD25 mAb ( Figure 6A). At day 0, mice were immunized s.c. with 1610 6 CFU of BCG. At day 30, all groups were challenged by < 100 CFU/mouse of M. tuberculosis H37Rv by aerosol route.
We analyzed anti-mycobacterial T-cell responses in these experimental groups. Antigen-experienced CD4 + T cells accumulating within mycobacteria-infected lung parenchyma can be divided into CD27 + and CD27 2 subsets. Only the CD27 2 population contains highly differentiated, IFN-c-producing T cells [33]. Thus, determination of percentages of activated CD27 2 cells represents an appropriate way to measure Th1 response at the site of mycobacterial infection. At one month post-challenge, we detected a markedly increased percentage of CD45RB 2 CD27 2 cells within the CD4 + Tcell subset of mice infected with M. tuberculosis, compared to the untreated controls ( Figure 6B). However, immunization with BCG or treatment with the control IgG or anti-CD25 mAb did not modify the percentages of CD45RB 2 CD27 2 cells within the lung CD4 + T subset of M. tuberculosis-infected mice. Moreover, the frequencies of IFN-c-producing CD4 + T cells from spleen or lungs, in response to stimulation with PPD or immunodominant epitopes of TB10.4 or Ag85A, were comparable in BCG-immunized and M. tuberculosisinfected mice that were injected with PBS, the control IgG or anti-CD25 mAb (data not shown). No difference was observed in the TB10.3/4:20-28-specific CD8 + T cell responses in these experimental groups (data not shown). Taken together, our results show that, although the attenuation of Treg function increases the antimycobacterial Th1 responses subsequent to BCG immunization and at the time of the challenge and at the chronic phase of the infection, the T-cell responses are comparable in all these experimental groups, presumably due to the sustained level of mycobacterial load and the consequent continuous T-cell triggering.
Anti-CD25 mAb treatment improves moderately but significantly the protective effect of BCG vaccination against infection with M. tuberculosis It has been already shown that anti-CD25 mAb (PC61) treatment has no consequence on the control of mycobacterial growth in the lungs, neither at early, nor at late time points after primary M. tuberculosis infection [15]. We thus focused on the effect of anti-CD25 mAb treatment on the control of M. tuberculosis infection subsequent to BCG immunization. We determined the mycobacterial load in the lungs of mice either non-immunized or immunized with BCG and treated with PBS, control IgG or anti-CD25 mAb, and challenged with M. tuberculosis, as detailed in Figure 6A. BCG immunization decreased the mycobacterial load by 2 log 10 ( Figure 7). Anti-CD25 mAb treatment, concomitant to BCG immunization, reproducibly led to a moderate, albeit statistically significant, decrease of mycobacterial load at the site of infection, i.e., the lung parenchyma (Figure 7, Experiments 1 and 2). In the experiment 1, the Mean6SD of the lung M. tuberculosis load in BCG-immunized mice, injected with the control IgG or anti-CD25 mAb was, respectively (1.9961.7)610 6 and (0.4560.40)610 6 , which means 4.4 fold less in the latter group. In the experiment 2, the mycobacterial load in BCG-immunized mice, injected with the control IgG or anti-CD25 mAb was, respectively (2.061.0)610 5 and (1.060.9)610 5 which means 2 fold less in the latter. Importantly, 5 out of 9 individuals in the last group displayed less than 1610 5 CFU/lungs. StatXact program and permutation test (two-sided p-value) applied on the results obtained from the experiments 1 and 2 showed that the differences between the anti-CD25 mAb-treated mice and non-Ig-treated or the control Ig-treated mice were statistically significant (p,0.0420).
It is noteworthy that based on our long experience, in BCGimmunized mice, the intensity of T-cell responses is of large variability among the individual of the same experimental group. In contrast, such variability was not observed in mice immunized by a single mycobacterial protein (TB10.4 or ESAT-6) in adjuvant.  Plausibly, the variability observed in mycobacterial load within the BCG-immunized groups reflects the variability in cell-mediated immune responses in different BCG-immunized individuals.
It is noteworthy that like C57BL/6 mice, BALB/c mice of specific pathogen free status, neither die nor display morbidity, even at 9 months following low dose H37Rv aerosol challenge. Therefore, a possible benefic effect of Treg attenuation during BCG immunization on M. tuberculosis-induced host mortality cannot be evaluated in mouse experimental model. Histopathological analyses showed that the anti-CD25 treatment in mice immunized with BCG and then infected with M. tuberculosis did not lead to enhanced inflammation or exacerbated lung tissue damage (data not shown). This indicates that dampening the antiinflammatory Treg subset by anti-CD25 mAb treatment did not increase the inflammation due to mycobacterial infection. These results show that the intervention of Treg, following BCG immunization, contributes partly in limitation of the efficiency of this vaccination in anti-mycobacterial protection.

Discussion
In the present study, we examined whether a recruitment/ expansion of Treg, concomitant to the induction of antimycobacterial T-cell responses, could limit the efficiency of BCG vaccine against infection with M. tuberculosis in mice. We first observed that, during the four weeks following s.c. immunization with BCG, despite the fact that the percentages of CD25 + FoxP3 + cells within the CD4 + subset remains constant, the total number of these cells increases proportionally to that of the other cells in the inguinal draining lymph nodes, but not in the spleen. The same holds true for the total number of Treg in the lung parenchyma of mice infected with M. tuberculosis via the aerosol route. These observations are in accordance with the data published by Quinn et al. in which mice, infected with 1610 4 CFU of BCG by nasal route or with <100 CFU of M. tuberculosis by aerosol route, displayed increased size of the Treg compartment, proportional to that of the other lymphocyte subsets [15]. As mentioned in the Results, the increase of the Treg population in these cases may be due to their recruitment, local proliferation/retention or conversion from other CD4 + T cells. It is noteworthy that the attenuated BCG, inducer of weak infiltration and inflammation, or virulent and highly persistent M. tuberculosis, both induce recruitment/ expansion of Treg, while one should expect that virulent mycobacteria would elicit more efficient induction of Treg to avoid tissue injury. This may suggest that induction of Treg is not a consequence of inflammation.
In this study, our approach to evaluate the effect of Treg on anti-mycobacterial adaptive immunity and protection was based on the use of anti-CD25 mAb (PC61) treatment. This strategy has  been shown to successfully attenuate in vivo Treg functions, without total depletion of this cell subset [19][20][21], in large numbers of various experimental models, including those studying T-cell responses to immunizations, to infections with virus, bacteria and parasites, as well as in anti-tumor immune responses [19,[22][23][24]. In most cases, the T-cell effector functions increase, suggesting that the effect of PC61 treatment is more pronounced on Treg cellswith great majority expressing high levels of CD25 -than on the action of recently activated CD4 + T-cell effectors -which transiently express intermediary levels of CD25. Accordingly, the fact that the PC61 treatment does not totally deplete the CD25 + T-cell population and that IL2 is not the major or the only one cytokine involved in the activation of CD4 + T-cell effectors may explain how PC61 treatment allows attenuation of Treg rather than inhibition of effector functions. Furthermore, the effect of PC61 treatment is relatively persistent, i.e., 12-15 days postinjection [19][20][21], making this treatment adaptable to the long experiments of anti-tuberculosis vaccination. At last, long-lasting total depletion of Treg leads to lymphoproliferative disorders [34,35], which can interfere with the adaptive responses mounted against M. tuberculosis.
In mice immunized with BCG and treated by anti-CD25 mAb (PC61), the frequency of IFN-c producing CD4 + T cells specific to the immunodominant epitopes of TB10.3 or TB10.4 immunogens increased significantly. Attenuation of Treg subset by PC61 treatment increased the intensity of IFN-c responses to all immunodominant and subdominant regions of TB10.4 antigen. Barely immunogenic regions in IgG control-treated, BCG immunized mice, become inducers of easily detectable IFN-c responses in their PC61-treated counterparts. No new immunogenic regions were revealed under the conditions of Treg attenuation, showing that the repertoire of T cells specific to the different epitopes of TB10.4 is not modified quantitatively. Hence, increase in T-cell responses to subdominant regions may have a direct positive impact on anti-mycobacterial immune-protection, as adaptive immune system becomes able to sense these epitopes more sensitively. Note that TB10.3 and TB10.4 belong to the highly immunogenic ESAT-6 protein family [25] and that TB10.4 is a powerful protective immunogen against M. tuberculosis infection [26,27]. Only a weak and not significant, effect of anti-CD25 mAb treatment was detected on the intensity of CD8 + T-cell responses specific to TB10.3 or TB10.4 immunogens, as judged by determination of percentages of such cells by the use of a pentamer of H-2K d restricting element combined with TB10.3/ 4:20-28 peptide [30,31].
We then studied the effect of anti-CD25 treatment on the control of the growth of virulent M. tuberculosis H37Rv in mice immunized with BCG. In accordance with previous reports [19][20][21], we observed that after injection of anti-CD25 mAb (PC61), the percentage of CD25 + FoxP3 + cells within the CD4 + subset remains decreased, as long as 2 weeks post injection, while at 4 weeks after injection, the Treg subset was totally reconstituted. Based on these data, we performed the anti-CD25 treatment regimen, shown in Figure 6A, to obtain an attenuated Treg compartment: (i) at the time of BCG injection, when anti-mycobacterial T-cell responses are initiated, (ii) during the month following this immunization, when anti-mycobacterial T cells expand, (iii) at the time of M. tuberculosis challenge, when the induced anti-mycobacterial T cells encounter the first infected cells, and (iv) at the beginning of the virulence phase of the infection, when the action of such cells controls, at least partially, the growth of mycobacteria. Therefore, in this study, although at 1 month after the challenge, i.e., at day 34 after the last anti-CD25 mAb injection, the Treg subset has been entirely recovered, the phases of initiation, expansion and effector function of anti-mycobacterial T cells occurred under effective attenuation of the Treg compartment. At 1 month post-challenge, when mice possessed substantial mycobacterial load, Th1 responses at the site of infection and in the spleen were comparable in all M. tuberculosis-challenged mice, independently of BCG immunization or anti-CD25 mAb treatment. Our most important finding is that the attenuation of Treg subset, by anti-CD25 mAb treatment, concomitant to BCG immunization, moderately but significantly improves the protective efficiency of this vaccine against infection with virulent M. tuberculosis. Indeed, BCG-immunized, anti-CD25 treated mice, challenged with virulent H37Rv by aerosol route, displayed a significantly reduced mycobacterial load at the site of infection, compared to their counterparts immunized with BCG alone or immunized with BCG and treated with a control Ig isotype.
Three different strategies have been already used to address the role of Treg in mycobacterial infection. (i) Quinn et al. showed that anti-CD25 mAb treatment during the early primary infection of mice with mycobacteria, given by nasal route, increased production of Th1 cytokines but did not reduce the mycobacterial load in the lungs [15]. (ii) Kursar et al. showed that, co-transfer of CD4 + CD25 + T cells with CD4 + CD25 2 T cells, at a 1:2 ratio, i.e., markedly higher than the physiological ratio, into RAG1% C57BL/6 mice, drastically decreased the capacity of CD4 + CD25 2 T cells to control M. tuberculosis load in spleen and lungs. Interestingly, the protective effect of CD4 + CD25 2 T cells in this model was not correlated with any increased IFN-c or TNF-a responses [16]. (iii) More recently, Scott-Browne et al. set up a mixed chimera C57BL/6 mouse model, reconstituted with bonemarrow cells from Thy1.1 FoxP3 +/+ and Thy1.2 FoxP3% mice [17]. In these chimeras, infection with M. tuberculosis by aerosol route was better controlled when FoxP3 + T cells were depleted by anti-Thy-1.1 mAb treatment. Importantly, in this model, the total depletion of Treg compartment induced a substantial polyclonal stimulation of T-cell compartment but did not increase Th1 responses specific to mycobacterial antigens. Therefore, in that study, it was not clear whether the cause of the improved control of M. tuberculosis growth was the inhibition of the effect of Treg on specific effector T cells, or a general bystander immuno-activation effect [17]. These three experimental models are very different in their detailed characteristics, yet in all of them, the mycobacterial infection was given into non-vaccinated individuals and therefore it was not evidenced whether prophylactic anti-mycobacterial immunization with BCG gives rise to induction of Treg, able to influence the outcome of the subsequent control of M. tuberculosis infection. Our work directly addresses this question in the context of a standard anti-tuberculosis vaccine assay and showed that the efficient attenuation of Treg by PC61 mAb treatment, concomitant to BCG immunization, has a statistically significant, albeit minor, effect on the fine tuning of T-cell immunity controlling M. tuberculosis infection.
During the preparation of this manuscript, a first analysis of anti-mycobacterial T-cell immunity and protection by BCG immunization in Treg-attenuation conditions, prior to M. tuberculosis challenge, has been published by Quinn et al [36]. This team observed that functional attenuation of Treg by PC61 mAb treatment during BCG vaccination increases Th1 responses to Ag85A or to culture filtrate proteins of M. tuberculosis but not protection against virulent M. bovis or M. tuberculosis. Compared to this work performed on C57BL/6 mice [36], our study concerns the effect of Treg in BALB/c mice immunized by BCG, in which we were able to study both CD4 + and CD8 + T-cell responses to TB10.3 and TB10.4 strong/protective immunogens. Moreover, we characterized the Th1 responses to all the immunodominant and subdominant regions of the TB10.4 antigen. By including high numbers of mice/group and by repeating the protection experiments, our work provides evidence that Treg influence the immune control of M. tuberculosis infection.
Several major differences between our experimental conditions and those of the work published by Quinn et al. [36] can explain the different results obtained in these two investigations of the effect of Treg on anti-mycobacterial protective capacity of BCG. Indeed, both the pathogen and the host used in these two studies are different. First M. tuberculosis Erdman strain has been used in the study by Quinn et al. [36] versus M. tuberculosis H37Rv strain in the present study. As previously described [37], M. tuberculosis Erdman is less virulent than H37Rv in mice and therefore the characteristics of mycobacterial infection are not comparable. Moreover, C57BL/6 mice have been used by Quinn et al versus BALB/c mice in the present study. Importantly C57BL/6 and BALB/c mice are genetically distant and therefore the characteristics of mycobacterial infection, i.e., repertoire of T-cell responses, immunodominance of mycobacterial antigens and H-2 restriction are distinct in these two strains. Moreover, BALB/c strain is well known to develop more readily Th2 responses while C57BL/6 mice are prototype of Th1proned mice. Most importantly, the frequency of Treg has been suggested to be modulated by genetic background [38]. Indeed, this study shows that the frequency of CD4 + CD25 + T cells in the thymus and peripheral lymphoid organs of BALB/c mice is higher than in C57BL/6 mice. Notably, CD4 + responder T cells of BALB/c mice show greater susceptibility to suppression of by Treg compare to C57BL/6 mice [38]. Considering all these notable differences, one cannot exclude that the effect of Treg on protective responses can also be quantitatively and qualitatively different in BALB/c and C57BL/6 genetic backgrounds. It is possible that BALB/c strain is a more favorable animal model to detect the effect of Treg in Th1controlled infectious diseases like tuberculosis. Interestingly, the pioneering work of Y. Belkaid et al describing the role of Treg in the control of Leishmmania infection was done in BALB/c mice [13].
Based on our observations, during rational design of improved BCG, it should be considered that, although the action of Treg does not represent the major cause of the limited efficiency of BCG, the impact of this regulatory cell population on the subsequent control of M. tuberculosis growth is significant and measurable. In contrast to the anti-CD25 mAb treatment, with mild autoimmune manifestations, germline FoxP3 mutations or general depletion of Treg performed in conditional FoxP3 mutants leads to lymphadenopathy, extensive T-cell activation and severe autoimmune lymphoproliferative disease in neonates and adults [35,39]. Therefore, it would be of particular interest to evaluate the control of M. tuberculosis, after BCG immunization, in an experimental model closer to physiological conditions, for instance under the action of TGF-b -which favors Treg development [39] -or under the combined actions of TGF-b, IL-6 and IL-21, detrimental to Treg development, yet favoring the development of pro-inflammatory Th17 [40]. It has been suggested that the latter can be involved in the recruitment of Th1 cells to the site of M. tuberculosis infection [41], even though a direct role of IL-17 in the control of M. tuberculosis growth remains to be confirmed [42].

Anti-CD25 in vivo treatment
Mice were injected i.p. with 0.5 or 1 mg of rat IgG 1 anti-mouse CD25 mAb (PC61) or of a control rat IgG 1 (CRL-1912). These Ig were prepared from ascitic fluids precipitated with 50% (NH 4 ) 2 SO 4 at 4uC, in endotoxin-free conditions. Absence of endotoxins in the Ig preparations was then checked by use of ''Limulus Amebocytes Lysate'' kit (Cambrex, Emerainville, France), with a detection limit of 0.01 IU/ml.
CD8 + T-cell assays Three weeks after immunization with BCG, splenocytes from mice were stimuled in vitro with 10 mg/ml of TB10.3/4:20-28 peptide [30] in RPMI complemented with 2 mM L-glutamax, 5610 25 M ß-mercapto-ethanol, 100 IU penicillin/ml and 100 mg streptomycin/ml and 10% FCS. At day 6, detection of CD8 + T lymphocytes specific to TB10.3/4:20-28 epitope was performed by cytofluorometry, by use of a PE-conjugated pentamer of H-2K d , complexed with TB10.3/4:20-28 peptide [31] (Proimmune, Oxford, UK). We established that in this pentamer assay, in vitro stimulation with TB10.3/4:20-28 peptide did not modify the frequency of the responding mice, compared to ex vivo analysis. Indeed, after in vitro stimulation, the percentages of the pentamer + cells increase in the same proportion in different individuals and with respect to the initial percentage of the pentamer + T cells detected ex vivo. Therefore, following in vitro stimulation, the sensitivity of this assay is improved without biasing the frequency of the responding mice or the relative importance of CD8 + T-cell responses in different mice.

Preparation of lymphocytes from lung parenchyma
Lungs were perfused with type IV collagenase (400 U/ml) and DNase I (Roche, Mannheim, Germany). After 45 min incubation at 37uC, 5% CO 2 , lungs were homogenized and passed through nylon filters with 100 mm-diameter pores (Cell Strainer, Becton, Dickinson, Falcon). Cells were then enriched in lymphocytes on a Ficoll gradient (Lympholyte M, Cedarlane Laboratories, Ontario Canada) before cytofluorometric analyses.

Protection assay against infection with M. tuberculosis
Mice were challenged, via the aerosol route, with M. tuberculosis H37Rv, by use of a home-made nebulizor. Two ml of a suspension containing 5610 6 CFU/ml were aerosolized to obtain an inhaled dose of <100 CFU/mouse. Infected mice were placed in isolators. One month post-challenge, lungs were homogenized by use of 2.5mm diameter glass beads and an MM300 organ homogenizer (Qiagen, Courtaboeuf, France). Serial 5-fold dilutions of homogenates were seeded on 7H11 Agar, supplemented with OADC (Difco, Becton Dickinson, Sparks, MD). CFU were counted after 21 days of incubation at 37uC. Two independents experiments were performed and statistical analyses were performed by use of StatXact program and Permutation test, as described previously [47].