A temperature sensitive Mycobacterium paragordonae induces enhanced protective immune responses against mycobacterial infections in the mouse model

Recently, we introduced a temperature sensitive Mycobacterium spp., Mycobacterium paragordonae (Mpg). Here, we checked its potential as a candidate for live vaccination against Mycobacterium tuberculosis and Mycobacterium abscessus. Intravenous infections of mice with Mpg led to lower colony forming units (CFUs) compared to infection with BCG, suggesting its usefulness as a live vaccine. The analyses of immune responses indicated that the highly protective immunity elicited by Mpg was dependent on effective dendritic maturation, shift of cytokine patterns and antibody production toward a Th1 phenotype, and enhanced cytotoxic T cell response. Compared to BCG, Mpg showed a more effective protective immune response in the vaccinated mice against challenges with 2 different mycobacterial strains, M. tuberculosis H37Ra or M. abscessus Asan 50594. Our data suggest that a temperature sensitive Mpg may be a potentially powerful candidate vaccine strain to induce enhanced protective immune responses against M. tuberculosis and M. abscessus.

(c) Growth of BCG and Mpg in the organs (lungs, liver and spleen) after intravenous inoculation into BALB/c_nu (nude, up panels) and BALB/c (down panels) mice (n = 3-4 per group) (*P < 0.05, **P < 0.01, ***P < 0.001; Student's t-test). (d) The viability of BMDCs was measured by MTS and 7-AAD staining assay after infection with 10 M.O.I. of BCG or Mpg overnight. Representative histograms of 7-AAD are presented (**P < 0.05, ***P < 0.001; Student's t-test). All data are mean ± standard deviation of CFU or intensity values in each group. days) after administering them into BALB/c and BALB/c_nu (nude) mice (1 × 10 6 CFU, intravenous route) ( Supplementary Fig. S2). In nude mice, the CFUs of Mpg in all the organs were significantly lower than those of BCG at each time point after infection. A similar trend was also found in the BALB/c mice, although some organ-specific differences were found (Fig. 1c). Even after 4 weeks of IV injection, the CFUs of Mpg were significantly lower than those of BCG in all the organs ( Supplementary Fig. S3). It suggests usefulness of Mpg as an attenuated live vaccine.
It has been reported that during infections with mycobacteria, resistance against cell death of the APCs, particularly DCs can increase the efficacy of the vaccine by extending the duration of antigen presentation to the T cells 22 . To address this issue, 24 hours after the bone marrow-derived dendritic cells (BMDCs) from BALB/c mice were infected with BCG or Mpg at 37 °C, the cytotoxicity levels of the infected BMDCs were compared by MTS (cell supernatants) and 7-amino-actinomycin D (7-AAD) staining assays (cell pellets) capable of distinguishing between live and dead cells 26,27 . Both methods showed that the rates of death in the BMDCs were reduced after infection with Mpg compared to infection with BCG (Fig. 1d), suggesting that Mpg could be superior to BCG in extending the life-span of the infected DC, thereby facilitating antigen presentation.

Enhancement of BMDC maturation by Mpg. To investigate the maturation of BMDCs after infection
with Mpg, the immature BMDCs were infected with 10 M.O.I. of BCG or Mpg for 24 hours at 37 °C. As a positive control, the BMDCs were incubated with LPS (1 μg/ml) for 24 hours. The surface expression of maturation markers (MHC II, CD40, CD80 and CD86) on the BMDCs was compared between BCG-and Mpg-infected cells using FACS analysis (Fig. 2). All the DC maturation markers in the Mpg-infected BMDCs were increased compared to those in the uninfected (negative control) and BCG-infected BMDCs. In the case of MHC II, CD80 and CD86 surface markers, the BCG-infected BMDCs exhibited expression levels almost similar to the negative control, strongly supporting the previous report that BCG down-regulated MHC II and CD80 expressions in BMDCs 28 . This suggests that Mpg could induce the maturation of BMDCs compared to BCG.
To compare the cytokine profiles of BMDCs after infection with BCG or Mpg, the cell culture supernatants were subject to ELISA (IL-10, IL-12, TNF-α or IL-6 cytokines). Notably, the Mpg-infected BMDCs produced significantly higher levels of IL-12 cytokines than the BCG-infected BMDCs. However, the Mpg-infected BMDCs secreted significantly lower levels of the other 3 cytokines TNF-α, IL-6, and IL-10 than the BCG-infected BMDCs (Fig. 3a). Given that the production of cytokines such as IL-10 and IL-12 during the process of BMDC maturation process can induce a Th1-or Th2-skewed immune response in the context of local inflammation [29][30][31] , our data suggest that Mpg could elicit a Th1-type immune response compared to BCG. These results were further supported by qRT-PCR analysis. The mRNA levels of IL-10 or IL-12 and two molecules related to MHC II expression, CIITA (MHC II transactivator) and H2DMb (MHC II-associated protein), were compared between the BCGand Mpg-infected BMDCs. Consistent with the ELISA data, IL-10 expression was significantly enhanced in the BCG-infected BMDCs, compared to the Mpg-infected BMDC, irrespective of the time after infection (about five-fold increase compared with Mpg-infected BMDCs at each time-point). In addition, the Mpg-infected BMDCs showed significantly higher expression levels of IL-12 than the BCG-infected BMDCs at 24 hours after infection (about one hundred-fold increase compared with the BCG-infected BMDCs) (Fig. 3b). A similar trend as shown in murine cell lines was also found in IL-10 and IL-12 cytokine profiles of Mpg infected human monocyte cell line, THP-1 (Fig. 3c). Additionally, the Mpg-infected BMDCs showed higher levels of the two MHC II-related targets, CIITA and H2DMb, than the BCG-infected BMDCs (about five-fold increase compared with the BCG-infected BMDCs 24 hours after infection) (Fig. 3b). In the case of CIITA, the CIITA mRNA level was increased in 24 hours, but reduced in 48 hours after infection with Mpg ( Supplementary Fig. S4), which is consistent with the previous finding that the CIITA mRNA level of Mtb infected cells was increased in 24 hours and decreased in 48 hours 32 .
Enhanced BMDC migration by Mpg. Since the migration of BMDCs to regional lymph nodes requires the expression of CCR7 as the receptor for lymphoid chemokines CCL19 and CCL2 33 , we compared the expression of CCR7 between BMDCs infected with either BCG or Mpg. To evaluate the migrational abilities of the infected BMDCs with BCG or Mpg, we compared the level of CCR7 expression on infected cell surfaces using FACS analysis. We found that the surface expression of CCR7 on the Mpg-infected BMDCs were significantly higher than in the BCG-infected BMDCs (Fig. 4a), and this result was also consistent with the qRT-PCR quantification of the corresponding mRNA level (Fig. 4b). To further evaluate the migrational capacity of BMDCs in response to Mpg, we compared the entry of BMDCs into lymphatic vessels between the BMDCs infected with either BCG or Mpg. We investigated the migration of the CFSE-labeled BMDCs stimulated with 10 M.O.I. BCG, Mpg or LPS (1 μg/ ml) after they were injected into the footpads of mice. Our data showed that the Mpg-infected BMDCs were more than twice as likely to migrate to the pLNs than the BCG-infected BMDCs (P < 0.05) (Fig. 4c). Taken together, our data suggest that Mpg could also induce enhanced BMDC migration compared to BCG.

BMDC infected with Mpg elicited enhanced mycobacteria specific CD4 or CD8 T cell response.
To test whether Mpg could have an effect on inducing adaptive immune responses, especially to promote T cell proliferation against mycobacteria, we conducted T cell proliferation assay by the CFSE dilution method 34 . The results showed that the BMDCs infected with Mpg could induce significantly increased proliferation levels in the both CD4 and CD8 T cells compared to the BMDCs infected with BCG ( Fig. 5a-c and e). In fact, the increases in proliferation in response to BMDCs infected with Mpg or BCG were more pronounced for the CD8 T cells (Fig. 5c and e). Additionally, the amounts of secreted IL-2 cytokine from the stimulated CD4 and CD8 T cells were compared. As shown by FACS analysis, both CD4 and CD8 T cells stimulated with the Mpg-infected BMDCs secreted significantly higher levels of IL-2 than the BCG-infected BMDCs ( Fig. 5d and f). This suggests   Proliferation of (a and c) CD4 and (b and e) CD8 T cells induced by BCG-or Mpg-infected BMDCs was examined by CFSE dilution method and detected by flow cytometry. The data represent the mean ± standard deviation of CFSE intensities in each group (*P < 0.05, **P < 0.01, ***P < 0.001; Student's t-test). (d and f) IL-2 cytokine amounts in the supernatants of the CD4 and CD8 T cells used for the proliferation assay were measured by ELISA (*P < 0.05, **P < 0.01, ***P < 0.001; Student's t-test).
that the BMDCs infected with Mpg could exert more efficient mycobacteria-specific T cell immune responses than the BMDCs infected with BCG.

Mpg elicits an enhanced cell-mediated immune response against Mab or Mtb infections in vaccinated mice.
The above findings demonstrated that Mpg has the potential to function as a live vaccine against mycobacterial infection. Therefore, to address the issue, we sought to check the protective effect of Mpg against infections with two different mycobacteria, Mab and Mtb. The mice were divided into 4 groups and immunized twice by prime-boosting (4 weeks after priming vaccination) protocol ( Supplementary Fig. S2).
To test whether Mpg could improve the cell-mediated immune (CMI) response after vaccination, the splenocytes of mice from each vaccinated group were subjected to IFN-γ ELISPOT analysis. In the vaccinated mice challenged with Mab, the two groups with Mpg in the vaccine module, groups III [BCG + Mpg] and IV [Mpg + Mpg], showed a significantly more robust IFN-γ secretion and yielded significantly higher numbers of IFN-γ spot forming cells (SFCs) compared to groups I [PBS + PBS] or II [BCG + BCG] at 3 or 14 days after infection. This trend was also seen in the case of Mtb challenge. Of note, group IV showed a significantly more robust IFN-γ secretion and yielded significantly higher numbers of IFN-γ SPCs than group III at 4 or 8 weeks after infection (Fig. 6a). Therefore, the increase in the use of Mpg to the vaccine module (0 vs. 1 vs. 2) was correlated with the potentiation of the CMI response in the Mtb-challenged mice, thereby strongly suggesting a significant role of Mpg in eliciting CMI response against mycobacteria. Additionally, secreted cytokine levels of IFN-γ, TNF-α, IL-2, IL-10 and IL-12 were detected by ELISA in the supernatants of the cultured splenocytes (stimulated with whole cell lysates of Mab and Mtb). The levels of secreted IFN-γ were correlated with the IFN-γ ELISPOT results. Of note, in the Mtb-challenged mice 8 weeks after infection, the difference between the vaccinated groups with or without Mpg as the vaccine module was the most pronounced. In the Mtb-challenged mice, the two vaccinated groups with Mpg as the vaccine module, groups III and IV, secreted significantly higher levels of TNF-α compared to the other groups without Mpg as the vaccine module, groups I and II at 8 weeks after infection. In the Mab-or Mtb-challenged mice, group IV always produced a significantly higher level of secreted IL-2 compared to other groups without Mpg, groups I and II. Also, in the case of IL-12 expression, group IV of Mtb-and Mab-challenged mice produced a significantly higher level of IL-12 compared to groups II and/or III. In the Mab-or Mtb-challenged mice almost similar levels of the anti-inflammatory cytokine IL-10 was found in groups II, III and IV (Fig. 6b). These data showed that Mpg vaccination enhanced the secretion of Th1 immune response-related cytokines (IFN-γ, TNF-α, IL-2 and IL-12) compared to BCG vaccination.
Mpg elicits a mycobacterial-specific Th1-biased humoral response in immunized mice. To examine whether Mpg vaccination also elicits a Th1-biased humoral immune response after the challenge with Mab or Mtb, we analyzed the levels of Mab-or Mtb-specific IgG2a and IgG1 isotypes in the serum samples from the vaccinated mice. IgG2a and IgG1 are known representatives of Th1 and Th2 immune response markers, respectively [35][36][37] . As shown in Fig. 6a, in the Mab-or Mtb-challenged mice, the two vaccinated group with Mpg as the vaccine module groups III and IV secreted significantly higher levels of IgG2a compared to the other groups without Mpg, groups I and II. The two vaccinated groups with Mpg as the vaccine module groups III and IV also secreted significantly higher levels of IgG1 than groups I and II in the Mab-challenged mice 14 days after infection or in the Mtb-challenged mice 4 weeks after infection. Of note, the highest IgG2a/IgG1 ratio indicative of a Th1-biased humoral immune response [35][36][37] was found in group IV in the Mab-challenged mice 14 days after infection or in the Mtb-challenged mice 8 weeks after infection (Fig. 7a). These results suggested that Mpg could elicit an enhanced Th1-biased humoral immune response in the vaccinated mice.

Mpg elicits a mycobacterial-specific cytotoxic T lymphocyte response in immunized mice.
Additionally, to test whether Mpg elicits an enhanced mycobacteria-specific cytotoxic T lymphocyte (CTL) response in the vaccinated mice, we analyzed CTL activities of splenocytes from mice vaccinated with BCG or Mpg via lactate dehydrogenase (LDH) cytotoxicity assay. The immunization procedure used a prime-boost protocol (boost vaccination 2 weeks after prime vaccination: BCG + BCG or Mpg + Mpg) as described in Supplementary Fig. 2c. The splenocytes from mice immunized with BCG or Mpg were stimulated in vitro with Ag85B protein (5 μg/ml) for 6 days and used as effector cells. P815 cells (H-2 d ) transfected with the plasmid harboring Ag85B-ESAT-6 fusion genes (pcDNA3.3-Ag85B:ESAT-6) served as target cells 38 . As shown in Fig. 7b, at the E/T ratio of 50:1, the CTLs from the Mpg-immunized mouse splenocytes elicited significantly higher levels of Ag85B specific target cell lysis compared to the BCG-immunized mouse splenocytes (P < 0.05). Our data indicated that Mpg can elicit an enhanced mycobacterial-specific CTL response in the immunized mice.

Mpg elicited an enhanced protective vaccine efficacy against Mab or Mtb infection. To exam-
ine the protective immune response against Mab and Mtb after vaccination with Mpg, organs (liver, lung and spleen) from mice challenged with Mab or Mtb strains after vaccination were homogenized at each time point (3 and 14 days after challenge with Mab; 4 and 8 weeks after challenge with Mtb), and their CFUs were calculated. Mpg vaccination resulted in significant reductions in bacterial burdens (both Mab and Mtb) in all the organs that compared BCG vaccination at all time-points. Notably, in the mice challenged with Mab or Mtb, group IV always showed significantly lower levels of bacterial burden in all organs or at both time-points compared to groups I or II, except for the lungs of the mice challenged with Mab 14 days after infection (Fig. 8a). Histopathological evaluation further showed superior protective immunity after Mpg vaccination compared to BCG vaccination. More resolved inflammation was found in the lung tissues form the vaccinated mice in the two groups vaccinated with Mpg, groups III and IV, compared to groups I or II (Fig. 8b). Notably, the recovery of inflammation was most pronounced in group IV that used Mpg both for priming and boosting vaccination (Fig. 8b). Taken together, our data suggest that Mpg can elicit an enhanced vaccine efficacy in immunized mice.

Discussion
In the current study, we have demonstrated that compared to BCG, a temperature sensitive Mpg plays a pivotal role as an attenuated live vaccine in protecting against mycobacterial infections with not only Mtb, but also Mab. Our finding that Mpg fails to infect macrophages at 37 °C and elicits a milder infection in mice compared to BCG (Fig. 1c) provides a novel insight into why its isolation frequency from clinical specimens is rare. Indeed, Mpg was serendipitously discovered from the sputum sample of a Korean patient who was infected with Mab by a culture at room temperature and not at 37 °C by our groups 25 . To the best of our knowledge, other Mpg strains except for Mpg JCM 18565 T have not been found to date, suggesting that Mpg may be a genuinely rare NTM that is generally present in the environment or in colder regions or tissues of body such as mucous membranes and lacks the capacity to infect deeper tissues of the human body. The bacteriological, epidemiological or clinical traits of Mpg led us to hypothesize that it could function as the first temperature sensitive strain for live vaccination to prevent mycobacterial diseases. The naturally present TS strain of Mpg may have a very low chance of reversion into a temperature-insensitive phenotype, which is a frequent occurrence with artificially modified TS mutants, and thus Mpg could be much safer as a live vaccine than the artificially modified strains.
Our findings supporting the superiority of Mpg to BCG in eliciting DC maturation and contributing into vaccine efficacy against mycobacterial diseases are noteworthy (Fig. 9). First, immature BMDCs infected with Mpg are more resistant to cell death, which could facilitate the presentation of the mycobacterial antigen to the T cells. This is consistent with a previous report, which demonstrated that the extended survival of DCs could enhance the efficacy of the vaccine 39 . Second, the reduced survival of Mpg within DCs at the physiological temperature of 37 °C could provide increased mycobacterial peptide production from the digested Mpg into the T cells, thereby leading to enhanced mycobacteria specific T cell proliferation and activation. Indeed, our data also showed that the BMDCs infected with Mpg could induce significantly higher levels of proliferation of the Mtb antigen-specific CD4 or CD8 T cells compared to the BMDCs infected with BCG (Fig. 5), thereby supporting the above notion. Furthermore, there are several reports supporting a possible role of the deleted mutant of BCG or Mtb interfering with its persistent survival in the APCs as a candidate for live attenuated vaccines [40][41][42][43][44][45] . Third, our data also showed that the innate immune response elicited by a more attenuated Mpg strain generates a cytokine milieu that favors the priming of Th1 cell responses, indicated by the enhanced ratio of the Th1 stimulator IL-12 to the immune regulatory IL-10 (Fig. 3). Since IL-10 can be acted as a negative regulator of induced IL-12 production 46 , both IL-10 and IL-12 are together induced in LPS treated BMDCs, in a TLR4 dependent manner 47 . However, in Mpg infected BMDC, the relative higher level of IL-12 compared to IL-10 is induced (Fig. 3). This IL-12 inducing capacity of Mpg, in turn, can lead to an enhanced adaptive immunity, which may be directed against the Mpg-encoded antigens that are shared with Mab or Mtb. Indeed, our data using vaccinated mice also showed that Mpg elicits a Th1-skewed adaptive immune response as shown by an enhanced mycobacterial-specific IFN-γ secretion (Fig. 6), IgG2a/IgG1 ratio (Fig. 7a) and CTL response (Fig. 7b). In particular, it has also been reported that the infection of BCG or Mtb into DCs could lead to enhanced secretion of immune regulatory IL-10 via TLR2-dependent ERK activation, reciprocally suppressing IL-12 secretion 48 . It can reduce the adaptive immune response against mycobacterial antigens, which is the main weakness of BCG as a live vaccine against mycobacterial diseases 48,49 . The higher level of IL-12 to IL-10 ratio found in the BMDCs infected with Mpg compared to those infected with BCG could lead to an increased expression of genes involved in antigen presentation, such as MHC class II, CIITA or H2DMb as shown by our data (Fig. 3). Fourth, we showed that Mpg could also induce an enhanced BMDC migration compared to BCG by measuring the surface expression of CCR7 in the Mpg-infected BMDCs (Fig. 4a) and by quantifying the DCs that had migrated from the footpad into the pLN (Fig. 4b). This can definitely increase the chances of antigen presenting DCs to encounter antigen-specific CD4 or CD8 T cells in the draining lymph node.
Notably, our finding that the addition of Mpg in our prime-booster immunized protocols positively correlated with improvement of the CMI response (Fig. 6) and protective immune response against Mab or Mtb reveals the superiority of Mpg over BCG in the efficacy of the vaccine against mycobacterial diseases (Fig. 8). Particularly, our result showing improved CMI or protective immune response in group III [BCG + Mpg] compared to group II [BCG + BCG] (Fig. 6) suggests a potential for Mpg to be developed not only as a priming vaccine in BCG non vaccinated (naïve) humans but also as a booster vaccine in BCG-vaccinated humans. Our IFN-γ ELISPOT data showed that the CMI enhancing effect of Mpg is more pronounced in the boosting stage than in the priming stage in Mab or Mtb infections compared to that of BCG (Fig. 6a). In particular, the boosting effect of BCG was almost marginal, suggesting that there may be a disparity between Mpg and BCG in inducing memory T cell responses. This further supports the notion of the potential of Mpg as a booster vaccine against Mtb infections in BCG-vaccinated adults. The issue of how Mpg could exert a CMI boosting effect via modulation of memory T cells should be addressed in the future.
The need for vaccines capable of protecting against Mab infections has increased because of the growing number of cases around a world that are failing treatment due to the natural resistance of Mab to most of the available antibiotics [16][17][18] . However, very few studies regarding vaccine development against Mab have been reported to date. The ones that have been reported mainly focus on DNA vaccine target genes, such as phospholipase C 50 or MgtC 51 , related to the virulence of Mab. Our data clearly demonstrated the utility of Mpg as a live attenuated vaccine against Mab and Mtb infections. Particularly, the strong inducing effect of Mpg in eliciting a CMI response against Mab infection in the vaccinated mice belonging to group III [BCG + Mpg] (Fig. 6) shown by our data suggests its potential role as a booster live vaccine in BCG-vaccinated adults, which is a major target population for vaccination. To the best of our knowledge, this is the first report showing the use of a live attenuated vaccine against Mab.
Usefulness of NTMs such as heat killed or live M. vaccae and M. indicus pranii as TB vaccine have been widely studies so far 52,53 . Both M. vaccae and M. indicus pranii showed enhanced CD8 + cytotoxic T lymphocyte (CTL) responses and the expression of IFN-γ against TB in mouse model 54,55 . Also, they contribute to reduce the TB burdens in mouse 55,56 . However, most of studies mainly focused on their immunotherapeutic capacity using heat killed strains 57,58 , but their potential as preventive vaccines using live attenuated strains are still controversial. In the current study, we clearly proved usefulness of Mpg as a live attenuated preventive vaccine against mycobacterial infections. We think that capacity to elicit an enhanced innate immune response in infected APCs and temperature sensitive trait of Mpg may provide a better potential for development as a live attenuated vaccine for mycobacterial infections, compared to other NTMs.
There are some noteworthy limitations in this study. First, the mice used for the evaluation of the efficacy of the vaccine against Mtb infections were infected with avirulent H37Ra strain instead of virulent strains such as H37Rv or clinical isolates. Second, an intravenous challenge model was used for vaccine evaluation instead of an aerosol challenge model, which is the main route of infection with Mab or Mtb. We have plans for evaluating the vaccine efficacy against Mtb with an aerosol challenge model using virulent Mtb strains in the future. We are also currently in the process of improving the vaccine efficacy of Mpg for its future implementation as a vaccine in human via its combination with other types of vaccine modules in prime-boost regimens or via using different immunization routes including the intranasal route.
In conclusion, our data suggest that a temperature sensitive Mpg may be a potentially powerful candidate vaccine strain for inducing enhanced protective immune responses against mycobacterial infections.

Materials and Methods
Mycobacterial strains and culture conditions. The mycobacterial strains used in this study are as follows: To detect the dead BMDCs, we used the 7-AAD staining assay. The BMDCs were infected as before. The infected BMDCs were washed three times with cold PBS and resuspended in 100 μl of cold FACS buffer (1% FBS and 5 mM EDTA in PBS). The suspended BMDCs were stained with 7-AAD for 5 min and immediately analyzed by flow cytometry (BD LSRFortessa).

DC migration assay.
To examine the migration of the infected BMDCs from the footpad to popliteal lymph node (pLN), the BMDCs were injected with LPS (3 mg/one-leg) one day before the infection. The BMDCs were infected with BCG or Mpg or treated with LPS (1 μg/ml) overnight or not infected at all. Then, the BMDCs were incubated with 5 μM of CFSE in PBS for 4 min in a 37 °C water-bath and for 4 min on ice for staining. The prepared BMDCs were resuspended to the same concentration of 10 8 cells/ml in PBS. The BMDCs in PBS (30 μl) were injected into the right footpads of mice. After 52 hours, pLNs from the injected mice were collected and ground with frosted glass to produce single cells. The single cells were stained with CD11c-APC-cy7. The populations of the migrated BMDCs in pLN were determined by detecting the CD11c + CFSE-labeled population using flow cytometry (BD LSRFortessa).

Quantitative real time-PCR (qRT-PCR). Total RNA from the BMDCs or THP-1 cells infected with
BCG or Mpg was extracted using RNA-spin TM Total RNA Extraction kit (iNtRON Biotechnology, Gyeonggi-do, Korea) according to the manufacturer's protocol. Purified RNA samples were used as templates for qRT-PCR with SensiFAST TM SYBR Lo-ROX One-Step kit (Bioline, Taunton, MA, USA). The primer pairs used to amplify interleukin-10 (IL-10), IL-12, H2DMb (MHC II-associated proteins), CIITA (MHC II transactivator) and CCR7 are given in Supplementary Table 1. The reaction was carried out with an ABI 7500 Real Time PCR System (Applied Biosystems, Foster City, CA, USA) and the data were analyzed using the Ct (cycle threshold) values.
T cell proliferation assay. To conduct T cell proliferation assay, CD4-and CD8-positive T cells from mice immunized with Mtb (H37Ra) (subcutaneous route at the base of the tail, 1 × 10 6 CFU in 100 μl of PBS) and BMDCs infected with 10 M.O.I. of BCG or Mpg were used. The schematic schedule for the T cell proliferation assay is described in Supplementary Fig. 2a. Proliferation assays were conducted with fluorescent cytoplasmic tracking dye, carboxyfluorescein diacetate succinimidyl ester (CFSE), CD4 BV421-conjugated anti-CD4 (Clone GK1.5, BD Horizon) and PE-conjugated anti-CD8a (Clone 53-6.7, eBioscience) as described previously 34,38 . The cell cycle profiles were determined using FACS LSRFortessa (BD Biosciences), and analyzed using FlowJo. All the experiments were performed in triplicate.
Mice and immunization procedures. Female BALB/c and BALB/c_nu (nude) (~25 g, 7-week-old) were purchased from Orient-Bio (Seoul, South Korea) and used for experiments at the age of 8 weeks. The mice were randomly divided into groups with five mice in each group.
For safety test, mice were infected with BCG or Mpg through the tail vein (1 × 10 6 CFU in 200 μl). At 1, 7, 14 and 28 days after the infection, the mice were euthanized by CO 2 , and their organs (liver, lungs and spleen) were removed for CFU enumeration (Supplementary Fig. 2b).
For CTL analysis, the BALB/c mice were subcutaneously vaccinated twice (with a 2-week-interval) with BCG or Mpg (1 × 10 6 CFU in 100 μl) at the base of the tail. Two weeks after final vaccination, the vaccinated mice were euthanized, and the spleens were removed for CTL analysis (Supplementary Fig. 2c).
For vaccination test, the BALB/c mice were subcutaneously vaccinated twice (with a 4-week-interval) with BCG or Mpg (1 × 10 6 CFU in 100 μl) at the base of the tail. The groups were as follows: group I, PBS-vaccinated group (two times); group II, BCG-vaccinated group (two times); group III, BCG-and Mpg-vaccinated group (first vaccination with BCG and second vaccination with Mpg); group IV, Mpg-vaccinated group (two times). Four weeks after the final vaccination, the mice were challenged with Mab or Mtb H37Ra strains via the tail vein (1 × 10 6 CFU in 200 μl). All the vaccinated mice were euthanized at each time point and their bloods and organs were removed for further immunological study ( Supplementary Fig. 2d). IFN-γ Enzyme-Linked ImmunoSpot (ELISPOT) assay. Splenocytes from the mice immunized with BCG or Mpg were used to conduct ELISPOT assays as described previously 62 . The ELISPOT assay was performed using the mouse IFN-γ (3 μg/ml, clone: AN-18) capture antibody (BD-Biosciences, San Diego, CA, USA) and biotin-labeled mouse IFN-γ (3 μg/ml, clone: XMG1.2) detection antibody (BD-Biosciences) as described previously 38 . Splenocytes (5 × 10 5 cells/well) were analyzed 24 hour after re-stimulation with whole lysates of Mab or Mtb at a final concentration of 5 μg/ml. The number of spot forming units (SFUs) per well was counted automatically using an ELISPOT reader (AID ELISPOT reader, Strassberg, Germany). CFU enumeration assay. Organs (liver, lungs and spleen) from the vaccinated and challenged mice were homogenized with strainers (70 μm mesh) and serially diluted with PBS. The bacterial loads in the organs were determined by plating the homogenized organs onto 7H10 agar plates (supplemented with OADC) for 7 days (in the case of Mab) or 4 weeks (in the case of Mtb) at 37 °C.

Serum antibody detection.
To detect serum antibody ratio, serum samples were collected by cardiac puncture method from the immunized mice after euthanasia with CO 2 . Whole lysates (5 μg/ml) of Mab or Mtb in 0.05 M carbonate-bicarbonate buffer (pH 9.6) were used to coat 96-well plates overnight at 4 °C. Serum antibodies were detected using mouse IgG2a and IgG1 antibodies (BD Biosciences, 1:1,000 dilution) as described previously 38 . The optical density (OD) was determined with a spectrometer at 450 nm 63 . Cytotoxic T lymphocyte (CTL) assay. The induced CTL responses were determined as described previously 38,64 . Briefly, the P815 cells (H-2 d ) were transfected with plasmids containing Ag85B-ESAT-6 (pcD-NA3.3-Ag85B-ESAT-6) by using Lipofectamine 2000 (Invitrogen, Carlsbad, USA), and then used as the target cells (5 × 10 4 cells) for the cytotoxic T lymphocyte (CTL) assay. The non-transfected cells were used as negative controls. Splenocytes (5 × 10 6 cells/well) from the vaccinated mice were co-cultured in RPMI 1640 medium (Biowest, Nuaille, France) with 10% FBS and antigens (Ag85B, 5 μg/ml) at 37 °C in a 5% CO 2 incubator for six days. These stimulated splenocytes were used as effector cells to evaluate the CTL responses. Cell cytotoxicity was evaluated with lactate dehydrogenase (LDH) in 96-well U bottom plates according to the manufacturer's protocol (CytoTox 96 Non-Radioactive Cytotoxicity Assay; Promega, Madison, USA) as described previously 38 . Hematoxylin and eosin (H&E) staining. For histopathological studies, the lungs were removed from the immunized mice. One fifth of the right lobe of the lung was fixed in 10% formaldehyde (Junsei Chemical Co., Tokyo, Japan) and embedded in paraffin. The paraffin blocks were sectioned at approximately 5 μm thickness and stained with hematoxylin and eosin (H&E).
Statistical analysis. The data represent the mean ± standard deviation. Student's t-test was used to compare the variance, and the differences were considered statistically significant when the probability values were less than 0.05.