Mycobacterium smegmatis Vaccine Vector Elicits CD4+ Th17 and CD8+ Tc17 T Cells With Therapeutic Potential to Infections With Mycobacterium avium

Mycobacterium avium (Mav) complex is increasingly reported to cause non-tuberculous infections in individuals with a compromised immune system. Treatment is complicated and no vaccines are available. Previous studies have shown some potential of using genetically modified Mycobacterium smegmatis (Msm) as a vaccine vector to tuberculosis since it is non-pathogenic and thus would be tolerated by immunocompromised individuals. In this study, we used a mutant strain of Msm disrupted in EspG3, a component of the ESX-3 secretion system. Infection of macrophages and dendritic cells with Msm ΔespG3 showed increased antigen presentation compared to cells infected with wild-type Msm. Vaccination of mice with Msm ΔespG3, expressing the Mav antigen MPT64, provided equal protection against Mav infection as the tuberculosis vaccine, Mycobacterium bovis BCG. However, upon challenge with Mav, we observed a high frequency of IL-17-producing CD4+ (Th17 cells) and CD8+ (Tc17 cells) T cells in mice vaccinated with Msm ΔespG3::mpt64 that was not seen in BCG-vaccinated mice. Adoptive transfer of cells from Msm ΔespG3-vaccinated mice showed that cells from the T cell compartment contributed to protection from Mav infection. Further experiments revealed Tc17-enriched T cells did not provide prophylactic protection against subsequent Mav infection, but a therapeutic effect was observed when Tc17-enriched cells were transferred to mice already infected with Mav. These initial findings are important, as they suggest a previously unknown role of Tc17 cells in mycobacterial infections. Taken together, Msm ΔespG3 shows promise as a vaccine vector against Mav and possibly other (myco)bacterial infections.

Mycobacterium avium (Mav) complex is increasingly reported to cause non-tuberculous infections in individuals with a compromised immune system. Treatment is complicated and no vaccines are available. Previous studies have shown some potential of using genetically modified Mycobacterium smegmatis (Msm) as a vaccine vector to tuberculosis since it is non-pathogenic and thus would be tolerated by immunocompromised individuals. In this study, we used a mutant strain of Msm disrupted in EspG 3 , a component of the ESX-3 secretion system. Infection of macrophages and dendritic cells with Msm espG 3 showed increased antigen presentation compared to cells infected with wild-type Msm. Vaccination of mice with Msm espG 3 , expressing the Mav antigen MPT64, provided equal protection against Mav infection as the tuberculosis vaccine, Mycobacterium bovis BCG. However, upon challenge with Mav, we observed a high frequency of IL-17-producing CD4+ (Th17 cells) and CD8+ (Tc17 cells) T cells in mice vaccinated with Msm espG 3 ::mpt64 that was not seen in BCG-vaccinated mice. Adoptive transfer of cells from Msm espG 3 -vaccinated mice showed that cells from the T cell compartment contributed to protection from Mav infection. Further experiments revealed Tc17-enriched T cells did not provide prophylactic protection against subsequent Mav infection, but a therapeutic effect was observed when Tc17-enriched cells were transferred to mice already infected with Mav. These initial findings are important, as they suggest a previously unknown role of Tc17 cells in mycobacterial infections. Taken together, Msm espG 3 shows promise as a vaccine vector against Mav and possibly other (myco)bacterial infections.

INTRODUCTION
Mycobacterium avium (Mav) is an opportunistic non-tuberculous mycobacterium (NTM) that mostly affects individuals with a compromised immune system (1)(2)(3). The drug regimen to eradicate Mav infections is arduous and often not successful and there are no vaccines available against Mav infections. However, a recent meta-analysis found evidence that vaccination with Mycobacterium bovis Bacillus Calmette-Guérin (BCG), the only available vaccine against Mycobacterium tuberculosis (Mtb), might exhibit cross-protection to infections with NTMs in immunocompetent individuals (4). Conversely, NTMs may exhibit anti-tuberculous resistance and also interfere with BCG vaccination (5). In addition to limited efficiency, an additional challenge with the BCG vaccine is that it is not well-tolerated by HIV-infected infants and other patients with a compromised immune system (6), leaving the people most vulnerable to mycobacterial infections without protection. Consequently, new vaccines with improved safety and efficacy profiles are needed to boost previously exposed or BCG-vaccinated individuals, and to combat emerging NTM infections for which we currently have no vaccines.
The development of new and improved vaccines against mycobacteria is challenging due to a lack of knowledge on the correlates of immune protection that could predict vaccine efficacy (7,8). Adaptive immunity to Mav is considered to be mediated mainly by CD4+ T helper (Th) 1 cells (9,10). In particular, the production of interferon (IFN) γ by CD4+ Th1 cells is important to control Mav infection, and mice genetically deficient in IFNγ have increased susceptibility to infection (9,11). In addition to the Th1 response, mycobacterial infections also elicit a Th17 response (12)(13)(14). Early studies suggested that the IL-23/IL-17 axis was not critical for protection against tuberculosis (TB) in mice (15,16). However, later studies in vaccinated mice provided evidence that Th17 cells may contribute to protection in mice that have been vaccinated with the Mtb antigen EsxA (13,14). A role for the Th1/Th17 balance in Mav infection was suggested, wherein mice deficient for the transcription factor T-bet, critical for Th1 cell differentiation, showed a shift from Th1 toward Th17 responses and were more susceptible to Mav infection (17). Regarding the importance of CD8+ cytotoxic T (Tc) cells, experiments in mice deficient in CD8+ T cells indicated that Tc cells play a minor role in Mav infection (10,18). However, as for Th cells, this could depend on the Tc cell subsets elicited. CD8+ T cells producing IL-17 (Tc17 cells) have been observed in pleural effusion of TB patients (19), and recently Loxton et al. (20) observed Tc17 cells in infants vaccinated with the strain BCG VPM1002. However, little is known about the functional role of Tc17 cells in mycobacterial infections.
Different approaches are used for the TB vaccine candidates currently under development and in clinical trials, either to replace BCG or to boost previously vaccinated or exposed individuals (7,8). One approach is to improve safety and efficacy by engineering BCG or other mycobacteria to interfere with phagosome maturation and to express Mtb antigens. Another strategy pursues administration of Mtb antigens as subunit booster vaccines together with adjuvants. Prominent MTb antigens that have been included in TB vaccine candidates are EsxA, EsxH, and MPT64 which aid in MTb immune evasion (21)(22)(23)(24)(25). These proteins are secreted by various secretion systems like the early-secreted antigenic target secretion system (ESX, or type VII secretion system). Five ESX secretion systems (ESX-1 to ESX-5) are described within various mycobacterial species (26). ESX-3 is involved in iron uptake and is conserved across all mycobacterial species (27,28). It has been shown that a modified strain of Mycobacterium smegmatis (Msm) in which the endogenous esx-3 was exchanged with the Mtb esx-3 locus, has potential as a vaccine against Mtb infections in mice (29). The vaccine strain elicited a pro-inflammatory milieu within mice and provided equal or superior protection, when compared to BCG, against subsequent Mtb challenge. However, it is not known if a Msm vaccine would provide protection against infections with NTM, such as bacteria of the Mycobacterium avium complex.
In this study, we created a Msm vaccine candidate strain that is deficient in the ESX-3 secretion-associated chaperone protein EspG 3 . We and others have previously shown that EspG 3 is an important component for ESX-3 function (27,30,31), and that Msm espG 3 functionally resembles the Msm esx-3 mutant used by Sweeney et al. (29). Our results show that Msm espG 3 was more efficient than Msm wild-type (WT) in activating CD4+ T cells specific to a model antigen, ovalbumin (OVA). Vaccination of mice with Msm espG 3 expressing a Mav antigen, MPT64 (Msm espG 3 ::mpt64), offered similar protection as BCG against Mav infection. A strong induction of IL-17-producing CD4+ and CD8+ T cells in response to Mav infection in the Msm espG 3 ::mpt64-vaccinated mice, revealing a possible role of IL-17-producing T cells in controlling Mav infection.

Cell Culture
Bone marrow cells were isolated from femurs of C57BL/6 mice. Red blood cells were lysed (RBC lysis buffer, eBioscience) before cells were resuspended in RPMI-1640 medium (Sigma) supplemented with 10% fetal calf serum (Gibco). To differentiate cells toward bone marrow-derived macrophages (BMDMs), the medium was supplemented with 20% supernatant from L-929 cells for 4-5 days. Ten nanograms per milliliters of mouse Granulocyte-macrophage colony-stimulating factor (Stem cell technologies) was added for 6-7 days to differentiate cells primarily toward bone marrow-derived dendritic cells (BMDCs), though the resulting cell population has been described to also contain other myeloid cell types (33).

In vitro Infection of BMDMs and BMDCs
BMDMs or BMDCs were seeded in a 96 well-plate (50,000 cells/per well in triplicate) in 200 µl RPMI-1640 medium (Sigma) supplemented with 10% fetal calf serum and cultivated overnight. Cells were infected with WT Msm and Msm espG 3 at a multiplicity of infection (MOI) of 10. Infection was performed for 2 h. To remove and kill extracellular bacteria, the cells were washed with Hanks balanced salt solution (Sigma), then incubated for 30 min in medium containing 10 µg/ml gentamicin (Sanofi) and washed again. Treatment of cells with 10 µg/ml gentamicin for 30 min prevented cells from overgrowth with extracellular bacteria and did not affect cell health. Directly after infection as well as 4, 10, and 24 h post-infection, cells were lysed with phosphate buffered saline (Sigma) containing 0.02% Triton X (Sigma). The lysate was plated in serial dilutions on 7H10 Middlebrook plates to quantify colony forming units (CFU). To assess major histocompatibility complex (MHC) class II and CD86 expression as well as nitric oxide production in Msm infected cells, BMDMs were seeded in 24 well-plates (500,000 cells/per well) and infected for 24 h as described above. To quantify nitric oxide production, cells were incubated for the last 30 min with 1 mM 4amino-5-methylamino-2 ′ ,7 ′ -difluororescein diacetate (DAF-FM diacetate, Invitrogen), control cells samples were treated with lipopolysaccharide (100 ng/ml, Sigma). To assess MHC class II and CD86 expression, cells were removed from the plates, treated with Fc block (anti-CD16/CD32, eBioscience) and stained with fluorescent monoclonal antibodies to MHC class II (I-Ab Alexa Fluor 488, BioLegend) and CD86 (Phycoerythrin, BD Biosciences). Nitric oxide production and MHC class II and CD86 expression was analyzed by flow cytometry on a BD LSR II flow cytometer (BD Biosciences) and analyzed with FlowJo_v.10 (FlowJo, LLC). GraphPad Prism 8 software (GraphPad Software, Inc.) was used to perform statistical analyses.

Plasmid Constructs
For the overexpression of mpt64, we used Gateway technology (Sigma) to clone our inserts. Entry clones for mpt64 and p750, the constitutive promoter, were made following a PCR with the following primers and a subsequent Gateway BP reaction.
B5r-p750 promoter: pDE43 (kindly provided by D. Schnappinger, Weill Cornell, NY) is a Gateway compatible destination vector that can be used in mycobacteria. Entry clones mpt64, p750, and pDE43 were fused to generate clones containing mpt64 under a constitutive promoter p750. Gateway reactions were performed as specified by the manufacturer. For over-expression of truncated ovalbumin (OVA), PCR primers were designed that amplified the region containing both the MHC class I antigenic epitope OVA 257−264 (SIINFEKL) and the MHC class II antigenic epitope OVA 323−339 (ISQAVHAAHAEINEAGR). Primer sequences: Truncated OVA forward primer 5 ′ GGA ATTCCATATGGGGATCCTGGAGCTTCCA 3 ′ , truncated OVA reverse primer 5 ′ ACATGCATGCCTAGTCTTCAGAGACGCT 3 ′ . Plasmids containing constitutive promoter pMV261 were used. Both the insert and plasmid were digested with NdeI and SphI (both New England Biolabs, Inc.). Clones generated were subsequently sequenced. qPCR RNA was quantified from both WT and espG 3 Msm either overexpressing truncated OVA protein or the Mav MPT64 protein (Supplementary Figure 1). Bacteria grown to exponential phase were pelleted down and resuspended in 600 µl Trizol (Invitrogen). The resuspended bacteria were added to tubes containing 0.1 mm beads (Bertin Technology) and bead-beated twice for 2 min with a FastPrep-24 instrument (MP biosystems, 4.0 M/S, 30 s ON and 30 s OFF). Onehundred and fifty microliters of chloroform was added to the supernatant after bead-beating. The aqueous phase containing RNA was carefully transferred into 1.5 ml RNase free Eppendorf tubes and 300 µl cold isopropanol (Teknisk) was added and kept at −20 • C for 15 min. RNA was isolated (Qiagen RNA purification Kit) and subsequently reverse transcribed to cDNA (Applied Biosystems). Quantitative realtime PCR was performed in duplicates on a StepOnePlus qPCR System (Applied Biosystems) with primers to truncated OVA (forward primer 5 ′ GTTGGTGCTGTTGCCTGATG3 ′ , reverse primer 5 ′ CTCTGCTGAGGAGATGCCAG 3 ′ ) and Mav mpt64 (forward primer 5 ′ GATCAGCCCTACCAGCTGAC 3 ′ , reverse primer 5 ′ TTCTGGACGACCTTGAGCAC 3 ′ ). RNA polymerase sigma factor SigA expression was analyzed as reference for normalization of qPCR data.

In vitro Antigen Presentation and T Cell Activation Assay
BMDCs were generated as described and seeded in 24 wellplates (500,000 cells per well) for in vitro analysis of MHC class I antigen presentation experiments or in 96 well-plates (50,000 cells per well) for in vitro analysis of CD4+ and CD8+ T cell cell activation. The cells were infected with WT Msm and Msm espG 3 overexpressing truncated OVA or the respective empty vector control strains at an MOI of 10. Treatment with full-length chicken OVA protein (225 µM, Sigma) plus lipopolysaccharide (100 ng/ml, Sigma) was carried out as positive control for antigen presentation. After Msm infection for 2 h, cells were washed and extracellular bacteria eliminated (30 min, 10 µg/ml gentamicin).
Analysis of CD4+ and CD8+ T cell activation: Directly after infection, BMDCs were washed and subsequently overlaid with 1 × 10 5 OVA-specific T cell hybridoma cells per well and incubated overnight at 37 • C. RF33.7 cells were used to detect MHC class I presentation of the OVA 257−264 (SIINFEKL) epitope and MF2.2D9 cell line to detect MHC class II presentation of the OVA 323−339 (ISQAVHAAHAEINEAGR) epitope, as described in Moura Rosa et al. (34) and Gopalakrishnan et al. (35). Supernatants from the BMDC:T cell hybridoma co-culture were collected and production of bioactive cytokines from activated RF33.70 or MF2.2D9 hybridoma cells was quantified in a bioassay (36). For this, IL-2 dependent HT-2 cells (1 × 10 4 cells/well) were cultured overnight in 50% supernatant from BMDC:T cell hybridoma co-culture before proliferation of HT-2 cells was analyzed using the CellTiter 96 Aqueous One Cell Proliferation Assay (Promega).

Mouse Vaccination and Infection
All protocols on animal work were approved by the Norwegian National Animal Research Authorities and carried out in accordance with institutional guidelines, national legislation, and the Directive of the European Convention for the protection of animals used for scientific purposes. Six to eight week old C57BL/6 mice were bred in house at the Comparative Medicine Core Facility (CoMed) at NTNU and used for vaccination and infection experiments.
Subcutaneous vaccinations were performed with Msm espG 3 ::mpt64 and BCG. 1 × 10 7 bacteria were injected in 100 µl PBS, sham-vaccinated mice received 100 µl of PBS. Msm espG 3 ::mpt64 vaccination was performed twice with a 15 day interval, BCG vaccination was performed with a single vaccination dose (1 × 10 7 bacteria, Figure 2A). An aliquot of the inoculum was plated in serial dilution on Middlebrook 7H10 plates to verify the bacteria number. Thirty days post the first vaccination, mice were challenged with the virulent Mav strain 104 (Figure 2A). Infection was performed by intraperitoneal injection of log-phase Mav strain 104 (1 × 10 9 CFUs in 200 µl PBS per mouse). Thirty days after infection, mice were killed, and spleen and liver were collected for CFU counting and T cell analysis. Bacterial load was measured by plating serial dilutions of organ homogenates (spleen, liver) on Middlebrook 7H10 plates.

Mycobacteria-Specific T Cell Cytokine Production and Memory T Cell Analysis
Mav-specific effector T cell cytokine production was analyzed from splenocytes of mice. Splenocytes from mice were isolated 30 days post Mav infection, stimulated overnight with Mav (MOI 3:1) and prepared for flow cytometry as previously described (37). Protein transport inhibitor cocktail (eBioscience) was added for the last 4 h of stimulation. Surface antigens were characterized by staining with fluorescence-labeled monoclonal antibodies against CD3 (Fluorescein isothiocyanate), CD4 (Brilliant Violet 605), CD8 (Brilliant Violet 785, all BioLegend). After fixation and permeabilization (eBioscience Intracellular Fixation and Permeabilization Buffer Set), intracellular cytokine production was analyzed by staining with fluorescent monoclonal antibodies to IFNγ (Phycoerythrin), TNFα (Allophycocyanin) and IL-17 (PE/Cy7, all Biolegend). For memory T cell phenotype analysis, unstimulated splenocytes were stained as described above for CD3, CD4 and CD8. Additionally, cells were stained with monoclonal antibodies against CD44 (Alexa Fluor 700) and CD62L (Brilliant Violet 510, both from Biolegend). Flow cytometry was performed on a BD LSR II flow cytometer (BD Biosciences) and data subsequently analyzed with FlowJo_v.10 (FlowJo, LLC). Statistical analyses were performed with GraphPad Prism 8 software (GraphPad Software, Inc).

Adoptive Transfer
A timeline of the procedures for the adoptive transfer experiments can be found in Vaccination of mice for adoptive transfer: Donor mice for adoptive cell transfer were vaccinated with Msm espG 3 ::mpt64 or sham-vaccinated as described above. On day 28, Msm espG 3 ::mpt64 and sham-vaccinated mice were infected with Mav strain 104 (1 × 10 7 in 200 µl PBS) to allow generation of fully active Mav-specific effector T cells. Two days later, donor mice were killed, and spleens harvested for isolation of CD3+, CD8+IL-17-, and CD8+IL-17+ T cells.
Isolation of T cell fractions for adoptive transfer: Total CD3+ T cells were purified from spleens of Msm espG 3 ::mpt64 or sham-vaccinated mice using the Dynabeads Untouched Mouse T Cell kit (Thermo Fisher). For Tc17-enrichment or Tc17depletion, untouched CD8+ T cells were isolated from spleens of Msm espG 3 ::mpt64-vaccinated mice (CD8a+ T cell isolation kit, Miltenyi Biotec). CD8+IL-17-and CD8+IL-17+ T cell fractions were enriched from CD8+ T cells using the mouse IL-17 Secretion Assay Cell Enrichment and Detection Kit according to the manufacturer's protocol (Miltenyi Biotec). The purity of adoptively transferred T cell fractions was assessed by flow cytometry and is exemplified in Supplementary Figure 6. CD3+ T cells for adoptive transfer experiments had a purity of >90%, Tc17-depleted cells contained <1% and and Tc17-enriched T cell fractions >10% of IL-17-producing CD8+ T cells.

EspG 3 -Deficient Msm Shows Reduced Survival in Antigen Presenting Cells and Increased T Cell Activation Compared to Wild-Type Msm
It has previously been shown that vaccination of mice with Msm in which the endogenous esx-3 locus is complemented with Mtb esx-3, promotes protective immunity against Mtb and thus has potential as a novel vaccine vector (29). We wanted to investigate if Msm deficient in the ESX-3 component, EspG 3 , by itself was superior compared to WT Msm in enabling antigen presenting cells (APCs) to present antigens produced within the bacterium. First, we confirmed that Msm espG 3 , Msm WT, and the complemented Msm strain ( espG 3 ::espG 3 ) had comparable growth rates when cultivated in vitro ( Figure 1A). Subsequently, we assessed survival of the Msm strains within APCs. Bone marrow-derived dendritic cells (BMDCs) and bone marrow-derived macrophages (BMDMs) were infected with WT Msm, Msm espG 3 , and the reconstituted Msm espG 3 ::espG 3 . Twenty-four hours of post-infection, the bacterial load of Msm espG 3 was one log (90%) lower compared to Msm WT or reconstituted Msm espG3::espG3 in BMDMs (Figure 1B, left graph). In BMDCs, the difference was less pronounced with 35% lower CFUs of Msm espG3 compared to Msm WT and the reconstituted strain ( Figure 1B, right graph). This reduced survival could result from restricted intracellular growth or from improved detection and clearance of Msm espG 3 compared to Msm WT, which could impact antigen presentation in different ways. We thus addressed activation status and nitric oxide production in BMDMs post Msm infection (Supplementary Figure 1A). Infection with all three Msm strains increased the surface expression of MHC class II on BMDMs. But while infection with Msm WT and Msm espG 3 ::espG 3 strongly increased expression levels of the T cell co-stimulatory molecule CD86, the Msm espG 3 strain failed to do so. None of the three Msm strains induced nitric oxide production in infected BMDMs.
To further assess the effectiveness of Msm espG 3 as an antigen-presenting vehicle, we introduced a truncated ovalbumin construct encoding the C57BL/6 mouse MHC class I epitope OVA 257−264 (OVA1) and the MHC class II epitope OVA 323−339 (OVA2) in Msm WT and Msm espG 3 . Only Msm espG 3 expressing the OVA2 epitope significantly increased activation of OVA2-specific CD4+ T hybridoma cells by infected BMDCs (Figure 1C, right). In contrast, OVA1-specific CD8+ T hybridoma cells were not significantly activated by BMDCs infected with the OVA1-expressing Msm espG3. Infection with Msm WT did not increase CD4+ or CD8+ T cell activation, independent of OVA expression. We further directly assessed antigen presentation of OVA1 peptides via MHC class I on BMDCs by antibody staining of MHC class I/OVA 257−264 complexes (Supplementary Figure 1C). OVA1 peptides were presented in BMDCs infected with OVA1 expressing Msm strains. However, we did not observe significant differences in MHC class I-restricted antigen presentation between the OVA1expressing Msm espG3 and WT strain. Taken together, the findings of reduced Msm espG 3 survival in APCs and increased CD4+ T cell activation suggest that Msm espG 3 could be a better vaccine vector than Msm WT.

Msm Vaccination Protects Against Mav Infection
To evaluate the effectiveness of Msm espG 3 as a vaccine vector in vivo, we replaced the OVA test-antigens with the Mav antigen MPT64, creating the strain Msm espG 3 ::mtp64. MPT64 is an immunodominant secretory antigen (38) and the Mtb homolog has been used as antigen in vaccine studies against TB (23)(24)(25). We vaccinated mice subcutaneously with Msm espG 3 ::mtp64 and included a booster vaccination at day 15, since Msm espG 3 is rapidly cleared from mice (29). BCG vaccination, which is commonly administered in a single dose, was performed for comparison. Sham-vaccinated mice received PBS (vaccination scheme shown in Figure 2A). Thirty days post the first vaccination, mice were challenged with Mav strain 104, and CFUs in liver and spleen were quantified 30 days after challenge ( Figure 2B). Organ bacterial load was significantly lower in the liver of both BCG and Msm espG 3 ::mtp64vaccinated mice compared to sham-vaccinated mice. In spleens, reduction of organ bacterial load was not statistically significant in any vaccination group compared to sham-vaccinated mice.  Figure 2A) Figure 4).
We first examined if the total frequency of CD4+ and CD8+ T cells producing effector cytokines (IFNγ, TNFα, or IL-17) in response to Mav stimulation varied between the different vaccines in our study ( Figure 3A). We observed Mavspecific effector cytokine production in about 13% of CD4+ T cells from BCG-vaccinated mice, compared to about 7% in both Msm espG 3 ::mtp64-and sham-vaccinated mice. The frequencies of cytokine-producing CD8+ T cells were higher in mice vaccinated with BCG and Msm espG 3 ::mtp64 compared to sham-vaccinated mice (7 vs. 5 and 4%, respectively). We further addressed multifunctionality of Mav-specific CD4+ and CD8+ T cells by analyzing the distribution of single-, double-and tripleproducers of the effector cytokines IFNγ, TNFα, or IL-17 within the total population of cytokine-producing CD4+ and CD8+ T cells (Figure 3B). A difference could indicate which effector T cell subsets and mechanisms are involved in controlling the Mav infection. The majority of the CD4+ T cells in BCG-and sham-vaccinated mice were typical Th1 cells producing IFNγ or IFNγ/TNFα (combined more than 80% of cytokine-producing cells, Figure 3B upper charts). In contrast, increased frequencies of IL-17 or IFNγ/IL-17 (combined ∼35% of cytokine-producing cells) expressing CD4+ cells were seen in Msm espG 3 ::mtp64vaccinated mice. A small fraction of triple-producing CD4+ cells (IFNγ, TNFα, and IL-17) was seen in all groups and most in Msm espG 3 ::mtp64-vaccinated mice (10% of the total cytokineproducing CD4+ cells compared to 2% in BCG and 2% in sham-vaccinated mice).
Finally, memory cell generation was assessed as a protective correlate of the vaccines (Supplementary Figure 5). Msm espG 3 ::mtp64 induced significantly more CD4+ T cells with a T central memory phenotype (CD44 hi CD62L hi ) than BCG, no significant difference was observed in CD8+ T central memory cells between Msm espG 3 ::mtp64 and BCG vaccinated mice. No difference was observed for CD4+ and CD8+ T cells with a T effector memory cell phenotype (CD44 hi CD62L lo ) between Msm espG 3 ::mtp64 and BCG.
To address whether the espG 3 deletion or MPT64 expression in Msm are required for the effector T cell polarization toward Tc1 and Tc17 cells, we analyzed effector T cell responses in mice that received Msm WT, Msm espG 3 or Msm espG 3 ::mtp64 vaccination before Mav infection. Msm espG 3 ::mtp64-vaccinated mice showed the highest frequencies of IL-17-producing CD4+ (Th17) and CD8+ (Tc17) T cells (Supplementary Figure 2B).
Taken together our results suggest that even if all Msm vaccine strains mediated protection to Mav infection, only the EspG 3 disrupted Msm strain overexpressing the MPT64 antigen favored polarization of Mav-specific T cell responses toward CD4+ Th17 and CD8+ Tc17 T cells.

Vaccination With Msm espG 3 ::mtp64 Elicits T Cells With Therapeutic Potential in Mav Infection
Since we found that vaccination with Msm espG 3 ::mtp64 conferred protection to Mav infection and polarization toward IL-17-producing effector T cells, we wanted to demonstrate that the protective effect against Mav infection originates from the T cell compartment. Therefore, we purified CD3+ T cells from Msm espG 3 ::mtp64-or sham-vaccinated vaccinated mice and adoptively transferred them into unvaccinated mice that were pre-infected for 15 days with Mav strain 104 (Figures 4A,B). Thirty days after adoptive CD3+ T cell transfer from vaccinated mice, organ bacterial load, and effector T cell responses were analyzed in the recipient mice ( Figure 5). Bacterial load in spleen and liver of mice that received adoptive transfer of CD3+ T cells from Msm espG 3 ::mtp64-vaccinated mice was significantly reduced compared to mice that received CD3+ T cells from sham-vaccinated mice (Figure 5A). These results demonstrate that effector T cells from Msm espG 3 ::mtp64-vaccinated mice have a therapeutic potential and convey to protection from Mav infection in the recipient mice.
We further analyzed the proportions of IFNγ-or IL-17-producing Mav-specific effector T cells in spleens of the recipient mice. Adoptive transfer of CD3+ T cells from Msm espG 3 ::mpt64-or sham-vaccinated mice did not change the total T cell frequencies of IFNγ-producing CD4+ Th1 or CD8+ Tc1 cells (Figure 5B). However, the frequencies of IL-17 singleproducing or IL-17/IFNγ double-producing CD4+ Th17 and CD8+ Tc17 cells were significantly increased in mice receiving CD3+ T cells from Msm-vaccinated mice compared to mice receiving T cells from sham-vaccinated mice (Figure 5B). These results suggest that T cells confer the protective effect of Msm vaccination against Mav infection. It has to be elucidated if the significantly increased frequencies of IL-17-producing Th17 and Tc17 that we found in Msm-vaccinated mice contribute to the protective effect.

Tc17 Cells Have a Therapeutic Effect in Mav Infection
The most striking difference between BCG and Msm espG 3 ::mtp64 vaccination was a shift from an IFNγ-dominated (Th1/Tc1) effector T cell response in BCG-vaccinated mice toward an effector response with increased frequencies of IL-17-producing CD4+ Th17 and CD8+ Tc17 effector T cells in Msm espG 3 ::mtp64-vaccinated mice. Since very little is known about the effect of CD8+ Tc17 cells in mycobacterial infections, we set out to investigate if Tc17 cells could have a protective role during Mav infection. For this, we adoptively transferred Tc17-enriched or -depleted CD8+ T cell fractions from Msm espG 3 ::mtp64-vaccinated mice into unvaccinated recipient mice. We tested if transfer of Tc17 cells protects mice when administered before the Mav infection is established (prophylactic effect) as well as the effect of administering Tc17 effector cells to an ongoing Mav infection (therapeutic effect). CD8+ T cells were isolated from spleens of Msm espG 3 ::mtp64-vaccinated mice and separated into fractions enriched or depleted for CD8+ Tc17 cells. The CD8+IL-17+ T cell fraction contained >10% IL-17+ T cells, while the CD8+IL-17-fraction was devoid of IL-17+ T cells (Supplementary Figure 6). The Tc17-enriched and Tc17depleted cell fractions were then adoptively transferred to naïve mice prior to challenge with Mav (prophylactic) or 15 days post-infection with Mav (therapeutic) (experimental procedures shown in Figure 4). Due to the incomplete enrichment of Tc17 cells, recipient mice received less Tc17 cells than estimated.
Bacterial load and effector T cell responses in recipient mice were analyzed in both setups 30 days after adoptive T cell transfer. Adoptive transfer of Tc17-enriched or Tc-17-depleted CD8+ T cells did not significantly reduce organ bacterial loads in spleen ( Figure 6A) or liver (Figure 6B) of the recipient mice. This was the case for the prophylactic as well as the therapeutic experimental setup.
Finally, we analyzed the Mav-specific effector T cell composition in in recipient mice 30 days post transfer of Tc17-enriched or -depleted cells. Since we did not trace the adoptively transferred T cells, we could not distinguish if these cells originated from the donor or recipient mice. In the prophylactic setup, we observed no differences in the IFNγproducing CD4+ Tc1 or CD8+ Tc1 cells in mice receiving Tc17-enriched or Tc17-depleted cells (Figure 6C, left). However, upon therapeutic transfer to mice infected for 15 days with Mav, we found higher frequencies of total IFNγ-producing CD4+ and CD8+ T cells if mice received Tc17-enriched cells (significant for CD8+IFNγ+, Figure 6C, left). Moreover, transfer of Tc17enriched cells was found superior in activating Mav-specific IL-17 responses in recipient mice, both for prophylactic and therapeutic transfer (Figure 6C, right). Frequencies of both, CD4+ Th17 as well as CD8+ Tc17 cells were significantly increased in mice that that received Tc17-enriched cells. The frequencies of IFNγ/IL-17 double-producers were also found significantly increased after therapeutic transfer of Tc17-enriched cells ( Figure 6D).
Taken together our results suggest that CD8+ Tc17 cells may have a therapeutic role in mouse Mav infection, although they do not seem to be required for protection. Further studies are needed to clarify the role of Tc17 cells in mycobacterial infections.

DISCUSSION
In the current study, we show that Msm and Msm mutants might function as a vaccine against Mav infection similar to vaccination with M. bovis BCG. The BCG vaccine is recognized as safe by the WHO, and in addition to prevent tuberculosis, the BCG vaccine has been used to treat bladder tumors for more than 30 years. Despite the success, there is a concern with toxicity of the BCG cancer treatment and a risk of disseminated BCG disease upon vaccination of immunocompromised individuals, which is contraindicated (40). The vaccine we investigated in this study is based on Msm, a non-pathogenic mycobacterium which is cleared quickly from human macrophages (41). In model systems lacking functional NK cells and T cells, Msm is well tolerated (40,42). Recently, vaccination with a Msm mutant in esx-3 was shown to protect mice against Mtb infection, demonstrating cross-reactivity and cross-protection between Msm and other mycobacteria (29). The Msm esx-3 mutant was rapidly cleared from mice after intravenous infection with high numbers of bacteria, without any adverse effects in the animals (29). Despite a handful of cases with Msm infected individuals (43), the overall picture shows that Msm based vaccines would offer improved safety profiles compared to BCG.
A study by Sweeney et al. (29) showed that esx-3 deletion in Msm leads to decreased virulence, enhanced induction of IL-12 and IFNγ, and a strong Th1 response. Reduced organ bacterial loads could result from restricted growth, e.g., from iron starvation resulting from ESX-3 deletion (28), or from improved bacterial clearance due to loss of ESX-3 secreted factors modulating inflammatory responses, which could improve antigen presentation. Our study is in favor of the latter scenario as we found significantly reduced bacterial load and reduced expression of CD86 in APCs infected with the Msm espG 3 , which could indicate reduced inflammatory properties of the espG 3 mutant strain. This reduced virulence did not impair antigen-presentation capabilities of the Msm espG 3 mutant as we detected equal or increased presentation of ovalbumin model antigens to CD4+ and CD8+ T cells with Msm espG 3 when compared to wild-type Msm. In vitro antigen presentation and CD4+ and CD8+ T cell activation was not directly addressed by the Sweeney et al. (29).
We could show a protective effect of Msm espG 3 ::mpt64 vaccination on subsequent Mav infection in mice, as organ bacterial loads were decreased compared to unvaccinated mice. The level of protection was found to be comparable to BCG vaccination, suggesting the presence of shared antigens in Msm, BCG, and Mav. Additional expression of the Mav protein MPT64 did not enhance the protective effect of the Msm vaccine on organ bacterial load. However, we only observed a differential development of Mav-specific effector T cell subsets in mice that received vaccination with the Msm strain that overexpressed the MPT64 antigen. BCG-vaccinated mice showed a Th1-and Tc1dominated immune response with the majority of Mav-specific T cells producing IFNγ. In contrast, mice vaccinated with Msm espG 3 ::mpt64 showed a bias toward IL-17-producing CD4+ Th17 and CD8+ Tc17 T cells. In addition, increased frequencies of polyfunctional CD4+ T cells producing IFNγ, IL-17 as well as TNFα were observed in Msm espG 3 ::mpt64-vaccinated mice. Overexpression of the Mav antigen MPT64 seemed to be required, as no Th17/Tc17 bias was observed in mice vaccinated with Msm WT or the Msm espG 3 mutant that did not express the MPT64 antigen. These results indicate that Th17 and Tc17 responses could be induced by expression and secretion of the MPT64 antigen, but more studies are needed to confirm if this is the case. Junqueira-Kipnis et al. (44) demonstrated that mice vaccinated with Msm esx3 overexpressing fusion protein (Ag85c, MPT51, and HspX) produced high frequencies of Th17 cells in the lungs on challenge with Mtb, accompanied with lower organ bacterial loads. In addition, Matsuyama et al. (45) showed that IL-17 production increased in CD4+ T cells upon infection with Mav, a minor population of CD8+ Tc17 cells expressing IL-17 was also detected. We also observed low levels of Tc17 cells in unvaccinated and BCG-vaccinated mice in our study. However, vaccination of mice with Msm espG 3 ::mtp64 prior to Mav infection was found to increase the frequencies of IL-17-producing Th17, Tc17 as well as IFNγ/IL-17 double-producing CD4+ and CD8+ T cells. One explanation could be that a proinflammatory milieu (IL-6,IL-1β, IL-23, and IL-12) generated by the vaccine aided in the generation of IL-17-producing T cells (46). It might also be that Msm espG 3 ::mtp64 induces apoptosis of APCs in the draining lymph nodes and thereby engages CD8+ cells. This has previously been shown for dendritic cells that take up apoptotic macrophages carrying mycobacterial antigens and subsequently trigger the proliferation of CD8+ cells via cross-presentation (47)(48)(49). Finally, it might also be that the vaccine causes a contact hypersensitivity-like reaction, which has been shown to engage CD8+ T cells with production of IFNγ and IL-17 (50,51).
Most of the knowledge about IL-17 in Mtb pathogenesis comes from investigations on CD4+ Th17 cells (15) and γ/δ+ T cells (52). Protective effects of Th17 cells have also been highlighted in a vaccine model against pulmonary TB (12)(13)(14). Khader et al. (13) demonstrated that CD4+IL-17+ T cells from vaccinated mice upregulated chemokine production, neutrophil recruitment, and promoted tissue infiltration of IFNγ-producing CD4+ T cells, leading to restriction in mycobacterial growth. In our study, IL-17-producing T cells and polyfunctional CD4+ T cells (IFNγ+/IL-17+/TNFα+) could have a similar role by promoting cell infiltration leading to reduced bacterial growth. Moreover, a recent study showed that BCG or Mtb infection can prime fully differentiated IFNγ+/TNFα+ effector CD4+ memory T cells with lower lung homing capacity than H56/CAF01 subunit vaccination, which induces central memory T cells dominated by IL-2-and IL-17-producing CD4+ T cell subsets (53). It could be that the Msm espG 3 ::mtp64 vaccine resembles the H56/CAF01 subunit vaccine in yielding memory T cells with superior homing capacity to infected tissues compared to BCG. Our finding that Msm espG 3 ::mtp64 induced more CD4+ central memory T cells than BCG supports this hypothesis, although it needs to be experimentally confirmed.
It is currently unknown if Tc17 cells have a protective role in mycobacterial infections. Interestingly, a recent study showed that infants vaccinated with BCG strain VPM1002 had high proportions of Tc17 cells (20). In Mav infection, there has only been observed small populations of Tc17 cells and no specific function has been tested (45). The role of Tc17 cells is better investigated in other diseases like cancer, fungal and viral infections (29,54,55). Tumor studies have shown that Tc17 cells, though less cytotoxic, are highly plastic and actively convert to cytotoxic IFNγ/IL-17 double producers in the presence of IL-12 (55,56). Satoh et al. (57) have suggested that in the presence of IL-12, suppression of SOCS3 in Tc17 cells permits the induction of both IL-17 and IFNγ, yielding cytotoxic Tc17 cells (57). In our study, we tested for a direct role of Tc17 cells in Mav infection. We saw no protective effect from prophylactic transfer of Tc17 cells from Msm espG 3 ::mtp64 vaccinated mice to naïve mice prior to Mav infection. We contemplate several reasons as an explanation for the failed protective effect of Tc17 cell transfer. First, Tc17 cells may need an adequate inflammatory milieu to survive and expand, which is not provided by naïve mice. Cells could thus be dead before mice are infected with Mav one day after adoptive transfer. In addition, the Tc17 cells could only be enriched to >10% of CD8+IL-17+ T cells in this study. Alternatively, we speculate that Tc17 cells, when reaching the immunological niche, may turn anergic due to low density of stimuli (58)(59)(60)(61). We did not trace the adoptively transferred Tc17-enriched and Tc17depleted T cells in our study and were thus unable to distinguish if the effector T cell responses in the recipient mice originated from the adoptively transferred cells or from immune cells of the recipient mice. The experiments are thus to be regarded as initial proof of concept studies for future follow-up, for example with mice expressing different CD45 isoforms. Tracking of Tc17-enriched donor cells could then allow analysis of survival, tissue homing and plasticity of the transferred cells and answer the question if the observed effect is due to a direct effector function of the transferred cells or if the transferred cells shape effector responses of the immune cells in the recipient mouse. In our model system, Tc17 cells were transferred into naïve mice without prior irradiation. It has been demonstrated that transfer of both BCG induced immunity and antitumor efficacy of Tc cells require depletion of immune cells through irradiation (62,63). However, Sweeney et al. (29) observed that memory CD4+ cells from Msm esx3::Mtbesx-3 vaccinated mice were able to provide protection of naïve mice without irradiation prior to transfer. On the other hand, when the Tc17 cells were injected into mice pre-infected with Mav (therapeutically), we observed increased protection with lower organ bacterial loads, as well as higher frequencies of CD8+IFNγ+, CD8+IL-17+, and CD8+IFNγ+IL-17+ T cells when compared to mice receiving Tc17 as a prophylactic measure. This suggests that Tc17 cells may require an inflammatory milieu to provide adequate protection. It could be that the Tc17-enriched cells were terminally differentiated at the time of the transfer and could not be re-activated to become effector cells upon a later challenge with Mav in the prophylactic setting. Lastly, it might be that adoptive transfer of an isolated T cell subset such as CD8+ Tc17 cells is not enough to protect the mice from infection and that other T cell subsets such as CD4+ Th17 cells or IFNγproducing Th1 and Tc1 cells are required for protection. This is supported by our observation that transfer of total CD4+ and CD8+ T cells, but not Tc17-enriched T cells alone, from Msm espG 3 ::mtp64-vaccinated mice mediated protection from Mav infection.
In this study, we have demonstrated the potential of Msm espG 3 ::mtp64 as a safe vaccine that can be further developed against mycobacterial infections. Msm espG 3 ::mtp64 vaccination elicited a higher proportion of IL-17-producing Th17 and Tc17 in comparison to BCG vaccination, which elicited a Th1/Tc1-dominated immune response. In this study, we used i.p. infection of mice as infection model, which leads to reproducible systemic infection of mice, but is of less physiological relevance than other routes of infection. Future studies are needed to substantiate the importance of Th17/Tc17 cells in protecting against Mav infection, e.g., with different and more physiological routes of Mav infection, by tracing the Tc17 cells in the recipient mice or by neutralizing IL-17 antibodies, and to establish their function and possible protective capabilities.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The animal study was reviewed and approved by the Norwegian National Animal Research Authorities.

FUNDING
This work was supported by the Research Council of Norway through its Centers of Excellence funding scheme, project number 223255/F50, and the Liaison Committee between NTNU and the Central Norway Regional Health Authority to TF, MH, and MS.