Mucosal delivery of ESX-1–expressing BCG strains provides superior immunity against tuberculosis in murine type 2 diabetes

Significance

Tuberculosis (TB) susceptibility and disease are significantly exacerbated in people with type 2 diabetes. The underlying mechanisms are incompletely understood, and it is not known if new TB vaccine candidates will be safe and provide protection in the context of diabetes. Using a long-term diet-induced murine model of type 2 diabetes, we demonstrate that increased susceptibility to TB is caused by impaired mycobacterial recognition and killing in the diabetic lung. Importantly, we show that mucosal vaccination of diabetic mice with Bacillus Calmette–Guérin (BCG) strains expressing the ESX-1 secretion system from Mycobacterium tuberculosis can overcome this defect and provide superior immunity against TB. Our data warrant a consideration of ESX-1–containing BCG strains as effective TB vaccines in older individuals and diabetics.

Abstract

Tuberculosis (TB) claims 1.5 million lives per year. This situation is largely due to the low efficacy of the only licensed TB vaccine, Bacillus Calmette–Guérin (BCG) against pulmonary TB. The metabolic disease type 2 diabetes (T2D) is a risk factor for TB and the mechanisms underlying increased TB susceptibility in T2D are not well understood. Furthermore, it is unknown if new TB vaccines will provide protection in the context of T2D. Here we used a diet-induced murine model of T2D to investigate the underlying mechanisms of TB/T2D comorbidity and to evaluate the protective capacity of two experimental TB vaccines in comparison to conventional BCG. Our data reveal a distinct immune dysfunction that is associated with diminished recognition of mycobacterial antigens in T2D. More importantly, we provide compelling evidence that mucosal delivery of recombinant BCG strains expressing the Mycobacterium tuberculosis (Mtb) ESX-1 secretion system (BCG::RD1 and BCG::RD1 ESAT-6 ∆92–95) are safe and confer superior immunity against aerosol Mtb infection in the context of T2D. Our findings suggest that the remarkable anti-TB immunity by these recombinant BCG strains is achieved via augmenting the numbers and functional capacity of antigen presenting cells in the lungs of diabetic mice.

Tuberculosis (TB) is caused by infection with Mycobacterium tuberculosis (Mtb) and is the leading infectious cause of death globally. Approximately 10 million new TB cases were reported in 2018 with a further 1.7 billion people worldwide latently infected and at risk for reactivation (1). Despite recent advances in diagnostics, treatment options, and control measures, TB still kills an estimated 1.5 million people each year (1). Reactivation of latent TB infection (LTBI) is strongly associated with comorbid immunosuppressing conditions, most notably HIV coinfection/AIDS and diabetes mellitus (DM) (2). It is now recognized that the influence of DM, particularly type 2 diabetes (T2D) on TB burden is greater than HIV coinfection, because of its higher prevalence (∼463 million people currently live with DM and the numbers are expected to escalate to 700 million by 2045) (3), with the majority of diabetics living in TB endemic countries. Based on recent metaanalyses, individuals with DM have a three- to fourfold increased risk of developing TB while ∼10% of TB patients have comorbid DM (4). Furthermore, the risk of reactivation of LTBI is significantly increased in TB/T2D comorbid patients (5). TB/T2D comorbidity is not limited to low- to middle-income countries but also exists in developed nations. As a result, TB/T2D comorbidity poses a significant challenge to the global eradication of TB.

Although the mechanisms underlying this increased susceptibility to TB are not well understood, multiple animal models of DM, including T2D, show defective innate immune recognition (6) and delayed adaptive immune priming (7) relative to standard models. However, the majority of these animal models lack many features of T2D. We have recently described a robust diet-induced animal model for T2D encompassing the cardinal features of human T2D such as obesity, glucose intolerance, chronic inflammation, hyperinsulinemia, progressive insulin resistance, and adipocyte and glomerular hypertrophy (8). Using this model, we demonstrated increased bacterial burden, lung immunopathology, and greater mortality following infections with Mycobacterium fortuitum (6) and Mycobacterium bovis Bacillus Calmette–Guérin (BCG) (9). However, the precise defects that predispose the diabetic lung to TB disease remain unknown.

While the immunological correlates of TB protection are not well defined, the role of CD4⁺ T cells in Mtb immunity is well established in animal models and TB patients. Depletion of Th1 cells results in early disease reactivation from LTBI as seen in HIV patients (10). However, accumulating evidence suggests a substantive role for CD4⁺ T cell-independent protective immunity (11). For example, we and others have recently shown that vaccine-induced CD4⁺ T cells are not necessary to prevent the reactivation of LTBI in murine (12) and nonhuman primate (NHP) models (11). Understanding which immune responses truly correlate with protection will be critical for the development of an effective TB vaccine. BCG, the only approved TB vaccine to date does not provide sufficient protection against pulmonary TB in adults (13). Current experimental TB vaccine strategies include: boosting BCG with improved and more immunogenic recombinant BCG (rBCG) strains; live attenuated Mtb vaccines; and subunit vaccines that are safe to use in immunocompromised individuals (14). There is also renewed interest in intravenous (i.v.) (15) and mucosal delivery of TB vaccines, including BCG, primarily due to the increased protection afforded by pulmonary resident memory T cells (T_RM) (16). rBCG strains engineered to incorporate immunodominant Mtb regions, such as the virulence-associated ESX-1 locus, cytokines, toxin-derived antigens, and genes important for antigen presentation enhance and broaden the vaccine-induced immune response (17). Furthermore, strategies that allow the vaccine strain to reach the cytosol via the incorporation of phagosome perforating molecules, such as the ESX-1 system (18) or listeriolysin (Hly) from Listeria monocytogenes (19) afford superior protection against an Mtb challenge. A number of rBCG strains including BCG ΔureC::hly (VPM1002), which secretes membrane-perforating Hly, are currently undergoing clinical trials (17). However, whether these novel live-recombinant vaccine candidates will be safe and efficacious in the context of T2D is not known.

The goal of global eradication of TB by 2050 requires an effective, safe vaccine that works in all individuals vulnerable to developing TB disease, including those with diabetes. Using our diet-induced murine model of T2D, here we investigated the mechanistic basis of TB/T2D comorbidity and evaluated the safety and efficacy of experimental rBCG strains. We analyzed the lung microbiome profiles as well as the gene expression patterns in control (CON) and T2D mice to investigate the impact of T2D on lung immunity that could potentially render diabetic animals more vulnerable to Mtb infection. We compared the conventional BCG Pasteur strain with rBCG strains that secrete immunodominant ESX-1 antigens derived from the Mtb “region of difference 1” (RD1). Our findings demonstrate that T2D is associated with altered functional capacity of lung immune cells. Most significantly, we found that mucosal administration of ESX-1–containing rBCG strains not only offers superior safety in the context of T2D but also confers outstanding protection against Mtb aerosol challenge. Immunological analyses indicate that rBCG-mediated anti-TB immunity was achieved via augmenting the antimycobacterial function of lung innate antigen presenting cells (APCs).

Results

Murine T2D Mimics Increased Human Susceptibility to Aerosol Mtb Infection.

To induce murine T2D, male C57BL/6 mice were subjected to extended feeding of an energy dense diet (EDD; Fig. 1A) as shown previously (8). Prior to diet intervention, the body mass of mice was comparable between standard rodent diet (SD)-fed age-matched control and EDD-fed T2D groups (SD, 17.94 ± 0.1824 g vs. EDD, 18.01 ± 0.3365 g; mean ± SEM, P = 0.8620). After 30 wk of diet intervention, mice fed with SD or EDD demonstrated an overall body weight increase of 83.55% and 152.24%, respectively. Body weight was significantly greater in the EDD group compared to the SD group (SD, 32.93 ± 0.5234 g vs. EDD, 45.43 ± 0.5604 g, P < 0.0001; Fig. 1B). In addition, mice fed the EDD had significantly higher fasting blood glucose (SD, 6.476 ± 0.3 mmol/L vs. EDD, 10.11 ± 0.2981 mmol/L, P < 0.0001; Fig. 1C) and impaired glucose tolerance as reflected by higher area under the curve (AUC) values (SD, 1,398 ± 64.13 vs. EDD, 2,203 ± 62.26, P < 0.0001; Fig. 1D), key metabolic features associated with the development of T2D.

Fig. 1.

Murine T2D mimics increased human susceptibility to aerosol Mtb infection. (A) Four- to 6-wk-old C57BL/6 male mice were fed with EDD and SD (control mice) for 30 wk to induce murine T2D. After the dietary intervention, mice were assessed for (B) body weight, (C) fasting blood glucose levels, and (D) glucose tolerance. (A) T2D confirmed naiÏve and control C57BL/6 mice were infected with very-low dose of aerosol Mtb H37Rv (10 to 20 CFUs). (E) At 1 d p.i., 5 mice from each group were killed to confirm the initial infectious dose. At 45 d p.i., infected lungs and spleens were assessed for (F) viable Mtb and (G) percent of lung affected. Results are presented as B–G individual data points, (D) pooled data means, and (G) representative images (magnification, 25×) from [B] 50, [C and D] 25, and [E–G] 5 mice per group. (Scale bar, 500 μm.) *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns: not significant by unpaired two-tailed Student’s t test. Data are means ± SEM.

To determine if the EDD-fed T2D mice are more prone to Mtb infection, we exposed T2D and control mice to a very-low dose (10 to 20 colony-forming units [CFUs]) of Mtb H37Rv. There was no difference in lung Mtb burden between control and T2D mice at 1 d after Mtb challenge (Fig. 1E), indicating that bacterial inhalation is not affected by T2D and all mice received a comparable dose. At 45 d after Mtb challenge, however, T2D mice displayed significantly higher lung and spleen CFU loads (Fig. 1F) accompanied by increased pathological damage to the lung tissue (Fig. 1G and SI Appendix, Fig. S1). Collectively, these data demonstrate that the diet-induced murine model of T2D mimics the cardinal features of human T2D and the increased susceptibility to aerosol Mtb infection, further confirming the appropriateness of this model to study TB/T2D comorbidity.

Compositional Changes in the Lung Microbiota following Mtb Infection.

The resident microbiota has a pivotal role in the onset of T2D and its plethora of complications while perturbed microbiome compositions and altered immune responses have both been implicated in many metabolic diseases (20). Although the gastrointestinal tract remains the most thoroughly investigated organ–microbiome interaction, we hypothesized that the lung microbiome may also play a significant role in shaping the host immune responses in the diabetic lung as the portal of entry of Mtb. To assess whether an altered microbiota in T2D contributes to increased susceptibility to TB disease, we compared the lung microbiome profiles of naïve and infected control and T2D mice (Fig. 2 A and B and SI Appendix, Fig. S2). Actinobacteria, Firmicutes, and Proteobacteria were the most abundant phyla in naïve-control and T2D lungs (SI Appendix, Fig. S2A). Mtb infection, however, led to a phylum-wide shift associated with a significant reduction of Actinobacteria in both groups and an increase in Bacteroidetes and Firmicutes (SI Appendix, Fig. S2A). The Shannon diversity index revealed an increase in overall lung community diversity in both control and T2D mice following the infection (SI Appendix, Fig. S2B). The significant expansion of families Muribaculaceae and Lachnospiraceae and near complete absence of Propionibacteriaceae following infection was prominent (Fig. 2A). Importantly, the increased TB burden in diabetic mice was detectable, with Mycobacteriaceae levels being significantly increased in T2D mice (CON, 7.313 ± 2.726% vs. T2D, 34.17 ± 6.851%, P = 0.0219; Fig. 2B). In addition to Mycobacteriaceae, pairwise comparison at family level uncovered six other differentially expressed families (DEFs) that were mutually increased in both TB and TB/T2D comorbid lungs compared to naïve control: Muribaculaceae, Lactobacillaceae, Lachnospiraceae, Ruminococcscrse, Erysipelotrichaceae, and Akkermansiaceae (SI Appendix, Fig. S2C).

Fig. 2.

Changes in lung microbiota and immune-related gene expression following aerosol Mtb infection. Relative abundance of (A) total bacterial families and (B) family Mycobacteriaceae in lung tissues from control and T2D mice prior to and 45 d post-Mtb infection. (C) PCA plot and a heatmap demonstrating unsupervised gene clustering of different clinical phenotypes. (D) The Venn diagram shows the DEGs in mice with TB and/or T2D compared to naiÏve control mice (adjusted P < 0.05). Results are presented as (A) relative proportions and (B) pooled data means from three to four mice per group. *P < 0.05 by unpaired two-tailed Student’s t test (B). Data are means ± SEM.

Collectively, these findings suggest that the lung microbiota in T2D mice may not directly facilitate the increased susceptibility to aerosol Mtb infection.

Differential Immune Gene Expression Patterns following Mtb Infection.

To examine possible inherent immune defects that might be associated with TB/T2D comorbidity, we quantified expression levels of 548 immune-related transcripts in the lungs of control and T2D mice pre- and post-Mtb infection using NanoString nCounter. Principal component analysis (PCA) of the gene expression profiles separated samples into distinct clusters by both infection and disease status (Fig. 2C). Unsupervised hierarchical clustering further confirmed distinct expression patterns among each clinical phenotype (Fig. 2C). The highest variation in gene expression was seen between the naïve control and TB/T2D mice. Although, Mtb infection significantly altered gene expression patterns, there was less separation between TB (Mtb-infected control mice) and TB/T2D subgroups. A total of 389 differentially expressed genes (DEGs; adjusted P < 0.05) were identified in the TB/T2D, TB, and T2D compared to naïve control mice (Fig. 2D and SI Appendix, Table S1). A total of 335 DEGs were identified in TB/T2D (vs. control). Among those, 312 genes were mutually differentially regulated in TB/T2D and TB mice. These include genes previously found in blood and lung transcriptome profiles from TB patients and Mtb-infected mice, respectively (21, 22). In addition, 124 DEGs were differentially expressed between T2D and naïve control mice. Among those, 87 genes were shared between all three clinical phenotypes while 19 and 34 genes were uniquely differentially expressed only in T2D and TB/T2D, respectively. Interestingly, most of these DEGs are important for innate immune recognition and were down-regulated in T2D and TB/T2D mice (i.e., Clec4a, Clec5a, Nod2, and CD14). In addition, a direct comparison of transcriptomes derived from the TB/T2D and the TB subgroups revealed 55 DEGs (adjusted P < 0.1, SI Appendix, Fig. S3). Collectively, the gene expression profiles identified here suggest that Mtb infection drives the majority of transcriptomic changes observed in infected mice, while T2D appears to cause the down-regulation of genes that are crucial for Mtb recognition, control, and for bridging innate and adaptive immune responses.

ESX-1–Containing BCG Strains Are Safer than Parental BCG in the Context of T2D.

The apparent defect in mycobacterial recognition in the lung in T2D prompted us to investigate the hypothesis that mucosal delivery of more immunogenic rBCG strains may provide a stronger immune response and, hence, better protection against TB in T2D. To this end we made use of experimental rBCG strains that have been shown to induce a more robust immune response (18). We vaccinated both nondiabetic control and T2D mice with three different strains of BCG: parental BCG Pasteur (hereafter referred to as BCG), BCG::RD1, or BCG::RD1 ESAT-6 Δ92–95, which both contain the extended RD1 locus from Mtb and secrete a full-length or a C-terminal truncated ESAT-6 protein, respectively. Each mouse received 5 × 10⁵ CFUs of the vaccine strain directly into the lung via the intratracheal (i.t.) route (Fig. 3A). The superiority of mucosal BCG vaccination in mouse and NHP models (16, 23) was reconfirmed in a pilot study (SI Appendix, Fig. S4). Throughout the 60-d vaccination period, animal wellbeing and body weight were monitored. At 30 d postvaccination (p.v.) all vaccinated animals except BCG::RD1 ESAT-6 Δ92–95 groups (both control and T2D) showed significant weight loss compared to unvaccinated animals. Interestingly, the most pronounced weight loss was seen in T2D mice that received conventional BCG (Fig. 3B and SI Appendix, Table S2). During the next 30 d, all groups started to regain lost weight. However, even at 60 d p.v. BCG-vaccinated T2D mice demonstrated significantly less body weight regain compared to all other groups (Fig. 3B).

Fig. 3.

ESX-1-containing BCG strains are safer than parental BCG in the context of T2D. (A) Control and T2D C57BL/6 mice were vaccinated with 5 × 10⁵ CFUs of BCG Pasteur, BCG::RD1, and BCG::RD1 ESAT6 ∆92–95 via the i.t. route. (B) Body weight of the mice was measured during the p.v. period of 60 d. At 60 d p.v., mice were killed and assessed for (C) clearance of vaccine strains in both lung and spleen and (D) immune cell infiltration in lung. Results are presented as (B) pooled data means, (C and D) individual data points, and (D) representative images (magnification, 25×) from 8 to 10 mice per group from two pooled independent experiments. (Scale bar, 500 μm.) *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns: not significant by one-way ANOVA followed by Tukey’s (B), Dunnett’s multiple comparison test (C and D), and unpaired two-tailed Student’s t test (C and D). Data are means ± SEM.

At 60 d p.v., mice from each vaccine group were killed to assess bacterial clearance, lung tissue inflammation, cellular immune response, and systemic cytokine/chemokine profiles. The highest recovery of viable bacteria in lung and spleen tissues was observed from BCG-vaccinated T2D mice, suggesting that T2D may have impaired the clearance of BCG from the lung, which led to a widespread dissemination of bacilli to extrapulmonary organs (Fig. 3C). In contrast, control and T2D mice receiving either of the rBCG strains had significantly fewer bacteria in their lungs and spleen (Fig. 3C). As expected, tissue inflammation was absent in nondiabetic unvaccinated lungs (Fig. 3D). However, hematoxylin and eosin (H&E) staining revealed that unvaccinated T2D lungs were significantly more inflamed compared to unvaccinated control lungs, confirming the chronic inflammatory nature of T2D (CON, 0.6270 ± 0.1949% vs. T2D, 5.113 ± 1.038%, P = 0.0004). There was no histological evidence of any organized lymphoid structures resembling inducible bronchus-associated lymphoid tissue (iBALT) in unvaccinated T2D lungs when compared to vaccinated groups. The highest degree of lung tissue infiltration was observed in mice vaccinated with BCG and BCG::RD1 strains. Overall, the mildest histopathological changes were observed following mucosal vaccination with BCG::RD1 ESAT-6 Δ92–95 (Fig. 3D). Hence, BCG::RD1 ESAT-6 Δ92–95 had less impact on overall body weight (Fig. 3B) and was more efficiently cleared from lung and spleen (Fig. 3C) in both control and T2D mice.

Overall, serum cytokine and chemokine levels at day 60 p.v. were comparable among all vaccine groups in both control and T2D mice (SI Appendix, Fig. S5A). Increased fold changes in IL-10, IL-12p40, and KC (CXCL1) were seen across all vaccinated groups in both control and T2D mice compared to unvaccinated mice (SI Appendix, Fig. S5B). The most noticeable fold decrease among all three vaccinated T2D groups was observed in MIP-1α (CCL3), a cytokine with a known function of recruiting inflammatory cells (24). Collectively, the absence of weight loss, the more efficient bacterial clearance, and the reduced lung pathology indicate that genetically modified rBCG strains carrying ESX-1 are safer than conventional BCG in age-matched control and T2D mice.

ESX-1–Containing BCG Strains Confer Superior Protection against Aerosol Mtb Challenge in Both Control and T2D Mice.

In the vaccination studies described above, the ESX-1–containing BCG strains persisted for a shorter time and showed less lung pathology than the parental BCG. However, to assess if this enhanced safety also translated into superior protection, we infected mice with a very low aerosol dose of Mtb H37Rv (10 to 20 CFUs) 60 d after vaccination (Fig. 4A). At 45 d after challenge, unvaccinated T2D mice had the highest number of detectable viable Mtb in lung and spleen tissues (Fig. 4B). Parental BCG reduced lung bacterial loads in both nondiabetic and T2D mice by 5- to 10-fold. This reduction has previously been shown in multiple studies in mice (16) and NHPs (23) following mucosal BCG vaccination. Intriguingly, when compared to unvaccinated and BCG-vaccinated groups, both rBCG strains significantly reduced lung Mtb burden not only in control, but also in T2D mice (Fig. 4B). More strikingly, systemic spread of bacteria to the spleen was almost completely reversed by vaccination with ESX-1–containing BCG strains, with most mice showing sterile immunity in splenic tissues in both control and T2D mice. BCG-vaccinated mice presented with the greatest histopathological changes in both control and T2D lungs (Fig. 4C). Consistent with bacterial loads, increased pathology was observed in unvaccinated and BCG-vaccinated groups while rBCG groups showed only minor additional lung pathology postinfection (p.i.) (Fig. 4C). The observed immunopathology among BCG-vaccinated animals, particularly in T2D, may be attributed to the unresolved inflammatory lung microenvironment caused by mucosal BCG administration prior to aerosol Mtb infection (Fig. 3D). Although systemic cytokine and chemokine levels were largely comparable between vaccine groups (SI Appendix, Fig. S5C), the majority of analytes were increased in T2D animals, among which IL-6, MIP-1β (CCL4), and TNF-α levels were significantly higher in rBCG-vaccinated T2D mice compared to the BCG group (SI Appendix, Fig. S5D). However, the reverse was observed in control mice (SI Appendix, Fig. S5D). Together these findings further substantiate the superiority of rBCG strains over conventional BCG in the context of T2D with superior protection and lack of systemic dissemination following aerosol Mtb infection.

Fig. 4.

ESX-1–containing BCG strains confer superior protection against aerosol Mtb infection in both control and T2D mice. (A) At 45 d postaerosol Mtb infection (10 to 20 CFUs), vaccinated and unvaccinated mice from both control and T2D groups were killed and assessed for (B) viable bacteria in lung and spleen and (C) lung immunopathology. Results are presented as (B and C) individual data points and (C) representative images (magnification, 25×) from 8 to 10 mice per group from two pooled independent experiments. (Scale bar, 500 μm.) *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns: not significant by one-way ANOVA followed by Dunnett’s multiple comparison test (B and C) and unpaired two-tailed Student’s t test (B and C). Data are means ± SEM.

Superiority of ESX-1–Containing BCG Strains Is Associated with Increased Age.

Given that near-sterile immunity against Mtb infection has not been reported previously with parenterally delivered (vaccination routes other than oral and/or mucosal) BCG::RD1 or BCG::RD1 ESAT-6 Δ92–95 (18, 25), we reasoned that our results could be related to the mucosal delivery of vaccines, the applied infectious dose, and/or the advanced age of the mice used in the challenge studies. Both control and T2D mice had been on a dietary intervention before being vaccinated, meaning that they were ∼11 mo old at the time of Mtb challenge (Fig. 1A). To dissect the possible contribution of age, we vaccinated young (6 wk) and old nondiabetic (36 wk) mice with BCG, BCG::RD1, or BCG::RD1 ESAT-6 Δ92–95 intratracheally and challenged them with a very-low dose of Mtb 60 d later (Fig. 5A). At 45 d after Mtb infection, all vaccinated mice showed significantly reduced organ bacterial loads in lung tissues compared to unvaccinated mice (Fig. 5B). However, aged mice that received BCG::RD1 showed a ∼2-log reduction in lung CFU burden compared to unvaccinated mice, whereas young mice displayed only a ∼1.3-log reduction (Fig. 5B). Similarly, compared to BCG-vaccinated mice, BCG::RD1 in aged mice had reduced lung CFU levels compared to young mice (∼1.5- vs. ∼0.5-log reduction, respectively). Age-related differences in lung pathology following infection were also noted with more inflammation seen in young compared to aged mice (Fig. 5C). In this set of experiments, though we detected comparable CFU levels in both young and aged BCG::RD1 ESAT-6 Δ92–95-vaccinated animals, lung pathology was significantly reduced in aged mice compared to young mice (Fig. 5C). These results suggest that the superiority of ESX-1–containing BCG against TB is enhanced by advanced age and mucosal delivery of the vaccine.

Fig. 5.

Superiority of ESX-1–containing BCG strains is associated with advanced age. (A) Young (6 wk) and old (36+ weeks) C57BL/6 mice were vaccinated with 5 × 10⁵ CFUs of BCG Pasteur, BCG::RD1, and BCG::RD1 ESAT-6 ∆92–95 via i.t. route. At 60 d p.v., mice were infected with a very-low dose (10 to 20 CFUs) of Mtb H37Rv. At 45 d p.i., mice were killed and assessed for (B) viable bacteria in lung and (C) lung pathology. Results are presented as (B and C) individual data points and (C) representative images (magnification, 25×) from 6 to 7 mice per group. (Scale bar, 500 μm.) *P < 0.05; **P < 0.01; ns: not significant by unpaired two-tailed Student’s t test (B and C). Data are means ± SEM.

ESX-1–Containing BCG Strains Alter the Immune Cell Composition and Augment Antimycobacterial Function of Innate APCs in the Lung Microenvironment.

To investigate if the superiority of ESX-1–containing BCG strains was due to differential expansion or recruitment of immune cells in the respiratory tract, we performed comprehensive flow cytometry-based cellular phenotyping in both lung parenchyma (Fig. 6A) and airways (SI Appendix, Fig. S6A) 60 d after i.t. vaccination. We analyzed 29 lymphoid and 8 myeloid cell subsets. In addition to total CD4⁺ and CD8⁺ T cells, we quantified central memory (T_CM), effector memory (T_EM) and T_RM cells from each subset as they appear to play a key role in vaccine-induced immunity (16, 26). Cells were also stained with CD4 and CD8 MHC-peptide tetramers derived from the immunodominant Mtb antigens, ESAT-6 and TB10.4, respectively, to measure mycobacteria-specific T cell responses. We also enumerated innate and innate-like cells, including polymorph-nuclear neutrophils (PMNs), dendritic cells (DCs), alveolar macrophages (AMs), interstitial macrophages (IMs) and mucosal-associated invariant T (MAIT) cells. We measured the fold change of each cell subset compared to the BCG-vaccinated group in both control and T2D mice (Fig. 6A and SI Appendix, Fig. S6A; raw data are shown in SI Appendix, Figs. S6–S9).

Fig. 6.

ESX-1–containing BCG strains alter the immune cell composition and augment antimycobacterial function of innate APCs in the lung microenvironment. (A) Lung resident immune cell subsets at 60 d p.v. (B) Schematic representation of in vivo and in vitro APC functional assays performed. Costimulatory marker expression of mLN CD11c⁺ DCs from (C) unvaccinated and (D) vaccinated mice 24 h after in vivo LPS instillation. CD11b⁺ F4/80⁺ macrophages from (E) unvaccinated and (F) vaccinated mice were also assessed for costimulatory marker expression. (G) Total iNOS⁺ lung cells 60 d p.v. and (H) nitrite production following 24-h in vitro Mtb infection (MOI 1). Results are presented as a (A) heatmap indicating log₂ fold changes relative to BCG Pasteur group and (C–H) pooled data means from (A, C, and E) 8 to 10 and (D, F, and H) 4 to 5 mice per group. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns: not significant by unpaired two-tailed Student’s t test (C and E), two-way ANOVA (D and F), and one-way ANOVA followed by Dunnett’s multiple comparison test (G and H). Abbreviations: ESAT6 Δ92–95; BCG::RD1 ESAT6 Δ92–95, DN; double negative, T; T cells. Data are means ± SEM. See SI Appendix, Figs. S6–S10 for FACS gating strategies, raw data, and representative FACS plots.

Apart from the expected induction of ESAT-6-specific CD4⁺ T cells, analysis of T cell subsets did not reveal consistent patterns across both rBCG vaccines and both groups of mice that might provide an explanation for the apparent protective efficacy and safety of these recombinant strains described above. B cells and MAIT cells showed a trend toward up-regulation in all mice vaccinated with ESX-1-containing BCG strains. Strikingly, however, vaccination with these strains resulted in significantly increased numbers of DCs and macrophages (Fig. 6A and SI Appendix, Fig. S9 E and F) in lungs from control and T2D mice. These results suggested that the superiority of mucosally administered ESX-1–proficient rBCG strains might be a consequence of altered innate APC function rather than T cell immunity. To further assess how rBCG vaccination may enhance the functionality of lung APCs, multiple in vivo and ex vivo functional/immunophenotyping assays were conducted (Fig. 6B).

First, we tested if rBCG strains may overcome the defect in up-regulation of costimulatory molecules in diabetes via preferential manipulation of costimulatory molecules. To this end, we used a previously reported lipopolysaccharide (LPS) instillation model (27). Twenty-four hours following i.t. LPS instillation, unvaccinated T2D mice displayed a significant down-regulation of antigen recognition (i.e., CD209) and costimulatory molecules (i.e., CD40, CD80, and MHCII) on DCs and macrophages from mediastinal lymph nodes (mLNs) (Fig. 6 C and E) and lung parenchyma (SI Appendix, Fig. S10D) compared to unvaccinated control mice. However, vaccination with rBCG strains did not overcome this reduced expression compared to vaccination with BCG neither in T2D nor in control mice (Fig. 6 D and F and SI Appendix, Fig. S10E). Similar results were obtained with in vitro infected bone marrow-derived dendritic cells (BMDCs) (SI Appendix, Fig. S10 F and G).

Next, we investigated if the superior protection delivered by rBCG strains was due to increased T cell proliferation as a result of enhanced APC–T cell interaction. Using vaccine-stimulated APCs derived from control and T2D mice, we performed in vitro T cell proliferation assays with the well-established model system of ovalbumin specific CD4⁺ (OT-II) and CD8⁺ (OT-I) T cells. No significant differences in OVA-specific T cell proliferation between groups were observed (SI Appendix, Fig. S10H). In addition, the number of ESAT-6 and TB10.4-specific CD4⁺ and CD8⁺ T cells in the lymphatics prior to Mtb infection were not significantly different between nondiabetic and T2D mice (SI Appendix, Fig. S10I), suggesting that enhanced anti-TB immunity is likely also independent of APC-mediated T cell activation and proliferation.

Finally, we assessed if the ESX-1–containing BCG strains may impact on the ability of innate APCs to produce antimycobacterial compounds, such as reactive nitrogen species (RNS). Indeed, mice vaccinated with rBCG strains harbored significantly more inducible nitric oxide synthase⁺ (iNOS⁺) cells including AMs and IMs (Fig. 6 A and G and SI Appendix, Fig. S9G), suggesting an improved ability to generate oxygen and nitrogen species. The increased capacity of rBCG-vaccinated mice to generate nitric oxide (NO) was mirrored by increased nitrite concentrations in culture supernatants after infecting primary lung cells with Mtb for 24 h (Fig. 6H). In addition, in vitro rBCG infection also led to the generation of higher nitrite levels by bone marrow-derived macrophages (BMDMs) from control and T2D mice (SI Appendix, Fig. S10J).

Collectively, these findings indicate that mucosal vaccination with ESX-1–containing BCG strains not only generates more myeloid-derived immune cells in aged nondiabetic and T2D mice, but more importantly these cells appear to be endowed with an increased ability to mount antimycobacterial effector functions.

Discussion

It is now widely accepted that the intersection between TB and DM poses a significant risk to the global TB eradication strategy (1). Rationally designed new TB vaccine candidates are urgently required to meet the challenges of global TB eradication. Although there has been significant progress over the last two decades to improve BCG by introducing Mtb-specific antigens and/or deletion of various genetic elements (14), to our knowledge BCG or BCG-derived strains have not yet been evaluated as vaccines for TB in the context of T2D. Using our robust long-term diet-induced murine model of T2D, which embodies significant alimentary and polygenic factors involved in the development of clinical T2D, here we demonstrate increased susceptibility to TB in T2D following a very-low dose aerosol Mtb challenge that is thought to mimic natural TB infection in humans. Moreover, we provide evidence that mucosal delivery of rBCG strains that express RD1-encoded immunodominant T cell antigens (18) confer superior protection against pulmonary TB in T2D while maintaining heightened safety levels over conventional BCG. Our data point toward an important role of mucosal rBCG vaccination in inducing Mtb-specific lung- and airway-resident T cells and most importantly augmenting the antimycobacterial capacity of innate APCs in the lung.

Despite our expectation that parental BCG would have a better safety profile characterized by less bacterial persistence and dissemination, reduced tissue inflammation and immunopathology, compared to the ESX-1–containing rBCG strains (28), i.t. BCG vaccination resulted in significantly higher organ bacterial loads and more rapid body weight loss in T2D mice. We have previously demonstrated that 14 d after i.v. BCG infection, T2D mice had a higher bacterial burden in lungs, spleen, and liver compared to control animals (9). This was associated with impaired phagocytosis followed by defective BCG killing by alveolar macrophages (6, 9). The quantitative increase in T cell response and increased lung inflammation in BCG-vaccinated T2D mice may be a consequence of persistent organism/antigen due to inadequate clearance. Unlike conventional BCG, bacterial persistence, and vaccine-induced total lymphocyte counts in T2D mice that received the rBCG strains were mostly comparable to the control group, suggesting that these experimental vaccines act with similar efficacy in both conditions.

Live attenuated vaccines need to be adequately invasive and persistent to induce an effective expression of immunogenic antigens and to prime the immune system for subsequent exposures (29). Using an NHP model of TB, Cadena et al. have recently reported that local immune responses induced by primary Mtb infection were associated with a remarkable protection against reinfection with Mtb (30). This raises the important question if concomitant immunity should be the ultimate goal of live attenuated TB vaccines. Most importantly, these results further underscore that adding secreted Mtb antigens into BCG and mimicking the natural infection under attenuated conditions may be the best way to achieve vaccine-induced sterile immunity against TB infection.

RD1 is required for the full virulence of Mtb and encodes components of ESX-1, a type 7 secretion system, and the absence of RD1 is vital for the attenuated virulence of the vaccine strain BCG (28). However, the RD1 region is not the only virulence factor that differentiates Mtb from BCG (31). Parenterally delivered BCG provides some protection against TB in animal models (32). Although BCG confers greater protection against severe forms of miliary and meningeal TB in infants and children, incomplete protection against pulmonary TB in adolescents and adults has been one of its major failures (13). In contrast, BCG::RD1 has shown improved vaccine efficacy against aerosol TB infection in murine and guinea pig models and the increased protection was attributed to ESAT-6 and CFP10 being crucial for mounting a maximal T cell response (18). Despite the superior protection, BCG::RD1 and Mycobacterium microti::RD1 knockins were associated with increased virulence in immunocompromised animals and longer persistence in immunocompetent hosts following a very-high dose of i.v. rBCG (10⁶ to 10⁷ CFUs) immunization (28), and hence were deemed unsuitable as human vaccines. Since then, numerous studies have provided compelling evidence that mimicking the natural route of infection is crucial for vaccine-induced anti-TB immunity. Intranasal or i.t. delivery of BCG provides increased protection against TB compared to traditional subcutaneous (s.c.) administration, and the enhanced protection is largely correlated with increased number of lung and airway T_RM cells (16, 26).

Our current findings demonstrate that i.t. administration of ESX-1–containing rBCG induce equivalent protection against a challenge with virulent Mtb in both nondiabetic and T2D mice and the number of bacteria recovered was significantly lower than from BCG-vaccinated mice. The superior protection was associated with reduced lung pathology with comparable serum cytokine/chemokine levels across vaccine groups. Increased lymphocytes particularly Mtb-specific activated T cells and higher frequencies of lung and airway T_RM cells may induce rapid recall response upon exposure to Mtb-containing aerosols, thus restricting bacterial replication and dissemination. Here in our long-term infection model, it is likely that the virulence-mediated adverse effects of ESX-1 are outmatched by the immune-stimulatory effect in mucosal rBCG-vaccinated mice.

ESX-1–containing BCG vaccination induced a substantial influx of CD11c⁺ DCs and CD11b⁺ macrophages not only to nondiabetic control but also to T2D lungs. APCs are crucial for the activation and expansion of T cells. BCG::RD1 strongly enhances the ability of DCs to produce proinflammatory IL-1β, TNF-α, and more importantly the capacity to expand IFNγ⁺ T cells (33). However, we couldn’t detect any increase in DC costimulatory marker expression, induction of Mtb-specific T cells prior to Mtb infection, or increased T cell proliferation in vitro following vaccination/stimulation with rBCG strains. In fact, a substantial reduction in almost every costimulatory surface marker compared to vaccination with BCG Pasteur was observed in primary lung and BMDCs following in vivo vaccination or in vitro infection, respectively, mirroring the active manipulation of APCs observed in natural Mtb infection (34). Although we did not assess global cytokine or gene expression levels in innate cells, iNOS expression and antimicrobial NO production were elevated in lung-resident AMs and IMs following rBCG vaccination. Conformably, rBCG’s increased capacity to generate NO was also evident at the bone marrow level. This suggests that rBCG vaccination may have imprinted unique epigenetic changes into these innate APCs, which alters the rapidity and magnitude of the immune response to subsequent Mtb challenge. It has recently been shown that hematopoietic stem cells (HSCs) derived from BCG-vaccinated mice give rise to pulmonary and circulating monocytes and macrophages imprinted with Mtb-specific memory-like phenotype (35). Although these results were obtained from a very-high dose i.v. vaccination, and our study utilized a very-low dose mucosal vaccination, the impact on APCs in both situations could point toward an important role for trained immunity and need further investigation. Collectively, this emphasizes the pivotal role of RD1 in activation, recruitment, and memory formation of both innate and Mtb-specific adaptive immune cells during TB infection in T2D mice.

Our results suggest that superior safety and efficacy seen for ESX-1–containing BCG against TB throughout this study, appears to be a consequence of advanced-age and mucosal delivery of the vaccine. The majority of mice currently being used in TB research are only 6 to 12 wk old, an age when animals are still undergoing physiological and developmental changes including in the immune system (36). Although a definitive mechanism cannot be proposed, the increased safety and efficacy of ESX-1 strains in aged mice could be attributed to a number of age-related immune mechanisms, including increase in age-associated B cells (37), increased exposure to innate immune training (38), altered cytokine/chemokine milieu (39), and changes in microbiota (40). Interestingly, our vaccination strategy in aged mice likely represents booster or revaccination in human adults rather than traditional neonatal BCG vaccination. This approach provides evidence of potential translational implications/applicability such as improving hyperglycemia in type 1 diabetes patients (possibly in T2D as well) (41) and enhancing trained immunity to exert nonspecific protection against TB and/or TB-unrelated infections in adults (40). However, given that the development of T2D in our model takes 30+ weeks, our current study does not provide any experimental evidence for the robustness of BCG-induced immunity when administered neonatally prior to the induction of T2D. Furthermore, our study did not evaluate the efficacy of vaccination beyond 60 d in this model. These translational shortcomings and the impact of the age for TB vaccine efficacy should be addressed in future studies.

Although we detected very little differences in the lung microbiota, previous animal models (42) and T2D patients (43), showed a markedly altered gut microbiota. Interestingly, gut microbiota dysbiosis is also associated with increased early susceptibility to Mtb infection and bacterial dissemination in mice (44). How an altered intestinal microbiota in T2D influences anti-TB immunity is still unclear, but the concept of cross-talk between gut and lung microenvironments via the “gut–lung axis” is exciting and could be investigated in future studies (45). Comparative immune gene expression analysis revealed that 16 out of 19 uniquely DEGs in T2D were to be down-regulated and included genes associated with cell recruitment, pathogen recognition, cytokine production, and APC function. For an instance, Clec4e (Mincle) and Clec5a are important for the innate control and recognize carbohydrate-based mycobacterial cell wall components tetrahalose 6,6 dimycolate (TDM) and lipoarabinomannan (LAM) (46). In addition to reduced lung CD14 transcripts, diminished MARCO expression in mLN macrophages and DCs was also detected (SI Appendix, Fig. S11), essential coreceptors for the TDM-induced immune response which have been associated with impaired innate recognition of Mtb in T2D mice (47). Furthermore, CCL12, intracellular NOD2, and lysosomal cathepcin C, all of which have been implicated in immunity to Mtb (22, 48, 49), were also found to be significantly down-regulated in T2D mice. Major transcriptomic changes were observed following Mtb infection. Among the 34 DEGs, 18 genes, the majority being innate immune response related, CXCL15, CCR1, IL18R, and CD209 (DC-SIGN) (25, 50, 51) were down-regulated in TB/T2D, suggesting that T2D-associated innate defects are further exaggerated following Mtb infection. Similarly, recent integrative analysis of gene expression on South Indian TB/T2D comorbid patients revealed a unique gene signature also reflecting defective antigen processing/presentation pathways (52). Although Mtb strains in South India are often of lineage 1, these similarities further strengthen the resemblance between comorbid human TB/T2D patients and our TB/T2D mice. Furthermore, the significant down-regulation of these genes indicates that increased susceptibility and disease exacerbation seen in T2D are likely associated with impaired early recognition and control of Mtb infection by innate APC subsets as evident by our functional APC assays.

In summary, our study provides a useful mouse model to investigate human TB/T2D comorbidity and to test TB vaccine efficacy against aerosol Mtb infection. Our findings accentuate that impaired innate recognition of Mtb in T2D within the lung exacerbates mycobacterial replication and TB disease. Moreover, mucosal vaccination with rBCG strains that express RD1-encoded Mtb antigens combines low virulence and improved protection against TB in aged nondiabetic and T2D mice. Together with our recent study showing superiority of ESX-1–containing strains in a systematic approach (53), these findings warrant a reconsideration of “virulent” BCG strains as effective vaccines in certain populations and a renewed focus on RD1-like BCG strains for TB vaccine development (18, 54). In addition, our study raises the important future questions of 1) understanding the impact of age on vaccine-induced immunity to TB; and 2) how new TB vaccine candidates interact with and “train” innate APC subsets to augment anti-TB immune responses in the lung.

Materials and Methods

Murine Model of T2D.

Male C57BL/6 mice were bred and maintained in specific-pathogen free (SPF) animal facilities within the Australian Institute of Tropical Health and Medicine (AITHM), James Cook University, Australia. At 4 to 6 wk of age, mice were randomly divided into two dietary groups. One group of mice received ad libitum access to an EDD (SF03-030, Specialty Feeds, Western Australia, high glycemic index semi-pure rodent diet containing 23% fat, 19.4% protein, 50.5% dextrose, and 9.32% fiber), while the control group received an isometric quantity of SD (SF08-020, Standard AIN93M rodent diet (55) containing 4% fat, 13.8% protein, 64.8% carbohydrate, and 9.4% fiber) for 30 wk. This murine T2D model was developed and extensively characterized using male C57BL/6 mice (8). Moreover, estrogen protects female mice from developing high-fat diet-induced adipocyte hypertrophy, liver steatosis, and from becoming insulin resistant (56) and thus female mice were not included. Glucose tolerance test was performed by measuring glucose concentrations at 15, 30, 60, and 120 min post-intraperitoneal (i.p.) glucose administration (2 g/kg) in 6-h fasted mice. EDD-fed mice with AUC levels higher than upper 99% confidence interval (CI) of the mean of age-matched control group (AUC^[EDD] > upper 99% CI of mean AUC^[SD]), were considered diabetic (8).

Bacterial Strains.

BCG Pasteur, BCG::RD1, BCG::RD1 ESAT-6 ∆92–95, and Mtb H37Rv were cultured in Middlebrook 7H9 broth (BD Biosciences) supplemented with 0.2% glycerol, 0.05% Tween 80, 10% albumin dextrose catalase (ADC) enrichment, and appropriate antibiotics (see below). Midlogarithmic cultures were harvested, washed in sterile phosphate-buffered saline (PBS), and stored in 15% glycerol/PBS at −80 °C until required for vaccination or infection. The approximate number of CFUs in inocula was determined as described below.

Vaccination and Infection.

Prior to vaccination and infection, frozen bacterial stocks were thawed, centrifuged, and pellets were resuspended in sterile PBS to achieve the appropriate vaccination or infection dose. Diabetic and control C57BL/6 mice were immunized via i.t. administration of 5 × 10⁵ CFUs per mouse. In a pilot study, mice were vaccinated with BCG via intranasal (i.n.), intramuscular (i.m.), and s.c. routes to determine the best route for vaccination (SI Appendix, Fig. S4). For i.t. vaccination, mice were anesthetized using isoflurane (induction 5% and maintenance 3%), the tongue was drawn out of the oral cavity with blunted forceps and 50 µL of inoculum was administered into the oropharynx with the nostrils covered until the inoculum was inhaled. Sixty days p.v., mice were challenged with an ultra-low aerosol dose of Mtb H37Rv (10 to 20 CFUs) using a Glas-Col inhalation exposure system. The actual initial infectious dose was determined by homogenization of lung tissues collected from five mice 1 d p.i. and cultured on 10% oleic acid albumin dextrose catalase (OADC)-enriched 7H11 agar plates.

Intratracheal LPS Instillation.

Anesthetized mice received 5 μg of LPS in 50 μL sterile PBS intratracheally, as described above.

Sample Collection.

At designated p.v. and p.i. time points, mice were killed using carbon dioxide asphyxiation or cervical dislocation. Blood was collected via cardiac puncture into Z-gel tubes (Sarstedt). Serum was prepared by centrifugation of clotted blood, filtered using 0.2 µm SpinX columns (Sigma), and stored at −20 °C. Bronchial lavage (BAL) was used to recover intra-airway luminal cells. Lungs and spleen were aseptically removed for CFU enumeration, histology, and for the preparation of single-cell suspensions. In addition, lung tissue sections and fecal pellets were stored in RNAlater (Invitrogen) for downstream DNA and RNA extractions.

CFU Enumeration.

Lung and spleen tissues were homogenized in sterile sample bags containing 1 mL of sterile PBS/0.05% Tween 80 and serial dilutions were plated on 10% OADC-enriched 7H11 agar plates supplemented with 10 µg/mL cycloheximide and appropriate antibiotics (50 µg/mL hygromycin and 20 µg/mL ampicillin for rBCG strains and Mtb, respectively). Thiophene-2-carboxylic acid hydrazide (TCH, 2 µg/mL; Sigma) was added to 7H11 agar plates to restrict the growth of BCG strains for Mtb culture. Agar plates were sealed and incubated aerobically for 3 to 4 wk at 37 °C. Colonies were counted and the total CFUs per organ was calculated based on dilution factors and organ size.

Cell Isolation.

Lungs were perfused with 10 to 20 mL of sterile PBS, excised (either other half of the right lung containing inferior and postcaval lobes or the whole lung), mechanically disrupted and digested for 30 min at 37 °C with sterile RPMI medium 1640 supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Gibco), 100 U/mL penicillin, 100 µg/mL streptomycin (Gibco), 7.5 µg/mL Collagenase D (Sigma), 1.75 µg/mL Collagenase VIII (Sigma), and 200 µg/mL DNaseI (Sigma). Single-cell suspensions were prepared by passing digested organs through a 70-µm cell strainer followed by erythrocyte lysis using ACK buffer.

BMDMs and BMDCs.

Cells from femurs and tibiae were recovered and grown in RPMI 1640 supplemented with 10% FBS, 2 mM l-glutamine, 100 U/mL penicillin, 100 µg/mL streptomycin, 1 mM sodium pyruvate, 50 µM 2-mercaptoethanol, and either 20 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF; Invitrogen) for BMDCs or 40 ng/mL macrophage colony stimulating factor (M-CSF; Invitrogen) for BMDMs at 37 °C (5% CO₂). After regular medium changes, adhered cells (BMDMs) were removed using TrypLE express (Gibco) on day 7 and nonadherent BMDCs were collected by gentle pipetting on day 8. CD11b⁺ and CD11c⁺ cell populations were enriched using the EasySep mouse CD11b- and CD11c-positive selection kit II, respectively (Stemcell Technologies). Macrophage and DC purity was typically >90 to 95% as determined by flow cytometry.

In Vitro Cell-Based Assays.

Single-cell lung suspensions in Dulbecco's Modified Eagle Medium supplemented with 10% FBS and 3 µg/mL Brefeldin A (eBioscience) were added to 96-well plates precoated with functional grade anti-CD3e antibodies (4 µg/mL, clone 145–2C11, Invitrogen) and incubated for 5 to 6 h at 37 °C (5% CO₂). Lung cells were also infected with Mtb H37Rv (multiplicity of infection [MOI] of 1 for 4 and 24 h) to detect proinflammatory cytokines and nitrite (NO₂⁻) levels by Griess reagent kit (Invitrogen), respectively. In addition, BMDMs and BMDCs were infected with BCG strains in vitro (MOI 10) to assess the cellular nitrite production and expression of costimulatory surface markers.

T Cell Proliferation Assay.

CD11c⁺ BMDCs were infected with BCG strains for 2 h at an MOI of 10 in RPMI 10% FBS. Cells were treated with 200 µg/mL amikacin for an additional 2 h to kill extracellular bacteria. After 16 h, cells were harvested and incubated with 10 µg/mL OVA_257–264 (SIINFEKL) and OVA_323–339 (ISQAVHAAHAEINEAGR) peptides for 1 h at 37 °C. OVA-specific CD4⁺ and CD8⁺ T cells were purified from C57BL/6 OT-I/II mice using EasySep Mouse CD4⁺ and CD8⁺ T Cell Isolation Kits (Stemcell Technologies) and stained with 2 µM Violet Proliferation Dye 450 (VPD450) (BD Bioscience). Peptide-pulsed DCs were then cocultured with VPD-stained T cells at a ratio of 1:1. Following 72-h incubation cells were stained for T cell lineage markers and acquired using flow cytometry.

Flow Cytometry.

Immune cell phenotyping of lung and BAL cell suspensions was done using the following antibodies. All antibodies were purchased from BD Biosciences, unless otherwise stated: CD3e (500 A2), NKp46 (29A1.4), CD4 (GK1.5), CD8a (53-6.7), CD44 (IM7), CXCR3 (CXCR3-173), CD69 (H1.2F3), CD62L (MEL-14), CD103 (M290), TCRβ (H57-597), KLRG1 (2F1), LFA-1 (H155-78), CD11c (HL3), CD11b (MI/70), F4/80 (BM8), I-A/I-E (M5/114.15.2), CD64a/b (X54-9/7.1), CD209 (5H10), CD282 (6C2), CD284 (MTS510), CD40 (3/23), CD80 (16-10A1), CD86 (GL-1), MerTK (DS5MMER, eBioscience), Ly6G (1A8), Siglec-F (E50-2440), CD19 (1D3), TCR Vα2 (B20.1, eBioscience), MARCO (ED31, BioRad), IFNγ (XMG1.2), IL-17A (TC11-18H10), iNOS (CXNFT, Invitrogen), and K_i-67 (SolA15, Invitrogen). Mtb and MAIT cell-specific tetramers were provided by the NIH Tetramer Core Facility and included ESAT-6 I-A(b) QQWNFAGIEAAASA, TB10.4 H-2K(b) IMYNYPAM, and mMR1 5-OP-RU, mMR1 6-FP. Fixable viability stain 780 (BD Biosciences) was used to exclude dead cells. The Foxp3/Transcription Factor Staining Buffer Set (eBioscience) was used for intracellular cytokine staining according to the manufacturer’s instructions. Cells were enumerated using CountBright Absolute Counting Beads (Invitrogen). The BD Fortessa ×20 and FACSCanto II with FACSDiva were used for the acquisition of cells in FACS buffer (PBS/5% FBS and 0.1% NaN₃) and data were analyzed using FlowJo software version 10 (Treestar). Gating strategies for different immune cell populations are shown in SI Appendix, Figs. S6–S9.

Serum Cytokine and Chemokine Analysis.

Frozen serum samples were thawed and prepared according to the Bio-Plex Pro Mouse Cytokine 23-Plex assay (BioRad) specifications and analyzed using a MagPix (Luminex) instrument. Data analysis and heatmap visualization (log₂ concentration) of cytokine levels were done using Prism version 8.3.0 (GraphPad).

Lung Histology.

Left lung lobes were fixed overnight with 4% paraformaldehyde and transferred into 70% ethanol the following day until processing. Processed lungs were embedded in paraffin and 4-µm sections were cut and stained with H&E. ImageJ software was used to calculate the percent of the tissue area affected by measuring the total surface area followed by the areas of dense cell infiltration, as described previously (12).

Total RNA Extraction and NanoString Assay.

Total RNA from fresh or RNAlater-preserved lung tissue samples (50 to 100 mg) from three to five naïve and Mtb-infected (45 d p.i.) control and T2D mice were extracted using the TRIzol Plus RNA purification kit (Invitrogen) according to the manufacturer’s protocol. Tissue samples were homogenized in M tubes (Miltenyi Biotec) with TRIzol reagent using a GentleMACS tissue dissociator (Miltenyi Biotec). Purified RNA was quantified using the Qubit 4.0 Fluorometer (Invitrogen) and RNA integrity number (RIN) was assessed using the Bioanalyser 2100 (Agilent). NanoString nCounter assay was performed using the mouse immunology panel v1 according to the manufacturer’s protocol (NanoString Technologies). Initial NanoString data quality control and normalization were performed in nSolver software (NanoString Technologies). The R software package limma was used for further differential gene expression and comparative analyses (57).

DNA Extraction and 16S rRNA Microbiome Sequencing.

Microbiome DNA from lung samples of naïve and Mtb-infected (45 d p.i.) control and T2D mice (three to five mice per group) was purified using a commercially available microbiome DNA purification kit (Invitrogen) with M tubes and a GentleMACS tissue dissociator (Miltenyi Biotec) used for initial mechanical tissue disruption. The purity of extracted DNA was assessed using the NanoDrop (Thermo Scientific), quantified with Qubit 4.0 Fluorometer (Invitrogen) and the final DNA concentration was adjusted to 5 ng/µL. The 16S rRNA gene encompassing the V5 to V8 regions was targeted using 803F and 1392wR primers (58) and 16S library was performed at the Australian Centre for Ecogenomics following Illumina best practices. Indexed amplicons were sequenced on the MiSeq Sequencing System (Illumina) using paired end sequencing with V3 300-bp chemistry according to the manufacturer’s protocol. FastQC, Trimmomatic (59), QIIME (60), and BLAST (61) software were used for quality control, adapter and low-quality base removal, operational taxonomy units (OTUs) generation, and for OTU identification. Host genome sequences were aligned to the mouse reference genome and all aligned reads were removed using bowtie2 (62) and SAMTools (63), respectively. R packages phylosEq (64) and DESeq2 (65) were used to generate a list of most differentially expressed bacteria for all possible pairings of the four sets of conditions (control, T2D, TB, and T2D/TB). All subsequent plots were generated using ggplot2.

Statistics.

Statistical analysis was performed, and graphs were generated using Prism version 8.3.0 (GraphPad). Two and multiple parametric group analyses were carried out using Student’s t test and one-way analysis of variance (ANOVA), followed by Dunnett’s or Tukey’s multiple comparison tests, respectively. P < 0.05 was considered significant, unless otherwise stated.

Study Approval.

This study was conducted in accordance with the National Health and Medical Research Council (NHMRC) animal care guidelines with all procedures approved by the animal ethics committee (A2400) of James Cook University, Australia.

Data Availability.

All data are available in the manuscript and SI Appendix.