Deletion of nuoG from the Vaccine Candidate Mycobacterium bovis BCG ΔureC::hly Improves Protection against Tuberculosis

ABSTRACT The current tuberculosis (TB) vaccine, Mycobacterium bovis Bacillus Calmette-Guérin (BCG), provides insufficient protection against pulmonary TB. Previously, we generated a listeriolysin-expressing recombinant BCG strain, which to date has successfully completed phase I and phase IIa clinical trials. In an attempt to further improve efficacy, we deleted the antiapoptotic virulence gene nuoG, encoding NADH dehydrogenase 1 subunit G, from BCG ΔureC::hly. In vitro, deletion of nuoG unexpectedly led to strongly increased recruitment of the autophagosome marker LC3 to the engulfed vaccine, suggesting that nuoG also affects xenophagic pathways. In mice, BCG ΔureC::hly ΔnuoG vaccination was safer than BCG and improved protection over that of parental BCG ΔureC::hly, significantly reducing TB load in murine lungs, ameliorating pulmonary pathology, and enhancing immune responses. Transcriptome analysis of draining lymph nodes after vaccination with either BCG ΔureC::hly or BCG ΔureC::hly ΔnuoG demonstrated earlier and stronger induction of immune responses than that with BCG SSI and suggested upregulation of inflammasome activation and interferon-induced GTPases. In summary, BCG ΔureC::hly ΔnuoG is a promising next-generation TB vaccine candidate with excellent efficacy and safety.

tors p62 and Ndp51 (11)(12)(13). BCG lacks Esx-1, does not rupture the phagosome, and is not targeted by autophagy under normal conditions (7,11,14,15). Apoptosis controls M. tuberculosis replication and spreading (16), while phagocytosis of apoptotic host cell-derived vesicles by dendritic cells boosts T cell responses via cross-presentation (17). M. tuberculosis has evolved strategies to inhibit these defense mechanisms. Screening for antiapoptotic genes in M. tuberculosis identified nuoG, encoding a subunit of the dispensable respiratory enzyme complex NADH dehydrogenase 1 (18,19). Disruption of nuoG increased M. tuberculosis-induced apoptosis via a tumor necrosis factor alpha (TNF-␣)-dependent mechanism and decreased the virulence of M. tuberculosis (18). As apoptosis is thought to enhance adaptive immune responses through cross-presentation (17,20,21), we deleted nuoG in BCG ⌬ureC::hly. We were able to surmount the high bar, further increasing efficacy against TB in a vaccine that already expresses 100-fold-higher protection than BCG, without loss of its excellent safety profile.
Further experiments in THP-1 macrophages demonstrated that, unexpectedly, knockout of nuoG from BCG strains drastically enhanced colocalization of the vaccine with the autophagy protein LC3 ( Fig. 2A), from 4 to 8 h p.i. up to 48 h p.i. ( Fig. 2A and B). While LC3 was previously shown to be increased in THP-1 macrophages after BCG ⌬ureC::hly infection (7), it did not specifically colocalize with the vaccine as seen after infection with BCG ⌬ureC::hly ⌬nuoG. This suggests an intriguing new role for the mycobacterial gene nuoG in suppressing host cell xenophagic responses, which may involve either the canonical autophagy pathway or LC3-associated phagocytosis (LAP), two mechanistically distinct processes involving autophagy proteins (22,23). Deletion of nuoG improves vaccine-induced protection. To assess the specific influence of nuoG deletion on vaccine efficacy, we immunized mice with BCG ⌬nuoG and determined bacterial loads over 180 days post-M. tuberculosis challenge (Fig. 3A). Vaccination with BCG ⌬nuoG consistently reduced the M. tuberculosis burden in lungs of mice over that after vaccination of mice with BCG SSI, with a similar, less pronounced trend in spleens (Fig. 3B). Having demonstrated that nuoG disruption improved protective efficacy, we then investigated whether this effect synergizes with the apoptosis-inducing phenotype of BCG ⌬ureC::hly. Consistent with previous reports, BCG ⌬ureC::hly-vaccinated mice were better protected than BCG SSI-vaccinated mice ( Fig. 4A and B) (5,24). Importantly, additional deletion of nuoG further improved efficacy against challenge with both an M. tuberculosis laboratory strain (H37Rv) (Fig. 4A) and a clinical M. tuberculosis isolate (Beijing/W lineage) (Fig. 4B). Protection was particularly improved in lungs, which also benefited from markedly ameliorated gross pathology (Fig. 4C) and histopathology (Fig. 4D) at 180 days p.i. Differences in bacterial counts were less pronounced but statistically significant in spleens ( Fig. 4A and B). Thus, BCG ⌬ureC::hly ⌬nuoG conferred increased protection compared not only to BCG but also to BCG ⌬ureC::hly, against both pulmonary and disseminated TB. In order to assess the safety of the recombinant BCG vaccine candidates, we examined persistence and dissemination in the months following vaccination. The two recombinant vaccine candidates were comparable, i.e., more quickly cleared from the lymph nodes than BCG SSI and with lower degrees of dissemination to the spleen and no dissemination to the lung (see Fig. S1 in the supplemental material). Attenuation of the recombinant strains compared to BCG SSI was confirmed by  studies in severe combined immunodeficiency (SCID) mice, in which BCG ⌬ureC::hlyand BCG ⌬ureC::hly ⌬nuoG-vaccinated mice survived twice as long as BCG SSI-vaccinated animals. Together, our data demonstrate that deletion of nuoG sustained the notable preclinical safety profile of BCG ⌬ureC::hly ( Fig. 4E) (5). In summary, deletion of nuoG from BCG ⌬ureC::hly improved efficacy against TB, paralleled by excellent safety.
Vaccination with BCG ⌬ureC::hly ⌬nuoG enhanced immune responses. Because BCG ⌬ureC::hly ⌬nuoG-vaccinated mice had the lowest bacterial burdens following M. tuberculosis challenge, we aimed to determine which immune responses were associated with protection. The more rapid increase in size of dLNs of BCG ⌬ureC::hly ⌬nuoG-vaccinated mice in the days following vaccination suggested increased stimulation of the acquired immune response (Fig. 5A). Previously, we had demonstrated significantly increased Ag85B-specific central memory CD4 ϩ T cells in dLNs of BCG ⌬ureC::hly-vaccinated mice compared to BCG SSIvaccinated mice at 14 days (6), and here, we observed the same trend, with similar numbers of Ag85B-specific CD4 ϩ T cells in BCG ⌬ureC::hly ⌬nuoG-vaccinated mice at day 14 (Fig. 5B). In addition, at day 21, frequencies were significantly increased for T follicular helper cells (Fig. 5C), central memory and effector memory CD4 ϩ T cells ( Fig. 5D and E), germinal center B cells (Fig. 5F), and gamma interferon (IFN-␥)-producing CD4 ϩ T cells (Fig. 5G) in dLNs of BCG ⌬ureC::hly ⌬nuoG-vaccinated mice compared to BCG SSI-vaccinated mice. No significant effect was found in frequencies of central memory and effector memory CD8 ϩ T cells ( Fig. 5H and I) in comparison to BCG SSI-vaccinated mice (Fig. 5B). Frequencies of T follicular helper cells, effector memory CD4 ϩ T cells, and IFN-␥-producing CD4 ϩ T cells were also increased in spleens of BCG ⌬ureC::hly ⌬nuoG-vaccinated mice compared to BCG SSI-vaccinated mice, whereas germinal center B cells, central memory CD4 ϩ T cells, and central and effector memory CD8 ϩ T cells were not significantly different (see Fig. S2A to G in the supplemental material). Mycobacterium-specific immunoglobulin G (IgG) levels ( Fig. 5J) were markedly increased after vaccination with both recombinant strains compared to vaccination with the current vaccine strain BCG SSI. We have already observed higher antibody levels induced by BCG ⌬ureC::hly than by BCG, both in mice and in humans (6,8). Overall, the trend was qualitatively similar for the two recombinant BCG strains, but BCG ⌬ureC::hly ⌬nuoG-vaccinated mice had increased CD4 ϩ T cell responses compared to BCG ⌬ureC::hly-vaccinated mice, as well as increased germinal center B cells, suggesting synergism between the mechanisms of efficacy of the two genetic strain modifications.
Gene expression analysis of dLNs in BCG ⌬ureC::hlyand BCG ⌬ureC::hly ⌬nuoG-vaccinated mice. Microarray analyses were performed to assess global host gene expression profiles in dLNs of mice in response to vaccination (Fig. 5K). Overall, results revealed earlier and stronger induction of immune responses by both recombinant BCG strains, particularly BCG ⌬ureC::hly ⌬nuoG, with vastly more genes differentially expressed (P Ͻ 0.05; fold change, Ͼ2) than after BCG vaccination (see Fig. S3A in the supplemental material). Genes in R1-to R8-labeled gene clusters are listed in Table S1 in the supplemental material. Because BCG ⌬ureC::hly ⌬nuoG was derived from BCG ⌬ureC::hly, numerous genes showed similar expression patterns in response to the two vaccine strains compared to BCG, e.g., IL-1␤ and IL-18, previously found to be upregulated 1 day after BCG ⌬ureC::hly vaccination (7) (see Fig. S3B and C). In contrast, gene expression levels of IFN-inducible GTPases (Gbps, Irgs, and Givns), often associated with phagosomal or autophagic vacuoles and inflammasome activation (25)(26)(27)(28), and ubiquilin, a key player in xenophagic re-  Table S1 in the supplemental material). sponses to M. tuberculosis (13), were increased at earlier time points and to greater levels in BCG ⌬ureC::hly ⌬nuoG-vaccinated mice (see Fig. S3C), which is in line with the observed increase in bacterium-associated LC3 responses in vitro. Table S2 in the supplemental material lists genes significantly upregulated (P Ͻ 0.05) Ն2-fold versus the naive group specifically in BCG ⌬ureC::hly ⌬nuoG-immunized mice at days 1, 3, and 7 postvaccination. By day 7, expression of IFN-inducible GTPases had also increased in BCG-and BCG ⌬ureC::hly-vaccinated mice but tended to remain slightly lower than that after BCG ⌬ureC::hly ⌬nuoG vaccination in most cases (see Fig. S4). Early upregulation of genes for inflammasome-associated interleukin-1␤ (IL-1␤), IL-18, cytosolic DNA sensor Ifi204, and Gbps was confirmed by reverse transcription-PCR (RT-PCR) (see Fig. S5). Gene ontology (GO) analysis of differentially expressed genes highlighted involvement of acute inflammatory responses at day 1 postvaccination and immune cell activation and differentiation at day 3, while concurrently with enhanced dLN enlargement in BCG ⌬ureC::hly ⌬nuoG-vaccinated mice, cell cycle and developmental pathways featured prominently at days 3 and 7 (see Table S3). Due to the overwhelming dominance of cell cycle and tissue development pathways obtained in GO analysis of day 7 gene expression, only the top 20 pathways are listed in the table.

DISCUSSION
An estimated 9.6 million new active TB cases and 1.5 million deaths occurred in 2014 (1), emphasizing the need for a more efficacious vaccine. The current TB vaccine, BCG, shows variable efficacy against the pulmonary form of the disease, although it has 60 to 80% protective efficacy against severe disseminated forms of disease in infants, such as meningitis (29). A recombinant live vaccine, BCG ⌬ureC::hly, which expresses listeriolysin, is the most advanced BCG replacement vaccine candidate in clinical trials, having completed phase I and phase IIa safety and immunogenicity trials successfully (NCT01479972, NCT01113281, and NCT00749034) and currently undergoing a phase II safety and immunogenicity trial in HIV-exposed newborns (NCT02391415). Our previous studies suggest that increased preclinical efficacy of BCG ⌬ureC::hly is based on (i) high egression of BCG-derived protein antigens and (ii) release of bacterial DNA into the host cell cytosol, subsequent induction of apoptosis and inflammasome activation, and increased generation of central memory CD4 ϩ T cell responses (6)(7)(8). While this vaccine awaits phase IIb efficacy trials, next-generation vaccines are being designed and tested in preclinical models aimed at optimizing efficacy and/or safety.
Recently, an antiapoptotic virulence gene, nuoG, was identified in M. tuberculosis (18,19). As apoptosis is thought to enhance adaptive immune responses through cross-presentation (17,20,21), we aimed to augment the efficacy and safety of BCG ⌬ureC:: hly by deleting nuoG. Because BCG ⌬ureC::hly already induces 100-fold-better protection than the sham control and 10-fold protection over BCG in mice, which has not been achieved by other recombinant BCG vaccine candidates to date (5,24), it sets a high bar for further improvement. Yet, our results demonstrate that vaccination with BCG ⌬ureC::hly ⌬nuoG further increased protection about 5-fold in lungs of mice challenged with the M. tuberculosis laboratory strain (H37Rv) and a clinical M. tuberculosis isolate (Beijing/W lineage) at 90 and 180 days p.i. while maintaining excellent safety in immunodeficient SCID mice. Note that in-creased protection was sustained against the clinical isolate M. tuberculosis Beijing/W, which is considered notoriously resistant against BCG vaccination. Deletion of nuoG from unmodified BCG also resulted in decreased pulmonary pathogen loads, suggesting a standalone function for nuoG in protective efficacy.
The increased efficacy of BCG ⌬ureC::hly ⌬nuoG versus BCG ⌬ureC::hly was associated with a numerical increase in CD4 ϩ T EM cells, T FH cells, and germinal center B cells and a trend toward an increase of CD4 ϩ T CM cells. T EM cells, which appear early after infection and can secrete effector cytokines such as IFN-␥ and tumor necrosis factor alpha (TNF-␣), provide immediate protection, while T CM cells proliferate in the LN and generate new waves of effector cells upon reexposure to antigen (6,30,31). Recently, T CM cells were found to be associated with protection after vaccination (6,31). We demonstrated previously that vaccination with BCG ⌬ureC::hly increased T CM responses as well as T FH responses and antibody production (6), and these responses seem to be further enhanced after BCG ⌬ureC::hly ⌬nuoG vaccination. Transfer studies demonstrated that protection against TB was conferred by the T CM cell population (6). Both T CM and T FH express CXCR5, and CXCR5-expressing T cells have previously been correlated with decreased lung pathology following vaccination and challenge with M. tuberculosis (32). The T FH population, which decreases more quickly than the long-lived T CM population, stimulates germinal center B cell responses (33), but the role of B cells and antibodies in TB remains unclear. Both T CM and T FH cells have been associated with enhanced antibody responses (30). Vaccination with BCG ⌬ureC::hly and BCG ⌬ureC::hly ⌬nuoG leads to increased antibody responses in mice, and enhanced production of Mycobacterium-specific antibodies was found in a phase I clinical trial in the BCG ⌬ureC::hly group over the BCG group (8). Although it is difficult to foresee a protective role of antibodies to M. tuberculosis once it is hidden inside host cells, vaccine-induced preexisting antibodies could participate in prevention of infection with M. tuberculosis (34). Apart from their role in antibody production, B cells can also present antigen to T cells and enhance T CM and T FH cell development (35).
In accordance with improved immune responses, dLNs were found to increase in size earlier in BCG ⌬ureC::hly ⌬nuoGvaccinated mice. Transcriptome analysis revealed similar changes in gene expression in both BCG ⌬ureC::hlyand BCG ⌬ureC::hly ⌬nuoG-vaccinated mice, with induction of genes such as IL-1␤, IL-18, Gbps, and other GTPases, although the expression of genes associated with GTPase activity, intracellular resistance, inflammatory responses, cell activation, and cell proliferation tended to be higher in BCG ⌬ureC::hly ⌬nuoG-vaccinated mice. Genes significantly differentially expressed between BCG ⌬ureC::hly ⌬nuoGand BCG ⌬ureC::hly-vaccinated mice included Ifng-and IFN-␥-induced genes, suggesting an improved antimicrobial Th1-type response. Overall, nuoG deletion appeared to synergize with, and enhance, the protective effects of the ⌬ureC::hly mutation in BCG against M. tuberculosis, since most responses were quantitatively but not qualitatively different. Unexpectedly, the present study also uncovered a novel potential role for the mycobacterial gene nuoG in suppressing host cell LC3-mediated pathways, in addition to its previously reported role in inhibition of apoptosis (18,19). For analysis of nuoG-mediated effects, we employed THP-1 cells, shown to be an appropriate model for human alveolar macrophage responses to mycobacterial infection (36) and used for investigation of antiapoptotic functions of nuoG (18,19). Previously, we have shown that BCG ⌬ureC::hly induced increased overall levels of the autophagy marker LC3 in infected THP-1 cells in an AIM2-and STING-dependent manner compared to BCG SSI (7). However, colocalization of LC3 with bacteria was not observed. Here, we demonstrate that both BCG and BCG ⌬ureC::hly strains deficient in nuoG colocalized with LC3 within THP-1 cells, which did not occur in the parental strains. This effect was observed to begin between 4 and 8 h p.i.; at 24 h p.i., most bacteria were completely surrounded by LC3, and up to 48 h p.i., LC3 was still associated with the bacteria. It has been shown that artificially inducing autophagy during BCG vaccination increases antigen processing, leading to improved Th1 responses and vaccine efficacy (14). It remains to be seen whether the strong colocalization of the autophagy protein LC3 mediates enhanced destruction of the bacteria through canonical autophagic pathways or through LAP pathways (22) and whether this provides a link to the enhanced protective immune responses seen in BCG ⌬ureC::hly ⌬nuoG-vaccinated mice. During autophagy, proteins such as LC3 form a double-membraned autophagosome, which captures cytoplasmic components and transports them to the lysosome for degradation (37). During LAP, autophagy components such as LC3 are translocated to the phagosomal membrane and promote fusion with the lysosome, which does not involve the formation of a double-membrane autophagosome (22). Conjugation of LC3 to phagosomes and subsequent association with lysosomes require the activity of NADPH oxidase (NOX2) and the production of reactive oxygen species (ROS). Intriguingly, previous studies on nuoG in M. tuberculosis revealed that inhibition of apoptosis was related to its ability to neutralize NOX2-dependent ROS (18). Therefore, the ability of nuoG to neutralize NOX2dependent ROS could also impact the induction of LAP, explaining why nuoG-deficient BCG has increased LC3 colocalized to phagocytosed bacteria. NOX2 activity is also required for efficient cross-presentation by human dendritic cells (38). Therefore, we speculate that the nuoG gene of M. tuberculosis plays multiple roles in inhibiting optimal host immune responses and antigen presentation.
The mechanism by which nuoG deletion leads to increased targeting of the bacteria is curious, as it implies a role for nuoG in inhibiting LAP, and possibly other autophagic pathways involving LC3. However, BCG ⌬ureC::hly and BCG ⌬ureC::hly ⌬nuoG had similarly decreased survival times and less dissemination in immunocompetent mice, suggesting that inflammasome-or apoptosis-mediated mechanisms induced by both vaccine strains are primarily responsible for eliminating BCG. Apoptosis of infected macrophages is considered important for immunity to pulmonary TB (18)(19)(20)39). Similarly to M. tuberculosis (19), we found that BCG did not induce elevated levels of apoptosis in THP-1 macrophages, and deletion of nuoG did not enhance this. As nuoG-inhibited apoptosis in M. tuberculosis relies on neutralization of NOX2-dependent ROS (18), it is possible that nuoGdeficient BCG induces lower levels of ROS. In vivo, apoptosis was increased only at day 14 in mice vaccinated with BCG ⌬ureC::hly ⌬nuoG, suggesting that increased apoptosis occurs downstream of altered intracellular mechanisms initiated in the absence of nuoG, including increased oxidative stress or autophagy.
In summary, our data reveal a potential novel role for mycobacterial nuoG in inhibition of LC3-mediated autophagic pathways, with relevance for protective immunity against M. tuberculosis. Even though the prominent long-term protective efficacy of BCG ⌬ureC::hly of up to 2 logs over that of BCG set the bar high for further improvement, we demonstrate a significant increase in vaccine efficacy as reflected by a 5-fold-lower pulmonary M. tuberculosis burden upon deletion of nuoG from BCG ⌬ureC::hly, corresponding with enhanced immune responses after vaccination and paralleled by an excellent safety profile.
Histology and ex vivo apoptosis assays. Lymph nodes were collected at days 0, 3, 7, and 14 postvaccination. Tissues were fixed in 4% formaldehyde in PBS and embedded in paraffin wax. Tissue sections were stained with hematoxylin and eosin (H&E), and terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) staining was performed for apoptosis (APO-BrdU TUNEL staining kit; Life Technologies). Apoptotic cells were counted per field of view at ϫ200 magnification.
Transcriptome analysis. Mice were vaccinated with BCG SSI, BCG ⌬ureC::hly, or BCG ⌬ureC::hly ⌬nuoG, and lymph nodes were collected in RNAlater RNA stabilization reagent (Qiagen) at days 1, 3, and 7 postvaccination. Following the collection of all lymph nodes, samples were removed from RNAlater, homogenized in Trizol (Qiagen) using a gen-tleMACS dissociator (Miltenyi), and then frozen at Ϫ80°C. After thawing, samples were allowed to stand at room temperature for 5 min, and then precipitation was performed using isopropanol, ammonium acetate (Ambion AM9070G), and glycogen (Ambion AM9510), and pellets were washed with 70% ethanol and resuspended in RNase-free water on ice. The concentrations were measured on a NanoDrop spectrophotometer, and the quality of the RNA was assessed using a Bioanalyzer. Agilent whole-mouse-genome microarrays were performed using RNA samples labeled with a one-color Quick Amp labeling kit (Agilent Technologies) according to the manufacturer's instructions. To avoid batch-specific effects, we spread samples from different groups and time points between microarray chips. Scanning of microarrays was performed with 3-m resolution and 20-bit image depth using a G2565CA high-resolution laser microarray scanner (Agilent Technologies). Microarray image data were analyzed and extracted with the Image Analysis/Feature Extraction software G2567AA v.A.11.5.1.1 (Agilent Technologies) using the protocol GE1_1105_Oct12 and recommended settings. Analysis of transcripts obtained at day 1 after vaccination with BCG and BCG ⌬ureC::hly only, in comparison to naive mice, was previously performed and published (7). Here, microarray data obtained from all days and with all vaccine strains were analyzed using GeneSpring 12.6 GX (Agilent Technologies), with quality control filters, normalization, and one-way analysis of variance (ANOVA). Naive mice were selected as the control group for comparative analysis. The P values were corrected for multiple comparisons, and values of P Ͻ 0.05 were considered statistically significant. Subgroups of differentially expressed genes with change greater than 2-fold from the comparison group (P Ͻ 0.05) were used for generation of heat maps, Venn diagrams, pathway analysis, and GO analysis. RT-PCR was performed to validate expression of selected genes. cDNA was generated by reverse transcription using the iScript cDNA synthesis kit (770-8897; Bio-Rad), according to the manufacturer's instructions, on a Gene Amp PCR System 9700 machine (Applied Biosystems). PCR was performed on a Step One Plus real-time PCR machine (Applied Biosystems) using the SYBR green Fast mix (Thermo Fisher Scientific) with the Fast protocol and the primers listed in Table S4 in the supplemental material. Primers were designed using Primer3Plus software (43). Analysis was performed using the threshold cycle (C T ) comparative method (44), with the housekeeping gene Ywhaz used for normalization.
Animal experimentation. Nine-to 10-week-old female mice (BALB/c and CB-17/Icr-Prkdc SCID /Rj [Janvier]; C57BL/6 [Charles River Laboratories]) were kept under specific-pathogen-free conditions in groups of five in individually ventilated cages. Animals were vaccinated subcutaneously in the tail base with 10 6 CFU of BCG strains. At designated time points postvaccination, mice were euthanized and tissues of interest were removed and homogenized in PBS-0.05% Tween 80 prior to CFU enumeration or processed otherwise. For protective efficacy studies, mice were aerosol challenged 90 days postvaccination with a low dose of 100 to 200 CFU of M. tuberculosis. All animal studies have been ethically reviewed and approved by the State Office for Health and Social Services, Berlin, Germany. Experimental procedures were carried out in accordance with the European directive 2010/63/EU on Care, Welfare and Treatment of Animals.
Statistical methods. GraphPad Prism 6.04 (GraphPad Software, Inc.) was used for statistical analysis. Survival curves were calculated by using the Mantel-Cox log rank test. Vaccine efficacy was evaluated by one-way ANOVA with Tukey's multiple-comparison test. Similarity of variances between groups compared was determined by the Brown-Forsythe test. For autophagy quantification, the Mann-Whitney test was used for pairwise comparison. Two-way ANOVA with Tukey's multiple-comparison test was used to evaluate immunology data.
Microarray data accession number. Microarray data are available from the NCBI GEO database under accession code GSE74282.