mRNA-based tuberculosis vaccines BNT164a1 and BNT164b1 are immunogenic, well tolerated and efficacious in rodent models.

Tuberculosis (TB) remains a global crisis. It was responsible for approximately 1.23 million deaths in 2024. While the Bacillus Calmette–Guérin (BCG) vaccine is the current standard, it suffers from a critical failure mode. It provides robust protection against severe childhood forms of the disease. However, it offers little to no efficacy against pulmonary TB in adolescents and adults. These are the very populations responsible for the majority of transmission.

The central challenge in TB vaccinology is the biological heterogeneity (diverse characteristics) of Mycobacterium tuberculosis (Mtb). The bacteria transition through various metabolic states. They move from active replication to dormancy and even resuscitation. Because a single-antigen vaccine might only catch the bacteria in one specific state, developers have struggled to create a universal shield. This paper introduces two mRNA-based vaccine candidates, BNT164a1 and BNT164b1. They are designed to bypass this limitation by targeting eight distinct antigens expressed across different stages of the infection cycle.

The failure of single-target strategies

Current TB prevention relies heavily on BCG. This is a live-attenuated vaccine (a weakened version of the pathogen) derived from Mycobacterium bovis. However, BCG's inability to prevent adult pulmonary disease creates a massive gap in global health security. Researchers have traditionally looked for "silver bullet" antigens. These are single proteins that, if targeted, would neutralize the pathogen.

The problem is that Mtb is a moving target. As the bacteria adapt to the host immune system, they shift their protein expression. A vaccine targeting only an active-stage protein will miss the bacteria when they enter a dormant, non-replicating state. To achieve broad coverage, a vaccine must be "octavalent" (targeting eight different proteins). This ensures that regardless of the bacteria's metabolic state, the immune system has a recognizable target. This paper moves toward a multi-antigen, mRNA-delivered architecture.

Architecture of an octavalent mRNA platform

The researchers designed BNT164a1 and BNT164b1 to deliver a complex payload of eight Mtb antigens. These include Ag85A, Hrp1, ESAT-6, RpfD, RpfA, HbhA, M72, and VapB47. Instead of delivering eight separate mRNA strands, the authors used a chimeric fusion strategy. They distributed the eight antigens across four distinct mRNAs. Each mRNA encodes a fusion of two antigens [Figure 1a].

The design process involved several critical layers: 1. Structural Integrity: To ensure these fused proteins would not misfold, the authors used AlphaFold2 to predict 3-D configurations. They found that the predicted structures closely matched the native shapes of the individual antigens. This suggests they would effectively trigger the correct immune responses. 2. Chemical Variation: The two candidates differ only in their RNA chemistry. BNT164a1 uses nucleoside-unmodified RNA (uRNA). This typically triggers stronger innate immune signals via Toll-like receptors. BNT164b1 uses $N^1$-methylpseudouridine-modified mRNA (modRNA). This is the same chemistry used in COVID-19 vaccines. It is optimized for high protein expression and lower inflammatory side effects. 3. Delivery: Both candidates utilize lipid nanoparticles (LNPs). These are tiny fat bubbles that protect the mRNA and ferry it into host cells.

Robust immunogenicity and bacterial clearance

The study demonstrates that this multi-antigen approach engages both the humoral (antibody-based) and cellular (T cell-based) arms of the immune system. In various mouse strains, including humanized HLA-A2.1/DR1 mice, the vaccine elicited antigen-specific T cell responses against all eight targets [Figure 2c,d].

The authors measured the efficacy of the vaccine in aerosol challenge models. In these models, mice were exposed to Mtb via inhalation. The results were significant: * Bacterial Burden: In mice infected with the reference strain H37Rv, the BNT164 candidates achieved a ~0.55–0.85 log reduction in bacterial loads [Figure 5b]. This means the bacterial count was reduced by roughly 4 to 7 times compared to saline controls. * Hypervirulent Strains: Against the more aggressive HN878 strain, the reduction reached ~0.8–1.0 log reduction [Figure 5c]. This represents an 80% to 90% reduction in the bacterial load. * Antibody Durability: Antibody levels showed a modest decline over time. However, a third "booster" dose successfully restored IgG levels to their peak [Figure 3b]. Following a third dose, the binding of antibodies to actual Mtb lysate increased 40-fold in the BNT164b1 group [Figure 3c].

A striking finding involves the quality of the immune cells recruited to the lungs. Protection correlated with the infiltration of CD8+ T cells possessing a "memory precursor" phenotype (MPEC/LLMP). Unlike short-lived effector cells that burn out quickly, these cells are primed to provide long-term immunity in the respiratory tract [Figure 6f,i].

Assessing the limits of the model

Several technical caveats must be acknowledged. First, the study relies on murine (mouse) models. These models have inherent limitations. Researchers note that mice may have lower sensitivity to certain innate immune signals compared to humans. This might mask differences between the uRNA and modRNA platforms.

Second, there is a biological mismatch in antigen expression. In the mice used, certain antigens may not be expressed in the same temporal window as in humans. Specifically, the hypoxia-associated Hrp1 and VapB47, and the resuscitation-associated RpfA and RpfD, might not be expressed during the studied timeframe. Consequently, the vaccine's ability to catch bacteria during these specific life stages remains unproven in this model. Finally, the researchers note that the effectiveness of these candidates in humans remains to be determined.

The verdict: A scalable leap forward

The BNT164 candidates represent a sophisticated evolution in vaccine design. By moving to an octavalent mRNA platform, the researchers addressed the problem of Mtb's metabolic plasticity. The data show that the vaccine is immunogenic and reduces bacterial loads across different genetic lineages.

The transition of these candidates into Phase 1/2 clinical trials (NCT05537038, NCT05547464) is the ultimate test. If the correlation between lung-resident memory CD8+ T cells and bacterial control holds true in humans, this approach could provide the durable protection that BCG has failed to deliver for decades.

Figures from the paper