A point-mutated Ag85B antigen improves recombinant bacterial expression and protects mice from aerosol M. tuberculosis challenge

Abstract

Tuberculosis (TB) remains a global health problem, providing motivation for improved vaccine approaches, such as subunit vaccines targeting specific Mycobacterium tuberculosis (M. tuberculosis) antigens. Ag85B is a protein involved in cell wall biosynthesis, is abundant in M. tuberculosis culture supernatants, and has been incorporated in several TB vaccines candidates. We observed low expression yields of Ag85B when expressed recombinantly in E. coli using a histidine-tag purification approach. To address this, we utilized the ThermoMPNN protein structure algorithm to predict several stabilizing mutations in Ag85B. Of these, a single mutation, D52W, significantly enhanced expression yield in E. coli with good storage stability. Ag85B-52W exhibited rapid binding to liposomes incorporating cobalt-porphyrin via his-tag interaction, resulting in suppression in reactivity with an anti-his-tag antibody (due to anchoring of the his-tag in the bilayer), while the surface-displayed protein remained reactive towards anti-Ag85B antibodies. Immunization with Ag85B-52W in a liposomal vaccine elicited antigen-specific antibody and T cell responses, resulting in reduced lung bacterial burden in mice following aerosol M. tuberculosis challenge.

Introduction

Mycobacterium tuberculosis (M. tuberculosis) is the etiological agent of tuberculosis (TB) and represents a major global health burden, resulting in approximately 1.3 million deaths worldwide annually [1]. Despite the development of the bacille Calmette-Guérin (BCG) vaccine over a century ago, TB remains a leading cause of death from infectious diseases. In 2022, an estimated 410,000 individuals were diagnosed with multidrug-resistant or rifampicin-resistant TB (MDR/RR-TB) [1], underscoring a new threat of TB. Currently, there is no licensed vaccine that effectively prevents TB in adults, and the protective efficacy of the BCG vaccine varies greatly [2]. This highlights the urgent need for the development of more effective TB vaccines, a goal that is actively being pursued.

One frequently studied antigen in TB vaccine research is antigen 85B (Ag85B), also known as the 30 kDa major secretory protein or Rv1886c. Ag85B is one of the most abundant exported protein by M. tuberculosis, and functions as a mycolyl transferase in the Ag85 complex (Ag85A, B and C) used for biosynthesis of the mycobacterial cell wall [3]. Ag85B is inherently immunogenic, containing immunodominant T cell epitopes, and induces specific cell-mediated immune responses in infected patients [4, 5]. These properties make Ag85B an attractive antigen candidate and a promising drug target. However, despite its potential, Ag85B tends to be thermally unstable [6, 7], posing challenges for vaccine development and storage of recombinant Ag85B. Numerous Ag85B vaccines have been explored, including DNA vaccines [5, 8–10], mRNA vaccines [11], fusion protein vaccines [12–15], and nanoparticle vaccines [6, 16–21]. These vaccines have shown partial protection against TB infection in murine and guinea pig models. Another Ag85 complex member, Ag85A, is also a popular candidate for TB vaccine development. However, the MVA85A, a viral vectored vaccine constructed using modified vaccinia Ankara virus expressing Ag85A, failed the Phase 2b clinical trial in 2013 [22].

Our group has developed cobalt porphyrin-phospholipid (CoPoP)-based liposomal vaccines targeting several infectious diseases, including malaria [23–26], Lyme disease [27], and SARS-CoV-2 (COVID-19) [28, 29]. CoPoP liposomes carrying recombinant antigens are serum-stable and induce strong, balanced antibody responses, and the CoPoP-based SARS-CoV-2 vaccine (EuCorVac-19) has successfully completed Phase 2 and 3 clinical trials [30–33]. For this study, we selected the CPQ (CoPoP/PHAD-3D6A/QS-21) adjuvant, which combines potent and well-characterized immunostimulatory components. PHAD-3D6A, a synthetic monophosphoryl lipid A (MPLA) and Toll-like receptor 4 (TLR-4) agonist, activates antigen-presenting cells in a non-pyrogenic manner [34, 35], while QS-21, a saponin-based adjuvant, synergistically enhances cytokine secretion and antibody responses when co-formulated in the liposomal bilayer [36]. Comparative studies evaluating CoPoP/PHAD-3D6A (CP) against several commercial adjuvants, including Alum, Titer Max, AdjuPhos, QuilA, Sigma Adjuvant System, Addavax (MF59-like), AS01, and Freund’s adjuvant (CFA/IFA), demonstrated that CP elicited the highest antigen-specific antibody titers against malaria antigens while maintaining the lowest local reactogenicity, highlighting the critical role of antigen particleization in driving robust immune activation [23]. Moreover, inclusion of QS-21 to generate CPQ further enhanced immunogenicity against malaria antigens, as evidenced by significantly higher antigen-specific IgG titers and increased frequencies of multifunctional (IL2⁺TNFα⁺IFNγ⁺) CD4⁺ and CD8⁺ T cells compared with CP and Alum [26].

Advances in structural biology and computational frameworks have led to structure-based antigen design becoming increasingly important in vaccine design [37]. Stabilization of vaccine antigens has emerged as important topic with demonstrated potential in a wide range of diseases including RSV [38], COVID-19 [39] and malaria [40]. To facilitate generation of stabilized proteins such as antigens, we developed the ThermoMPNN deep neural network trained to predict stability changes for protein point mutations given an initial structure, which only requires the input of a pdb file and can make amino acid mutation predictions for improved thermostability [41].

In this study, we utilized CoPoP liposomes containing PHAD-3D6A, and QS-21 (CPQ liposomes) to deliver a point-mutated recombinant Ag85B protein with a C-terminus His-tag via a cobalt-histidine interaction. To improve the thermal stability of recombinant Ag85B expressed in E. coli B21, we assessed several single mutations using ThermoMPNN algorithm. The Ag85B-D52W single mutation increased expression yield. Importantly, when applied as a vaccine with the CPQ liposome platform, the stabilized version of Ag85B elicited antigen-specific antibody production, T cell responses, and reduced bacterial burden and pathology in the lungs of mice infected with M. tuberculosis.

Results

Prediction of Stabilizing Mutations

To identify putative stabilizing mutations for Ag85B, the crystal structure of the mature protein (PDB code: 1F0N, [3]) was used as input to the neural network ThermoMPNN [41] (Fig. 1a, b). ThermoMPNN predicts the change in thermostability (ddG) of a protein caused by point mutations in its sequence. From the ThermoMPNN predictions, three Ag85B point mutations were selected for experimental characterization: P14Y, D52W, and E277I (the numbering starts after removing the first 40 amino acids of the signal peptide). P14 is located at the end of a b-strand and is partially solvent exposed. A tyrosine at this position is predicted to pack against the side chain carbons of P16. The mutant tryptophan at residue 52 is predicted to pack against Y58, Y10, and F1. The mutant isoleucine at residue 277 is expected to expand the hydrophobic core by making interactions with W208 and Y145.

Fig. 1: Selected Ag85B point mutations for experimental characterization.

(a) Crystal structure of Ag85B (PDB code: 1F0N) showing the locations of residues selected for site-directed mutagenesis. (b) Depiction of recombinant Ag85B proteins indicating the site-specific mutations and the corresponding predicted change in ddG stability parameter for each mutation site.

Expression Yield and Thermal Stability of Ag85B Mutants

To assess the impact of point mutations on the expression yield and thermal stability of Ag85B, the wildtype and mutant proteins were recombinantly expressed in E. coli BL21 and subsequently purified. The D52W mutant (Ag85B-52W) exhibited the highest expression yield (Fig. 2a, b), producing 1.12 mg of protein from a 50 mL culture using a standardized purification protocol. The other mutations, Ag85B-14Y and Ag85B-227I, also demonstrated significantly higher yield (0.82 mg and 0.64 mg, respectively) compared to the wildtype Ag85B, which yielded 0.36 mg. These results were consistent with densitometry analysis of the Ag85B band visualized by Coomassie staining of SDS-PAGE gel (Fig. 2c).

Fig. 2: Improved expression yield of Ag85B with the D52W point mutation.

(a) SDS-PAGE analysis of purified wildtype and mutant Ag85B proteins. (b) Protein expression yields determined by BCA assay following standard purification. (c) Densitometric quantification of Ag85B protein bands using Image Lab software. (d) Thermal stability of wildtype and mutant Ag85B proteins measured by Nano-DSF, showing melting temperature (T_m) values. Data represent mean ± SD from three independent single-colony preparations per construct. Statistical analysis was performed by one-way ANOVA followed by Tukey’s test. *p < 0.05, **p < 0.01, ***p < 0.005, and ****p < 0.001.

To evaluate the thermal stability of these proteins, Nano-DSF thermal unfolding experiments were conducted to determine their melting temperatures (T_m). The results show that the T_m values for Ag85B wildtype, Ag85B-14Y, Ag85B-52W, and Ag85B-227I were 64.5°C, 64.8°C, 62.7°C, and 63.9°C, respectively (Fig. 2d). These findings indicate that the point mutations resulted in at least a two-fold increase in expression yield compared to the wildtype protein, although there was no significant enhancement in thermal stability.

Characterization of liposome-bound Ag85B antigens

Recombinant Ag85B proteins bearing a C-terminus His-tag were analyzed for their ability to form particles with CPQ liposomes via a cobalt-histidine interaction. Control liposomes, lacking cobalt in the porphyrin ring (“2HPQ” liposomes), were also evaluated. The formation of Ag85B-liposome was assessed using a competition assay with Ni-NTA beads. In this assay, unbound (free) protein was captured by the beads (“B”), while liposome-bound protein remained in the supernatant (“S”). The binding patterns for all four Ag85B point-mutant proteins were nearly identical (Fig. 3a). Ag85B-52W and Ag85B-227I mutants demonstrated complete binding to CPQ liposomes but not to 2HPQ liposomes, indicating that the recombinant Ag85B proteins were successfully loaded onto CPQ liposomes.

Fig. 3: Characterization of CPQ liposomes decorated with recombinant Ag85B antigens.

(a) Ni-NTA bead competition assay. Recombinant Ag85B proteins were incubated with liposomes for 3 hours, followed by Ni-NTA bead isolation. Proteins captured by Ni-NTA beads (marked as “B”) represent unbound soluble proteins, while liposome-bound proteins remained in the supernatant (marked as “S”). (b) Hydrodynamic size and (c) polydispersity index of CPQ liposomes after Ag85B binding were determined by dynamic light scattering (DLS). (d) Slot blot detection of Ag85B proteins. Binding of Ag85B protein to CPQ liposomes was detected using polyclonal anti-Ag85B antibodies. HRP signals were visualized in chemiluminescence mode, while the color of the CPQ liposomes (green) was captured in colorimetric mode, indicating the absorption of the liposomes on the membrane. (e) Slot blot detection of His-tag reactivity. His-tags were detected by monoclonal anti-His tag antibodies. Data represent mean ± SD from triplicates measurements (n=3 per group).

The particle sizes and polydispersity index of the liposomes were analyzed (Fig. 3b, c). The size of the CPQ liposomes remained consistent at ~100–110 nm, regardless of whether Ag85B was bound to the liposomes. The particleization kinetics of Ag85B was further examined using a fluorescent resonance energy transfer assay. Approximately 50% of the point-mutated Ag85B proteins converted into particle form with CPQ liposomes within 30 minutes of incubation (Supplementary Fig. 1).

Next, we assessed the conformational integrity of the Ag85B proteins in particle form. Slot blot analysis revealed that all CPQ-bound Ag85B proteins were recognized by the anti-Ag85B polyclonal antibody (Fig. 3d, e). In contrast, the His-tag reactivity was undetectable with the anti-His tag monoclonal antibody following Ag85B particleization. This suggests that the His-tags of the Ag85B proteins were fully embedded into the CPQ liposomes and inaccessible to the antibody.

Overall, these findings demonstrate that His-tagged, point-mutated Ag85B proteins rapidly formed particles with CPQ liposomes while retaining their structural integrity.

Stability of CPQ Liposomes Bound to Recombinant Ag85B

Among the mutated recombinant Ag85B proteins, Ag85B-52W demonstrated the highest expression yield while maintaining strong binding efficiency to CPQ liposomes, making it the selected antigen candidate for further TB vaccine formulation. To evaluate the storage stability of this vaccine candidate, both wildtype Ag85B and Ag85B-52W proteins were admixed with either PBS or CPQ liposomes and stored at 25°C for 30 days.

Ag85B-52W bound CPQ liposomes exhibited no signs of proteolysis after 30 days of storage at 25°C (Fig. 4a). In contrast, CPQ-bound wildtype Ag85B and proteins stored in PBS showed reduced signal intensities on the SDS-PAGE gel, indicating decreased protein stability. The SDS-PAGE results demonstrate that Ag85B-52W remained highly stable in both free and CPQ-bound forms, with no significant degradation observed over the 30-day period (Fig. 4b).

Fig. 4: Protein stability of CPQ liposomes binding to recombinant Ag85B proteins.

(a) SDS-PAGE analysis of Ag85B protein stability. Recombinant Ag85B proteins, admixed with PBS or CPQ liposomes, were stored at 25°C for 30 days. Samples were stored in Laemmli buffer until analyzed by SDS-PAGE. (b) Stability of Ag85B-52W. Recombinant Ag85B-52W protein, admixed with PBS or CPQ liposomes, was stored at 25°C or 4°C for 30 days. Samples were collected on Day 0–30 and stored in Laemmli buffer and analyzed by SDS-PAGE. (c) Slot blot detection of Ag85B stability. Detection of Ag85B protein stability in CPQ liposomes, stored at 4°C or 25°C, using polyclonal anti-Ag85B antibodies. (d) Slot blot detection of His-tag reactivity. Detection of His-tag reactivity in CPQ liposomes using monoclonal anti-His tag antibodies.

The stability of Ag85B-52W binding to CPQ liposomes was further confirmed through slot blot analysis. The binding accessibility of Ag85B-52W to polyclonal anti-Ag85B antibodies remained unchanged after 30 days of incubation at both 4°C and 25°C (Fig. 4c). In parallel, the His-tag of Ag85B-52W remained fully embedded within the CPQ liposomes throughout the storage period (Fig. 4d), as indicated by the lack of detectable His-tag reactivity by monoclonal anti-His tag antibodies.

These findings suggest that the D52W point mutation significantly enhanced the storage stability of Ag85B, and encapsulation in CPQ liposomes provided additional protection, maintaining antigen integrity. This stability improvement could enable the storage and transportation of the TB vaccine without requiring cold-chain logistics, enhancing the flexibility and feasibility of vaccine distribution in diverse settings.

Immunization of CB6F1/J Mice With CPQ/Ag85B-52W Vaccines Induced Antigen-Specific Antibody and T Cell Responses

In our previous work, we have shown that a 2 μg antigen dose provides strong immunogenicity for CoPoP-based vaccines targeting malaria and SARS-CoV-2 [26, 29, 42]. Therefore, we selected a 2 μg antigen dose for the Ag85B vaccine formulation. CB6F1/J mice were immunized intramuscularly with 2 μg of Ag85B proteins admixed with either CPQ liposomes or Alum on days 0 and 14, and sera were collected on day 28. The doses of CoPoP, PHAD-3D6A, and QS-21 in the CPQ formulation were 8, 3.2, and 3.2 μg, respectively. Mouse body weight was monitored throughout the study (Fig. 5a). No significant changes in body weight were observed after immunization, and mice exhibited steady weight gain, indicating that the vaccine formulations did not exert any overt toxicity. ELISA analysis revealed that mice immunized with CPQ/Ag85B-52W liposomes had significantly higher anti-Ag85B IgG titers (4.8×10⁴) than those receiving 2HPQ/Ag85B-52W (3.6×10³), Alum/Ag85B-52W (3.8×10³), or untreated controls (Fig. 5b). These results underscore the critical role of liposome-based display in eliciting strong humoral responses for this antigen. We further compared Ag85B-specific immunogenicity between BCG- and Ag85B-52W-immunized mice. ELISA analysis showed that the BCG did not induce significantly higher anti-Ag85B IgG titers relative to untreated or saline controls. (Supplementary Fig. 2).

Fig. 5: CPQ/Ag85B vaccines induce antigen-specific antibody and robust T cell responses in mice.

CB6F1/J mice were immunized intramuscularly with Ag85B vaccines (2 μg of antigen) on days 0 and 14. (a) Body weight of mice monitored throughout the study. (b) Anti-Ag85B IgG titer measured by ELISA. Serum samples collected on day 28 were analyzed for anti-Ag85B IgG titers. The groups tested included: untreated (n=4), Alum/52W (Ag85B-52W with Alum; n=4), 2HPQ/52W (Ag85B-52W with 2HPQ liposomes; n=4), and CPQ/52W (Ag85B-52W with CPQ liposomes; n=4). Splenocytes harvested 14 days after the second immunization were restimulated with an Ag85B peptide pool. (c) Representative IFNγ/ TNFα double-color ELISpot results. (d) Quantitative analysis of single-positive (IFNγ or TNFα) and double-positive (IFNγ⁺TNFα⁺) spots. (e) Representative IFNγ/IL2 double-color ELISpot results. (f) Quantitative analysis of the single-positive (IFNγ or IL2) and double-positive (IFNγ⁺IL2⁺) spots. Data are presented as (b) geometric mean ± geometric SD or (a, d, f) mean ± SD (n=4 per group). Statistical significance was determined using ordinary one-way ANOVA (b) or two-way ANOVA (d, f) followed by Tukey’s multiple comparisons test. Significance levels are indicated as *p < 0.05, **p < 0.01, ***p < 0.005, and ****p < 0.001.

To further characterize antigen-specific T-cell responses, IFNγ/TNFα double-color and IFNγ/IL2 double-color ELISpot assays were performed using Ag85B peptide pools for restimulation. Splenocytes from mice immunized with CPQ/Ag85B-52W exhibited significantly higher secretion of TNFα, IFNγ, IFNγ/TNFα, IL-2, and IFNγ/IL-2 compared with those from the 2HPQ/Ag85B-52W, Alum/Ag85B-52W and untreated groups (Fig. 5c, d, e, f). These results show that CPQ/Ag85B-52W vaccine induced antigen-specific TNFα, IFNγ, and IL-2-producing splenocytes.

Previous studies have demonstrated that these cytokines are secreted by polyfunctional CD4⁺ T cells, which play a central role in vaccine-mediated protection against M. tuberculosis [43, 44]. As Type 1 T helper (Th1) cells, these CD4⁺ T cells produce IL-2 and TNFα to support dendritic cells during T cell priming and B cell help, while IFNγ activates macrophages to eliminate intracellular M. tuberculosis and enhances CD8⁺ T cell responses [45]. IL-2 additionally promotes proliferation and survival of TCR-activated T cells and supports the memory T cell differentiation during primary infection [46]. In murine TB models, the induction of IL-2 producing CD4⁺ T cells correlates with vaccine-induced protection against M. tuberculosis [47].

Together, these findings demonstrate that the CPQ/Ag85B-52W vaccine elicits strong antigen-specific antibody and multifunctional T cell responses, supporting its potential as an effective TB vaccine candidate. In contrast, vaccination with Alum- or 2HPQ-based formulations induced good antibody responses but failed to generate comparable T cell responses, whereas the CPQ formulation promoted both balanced antibody and polyfunctional T cell responses.

CPQ/Ag85B Vaccines Induced Short Term Protection against M. tuberculosis Challenge in Mice

To investigate the protective efficacy of the novel Ag85B CPQ liposomal vaccine, we measured the bacterial burden in infected mice by assessing colony forming units (CFUs) in the lungs at 30 days post-challenge. Six- to eight-week-old female CB6F1/J mice (n=5 per group) were vaccinated with saline, BCG, CPQ/Ag85B, CPQ/Ag85B-14Y, or CPQ/Ag85B-52W. BCG was administered in the scruff of the neck at 1×10⁶ CFU, whereas all other vaccines were delivered intramuscularly with 2 μg antigen in 50 μl. Mice received a homologous boost three weeks later (BCG mice received saline), and three weeks after boosting, mice were transferred to BSL-3 and challenged by aerosol infection with 50–100 CFU of M. tuberculosis H37Rv ATCC 27294.

Bacterial counts indicated that mice vaccinated with CPQ/Ag85B-52W experienced a significant reduction in bacterial burden in the lungs compared to the unimmunized control mice (saline) (Fig. 6a).

Fig. 6: Protective efficacy of CPQ/Ag85B vaccination against M. tuberculosis aerosol challenge in mice.

CB6F1/J mice were assigned to the indicated vaccine groups subjected to aerosol M. tuberculosis challenge following. (a) Lung colony-forming units (CFUs). Data show geometric mean ± geometric SD (n=5 per group). (b) Percentage of lung lesions. Four lung tissue slides per mouse were analyzed, and each data point represents the average of the four slides for an individual animal. Data are presented as mean ± SD (n=5). Statistical significance was determined using ordinary one-way ANOVA with Dunnett’s multiple comparison test: *p < 0.05, **p < 0.01, ***p < 0.005, and ****p < 0.001.

Additionally, we evaluated lung histology to assess lesion formation by performing H&E staining of lung sections from the same mice as in Fig. 6a. CPQ/Ag85B immunization significantly reduced pathological damage in the lungs, whereas unvaccinated mice exhibited extensive lung damage following TB challenge (Fig. 6b, Supplementary Fig. 3). These findings suggest that the novel Ag85B CPQ liposomal vaccine confers short term protective immunity against M. tuberculosis infection in mice.

Discussion

Stabilized antigens could become useful tools for the development of next-generation TB vaccines. Given that challenges associated with the cold-chain transportation and storage of effective TB vaccines in developing countries, a stabilized vaccine that can be stored at room temperature with an extended shelf life would significantly benefit individuals exposed to the threat of TB infection globally. Our study demonstrates that the single mutation in the Ag85B protein, predicted by the ThermoMPNN algorithm, significantly enhanced the antigen’s E. coli expression yield and storage stability. An actual increase in thermal stability based on protein unfolding melting temperatures was not observed. Structural modeling using AlphaFold2 revealed that the tryptophan residue introduced at position 52 in Ag85B-52W variant is mostly buried (Supplementary Fig. 4), with solvent accessible surface area analysis (via get_sasa_relative in PyMOL) indicating that approximately 80% of the residue is inaccessible to water. The aspartic acid at the corresponding position in the wildtype protein is similarly buried, suggesting that this mutation is unlikely to substantially alter B cell epitope presentation or humoral responses.

The recombinant Ag85B proteins can form stable particles with CPQ liposomes. Immunization with the Ag85B CPQ liposomal vaccines in mice resulted in elevated antigen-specific antibody production and significantly increased IFNγ, TNFα, and IL-2 secretion by splenocytes, indicative of a robust polyfunctional CD4⁺ T cell (Th1) response, which play a key role in vaccine-induced protection against M. tuberculosis. Consistent with prior findings, other experimental TB vaccines have also been shown to activate antigen-specific CD4+ and CD8+ T cell responses [48, 49]. However, the limitation of this study is that we did not characterize antigen-specific CD4+ and CD8+ T cell populations with intracellular cytokine staining.

We additionally compared Ag85B-specific antibody responses following CPQ/Ag85B and BCG immunizations (Supplementary Fig. 2) and observed that BCG did not induce significantly higher anti-Ag85B IgG titers relative to untreated or saline controls. Although we did not assess Ag85B-specific IFNγ, TNFα, and IL-2 secretion by splenocytes after BCG vaccination, previous studies have shown that conventional BCG elicits only limited Ag85B-specific cytokine responses, consistent with its modest induction of Ag85B-directed immunity [50, 51].

Short-term protection studies (day 30 post-challenge) demonstrated that immunization with CPQ/Ag85B vaccines reduced bacterial burden and attenuated lung pathology following M. tuberculosis infection in mice. Despite these encouraging results, the efficacy of our CPQ/Ag85B vaccines remain inferior to that of the conventional BCG vaccine. Another limitation of this study is the lack of dose-ranging evaluation for the antigens. In addition, immunodominant T cell epitopes in certain mouse strains may be absent in others and in other species. Future studies should therefore evaluate vaccine performance in multiple antigen dosages in additional animal models to confirm cross-strain and cross-species protective immunity.

To further enhance efficacy, incorporating additional antigens into a multivalent formulation represent a promising strategy. The CoPoP liposome platform readily enables such antigen combination [23]. Multiple TB culture filtrate proteins, including ESAT-6 [52], TB10.4, CFP-10 [53], APA [54], MTB12 [55] and MTB39 [56], have been identified as strong vaccine candidates. We aim to determine the optimal antigen combination and liposomal formulation to develop a multivalent TB vaccine with the potential to surpass BCG vaccine in confering durable, broad-spectrum protection across diverse populations.

Conclusion

We demonstrated that a single mutation in the Ag85B protein, D52W, as predicted by ThermoMPNN algorithm, significantly enhanced antigen expression yield in E. coli and resulted in good storage stability. Additionaly, immunization of mice with the CPQ-formulated Ag85B liposomal vaccines resulted in strong antigen-specific antibody responses and robust IFNγ, TNFα, and IL-2 secretion by splenocytes. These immune responses correlated with protection against M. tuberculosis infection, as indicated by reduced bacterial loads and mitigated lung pathology. Collectively, these findings highlight the potential of this novel CPQ/Ag85B vaccine as a foundation for developing a multivalent, stabilized TB vaccine with improved protective efficacy.

Materials and Methods

Design of the Recombinant Proteins

Point mutations predicted to stabilize Ag85B were identified using the Colab implementation of ThermoMPNN [41]. ThermoMPNN is a graph neural network with message passing that was trained to predict changes in the free energy of folding for protein point mutations. The crystal structure (or model) of a protein serves as input to ThermoMPNN, and the model predicts a ddG of folding for all possible point mutations that can be made to the protein. For Ag85B, the crystal structure 1F0N was used as input. The wildtype sequence used for our studies was the Ag85B protein from M. tuberculosis (strain ATCC 25618/H37Rv), retrieved from UniProtKB/Swiss-Prot database (accession numbers: P9WQP1). The pET-3a plasmids containing our engineered constructs were manufactured by GenScript (Piscataway, NJ). The BL21 (DE3) E. coli strain was obtained from Intact Genomics (MO, USA). The protein expression construct included a 6-histidine tag at the C-terminus of the protein. Full amino acid sequences are shown in Supplementary Table 1.

Expression of the Soluble Recombinant Proteins

An E. coli DirectPlate^™ BL21 (DE3) chemically competent cell clone (Intact Genomics, Cat# 1019–36) harboring the target plasmid was streaked on a LB agar plate (Teknova, Cat# L1093) containing 100 μg/mL ampicillin (Quality Biological, Cat# 351-344-731) and incubated at 37°C for 16 hours. A single colony from each construct was inoculated into 2 mL of trypticase soy broth (BD Diagnostics, Cat# 292770) supplemented with 100 μg/mL ampicillin and cultured for 3 hours. The starter culture was then transferred into 50 mL of terrific broth (Gibco, Cat# A1374301) containing 100 μg/mL ampicillin and incubated at 37°C until OD₆₀₀ reached 1.0–1.2. Isopropyl-β-D-thiogalactopyranoside (IPTG) (Invitrogen, Cat# 15529019) was added to a final concentration of 1 mM for 3 hours of induction at 37°C. The cell pellet was collected after centrifugation at 4,000g for 30 minutes.

The pellet was lysed by incubation in B-PER Complete Bacterial Protein Extraction Reagent (Thermo Scientific, Cat# PI89822) according to the manufacturer’s instructions. The lysate was centrifuged at 16,000g for 20 minutes to remove cell debris, and the supernatant was collected for further protein purification steps. The lysate was mixed with 1X volume of Binding Buffer (50 mM Na₂HPO_4, 300 mM NaCl, 10 mM Imidazole, pH 8.0). A total of 0.3 mL of Ni-NTA resin (Thermo Scientific, Cat# 88221) was packed into a 5 mL column, rinsed with 1 mL of deionized H₂O, and equilibrated with 1 mL of Binding Buffer. The lysate was then added to the column and passed through it ten times. After washing with 25 mL of Wash Buffer (50 mM Na₂HPO_4, 300 mM NaCl, 20 mM Imidazole, pH 8.0), the target protein was eluted by adding 0.6 mL of Elution Buffer (50 mM Na₂HPO_4, 250 mM NaCl, 300 mM Imidazole, pH 8.0) three times. The eluted fractions were collected and dialyzed against 1 L of PBS three times. Lastly, the purified proteins were filtered through a 0.22 μm membrane and examined using SDS-PAGE analysis. Protein concentrations were determined by using a BCA assay. For each construct, protein expression and purification were repeated three times using independent single colonies. Band intensities corresponding to the expected molecular weight were quantified from SDS-PAGE gels using Image Lab software (Bio-Rad).

Thermal Stability Assay

Nano-DSF thermal unfolding experiments were performed on a Prometheus NT.48 system (Nanotemper). Proteins were diluted to 5.625μM in 1x PBS pH 7.4 and 10 μL of each protein was loaded into a capillary. The samples were heated from 20°C to 95°C at a rate of 1°C/min. The intrinsic fluorescence signals at 330 nm and 350 nm were measured for each experiment. The T_m was calculated automatically based on the first derivative of fluorescence signal by the software PR.Thermcontrol (Nanotemper). The T_m values reported in this study were extrapolated from the first derivative of 330 nm fluorescence signal. Each construct was measured in triplicate using protein samples expressed from three independent single colonies.

Liposome Preparation

Liposomes were prepared using ethanol injection and nitrogen-pressurized lipid extrusion as reported previously [23]. The prepared liposomes were dialyzed against PBS twice at 4°C to remove ethanol and passed through a 0.2 μm sterile filter. The final liposome concentration was adjusted to 320 μg/mL and stored at 4°C. Liposome sizes and polydispersity index were determined by dynamic light scattering (DLS) using a NanoBrook 90 plus PALS instrument after 500-fold dilution in PBS. The CoPoP/PHAD (CP) liposomes had a mass ratio of [DOPC:CHOL:CoPoP:PHAD-3D6A] [20:5:1:0.4]. PoP/PHAD-3D6A (2HP) liposomes, which contain hydrogen instead of cobalt in the PoP, served as control liposomes, with a mass ratio of [DOPC:CHOL:PoP:PHAD-3D6A] [20:5:1:0.4]. For liposomes containing QS-21, QS-21 (1 mg /mL) was added to the liposomes after formation at an equal mass ratio as PHAD-3D6A. The CoPoP/PHAD-3D6A/QS-21 (CPQ) had a mass ratio of [DOPC:CHOL:CoPoP:PHAD-3D6A:QS-21] [20:5:1:0.4:0.4], PoP/PHAD-3D6A/QS-21 served as control liposomes with the same mass ratio of [DOPC:CHOL:PoP:PHAD-3D6A:QS-21] [20:5:1:0.4:0.4].

Ni-NTA Bead Competition Assay

To assess recombinant protein binding stability in particle form, Ni-NTA magnetic beads (ThermoFisher, Cat # 88831) were added to the liposome-incubated antigens (1:4 mass ratio of total protein: CoPoP) or free proteins in PBS. The samples were incubated with the beads for 30 minutes at room temperature. Following incubation, the supernatant and beads were separated using a magnetic separator (ThermoFisher, Cat # 12321D). The beads were then resuspended in PBS, and denaturing reducing loading dye was added to all samples (supernatant and beads), followed by heating at 95 °C for 10 minutes. A total of 46 μL samples (1.5 μg protein) were loaded into each well of a Tris-Glycine gel (Bio-Rad, Cat# 4568084). PageRuler Prestained Protein Ladder (4 μL) (Thermo Scientific, Cat# 26616) was also loaded. Gels were run at a constant 200 V for 30 minutes and stained with SimplyBlue SafeStain (Invitrogen, Cat# LC6065) following the manufacturer’s instructions. Gel images were acquired using a Bio-Rad GelDoc Go Imaging System.

Slot Blot Assay

A 0.2 μm nitrocellulose membrane (Thermo Scientific, Cat# 88013) was affixed in a 48-well microfiltration apparatus (Bio-Rad, Cat# 170–6545). Each well was first loaded with 50 μL of PBS, followed by 50 μL of sample (50 ng), which flowed through the membrane by gravity. After blocking the membrane with 5% bovine serum albumin (VWR, Cat# 97061–422) for 1 hour at room temperature, the membrane was incubated with anti-M. tuberculosis Ag85B polyclonal antibody (Abcam, Cat# ab43019) or anti-His tag monoclonal antibody (GenScript, Cat# A00186) at a 1:1000 or 1:5000 dilution, respectively, in 2% BSA for 1 hour at 37°C. The membrane was washed three times with PBS, followed by incubation with anti-rabbit IgG antibody HRP (GenScript, Cat# A01827) or anti-mouse IgG antibody HRP (Cell Signaling, Cat# 7076S) at a 1:2000 dilution in 2% BSA for 1 hour at 37°C. The membrane was washed again with PBS and treated with VisiGlo HRP substrate mixture (VWR, Cat# 97063–148). Images were captured using a Bio-Rad ChemiDoc Imager.

Fluorescence Quenching Assay

The recombinant protein (200 μg) was dialyzed into 100 mM sodium bicarbonate buffer (pH 9.0) using a Slide-A-Lyzer Dialysis Cassette (Thermo Scientific, Cat# 66383) twice at 4°C. The protein was labeled with 10 μL of 1 mg/mL DY-490-NHS-Ester (Dyomics, Cat# 490–01) for 1 hour at room temperature with continuous mixing. The labeled protein was then dialyzed with PBS three times to remove excess dye. The fluorescence of the labeled protein (491 nm excitation, 527 nm emission) was measured using a TECAN Safire microplate reader to assess binding efficiency to liposomes. Upon binding to liposomes, fluorescence was quenched due to energy transfer from DY-490-Ag85B to the porphyrin in CPQ or 2HPQ [23]. Binding kinetics were monitored by the changes in fluorescence after incubating the same volume of protein (80 μg/mL) with liposomes in PBS (320 μg/mL) or PBS alone. Samples were diluted 10-fold prior to measurement in the plate reader. The binding percentage was calculated using the following equation:

$$\text{Binding\%}=1-(\frac{{\text{Fluorescence}}_{\text{test}\text{liposomes}}}{{\text{Fluorescence}}_{\text{free}\text{labeled}\text{protein}}})\times 100\%$$

Protein Storage Stability Test

Antigen samples in PBS or CPQ liposomes were stored at room temperature or 4°C. Samples were collected on days 0, 5, 10, 15, 20, and 30, boiled with reducing sample buffer (Thermo Scientific Pierce, Cat# 39000) for 5 minutes, and stored at −20°C. Prior to SDS-PAGE analysis, samples were re-boiled for 5 minutes.

Murine Immunization

Murine studies were performed according to protocols approved by the University at Buffalo or Colorado State University IACUC. Six- to eight-week-old female CB6F1/J mice (The Jackson Laboratory, JAX stock #100007) received intramuscular injections on days 0 and 14 containing 2 μg Ag85B proteins combined with the liposomal adjuvants: The CoPoP/PHAD-3D6A/QS-21 (CPQ) liposome formulation had a mass ratio of [DOPC:CHOL:CoPoP:PHAD-3D6A:QS-21] [20:5:1:0.4:0.4]. The PoP/PHAD-3D6A/QS-21 (2HPQ) liposome formulation had a mass ratio of [DOPC:CHOL:PoP:PHAD-3D6A:QS-21] [20:5:1:0.4:0.4]. The commercial adjuvant Alum (InvivoGen, Cat# vac-alu-50) was used as a comparison. CPQ and 2HPQ vaccines were prepared by incubating Ag85B protein (80 μg/mL) with liposomes (320 μg/mL) for 3 hours at room temperature. Alum vaccines were prepared by mixing antigen with Alum diluted to 1 mg/mL in PBS. The experimental groups included untreated controls (n=4), Alum/52W (Ag85B-52W with Alum; n=4), 2HPQ/52W (Ag85B-52W with 2HPQ liposomes; n=4), and CPQ/52W (Ag85B-52W with CPQ liposomes; n=4). On day 28 post-immunization, blood was collected under 5% isoflurane anesthesia, followed by humane euthanasia via cervical dislocation. Sera were separated by centrifuging the whole blood at 2,000g for 20 minutes at 4°C.

Antigen-Specific Antibody Titer Measurement

Anti-Ag85B IgG titers were assessed by ELISA using 96-well plates coated with 1 μg/mL of Ag85B in coating buffer (3.03g Na₂CO₃; 6 g NaHCO₃ in 1 L deionized water, pH 9.6) for 1 hour at 37°C. After washing and blocking with 2% BSA in PBS containing 0.1% Tween-20 (PBS-T), serial dilutions of mouse sera in PBS were added and incubated for 1 hour at 37°C. After washing with PBS-T, HRP-linked anti-mouse IgG antibody (Cell Signaling, Cat# 7076S) was added, followed by 3,3’,5,5’-tetramethylbenzidine (TMB) substrate (Surmodics, Cat# TMBW-1000–01). The reaction was stopped with 1M HCl, and absorbance was measured at 450 nm. Titers were defined as the reciprocal serum dilution at which the absorbance exceeded the background by 0.5 absorbance units. The experimental groups included untreated controls (n=4), Alum/52W (n=4), 2HPQ/52W (n=4), and CPQ/52W (n=4).

ELISpot

On day 28 post-immunization, CB6F1/J mice were anesthetized with 5% isoflurane and euthanized by cervical dislocation. Spleens were excised and passed through a 70 μm cell strainer (VWR, Cat# 76327–100) using ice-cold PBS. Red blood cells were lysed using RBC lysis buffer (BioLegend, Cat# 420301) to obtain single-cell splenocyte suspensions. A total of 5×10⁵ splenocytes were re-stimulated with 1 μg/mL of PepMix^™ M. tuberculosis Ag85B (JPT, Cat# PM-MYCTU-AG85B) for 24 hours at 37°C in mouse IFNγ/TNF-α double-color ELISpot plates (C.T.L., Cat# mu-IFNγ/TNF-α) or mouse IFNγ/IL-2 double-color ELISpot plates (C.T.L., Cat# mu-IFNγ/IL-2), following the manufacturer’s instructions. Wells containing medium alone served as negative controls. The experimental groups included untreated (n=4), Alum/52W (n=4), 2HPQ/52W (n=4), and CPQ/52W (n=4).

Mycobacterium tuberculosis challenge study

Six- to eight-week-old female CB6F1/J mice were purchased from the Jackson laboratory as described above and housed in a BSL2 at CSU during vaccinations. Experimental groups included saline (n=5), BCG (n=5), CPQ/Ag85B (n=5), CPQ/Ag85B-14Y (n=5), and CPQ/Ag85B-52W (n=5). For BCG vaccination, mice were injected SubQ with 50μl of 1×10⁶ CFU BCG Pasteur in the scruff of the neck. All other vaccinations were performed IM in the thigh region upon injection of 50μl saline or 2μg of CPQ/Ag85B. Three weeks later, mice were boosted with the corresponding priming vaccine, except BCG-vaccinated mice which received a saline boost. Mice were moved three weeks later into BSL3 conditions and infected via the aerosol route with a Glas-Col apparatus calibrated to deliver 50–100 CFUs/mouse of actively growing M. tuberculosis H37Rv ATCC 27294.

On day 0 and 30 post-infection, mice were humanely euthanized via CO₂ asphyxiation and organs harvested to enumerate bacterial burden in lungs as well as for histopathological analysis. Lungs were homogenized at 7,200rpm for 16 sec in PBS using 7ml Precellys tubes and plated on 7H11 nutrient agar. After incubating for 3–4 weeks at 37°C, CFUs were enumerated, log10 transformed and analyzed with Prism using one-way ANOVA and Dunnett post-test in comparison to saline vaccinated mice.

For histopathological analysis, lungs were fixed in PBS + 4% paraformaldehyde for at least 72 hours and submitted for paraffin embedding, sectioning and staining at CSU’s Diagnostic Lab. Hematoxylin and Eosin (H&E) stained tissue samples were scanned using Brightfield Imaging on a Vectra Polaris^™ Automated Quantitative Pathology Imaging System at 40x resolution. QPTIFF data files were uploaded to QuPath, an open-source software for digital pathology image analysis [57]. Annotations were applied to the negative control tissue samples to generate snapshots used as training images, highlighting background, lesioned, and non-lesioned tissue types. By manually defining these categories, the software was trained to recognize and differentiate between these elements as a pixel classification system. Training images from positive control tissues were used to refine the software before applied to the remaining H&E samples to generate density maps, distinguishing between background, lesioned, and non-lesioned areas of a sample. Area of a tissue sample dedicated to each of these elements was quantified and used to calculate the percentage of lesion burden per lung. Percent of lesion/lung was analyzed with Prism via one way ANOVA and Dunnett post-test in comparison to saline vaccinated mice.

Significance Statement.

A point mutation in the M. tuberculosis Ag85B protein, predicted by ThermoMPNN algorithm, enhanced its antigen expression yield in E. coli with good storage stability. Immunization with the Ag85B mutant with a liposome vaccine system in mice resulted in antigen-specific humoral and cellular response that protected mice against M. tuberculosis infection. This approach could facilitate the use of Ag85B in TB vaccine development.

Data Availability

All data are included in the manuscript and/or supporting material.