Recombinant Bacillus Calmette–Guérin Expressing SARS-CoV-2 Chimeric Protein Protects K18-hACE2 Mice against Viral Challenge

Since the outbreak of the COVID-19 pandemic in late 2019, massive loss has been experienced around the globe, and current death counts exceed 6 million worldwide (1, 2). Since SARS-CoV-2 was first identified in Wuhan, China (3), a tremendous effort has been put forward by the scientific community to contain its spread as governments struggle to deal with exhausted healthcare services (4, 5). Since COVID-19 was designated a pandemic in March 2020 by the World Health Organization, several actions have been taken to manage this critical situation (6). Lockdown countermeasures, physical distancing, home-office practices, financial aid packages, and drug repositioning studies (7, 8) are among the many strategies that have been implemented over the past 2 years.

The prompt availability of viral genome sequencing (9) was instrumental for the development of accurate diagnoses and experimental treatments (10). In that context, genomic surveillance programs have been playing critical roles in supporting direct governmental measures and epidemiological strategies (11). The SARS-CoV-2 genome consists of a single-stranded positive sense RNA that encodes among its structural proteins two major candidates for vaccine development: the nucleocapsid protein (N) (12, 13) and the spike glycoprotein (S) (14, 15). The latter is the primary vaccine candidate because it is essential for viral attachment to the angiotensin-converting enzyme 2 (ACE2) receptor and subsequent entry of the virus into the host (16, 17). This human respiratory disease presents clinical manifestations ranging from asymptomatic to severe. After an incubation period of 4–14 d, the most common symptoms are characterized by chills or fever, cough, headache, and fatigue, although intense pneumonia, ageusia, and anosmia (loss of taste and smell) have also been reported in some cases. Depending on the prognosis, severe cases may present cytokine storm followed by respiratory failure and death (16, 18).

Because SARS-CoV-2 is a highly transmissible viral infection, large-scale mass vaccination has become a critical management tool. According to World Health Organization, ∼160 vaccine candidates are currently in clinical evaluation (1–4, 19). Intervention strategies are for the most part based on inducing neutralizing Abs against S or other structural proteins, but vaccine designs aimed at the inhibition of proteases or RNA-dependent polymerases are also being evaluated (20–22). Multiple platforms have been assessed since December 2020 to develop a protective vaccine against SARS-CoV-2. So far, it is still unclear which vaccine methodology is the most effective. The mRNA-based vaccines (such as BNT162 from Pfizer/BioNTech and mRNA-1273 from Moderna [23, 24]), adenovirus-based vaccines (such as AZD1222/chAdOx1 developed by the University of Oxford and AstraZeneca and Sputnik V by the Gamaleya Research Institute of Epidemiology and Microbiology [25, 26]), and whole-virus-based vaccines (such as CoronaVac developed by Sinovac Research and Development and BBV152/Covaxin from Bharat Biotec [27, 28]) are among the current methodologies with approved vaccine candidates for population use. More than 11 billion doses of vaccines have been administered to the worldwide population (2). However, despite the mass production and distribution of vaccines, new variants of concern (VOCs) continue to emerge (6), causing worries about escape from the vaccine-induced immune protection. Several studies have shown that key mutations on the viral proteins compromise Ab neutralization (24, 29–31), jeopardizing population coverage and therefore threatening COVID-19 control. This highlights the feasibility of using other viral structural proteins (32), or even a combination of them (33), as candidates. Therefore, it is of the utmost importance to better understand the long-term protective mechanisms of currently used vaccines, as well as to invest in full characterization of new vaccine candidates to contain the spread of SARS-CoV-2 and its VOCs in a new wave of the global pandemic.

Another candidate that has been investigated due to its beneficial heterologous effects is bacillus Calmette–Guérin (BCG). This attenuated strain of Mycobacterium bovis is protective against tuberculosis, and nowadays it is the most applied vaccine in the world (34). A growing body of evidence indicates that BCG vaccination also provides different levels of protection against nonspecific pathogens (35–37), the result of a phenomenon known as “trained immunity” (38). This so-called heterologous protection induced by BCG immunization has been reported against viral infections, such as HSV, respiratory syncytial virus, yellow fever virus, human papillomavirus, hepatitis B virus, and influenza A virus (34, 39). These findings raised the possibility of BCG vaccination as a promising weapon against SARS-CoV-2 infection (40). Multiple randomized trials and retrospective cohort studies are being carried out worldwide to assess if BCG vaccination provides any level of protection against SARS-CoV-2 infection (41–43). A number of studies have suggested an association of prior BCG vaccination with lower incidence of COVID-19 disease (42, 44). Nevertheless, this concept remains controversial. Currently, the www.clinicaltrials.gov database records more than 20 clinical trials, the results of which are expected to shed some light on this question. In addition, recent preclinical studies have demonstrated that the use of BCG as an adjuvant enhances COVID-19 vaccine protection in a murine model (45–47).

Beyond any heterologous protective effects, this live bacterial vaccine can also be used as a powerful adjuvant-like tool to elicit strong cellular and humoral immune responses, enhancing vaccine responses against viral infections (35, 36, 48, 49). These immunological features laid the foundation for the development of BCG-based platforms to deliver heterologous Ags (also known as recombinant BCG [rBCG]) (50, 51). To date, several strategies for stable and controlled expression of foreign Ags by rBCG strains have been investigated with promising results, including viral, bacterial, and parasitic proteins (52–54). Considering these studies, a rBCG strain expressing SARS-CoV-2 Ags merits investigation as a vaccine candidate against COVID-19 (55). This strategy could provide a powerful combination of the beneficial effects elicited by the BCG upon the innate immune system (i.e., trained immunity) and also specific cellular and humoral responses against SARS-CoV-2 elicited by the specific recombinant Ags vectorized by this system. Herein, we describe a rBCG expressing a chimeric protein consisting of N- and S-derived immunogenic epitopes from SARS-CoV-2 (here termed rBCG-ChD6), previously identified by our team (12, 14). Furthermore, we demonstrated the protective effects of rBCG-ChD6 against SARS-CoV-2 infection in K18-hACE2 mice. A single administration of rBCG-ChD6 followed by a booster dose of the recombinant chimera protein with alum elicited strong cellular and humoral immune responses that resulted in protection against severe SARS-CoV-2 infection in the lungs of the K18-hACE2 mouse model.

All animal experiments were conducted in accordance with Brazilian Federal Law 11,794, which regulates the scientific use of animals, and institutional animal care and use guidelines to minimize the discomfort and suffering of animals. All protocols involving animals used in this study were approved by the ethics committee for animal experimentation at the Federal University of Minas Gerais under permit 93/2022. K18-hACE2 mice from The Jackson Laboratory [strain 034860-B6.Cg-Tg(K18-hACE2)2Prlmn/J] were housed in a ventilated rack system with standard 12-h/12-h light/dark cycles. In vivo experiments were carried out using 6- to 8-wk-old mice for each group (n = 5–7), including male and female animals evenly distributed among experimental groups. Standardized environmental enrichment (red mouse igloo and foraging) was used for animal welfare. Mice were properly transported to a biosafety level 3 facility at the University of São Paulo (Ribeirão Preto, Brazil) for SARS-CoV-2 infection and further experiments. All experiments not involving SARS-CoV-2-infected mice were conducted at an animal facility at the Federal University of Minas Gerais (Belo Horizonte, Brazil).

All peptides used in this work (termed Spep and Npep) were purchased from GenScript (Nanjing, China). Escherichia coli strains DH5α and CodonPlus (Invitrogen, Waltham, MA) were used for the cloning and expression protocols, respectively. The Mycobacterium bovis BCG Danish strain was used to generate the recombinant BCG strains (namely, rBCG-ChHsp and rBCG-ChD6) and also as a wild-type (WT) control for in vitro and in vivo immune experiments. The SARS-CoV-2 Wuhan strain was used for in vivo and in vitro assays. This WT strain (HIAE-02, SARS-CoV-2/SP02/human/2020/BRA) was isolated from a nasopharyngeal specimen of a COVID-19-positive patient from Hospital Israelita Albert Einstein, São Paulo, Brazil, which presents 99.9993% similarity to the original strain isolated in Wuhan, China (9). All reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise stated.

The data presented in this article have been submitted to the following repositories: SARS-CoV-2 HIAE-02 (GenBank MT126808.1; https://www.ncbi.nlm.nih.gov/nuccore/MT126808), S protein (National Center for Biotechnology Information reference YP_009724390.1; https://www.ncbi.nlm.nih.gov/protein/YP_009724390.1), and N protein (National Center for Biotechnology Information reference YP_009724397.2; https://www.ncbi.nlm.nih.gov/protein/YP_009724397.2).

The Chimera coding sequence (CDS) was designed as part of a functional cassette for direct cloning into E. coli– and M. bovis–specific expression vectors. The cassette contains the Chimera CDS, restriction sites for molecular handling alongside BCG regulatory regions (including a transcription terminator and a Shine-Dalgarno sequence). This cassette, herein termed “Chimera-cassette,” was cloned by GenScript (Nanjing, China) into the pET-28a expression vector between the NdeI and EcoRI restriction sites, and the coding sequences were optimized for E. coli and M. bovis codon use. The plasmid herein termed “pET-28a::Chimera” was designed in such a way that Chimera CDS is under the control of the T7 promoter and in frame with a polyhistidine-tag for further affinity chromatographic recovery. The Chimera-cassette was subcloned in-house into BCG expression vectors (pJ-Hsp60 and pJK-D6) with specific promoters described elsewhere (56). Subcloning was performed using NdeI and EcoRI restriction sites so that the Chimera CDS was under the control of each promoter for constitutive expression and in frame with the transcription terminator, so that other regulatory genetic regions would not disturb BCG expression fit. The Shine-Dalgarno sequence was inserted after the transcription terminator and in frame with an auxotrophic mark (present in plasmids pJ-Hsp60 and pJK-D6) for further experiments that would not require antibiotic resistance. E. coli DH5α was transformed with each recombinant plasmid individually (herein termed “pJ-Hsp60::Chimera” and “pJK-D6::Chimera”) and screened on Luria-Bertani agar plates containing kanamycin (50 μg/mL). Clones were confirmed by Sanger method sequencing at Fiocruz (Minas Gerais, Brazil).

NMR measurements were performed at 2 mM peptide in a DMSO-d6 solution. NMR experiments were recorded using a Bruker Avance III HD 800 MHz spectrometer (Federal University of Rio de Janeiro, Rio de Janeiro, Brazil) at 300 K. Two-dimensional experiments were recorded for sequential NMR assignment (57) and distance restriction derivation. These included ¹H-¹H total correlation spectroscopy using the decoupling in the presence of scalar interactions-2 pulse sequence (58) with water suppression. The spectral width was 7812.5 Hz, and 512 t1 increments with 40 transients of 1024 points were collected. Nuclear Overhauser effect spectroscopy spectra were acquired using a pulse sequence with water suppression by gradient tailored excitation(noesyesgpph) and a mixing time of 150 ms (59). The spectral width was 7812.5 Hz, and 512 t1 increments were collected with 40 transients with 4096 points for each free induction decay. ¹H-¹³C heteronuclear single quantum coherence spectra were collected in the processed mode with phase-sensitive, multiple-processed ¹H-¹³C heteronuclear single quantum coherence using the echo-antiecho sequence (hsqcedetgp) with F1 and F2 spectral widths of 72,010 Hz and 9,615 Hz, respectively (60). A total of 128 t1 increments with 300 transients with 2048 points were acquired. Chemical shift reference was performed with the internal control. All data were processed using TopSpin software 4.1 (academic license, Bruker), assigned using CCPNMR analysis software (61). Three-dimensional structure manipulation was performed using Chimera (University of California, San Francisco) (62). DANGLE software was used to derive dihedral constraints (63). Aria2.3 software was used for structure calculation (64). An ensemble of 50,000 structures was generated using a simulated annealing protocol as proposed by Krishnan et al. (65).

The pET-28a::Chimera construct was transformed into E. coli CodonPlus (Agilent Technologies, Santa Clara, CA) cells. For protein expression, it was cultured in 1 liter of Luria-Bertani medium supplemented with kanamycin (50 μg/ml) in 3-liter Erlenmeyer flasks on a rotary shaker at 180 rpm at 37°C until an OD_600nm of 0.8 was achieved. Gene expression was induced by adding isopropyl-β-d-thiogalactoside at a final concentration of 1 mM, and expression was carried out for 4 h at 37°C at 180 rpm. Cells were harvested by centrifugation at 3000 × g at 4°C for 15 min, and the pelleted cells were gently suspended in lysis buffer (10 mM Na₂HPO₄, 10 mM NaH₂PO₄, 0.5 M NaCl, and 10 mM imidazole). Mechanical disruption was achieved by four cycles of sonication (Branson Sonifier SLPe, Emerson Electric Co.) with pulses of 30 s with amplitude of 60% and intervals of 30 s, and the lysate was centrifuged at 3000 × g for 15 min. Since rChimera was recovered in inclusion bodies, it was resuspended in denaturing buffer (10 mM Na₂HPO₄, 10 mM NaH₂PO₄, 0.5 M NaCl, 40 mM imidazole, and 8 M urea) for solubilization. Protein recovery was achieved by affinity chromatography on an Ni-Sepharose column (HiTrap chelating 5 ml) using an AKTA explorer chromatography system (GE Healthcare, Sao Paulo, Brazil) previously loaded with appropriate buffer (10 mM Na₂HPO₄, 10 mM NaH₂PO₄, 0.5 M NaCl, 8 M urea, 40 mM imidazole, pH 7.5). Isocratic elution was performed with buffer containing 10 mM Na₂HPO₄, 10 mM NaH₂PO₄, 0.5 M NaCl, 8 M urea, 0.5 M imidazole, pH 7.5. Purified proteins were dialyzed against PBS, pH 7.0, at 4°C using a Spectra/Por membrane (MWCO 6-8,000 Da; Spectrum Medical Industries, Laguna Hills, CA). Protein concentration was determined using the BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions.

Human sera were used in ELISAs to measure specific IgG Abs against rChimera, Spep, or Npep. The protocol was adjusted from a previous report by our group (66). Briefly, microtiter plates (Sarstedt, Nümbrecht, Germany) were coated with 1–3 µg/mL of each Ag individually. Plates were incubated with sera diluted 1:40 or 1:100 in appropriate buffer. For the unpaired analysis, sera from healthy individuals harvested before the COVID-19 pandemic (in 2015) were used as negative individuals (n = 35), and sera from SARS-CoV-2-infected individuals before mass vaccination (in 2020; positive by molecular diagnosis with quantitative PCR [qPCR]) were used as positive individuals (n = 35). For the paired analysis, sera from 13 individuals (all negative by qPCR and ELISA) were assessed before and after CoronaVac immunization (3 wk after the second vaccine dose). Plate-bound Abs were detected using peroxidase-conjugated anti-human IgG (Promega, Madison, WI). Colorimetric reaction was induced by 3,3′,5,5′-tetramethylbenzidine and was stopped with H₂SO₄ solution. Plates were read at 450 nm in a microplate reader (Bio-Rad Laboratories, Hercules, CA).

The M. bovis BCG strain Danish (herein termed BCG WT) was cultured in supplemented Middlebrook 7H9 broth (0.05% Tween 80, 0.1% glycerol, and 10% oleic acid-albumin-dextrose-catalase) to midexponential phase as previously described (67). Cells were then harvested by centrifugation and suspended in 10% cold glycerol, and this procedure was performed twice for cell sensitization. BCG cells were aliquoted and stored at −80°C for in vitro and in vivo use and for electroporation of recombinant plasmids.

In order to transfect pJ-Hsp60::Chimera or pJK-D6::Chimera, 0.1 µg of each purified and dialyzed recombinant plasmid was individually electroporated into a single BCG WT aliquot with the following parameters. The recombinant plasmids and BCG WT were incubated together in 2-mm cuvettes on ice for 15 min prior to electroporation using a Gene Pulser Xcell Electroporation System (Bio-Rad Laboratories) exponential wave, a resistance of 1000 Ω, 2500-V voltage, and 25-µF capacitance. Immediately after the pulse, cells were transferred to 2-ml supplemented Middlebrook 7H9 and incubated for 16 h at 37°C and 5% CO₂ atmosphere. The culture was later seeded on supplemented selective Middlebrook 7H11 (0.5% glycerol, 10% oleic acid-albumin-dextrose-catalase, and 50 µg/ml kanamycin) agar plates and incubated at 37°C and 5% CO₂ atmosphere. A single colony for each transformant was cultured on selective supplemented Middlebrook 7H9 medium until midexponential phase. Cells were then harvested by centrifugation and suspended in saline buffer containing 10% glycerol and stored at −80°C until use.

rBCG lysate supernatant preparations were obtained as follows. Late-log phase cultures of BCG WT and rBCG were disrupted on ice with 20 sonication cycles (Branson Sonifier SLPe, Emerson Electric Co.) with pulses of 10 s, amplitude of 50%, and intervals of 10 s in lysis buffer (10 mM Tris, 1 mM EDTA, and 1% SDS). The lysate was subjected to centrifugation at 3000 × g for 20 min at 4°C, and the supernatant was collected for total protein concentration determination using a NanoDrop ND-1000 spectrophotometer. Purified rChimera and rBCG lysate supernatant preparations were analyzed by SDS-PAGE as previously described (68). Gels were subjected to electroblotting onto a nitrocellulose membrane using a semidry system (69). The membranes were blocked with TBST (TBS plus 0.05% Tween 20, pH 7.2) containing 5% nonfat dry milk for 2 h at room temperature. The membranes were then incubated with sera containing polyclonal anti-rChimera Abs (1:200) from mice in TBST for 1 h at 4°C. Membranes were later incubated with goat anti-mouse IgG conjugated to HRP (1:2000) at room temperature for 1 h. The reaction was developed using ECL Western blotting detection reagents (Amersham Biosciences, Piscataway, NJ) according to the manufacturer’s instructions and visualized in an Amersham Imager 600 device (GE Healthcare, Chicago, IL). Densitometric analyses were performed as needed using ImageJ software.

After C57BL/6 mice were sacrificed, femurs and tibias were aseptically removed and polished free from adherent tissues. Collected bones had their ends cut, and bone marrow was flushed out with PBS using a 23-gauge needle. BMDMs were generated by culturing the bone marrow cells in supplemented media in petri dishes for 7 d as previously described (67). Briefly, culture medium was changed at day 4, and, at the seventh day of cell differentiation, cells were obtained completely differentiated in macrophages. BMDMs (5 × 10⁵ cells/well) were seeded in 24-well plates (Thermo Fisher Scientific) for 24 h in supplemented media without antibiotics prior to BCG infection. Before use, BCG WT, rBCG-ChHsp, and rBCG-ChD6 aliquots were thawed and passed through a 27-gauge needle 10 times to disrupt bacteria clumps. Infection was carried out with BCG WT, rBCG-ChHsp, or rBCG-ChD6 at a 5:1 multiplicity of infection ratio. At 4 h postinfection (hpi), the monolayer of cells was washed with fresh medium to remove extracellular bacteria, and then supplemented medium was added. Infected cells were either lysed (4 hpi analysis) or incubated at 37°C with 5% CO₂ atmosphere for 72 h (72-hpi analysis). Lysis was performed with saponin 0.1% for 20 min, and cell lysates were plated onto Middlebrook 7H11 agar plates serially diluted. CFUs were determined within 2–3 wk.

Groups of 6- to 8-wk-old K18-hACE2 mice (n = 5–7) were immunized s.c. in the nape of the neck by the following prime-boost regimen. At day 0, mice were administered PBS, BCG WT, or rBCG-ChD6 (1 × 10⁶ CFU/animal). Four weeks later, mice were boosted with 25 μg rChimera associated with 100 μg Al(OH)₃ (Alum-Alhydrogel adjuvant 2%; InvivoGen, San Diego, CA), boosted with PBS formulations containing 100 μg of Alum, or boosted with rBCG-ChD6 (1 × 10⁶ CFU/animal). Twenty days after the second immunization, mice were either (1) challenged with SARS-CoV-2 (2 × 10⁴ PFU) intranasally at a biosafety level 3 facility or (2) sacrificed, and the spleens were harvested for cytokine analysis. Independent experiments were carried out for both outcomes (SARS-CoV-2 challenge or euthanasia for spleen collection). Challenged animals were observed daily for body weight assessment and at day 6 postinfection were sacrificed for blood and lung collection. Lungs were harvested and processed for viral load determination.

Sera were collected from all mice in each experimental group at day 50 before euthanasia by submandibular bleeding. In order to assess the levels of anti-rChimera Abs, indirect ELISA was performed as previously described (66). Briefly, microtiter plates (Sarstedt) were coated with rChimera (10 µg/ml). Serum samples were serially diluted and incubated at room temperature for 1 h. Plate-bound Abs were detected using peroxidase-conjugated anti-mouse total IgG (1:2000) or its isotypes IgG1 (1:2000) and IgG2c (1:5000) (Promega) diluted in PBS. The colorimetric reaction was induced by 3,3′,5,5′-tetramethylbenzidine, and sulfuric acid was used to stop the reaction. Absorbance was detected at 450 nm in a microplate reader (Bio-Rad Laboratories). Isotype switch was calculated by IgG2c/IgG1 ratio using detected absorbances at 1:100 dilution. Endpoint titers were estimated by detecting the dilution of each sample that reached the average of the control sera ±2 SD units.

Plasmids and HEK293T/ACE2 cells for generation of the pseudotyped virus neutralization assay were kindly provided by Drs. Paul Bieniasz and Frauke Muecksch at Rockefeller University (New York, NY) (70). Sera from the immunized mice were incubated for 1 h with an HIV-1-based pseudovirus at different dilutions. Pseudovirus sera solutions were used to infect 293T/ACE2 cells in 96-well plates, and NanoLuc (relative light units [RLU]) activity was detected. Neutralization rate was calculated using the following equation: [1 − (MRS − MRNC)/(MRPC − MRNC)] × 100, where MRS = mean RLU of the samples, MRNC = mean RLU of the negative control (cells without virus inoculation), and MRPC = mean RLU from the positive control (cells inoculated with virus without any serum). The readings from MRPC were used to assess the limit of detection of the experiment. Results were plotted as neutralizing rate and compared with the control group (samples incubated with sera from mice previously immunized with CoronaVac vaccine). Neutralizing assays were performed using the Wuhan WT strain as well as the Gamma, Delta, and Omicron variants.

Spleens of mice were harvested 20 d after boost with rChimera or PBS in formulation with alum, and spleen cell culture was performed as previously described (66). Briefly, spleens were mechanically disrupted, treated with ammonium-chloride-potassium buffer, washed, and resuspended in RPMI supplemented with 10% FBS and 1% penicillin-streptomycin. Ninety-six-well plates were seeded with 1 × 10⁶ cells/well, and cultures were treated with polymyxin B (30 μg/ml). Spleen cells were further stimulated with rChimera (25 μg/ml) and with Con A (ConA 5 μg/ml) or LPS (1 μg/ml) as positive controls. Cytokine levels were assessed using the Duoset ELISA kit (R&D Systems, Minneapolis, MN) according to the manufacturer’s directions. Readings from the analytical standard curve were used for assessment of the experimental limits of detection. Spleen cell culture supernatants were assayed for IL-6 and IFN-γ (after 24 and 72 h of stimulus, respectively).

Lungs were harvested at day 6 postinfection and homogenized in TRIzol (Thermo Fisher Scientific) reagent using a TissueLyzer LT (Qiagen, Hilden, Germany) to isolate total RNA according to the manufacturer’s instructions. Real-time qPCR was conducted for viral load determination using the TaqMan Fast Virus 1-Step Master Mix kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s recommendations. The reaction was performed as previously described (71), and the primers and probes used are described in Table I. Samples were analyzed in duplicate and compared with a standard relative curve using viral stocks from cell culture of SARS-CoV-2, the serial 10-fold dilution starting from 100 ng of viral RNA, and the limit of detection was established at a cycle threshold of 40 equivalent to 0.0001 ng of total RNA. PCR was performed in a QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific). Viral load cycle threshold values were interpolated to the standard curve and presented as log₁₀ of the RNA nanogram.

At 6 d postinfection, lungs were harvested and weighed, then disrupted in a TissueLyzer LT (Qiagen) after appropriate weight/volume standardization. Homogenates were centrifuged at 14,000 × g for 10 min at 4°C, and supernatant containing virus was collected and serially diluted to infect VERO-E6 cells previously seeded on 48-well plates (5 × 10⁴ cells/well) as previously described (71). Briefly, infection was carried out for 1 h, and supernatant was discarded. Carboxymethyl cellulose 4% was added to each well, and plates were incubated for 4 d at 37°C with 5% CO₂ atmosphere for plaque formation. Cells were fixed with 4% formaldehyde solution and stained with crystal violet. PFUs were counted in all dilutions and applied to the formula (PFU mean counts)/(dilution × used virus volume). Viable viral titer results are represented as PFU/g, and the limit of detection was the maximum dilution in which PFUs could be counted in the positive control wells.

At 6 d postinfection, lungs were harvested and fixed in phosphate-buffered 10% formaldehyde for 48 h and later dehydrated, diaphanized, and embedded in paraffin. Tissue sections measuring 3–4 μm were stained with H&E for further histological analyses. Overall lung architecture was analyzed in the slides, and 15–20 representative images for each slide were captured using an Olympus CX31 optical microscope attached to a JVC 1270/RGB microcamera using 10× or 20× magnification lens. Cell infiltration was depicted in representative pictures. The free alveolar space was measured using ImageJ software. Briefly, captured digital images were used to create binary images, and pixels were later converted into square micrometers. Free area was measured and expressed in relation to the total area of each section for percentage calculation. For free alveolar space determination, vascular vessels, bronchi, and bronchioles were excluded from the analysis.

All statistical analyses were performed using the Prism software package (GraphPad Software, La Jolla, CA). Kolmogorov-Smirnov normality tests were applied. Analyses with two groups were performed using the Student t test for parametric data or the Mann-Whitney U test for nonparametric data. Analyses with more than two groups were performed with one-way or two-way ANOVA, and Tukey or Bonferroni adjustments were included for multiple comparisons. p Values obtained by these methods were considered significant when less than 0.05. For humoral assays, ELISA data were submitted to receiver operating characteristic curve analysis to determine the cutoff points when necessary.

The rational design of an immunogenic vaccine against SARS-CoV-2 by immunoinformatic analysis of viral structural proteins has previously been reported by our group (12, 14). The N and S proteins were screened for major B and T cell epitopes for potential targets, and the most immunodominant epitopes were selected for further characterization. We selected two N- and S-derived peptides for characterization based on previous analysis. Peptide 1 (Fig. 1) comprises epitopes identified in the N-terminal domain (NTD) region of the N protein. This region is highly conserved and has high potential for T cell cross-reactivity with N proteins from other human coronaviruses. Another N-derived peptide (peptide 2 in Fig. 1) comprises epitopes from an unstructured serine-arginine-rich region between the NTD and C-terminal domain (CTD) regions and represents a strong immunodominant B cell epitope. The selected S-derived peptides (peptides 3 and 4 in Fig. 1) are both epitopes identified in the receptor binding domain region with high binding affinity for multiple human MHC classes I and II alleles. This region is also predicted to have several B cell epitopes that are likely to elicit a strong Ab response mediated by T cell help. Taken together, these immune features predict highly immunogenic potential for the use of these peptides in a chimeric multiepitope vaccine.

In order to validate our immune predictions and also to gain further insights into preferential structural conformations, peptides 1 and 3 (Fig. 1), herein termed Npep and Spep, respectively, were chemically synthesized and used in further analyses (Fig. 2). The three-dimensional structures were solved in DMSO to evaluate the peptide preferred conformations in a medium able to mimic the cellular environment. Both peptides were found to lack mid- and long-range constraints. The Npep forms a concavity in the peptide center, exposing residues 11P, 12R, 13W, 14Y, and 16Y, and seems to be highly flexible in the extremities (Fig. 2A). Similarly, Spep structures appear to form a concavity in this region, exposing some residues (14R, 16F, 17R, and 18R), whereas the NTD and CTD regions are more flexible (Fig. 2B). The best Spep root mean square deviation values are in the region of residues 9 to 10. The Spep concavity exposes an aromatic residue surrounded by highly positively charged residues, whereas the Npep exposes a positive large arginine surrounded by aromatic tyrosines and a tryptophan.

Predicted immunogenicity of Npep and Spep was assessed by indirect ELISA using human sera. Sera from individuals with a positive qPCR diagnosis for SARS-CoV-2 before mass vaccination programs (i.e., collected in June 2020, so individuals had no vaccine-induced Abs) were used as positive controls, and sera from healthy individuals (obtained before the COVID-19 pandemic) were used as negative controls. Both Npep (Fig. 2C) and Spep (Fig. 2D) were revealed to be reactive to sera from SARS-CoV-2-infected patients and nonreactive to sera from negative healthy individuals, validating our previous immunoinformatic analysis and highlighting their potential use in vaccine formulations.

The immunogenic peptides characterized in the N and S proteins from SARS-CoV-2 were used to guide a rational design of a chimeric recombinant protein suitable for use in vaccine formulation, herein termed “rChimera.” This recombinant protein was designed to contain two domains (Fig. 3A). Its NTD (consisting of N-derived peptides, herein termed “N-domain”) is thought to elicit strong humoral and cellular responses against SARS-CoV-2 and potentially to evoke cross-reactivity against other human coronaviruses (including peptides 1 and 2 listed in Fig. 1). This first domain comprises 161 aa, incorporating the NTD and also a very flexible, nonuniform, serine-arginine-rich region from the serine-arginine rich domain (peptide 2) as a linker to the second domain (S domain) of the rChimera. Its CTD (consisting of S-derived peptides, herein termed “S-domain”) exhibits several B cell epitopes, and it is designed to induce neutralizing Abs against the receptor binding domain from S, impairing viral attachment and subsequent genomic RNA delivery. This second domain comprises 294 aa, including peptides 3 and 4 listed in Fig. 1. The rChimera protein was designed to have 455 aa in length (52.2 kDa).

The coding sequence of rChimera was optimized for heterologous expression in E. coli and M. bovis systems as described in Materials and Methods. After heterologous expression in E. coli and affinity chromatographic recovery, the reactivity of rChimera against human sera was demonstrated by indirect ELISA (Fig. 3B) and was clearly able to distinguish SARS-CoV-2 naturally infected individuals from healthy donors. Interestingly, it was also reactive to the sera from CoronaVac-immunized individuals (who had not previously been infected), demonstrating the rChimera protein is also recognized by Abs generated from inactivated whole-virus immunization (Fig. 3C). Once rChimera was recognized by anti-SARS-CoV-2 Abs, its CDS was subcloned into suitable plasmids for expression in BCG.

The Chimera CDS was cloned in two different plasmids, pJ-Hsp60 and pJK-D6, under the control of two promoters, P_Hsp60 or P_D6, both previously described elsewhere (56), and the details are described in the Materials and Methods section. These constructs (Supplemental Fig. 1A) were designed with regulatory sites after Chimera CDS, containing a transcription terminator site and an auxotrophic marker (LysA). These molecular elements were added so that, in further experiments, the transcription terminator site could be enzymatically removed, and the plasmids would be able to be maintained by auxotrophic complementation in an appropriate BCG strain without requiring antibiotic selection. Recombinant plasmids were transformed into BCG Danish strain (herein termed “BCG WT”), and recombinant BCGs were handled accordingly. Two recombinant strains were generated and cultured for further experiments: rBCG-ChHsp and rBCG-ChD6, containing the recombinant plasmids with Chimera CDS under the control of P_Hsp60 and of a high-strength promoter (P_D6, obtained through error-prone PCR of the strong P_L5 promoter, originally from mycobacteriophage L5), respectively. Late-log phase cultures were disrupted by sonication in appropriate lysis buffer, and lysate supernatants containing soluble rChimera were analyzed by Western blotting (Supplemental Fig. 1B). Densitometric analysis demonstrates that the rBCG-ChD6 strain expresses 1.8-fold higher levels of rChimera than rBCG-ChHsp, observed at ∼52 kDa. As expected, BCG WT control did not reveal a protein band at this molecular mass under the incubation with sera from rChimera-immunized mice. The differential growth of the recombinant strains was assessed in vitro by incubation with BMDMs at a multiplicity of infection of 5:1 (Supplemental Fig. 1C). Bacterial growth was monitored at 4 and 72 hpi. Despite a slight significant increase in the uptake of rBCG strains by the macrophages when compared with BCG WT at 4 hpi, this phenotype was not observed at 72 hpi. Altogether, these results indicate that both rBCG strains replicate well in macrophages. As rBCG-ChD6 expresses higher levels of rChimera, this strain was selected as a suitable candidate for in vivo analysis and further use in vaccine formulation.

In order to assess the efficacy of our designed chimeric protein for vaccine formulation, a prime/boost immunization strategy was used (Fig. 4A). K18-hACE2 mice were immunized s.c. with BCG WT or rBCG-ChD6 or with PBS as a control. Four weeks later, mice were boosted with a formulation containing rChimera or PBS as a control (both with alum). Twenty days later, serum was harvested to measure total IgG responses, IgG2c/IgG1 ratio, neutralizing Abs, and cytokine (IFN-γ and IL-6) responses.

Sera from immunized mice were assayed by ELISA (Fig. 4B), and a strong humoral response was detected (at day 50) in the animals immunized with rBCG-ChD6 and boosted with purified rChimera associated with alum when compared with the other studied groups. Mice previously immunized with BCG WT or PBS before rChimera boost presented less anti-rChimera Ab titers when compared with the rBCG-ChD6 group. Significant Ab levels were not detected prior to the booster dose for all groups at day 30 (data not shown). We also assessed the Ab isotype profile measuring IgG2c/IgG1 ratio in vaccinated mice. We observed an increased IgG2c/IgG1 ratio in rBCG-ChD6-immunized mice compared with the other vaccinated groups (Fig. 4C). Because the IgG2c Ab isotype correlates with a Th1 type of immunity, an early IgG2c response could suggest an effective neutralizing Ab and cellular responses against SARS-CoV-2 in rBCG-ChD6-immunized animals. In parallel, we observed higher IFN-γ (Fig. 4D) and IL-6 (Fig. 4E) production by splenocytes of rBCG-ChD6/rChimera and BCG WT vaccinated groups when compared with animals that received rChimera or PBS with alum. Even though alum is known as an adjuvant that primes host immune responses to Th2, BCG has the ability to induce a strong Th1 immunity. It has previously been demonstrated that T cell responses in recovering COVID-19 patients are predominantly Th1 driven (72). Furthermore, the rBCG-ChD6/rChimera + Alum vaccinated group presented increased levels of neutralizing Abs against SARS-CoV-2 Wuhan strain when compared with the BCG WT/rChimera group (Fig. 4F). Alternatively, the same sera were assayed against SARS-CoV-2 VOCs but resulted in a decreased neutralizing rate for the Gamma variant and failed to neutralize the Delta and Omicron variants (Fig. 4F). Collectively, these results demonstrate the rBCG-ChD6/rChimera + Alum–vaccinated animals engendered a type of immune response that correlates with protective potential against COVID-19, but further investigation might be required regarding viral VOCs.

Next, we evaluated the protective potential of the rBCG-ChD6/rChimera + Alum immunization regimen employing a murine model of SARS-CoV-2 infection (Fig. 5A). Twenty days after the boost immunization, K18-hACE2 mice were intranasally inoculated with 2 × 10⁴ PFU of SARS-CoV-2. Control groups, comprising PBS/PBS + Alum and PBS/rChimera + Alum, demonstrated severe clinical deterioration beginning at day 4 postinfection, and animals were very weak at the sixth day of observation (Fig. 5B). This is suggestive that a single dose of rChimera protein might not have been sufficient to elicit a protective response against SARS-CoV-2 infection. A similar phenotype was also observed in the animals immunized with BCG WT/rChimera + Alum, suggesting the immune response elicited by BCG immunization also was not enough to protect against the infection, even with an rChimera protein booster dose (Fig. 5B). Remarkably, rBCG-ChD6-immunized mice demonstrated no apparent weight loss when observed for 1 wk (Fig. 5B), indicating that the immune response induced by this vaccine played an important role in the protection observed.

At day 6 postinfection with SARS-CoV-2, mice were euthanized, and their lungs were harvested for further analysis. Bulk RNA extracted from the tissues was used as a template for RT-qPCR determination of the viral load (Fig. 6A) using specific primers for SARS-CoV-2 structural proteins (Table I). The rBCG-ChD6-immunized group presented a marked decrease in the amount of detected viral RNA in the tissues from infected lungs, and statistically significant differences were observed for all assessed groups. Although the other experimental groups apparently presented less RNA viral load, no significant difference was detected (p = 0.2247 for BCG WT/rChimera + Alum immunized group versus PBS control group; p = 0.234 for PBS/rChimera + Alum immunized group versus PBS control group). Additionally, the PFU analysis revealed that all three vaccinated groups demonstrated reduced viral viability when compared with the PBS control group (Fig. 6B). However, full viral impairment in infected cells was observed in the rBCG-ChD6-immunized group, where no PFU was detected. Regarding lung pathology associated with infection, rBCG-ChD6/rChimera + Alum–immunized mice presented overall pathologic amelioration at 6 d postinfection, with reduced inflammatory infiltrates in the alveoli (Fig. 7A–7D), leading to alveolar space preservation to a greater extent (Fig. 7E). Alternatively, we observed that mice which received two doses of rBCG-ChD6 as a homologous prime boost strategy failed to be protected against viral infection. Body weight loss (Supplemental Fig. 2A) and viral load of rBCG-ChD6-vaccinated mice were not statistically significant (Supplemental Fig. 2B, 2C) when compared with mice immunized with PBS, highlighting the importance of an rChimera boost for the development of an effective and protective immune response. Collectively, these findings demonstrate that mice immunized with a single dose of rBCG-ChD6 and boosted with purified rChimera associated with alum are protected against COVID-19 clinical disease manifestations due to an early and rapid viral control and reduced pathology.

The COVID-19 pandemic has had a tremendous impact on society, with a high number of deaths and economic loss. A critical issue is ensuring the adequate supply of vaccines to low- and middle-income countries. The mRNA vaccines, for instance, require maintenance at low temperatures and have complex logistical requirements. An interesting strategy is to reposition existing licensed vaccines for use against COVID-19. In this context, M. bovis BCG, the most frequently applied vaccine worldwide, has drawn particular interest (73). Despite its original purpose as a tuberculosis vaccine, numerous studies have demonstrated the potential of BCG to provide effective protection against heterologous infections. Particularly, BCG vaccination is known to provide protection against other respiratory infections by inducing trained immunity (39). Our hypothesis was that recombinant BCG expressing immunodominant peptides from SARS-CoV-2 could be an effective strategy to induce protection against COVID-19 by combining the adjuvant-like properties of BCG, its ability to induce strong cellular and trained immunity, and anti-SARS-CoV-2-specific neutralizing Abs.

In this study, we describe the construction of a recombinant BCG-based vaccine that enables complete elimination of viable SARS-CoV-2 in the lungs of K18-hACE2 mice. We have designed an N/S chimeric protein based on highly immunogenic peptides identified by bioinformatic analysis (12, 14). These peptides were predicted to elicit strong humoral and cellular immune responses against two key structural proteins from SARS-CoV-2. The structures of these peptides were defined by NMR, and our results show that the N-peptide exhibits the typical random shape of spectra previously observed for longer β-sheet-forming peptides (74). These preferred conformations of the N-peptide correlate with the immunogenic character of T cell epitopes, which were generally found to be disordered structures with preferred conformations in specific regions (75, 76). However, we observed that the S-peptide appears to have a conformational folded preference that resembles those found in peptides bound to Ab molecules (77). The interaction between peptide and Ab is guided by many complex factors, and the structure and stability of the peptide in solution are among the most important features, demonstrating the importance of having these peptide domains in the chimeric recombinant protein.

The strategy here was to target not only the S protein but also key regions from the N protein, which could also confer broader cross-recognition of immunogenic peptides with other betacoronaviruses (12). A recent study demonstrates that cross-protective immunity could be of interest when assessing SARS-CoV-2 vaccines (78). Because the S protein is highly mutated in different viral variants and may account for loss of Ab-mediated protection as new variants appear (79), incorporating N-derived peptides in our vaccine design was an important strategy. The N-epitopes selected in this study were predicted to elicit T-cell-mediated responses, activating another arm of the immune system and not relying solely on neutralizing Abs. The rChimera protein was recognized by the sera of SARS-CoV-2-infected patients and by CoronaVac-immunized individuals, demonstrating its antigenic properties. K18-hACE2 mice were immunized with the rBCG-ChD6 or BCG WT as a control and boosted with rChimera protein associated with alum, with the aim being to induce strong humoral and cellular immune responses to N- and S-peptides. Recently, proteomic studies demonstrated that M. bovis BCG and SARS-CoV-2 have significant peptide homology (80, 81). Eggenhuizen et al. reported that CD4⁺ and CD8⁺ T cells in vitro primed with BCG-derived peptides present cross-reactivity to SARS-CoV-2 peptide homologs, which could partially account for the epidemiological observations of BCG-vaccinated individuals being partially protected against SARS-CoV-2 infection. Furthermore, BCG induces nonspecific heterologous protection due to “trained immunity” by reprogramming innate immune responses (39). These findings reinforce the use of an rBCG-system for the delivery of SARS-CoV-2-derived Ags to elicit efficient immune responses.

We observed that immunization with rBCG-ChD6/rChimera + Alum elicited the highest anti-rChimera total IgG and IgG2c Ab titers with neutralizing activity to SARS-CoV-2 Wuhan strain in prechallenge experiments when compared with BGC WT or with rChimera vaccinated groups. Additionally, rBCG-ChD6 or BCG WT immunized animals presented higher levels of IFN-γ and IL-6 by spleen cells than those immunized with rChimera with alum, highlighting the importance of the BCG-based system in evoking strong cellular responses. A similar profile has been seen in mice vaccinated with a recombinant BCG expressing an HIV-1 p24 Gag (82), which led to a Th1-like cellular response and higher IgG2a/IgG1 ratio (83, 84). Recently, another recombinant BCG expressing SARS-CoV-2 N protein was reported (85) to induce a Th1-cell response when spleen cells from immunized mice were stimulated with the N protein. However, these investigators showed that a single dose of rBCG expressing N Ag was insufficient to elicit a strong humoral response. This reinforces the need of a booster dose to enhance Ab titers with neutralizing activity in the rBCG platform vaccine. We also observed that a homologous prime boost protocol based only on rBCG-ChD6 failed to protect mice from SARS-CoV-2 infection, which corroborates the importance of a purified rChimera protein booster dose to induce a more robust and effective immune response.

In addition to several clinical trials being conducted to evaluate the impact of BCG vaccination against COVID-19, several groups have been assessing the heterologous protective effects of BCG vaccination on SARS-CoV-2 infection in a K18-hACE2 murine model, but divergent findings have been reported. Kaufmann et al. (86) have observed that BCG vaccination fails to protect against SARs-CoV-2. In their work, K18-hACE2 mice were immunized s.c. or i.v. with BCG before challenge, but no differences in morbidity or mortality were observed when compared with the BCG unvaccinated group. The same phenotype was observed in Syrian golden hamster model, which naturally expresses the ACE2 receptor. In contrast, Hilligan et al. (87) observed a different outcome for i.v. inoculation of BCG. In their work, they observed that mice immunized i.v. (but not s.c.) with BCG were protected against lethal SARS-CoV-2 challenge, with viral load being significantly reduced in the lungs from K18-hACE2 mice. They suggest that this outcome was due to changes in both the composition and the function of pulmonary cellular components induced by BCG, because i.v. inoculation allows the Mycobacterium to reach the lungs, inducing innate immune alterations. Similar findings were reported by Zhang et al. (47), who found that protection was achieved by i.v. inoculation of BCG. Despite the use of different BCG strains (Tice, Pasteur, and Tokyo, respectively) or different SARS-CoV-2 doses (1 × 10³ median tissue culture-infective dose [TCID₅₀], 4 × 10³ TCID₅₀, 10⁴ TCID₅₀, or 1.25 × 10⁴ PFU) employed in these studies, the key feature that accounted for protection was the i.v. inoculation of BCG; however, it is not a clinically accepted practice in humans.

In our study, immunizations were performed s.c. because this route is similar to intradermal BCG vaccination performed in humans, which makes this model more suitable for ready implementation in a clinical setting. Here, post-challenge analysis demonstrates that rBCG-ChD6/rChimera + Alum–immunized mice showed no apparent clinical manifestation of COVID-19, featured by no weight loss at day 6 postinfection and overall pulmonary alveolar preservation at histopathologic analysis, but they also showed diminished viral load detected in the lungs by qPCR and no viable virus detected by PFU analysis when compared with the BCG WT and rChimera control groups. These data together point to a prompt neutralizing response against SARS-CoV-2 elicited in rBCG-ChD6/rChimera + Alum–vaccinated animals, which led to full eradication of viable virus in the lungs of the mice. The neutralizing activity to SARS-CoV-2 Wuhan strain from Abs derived from rChimera’s S-domain could have had a significant impact on the early control of the infection. Together, a combination with a Th1 type of cellular response induced by vaccination (presumably by both N- and S-domains from rChimera) leads to the protective response observed against viral challenge; however, further investigation is required to define the mechanisms involved in protective immunity. Interestingly, although the BGC WT-immunized group has presented a significant decrease in viable virus in the lungs, mice had lost 20% of their body weight at day 6 postinfection. It is important to mention that despite this group having been boosted with rChimera with alum, they showed a severe deteriorated phenotype comparable to the PBS control group. Similar results have been reported by Counoupas et al. (45). In their study, K18-hACE2 mice were s.c. immunized with a single formulation containing PBS or BCG and later challenged with SARS-CoV-2. Their findings revealed that both these vaccinated groups presented body weight loss and elevated viral titers in the lungs, with no statistical difference. Protection was achieved only by immunizing mice with a formulation containing a mix of BCG WT, full-length trimeric S and alum.

In summary, our study revealed that a prime-boost immunization system based on an rBCG-ChD6 expressing a chimeric protein (rChimera) derived from the SARS-CoV-2 Wuhan strain followed by vaccination with rChimera protein plus alum fully protects mice against viral challenge. A booster dose of rChimera protein with alum provided efficient neutralizing Ab rates against the SARS-CoV-2 Wuhan strain, which led to prompt and early control of infection. This immunization system relies on the BCG platform and alum adjuvant, both clinically approved for use in human practice, which would confer rapid incorporation into healthcare systems where BCG is currently employed. Despite failing to neutralize the Delta and Omicron variants and with an inferior neutralization rate to the Gamma strain, further investigation is required to determine the effectiveness of this BCG-based system against the most recent SARS-CoV-2 VOCs, which could culminate in updating the rChimera protein by incorporating VOC immunogenic epitopes into the construct.

The authors have no financial conflicts of interest.