A Porcine-Isolated Mycobacterium bovis Strain Exhibits Hypervirulence in a Murine Pulmonary Tuberculosis Model

TL;DR

A pig-isolated strain of Mycobacterium bovis shows higher virulence than a known virulent strain in a mouse model of tuberculosis.

Contribution

Identification of a hypervirulent M. bovis strain from pigs and its distinct genomic and pathogenic features.

Findings

MB894 caused more severe disease and higher organ colonization than MB303 in mice.

Genomic analysis revealed candidate genes potentially linked to MB894's hypervirulence.

MB894 induced higher cytokine levels and greater dissemination compared to MB303.

Abstract

This study provides valuable insights into the pathogenic characteristics and immune response associated with Mycobacterium bovis strains, particularly focusing on a strain isolated from domestic pigs in Argentina. The research compares the local strain M. bovis strain 894 (MB894) with the highly virulent M. bovis MB303 strain from wild boar, contributing to understanding how M. bovis adapts and exhibits varying levels of virulence in different contexts. We used a murine infection model that allowed us to comprehensively evaluate several parameters critical to understanding the strain’s pathogenicity and performed whole genome sequencing to identify candidate genes that could explain the molecular mechanisms underlying the virulence of the strain. Further studies are needed to explore the virulence factors identified through genomic analysis.

Mycobacterium bovis is the causative agent of tuberculosis (TB) infecting a wide range of animal hosts, including humans. Domestic pigs (Sus scrofa domestica) are susceptible to different mycobacteria, particularly species within the Mycobacterium avium complex (MAC). However, in countries where bovine TB is endemic, such as Argentina, M. bovis is the most frequently reported species in pigs. This study aimed to evaluate the immune response and disease progression of a local strain (MB894) isolated from pigs and compare its pathogenicity with the highly virulent strain MB303, isolated from wild boar. Additionally, we sought to explore the genomic basis underlying the virulent phenotype of MB894. For this purpose, a murine infection model was used to assess pathogenicity, organ colonization, dissemination and cytokine induction. Whole-genome sequencing was performed to identify genetic features, including non-synonymous SNPs and INDELs, potentially associated with virulence. The severe immunopathogenesis produced by MB894, the higher multiplication rate in the evaluated organs, and the greater dissemination to other organs compared to MB303, combined with the cytokine levels induced by this strain, prompted us to classify MB894 as a hypervirulent strain. Genomic analysis revealed candidate genes that may be virulence factors contributing to this phenotype. In summary, MB894 represents a hypervirulent M. bovis strain with distinct pathogenic and genomic characteristics. These findings provide insights into the molecular determinants of virulence and highlight the need for further evaluation of identified gene candidates.

Figures

Full text

1. Introduction

Mycobacterium bovis is the causative agent of tuberculosis (TB) in cattle, a disease that accounts for economic losses and animal welfare. It also represents a threat to public health since M. bovis can infect humans and cause pulmonary disease that is clinically indistinguishable from Mycobacterium tuberculosis, the main agent of human TB [1]. M. bovis belongs to the M. tuberculosis complex (MTBC), which comprises ten mycobacterial species that cause TB [2] and can infect a wide range of animal hosts, including possums, voles, deer, badgers, seals, wild boars and pigs [3]. These hosts can be considered primary or maintenance hosts, or secondary or spillover hosts, depending on the epidemiology of the disease in the specific region [4,5,6,7]. Domestic pigs (Sus scrofa domestica) are susceptible to different mycobacteria. In Argentina, where bovine TB is endemic, the role of M. bovis in swine TB might have a substantial impact. Conversely, in countries where bovine TB is not endemic, Mycobacterium avium complex species become relevant [7,8,9,10].

In previous work, we isolated an M. bovis strain from TB-like lesions (TB-LL) in pigs at a slaughterhouse (MB894) and reconfirmed its virulence in an experimental infection trial in pigs [11]. In this work, we aim to evaluate the immune response induction and disease progression of MB894 in a mouse model in comparison with the highly virulent strain MB303 isolated from wild boar [12].

Genomic sequencing has been recently used to understand M. bovis transmission and evolution [13] and to explore the circulating genotypes, phylogenetic relatedness and underlying virulence and drug resistance genetic markers [14]. Although MTBC strains are highly conserved, they exhibit a variable degree of pathogenicity [15,16,17]. Increasing evidence indicates that genetic variation among mycobacterial isolates influences immune response and disease progression. This variability in MTBC pathogenicity is probably caused by the diversity in genetic and evolutionary backgrounds [18,19]. Therefore, it is necessary to understand the role of the genetic diversity of different pathogenic strains of M. bovis in disease development.

In this work, we performed whole-genome sequencing of MB894 and compared it with the genome of MB303 to identify features in the coding sequences, such as nonsynonymous SNPs and INDELs, potentially underlying the high virulence of MB894 observed in mice and previously in pigs.

2. Materials and Methods

2.1. M. bovis Culture and Inoculum Preparation

M. bovis strain MB894, spoligotype SB0153, was isolated in our laboratory from TBTB-LL from pigs at a slaughterhouse and only cultured for a few passages [11]. M. bovis strain MB303 was isolated from a wild boar in the province of La Pampa in Argentina in 2004 and characterized as spoligotype SB0140 [12]. This strain was kindly provided by Zumarraga, MJ, and Bigi, FB.

M. bovis strains were grown in Middlebrook 7H9 medium (BD DIFCO^TM^ Laboratories, Franklin Lakes, NJ, USA) supplemented with albumin (A) 0.5%, dextrose (D) 0.4%, pyruvate (P) 0.5% and Tween 80 (T) 0.05% (M7H9-AD-P-T) or Middlebrook 7H10 (Difco Laboratories, 262,710) supplemented with AD-P, for 3–4 weeks. Cells were harvested by centrifugation at 2500× g for 20 min and washed twice in sterile PBS. The bacterial pellet was re-suspended in PBS and declumped by 20 passages through a 25 G needle syringe. The concentration of this suspension was adjusted to an optical density of 10^3^ bacteria/mL, considering the formula DO600: 0.1 = 10^7^ CFU/mL [20].

2.2. Experimental Infection in a Pulmonary TB Mouse Model

The animal study protocol was approved by the Institutional Animal Care and Use Committee (CICUAE), the Statistical Subcommittee from the CICUAE and under the regulations of the Ethical Committee of the National Institute of Agricultural Technology (INTA), Protocol N°19/2019. We have used the experimental BALB/c model of progressive pulmonary TB that has been described in detail previously [21,22]. BALB/c female mice were acquired from the animal facility of the School of Veterinary Sciences, National University of La Plata, Buenos Aires, Argentina. A total of 31 animals, from 18 to 22 g in weight and 4 to 6 weeks of age, were adapted to the facilities for one week. Experimental groups (n = 12) and a control group (n = 7) were established, where each subject constituted a single experimental unit. Each group was housed in separate and labeled cages fitted with microisolators connected to negative pressure. Throughout the experiment, animals were provided with food and water ad libitum and maintained in an enriched environment. The animals were monitored twice daily by Med Vet Dr Cuerda and Med Vet. Dr Colombatti. Experimental groups were infected with MB894 and MB303, respectively, and a negative control group was inoculated with sterile PBS. The infections were performed inside the biosafety level 3 (BSL3) laboratories. Each animal was intratracheally inoculated with 0.1 mL of the inoculum (1 × 10^2^ CFU) [12].

2.3. Virulence Assay

The degree of virulence and immune response of the M. bovis strains MB894 and MB303 were analyzed at 1 (T0) and 45 (T45) days post-infection (dpi) by necropsy and the lungs, spleen and liver from all the animals were removed for colony-forming units (CFU) and cytokine determinations. During the necropsy, total body and individual organs’ weights were recorded. A thorough postmortem examination was performed to detect the presence of macroscopic lesions. The cytokine profile induced by each of the strains was determined at T45 in the cell suspension of splenocytes stimulated with bPPD and M. bovis total extract. The histopathological analysis of the livers, lungs and spleens was performed using hematoxylin/eosin (H/E) staining. The group inoculated with PBS was euthanized at T45 as a negative control for histopathology and cytokine analysis.

2.4. Colony Forming Units

Before infection, the inoculum was controlled for contamination on blood agar plates, and the inoculum doses were determined by serial dilutions and CFU counting on 7H10 plates supplemented with AD-P. The infective dose was estimated by the CFU recovered from the lungs and spleens of five mice per group at 1 dpi (T0). Seven mice per infected group and the negative control group were euthanized at 45 dpi (T45). The right lung and the spleen were removed and homogenized in 1 mL of RPMI 1640 without antibiotics (Invitrogen). Serial dilutions of each homogenate were spread onto duplicate plates and incubated at 37 °C. CFUs were determined 6 to 8 weeks later.

2.5. Quantification of Cytokines

The remaining of the splenocytes suspensions were used for cytokine quantification. One hundred µL of the homogenized splenocytes obtained at T45 were treated with ammonium chloride for erythrocyte lysis and subsequently concentrated. Splenocyte viability was determined by Trypan Blue staining using a Neubauer chamber. Splenocytes (1 × 10^6^/well; in RPMI 1640 supplemented with 10% fetal calf serum, 100 U of penicillin/mL and 100 μg of streptomycin/mL), were incubated in 96-well round-bottom plates at 37 °C in a 5% CO_2_ chamber and stimulated with: 20 μg/well of bPPD (Bovine PPD BOVIGAM^®^, Thermo Fisher Scientific Inc., Waltham MA, USA); 20 μg/well of MB894 total extract or 2 μg/well of Concanavalin A for 72 h. Cytokines IFNγ, TNFα and IL-10 were quantified using the BD Cytometric Bead Array (CBA) Flex Set system in a BD FACS Calibur flow cytometer, following the manufacturer’s instructions (BD^®^ Bioscience Co., Franklin Lakes, NJ, USA).

2.6. Histopathology

The left lung and the liver were fixed in 10% formalin solution and then embedded in paraffin following routine procedures. The 3 μm thick sections were stained with H/E and Ziehl Neelsen modified by Ellis and Zabrowarny (ZN) and observed at 400× in a Biological Microscope Olympus CX31(Evident, former Olympus Corporation, Miami, FL, USA). We performed a descriptive histopathology of the liver and lung. The microscopic changes in the liver were determined as the presence of inflammatory lesions, and the score was calculated as the number of lesions, size and hemorrhagic lesions. In the lungs, we evaluated individual lesions or cell type lymphocytes, neutrophils, active macrophages, hemorrhagic lesions, and bronchus-associated lymphoid tissue (BALT) hyperplasia, distribution and total surface affected. For each characteristic described in both organs, an individual score was defined ranging from 0 to 4, and the sum of the variables defined the final organ score.

2.7. Statistical Analysis

The Kruskal–Wallis and Dunn post-test with multiple comparisons were used to analyze the weight of the animals and organs, the CFU in the spleens and the levels of cytokines. A p-value below 0.05 was considered significant. Graphical representations were based on GraphPad Prism version 6.0 (GraphPad Software Inc., San Diego, CA, USA).

2.8. Genome Sequencing, Assembly, and Annotation

Whole genome sequencing was performed at the Genomic Unit of the Biotechnology Institute- INTA using MiSeq Illumina’s integrated next-generation sequencing instrument, using reversible-terminator sequencing-by-synthesis technology. Paired-end runs were adjusted to read lengths of 2 × 250 base pairs. Libraries were prepared using Nextera XT library preparation kit (San Diego, CA, USA). Quality analysis was done using FastQC v0.11..8, and filtering and trimming using Trimmomatic v0.39 (TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:60) [21]. De novo assembly was done using SPAdes v3.11.1 [22] and the genome of M. bovis AF2122/97 was used as a reference (LT708304.1). The genome was annotated using PROKKA v1.14 [23]. The assembly summary statistic is shown in Supplementary Table S1. Functional category annotation and GO terms were annotated using Sma3s tool [24]; all sequences were blasted against the M. bovis database from Uniprot. The Whole-Genome Shotgun project has been created in the NCBI database under the accession number PRJNA1212191. The sequences have been deposited under the accession number SRR32217824. The biosample number SAMN46293344 corresponds to MB894.

2.9. Variant Identification

To identify the single-nucleotide variants (SNVs), the reads filtered for each strain were first mapped using Burrow Wheeler Aligner MEM algorithm v. 0.7.17 (https://github.com/lh3/bwa) (accessed on 14 May 2019) to the reference genome M. bovis AF2122/97 (LT708304.1). To determine the variants (SNPs and INDELs) we used the genome of MB303 https://www.ncbi.nlm.nih.gov/datasets/genome/GCF_000488915.1/ (accessed on 21 May 2021) [25]. Sequence alignment files were sorted and indexed with Samtools v1.9 [26]. To perform variant calling (SNPs and INDEls) we first used samtools mpileup [27], and then Varscan v2.3.9 [28] was applied using pileup2snp and pileup2indels with parameters –min-reads2 20 -min-var-freq 0.2, allowing heterozygous mutations. Detected variants were annotated using SnpEff v4.3 [29]. We selected those variants that produce a change in the coding sequences, changing the amino acid or the open reading frame, which can produce premature stop codons. These variants were manually curated through visual inspection of the mapping of reads and reference annotations using Integrative Genomics Viewer (IGV) [30].

2.10. Phylogenetic Analysis

For the phylogenetic analysis, we used fifty freely available complete genomes of M. bovis from the National Center of Biological Information (NCBI) database (Supplementary Table S2). We included strains from different countries (Argentina N = 7; Brazil N = 8; Canada N = 4; Egypt N = 3; France N = 5; Paraguay N = 2; South Korea N = 3; Uruguay N = 7; USA N = 3 and 1 strain from each country Belgium, Bulgaria, China, India, Japan, Mexico and United Kingdom); and from different origins (cattle N = 36; human N = 8; pigs, wild boar N = 3; sheep N = 1; tapir N = 1; elk N = 1). Genome assembly completeness of each isolate was assessed using BUSCO v.5.3.2 [31].

The core genome alignment was generated with the Parsnp v1.2 tool [32], using the M. bovis AF2122/97 strain as reference. Subsequently, the core genome SNV alignment was extracted using the HarvestTools v1.2 [32] and submitted to the Gubbins tool [33] to remove potential recombination sites. The filtered core-genome SNV alignment was utilized as input to generate a maximum-likelihood (ML) phylogenetic tree using a GTRCAT model from RaxML v8 [34] with 1000 bootstrap replicates. Finally, the phylogenetic tree was visualized with iTOL [35].

2.11. Lipid Analysis

Total lipids from bacterial cells were extracted following procedures described previously [36,37]. Briefly, cultures were grown to late exponential phase (optical density (OD) 600 nm 0.8–1) and total cell wall lipids were extracted 3 times with chloroform: methanol. Lipids were analyzed by TLC on silica gel 60F254 by loading the same lipid quantities per lane (100 μg). TLCs were developed in petroleum ether: diethyl ether (98:2), three developments to evaluate phthiocerol dimycocerosate (PDIMs), and revealed by spraying with a CuSO_4_-phosphoric acid solution and heating. These experiments are representative of three independent lipid samples of each M. bovis strain. The relative abundance of PDIMA and PDIMB was estimated using the software ImageJ v1.53. The total lipid area for each strain, MB894 and MB303, was quantified using the ImageJ v1.53 program, followed by analysis of the bands corresponding to each PDIM fraction. The proportion of each fraction was calculated for each strain.

3. Results

3.1. Evaluation of Clinical Manifestations and Organ Weights in a Murine Model of Pulmonary TB

Animals from both infected groups started showing clinical signs by 40 dpi, evidenced by disordered hair, somnolence and lack of appetite. There were no differences in clinical signs between groups. In compliance with the regulations of the Institutional Animal Care and Use Committee of our institute, following the parameters of animal welfare standards, we set up the endpoint at 45 dpi, while several mice began to manifest clinical signs such as tachypnea, respiratory distress, piloreactions and consumption.

At necropsy, gross examination of organs revealed granulomatous lesions extended throughout the lungs in all animals from both infected groups. No gross lesions were observed in the livers of any of the animals (Supplementary Figure S2).

Total body weight showed no differences between animals of both infected groups and the control PBS group. An organ coefficient was obtained for each liver and spleen, following the formula: organ coefficient = organ weight (g)/body weight (g). Only the spleen coefficient of the MB894 group was statistically significantly higher than the PBS group, indicating a marked splenomegaly. No differences were found between the MB894 and the MB303 group (Figure 1A,B).

3.2. Bacillary Load and Histopathology in Necropsied Organs

CFU counts from lungs were performed at T0 (1 dpi) and T45 (45 dpi). Bacillary load was similar for both groups at T0, indicating that both strains were equally able to colonize the lungs after 24 h post infection. At T45, the bacterial burden increased 4 orders of magnitude related to the infective dose, with CFU counts reaching up to 10^6^ CFU/mL in both infected groups, indicating active proliferation. No statistically significant differences were observed between MB894 and MB303 (Figure 2A).

At T45, bacteria were isolated from the spleens of all the animals in both infected groups. In the liver, CFUs were recovered in all the animals of the MB894 group, and in 6/7 animals of the MB303 group. We observed a significant increase in the spleen bacillary load compared to the liver. However, no statistically significant differences were found between groups (Figure 2B).

The histopathological score was obtained for the lungs and liver. In the lungs, we observed characteristic granulomatous lesions involving all the lung parenchyma, with extended areas of necrosis in both groups (Figure 3A and Figure S1). Both strains showed more than 50% of the lungs affected; however, in the MB894 group, lungs were more severely compromised, with animals having affected almost 100% of the organ (Figure 3B). The study of individual lesions showed an increase in inflammatory cells such as lymphocytes, neutrophils, and active macrophages, with a higher rate of diffusible granulomatous lesions, BALT hyperplasia and hemorrhage that was superior in the MB894 group (Figure 3C), although there were no significant differences between strains. Considering these characteristics, the histopathological total lung score was calculated by the summary of the individual variables, and the MB894 groups had significantly higher mean score (16.14) compared to MB303 (12.57), indicating a severe granulomatous process given by this strain (Figure 3D).

Regarding the liver, the histopathological lesions were represented by a mixed inflammatory lympho-histiocytic-like, consisting mainly of histiocytes with incipient activity, lymphocytes and occasional presence of neutrophils. Unlike the lungs, no defined granulomatous lesions were observed, but the initiation of the granulomatous process was observed (Figure 3A and Figure S2). The distribution pattern was multifocal, with location in the portal triads. Mice from the MB894 had slight to moderate hemorrhage that was not observed in group MB303. Inflammatory lesions were found in all animals of both groups, but the size and number of lesions were significantly higher in animals of the MB894 group (Supplementary Figure S3).

3.3. Cytokine Expression in Stimulated Splenocytes

In order to evaluate the immune response induced by both M. bovis strains, we measured the production of cytokines IFNγ, TNFα and IL-10 in the supernatant of splenocytes at T45 following stimulation with MB antigens for 72 h. IFNγ levels were significantly higher in both infected groups compared to PBS controls after stimulation with bPPD and MB lysate, with no differences between strains. Notably, splenocytes from MB894-infected mice produced significantly higher IFNγ when stimulated with MB lysate than with bPPD (Figure 4A).

IL-10 expression was detected at basal levels with bPPD stimulation but increased markedly with MB lysate in both infected groups, showing significant differences compared to PBS controls. TNFα followed a similar pattern: low basal levels with bPPD and higher levels with MB lysate, with statistical significance observed only for MB894 compared to PBS. No differences were detected between MB894 and MB303 for TNFα or IL-10 (Figure 4B,C).

Stimulation with Con A confirmed cell viability and basal levels of cytokines were detected in non-stimulated cells.

3.4. Sequencing and General Features of the Genome of MB894

Whole Genome Sequencing of the MB894 strain resulted in 2 × 250 paired-end reads: 2 × 1,992,661 = 3,985,322 bp. After the filtering and trimming using Trimmomatic, a total of 2 × 910,394 = 1,821,868 paired reads of good quality (Phred quality score of Q > 30) were assembled in 91 contigs (N50 = 106,671 pb). Genome annotation was performed using Prokka, resulting in a total length of 4,284,489 pb organized in 4000 CDS with a GC content of 65.53%, according to the reference strain AF2122/97. Figure 5A shows the circular map of the MB894 genome generated by the GCView server [38]. Functional annotation was performed using the eggNOG tool. A total of 2854 genes, encoding for previously known proteins or characterized processes, mainly involved in lipid metabolism and cell wall remodeling; transcription; secondary metabolites biosynthesis, transport, and catabolism; energy production and conversion; amino acid transport and metabolism (Figure 5B). The remaining genes, corresponding to approximately 35% of the total CDS, belong to the group of unknown or hypothetical proteins.

A phylogenomic tree built from core chromosomal-derived SNPs showed the phylogenetic relation among MB894 with other M. bovis isolates from different hosts and countries (Figure 5C).

3.5. Comparative Genomics Analysis Between MB894 and MB303 Strains

Using the mpileup program, we mapped the reads of MB894 against the high-virulent strain MB303, finding 98.7% of mapping, indicating the correct identity of the strain. We looked for SNPs/INDELS between the strains, and found 335 SNPs, 18 INS and 16DEL (Supplementary Table S3). We then looked for the effect of these mutations on the amino acid sequence. Among the 369 differences, 10 produced severe changes at the protein level, mainly changes in the ORF, or premature STOP codon generating a truncated protein (Table 1). Most of these genes have similarities with virulence genes, putatively involved in lipid biosynthesis or transport lpt1, pks15/1, Mb894_03491, Mb894_03759, a transporter of the MFS family, a lipoprotein lppB and a protein belonging to the ESX-1 family.

We also looked for the presence of polymorphisms in virulence genes previously reported by Bigi and collaborators [39]. The authors compared the whole genomes of strain MB303 and MB534 (a low virulence strain) with the reference strain AF2122/97 (LT708304.1). They found NS-SNP in 9 genes of the high-virulent strain MB303 and in 5 genes of the low-virulent strain MB534. Among these genes, we found differences in 11 genes of MB894, compared to MB303 (esxA, pstS3, hpX, htrA, hrcA, mmpL3, mmpL7, mmpL8, Mb3114, ppsA and ppsE) and in 2 genes compared to MB534 (irtA and pks10). Interestingly, most of these genes are involved in lipid metabolism.

These findings prompted us to evaluate the total content of lipids and PDIMS in the two strains.

3.6. Lipid Analysis by TLC

We looked for the total content of the extractable lipids and PDIMs by thin-layer chromatography (TLC) analysis. We did not observe differences between the strains in the total lipids. PDIMs were accumulated in virulent strains, while in the reference strain were not detected (Figure 6A). We then quantified the bands using the ImageJv1.53 program and observed the accumulation of PDIMA and B in the MB894 strain, accounting for about 15% of the total lipid content (PDIMA 7.9%, PDIMB 8.9%) (Figure 6B). This accumulation was similar (or even greater) than the accumulation observed in the MB303 strain (PDIMA 3%, PDIMB 6%), which is characteristic of virulent strains [40].

4. Discussion

In a previous study, we evaluated the virulence of a local strain of M. bovis MB894 in an experimental infection trial in domestic pigs. We were able to reproduce the natural infection after oral inoculation, and the distribution of the granulomatous lesion in the lungs confirmed the tropism of M. bovis for the respiratory tract. These results suggest an adaptation to this route of spread from infected individuals to the environment and the role that domestic pigs play in the epidemiology of MBTB [11].

In order to deeply characterize the virulence of this strain, we performed an infection assay using the pulmonary TB model in BALB/c mice. We first conducted macrophage infections using porcine monocyte-derived macrophages (PMDMs). We observed significant variability in CFU counts from the inoculum and different time points, both between strains and across different experiments, which made it challenging to interpret the results reliably (Supplementary Figure S4). As a result, we opted to focus on the in vivo experiments using the Murine Pulmonary TB Model. We believe this approach provides more consistent and interpretable data regarding the infection dynamics. For this purpose, mice were intratracheally infected with the MB894 strain and compared with the strain MB303 that was previously characterized and classified as a hypervirulent strain [12].

In the murine model, hematogenous dissemination of M. tuberculosis is transient. Bacteria primarily spread through lymphatic routes to the draining lymph nodes between days 9–11 post-infection, and then systemically to the spleen and liver by days 11–14 [41]. At one day post-infection, bacteria are largely confined to the lungs, while by day 45, infection is chronic and bacteria are sequestered within tissues rather than circulating in the blood [42]. These dynamics suggest that blood CFU measurements at these time points would provide limited additional insight, which is why, for the murine model, the bacterial burden is measured in lungs, spleen, and liver [43]. A low infectious dose was used to induce latent infection. However, due to the high degree of virulence of both strains, a disease resembling progressive pulmonary TB was reproduced in mice, similarly to what was previously observed for the MB303 strain [12]. In addition, it was observed that strain MB894 multiplied significantly in the evaluated organs triplicating the infective dose in the lungs and colonizing the liver and spleen.

When analyzing the area of affected lung tissue, it was observed that both strains developed severe granulomatous lesions. The exacerbated inflammatory reaction, the massive pneumonia, along with the tissue damage produced by both strains, is consistent with what was observed for hypervirulent strains of M. bovis [12,44]. The MB894 strain also caused splenomegaly, a phenomenon not observed for the hypervirulent strain MB303. This could suggest a stronger inflammatory response of the spleen, due to a more potent immune response in animals infected with the MB894 strain. Moreover, hepatic histopathology revealed a mixed inflammatory response, greater in the MB894 group. Overall, although both strains proved to be hypervirulent, the MB894 strain showed severe tissue damage, greater dissemination and a generalized infection, suggesting a greater virulence.

The observed data were supported by the study of cytokines, crucial for the activation and modulation of the immune response. TB is a disease that can present different stages and depends not only on the pathogen but also on the immune response of the host, which plays a fundamental role in controlling the bacteria and the progression of the disease. Both in mice and humans, infection triggers a Th1-type cytokine profile, where the main cytokines involved are IFNγ and TNFα, which promote macrophage activation with the release of iNOS and nitric oxide, playing a fundamental role in the destruction of bacilli within cells [45]. Moreover, TNFα has been demonstrated as a crucial factor for TB latency [46,47], favoring the development and maintenance of the granuloma [48].

When evaluating cytokines in the culture supernatant of splenocytes, a clear Th1 profile was triggered by both strains, with a significant increase in IFNγ and TNFα, particularly in the MB894 group compared to the control. These cytokines work in synergy to facilitate granuloma formation, enabling the host to control the infection in a latent state [49]. In this direction, the greater production of TNFα by splenocytes in response to MB894 infection contributes to a pronounced inflammatory response, causing greater tissue damage, suggesting a greater virulence associated with this strain. Additionally, the higher production of IL-10 by both strains suggests a greater ability to modulate the immune response and potentially establish latency, supported by the massive necrosis observed in both groups. The cytokine profile observed in this study is consistent with the findings of other authors who have evaluated the immune response in experimental models of M. bovis infection [12,44].

Based on the severe immunopathogenesis produced by M. bovis strains in the murine model, we looked for genomic features that could be involved in the virulence of the strains. Other studies have also reported genetic differences between various M. bovis strains, which could contribute to variations in immune response and disease progression [16,44,50].

We performed whole-genome sequencing of the MB894 strain and compared it with the genome of the MB303 strain. This strain has been extensively characterized to understand its virulence, both in animal models [12,51] and through genomic and transcriptomic studies [39,52]. The authors identified differences, most of which were related to genes involved in lipid transport or metabolism [39]. In another work, differential mutations in virulence factors were identified in genes involved in the secretion and transport of cell surface components such as PDIMs and phenolglycolipids (PGLs), which might be linked to the observed differences in pathogenesis [44].

We found 369 variations at the SNP and INDEL levels, of which 10 resulted in severe changes that occurred in genes involved in membrane proteins and cell wall components. Among these, we can highlight Mb2813c, which encodes a potential lipid transporter protein, Lpt1. This mutation could affect the expression of the protein in the MB894 strain. However, the same mutation was identified in the low-virulence MB534 strain, and therefore, even though its role during infection is not yet understood, it would not explain the degree of virulence of the strain [39]. Other mutations reported in their work were also found in MB894 and shared with the MB534 strain, such as the MmpL, PstS3, PpsA, and PpsE. In all these cases, the variations occurred at a single amino acid position, making it challenging to assess their impact on the protein function and therefore were not considered in our analysis. On the other hand, other polymorphisms reported in the IrtA and Pks10 were conserved between the MB303 and MB894 strains but differed from MB534, indicating that the first two strains share more similarities than differences. IrtA is involved in iron acquisition, and the variation occurs in a conserved domain [39]. Pks10 is involved in PDIM synthesis, and thus, the authors suggest that it may contribute to the decrease in the PDIM levels observed in the attenuated strain. PDIMs are key virulence factors in both M. bovis and M. tuberculosis. Several genes are involved in the synthesis and transport of these complex branched lipids to the cell wall, such as acyl-CoA synthetases, dehydrogenases, and polyketide synthases (pks). Genomic analysis of the MB894 and MB303 strains revealed both differences and similarities in genes related to lipid metabolism and transport. While further experiments are needed to fully elucidate the contribution of each of these genes to the virulence of MB894, TLC analysis of the PDIMs indicated a similar accumulation of these lipids in the MB894 strain compared to MB303. These complex lipids are closely associated with the virulence of M. bovis, as evidenced by the significantly reduced PDIM production in the attenuated MB534 strain [39]. Thus, part of the virulent phenotype observed in the MB894 strain may be attributable to the accumulation of these lipids.

Moreover, we also identified mutations that introduced a premature stop codon in genes encoding Pks15 and an MFS family transporter protein. Additionally, we observed frame-shift changes in genes putatively involved in lipid metabolism, including dehydrogenase and hydrolase. In the latter, deletion of one nucleotide (T) in MB894 results in a frameshift that produces two proteins orthologous to Rv3337 and Rv3338 from M. tuberculosis strain H37Rv. In this strain, these proteins exist as two separate genes with an overlapping region. Rv3338 (EphH) belongs to the alpha/beta hydrolases superfamily and contains the AB hydrolase-1 domain with epoxide hydrolase activity [53]. EH enzymes have been identified in M. tuberculosis and play a crucial role in mycolic acid metabolism [54]. Contrastingly, in the reference strain M. bovis AF2122/97, a single base insertion leads to a single product (Mb3370).

We found another variation in MB894 consisting of an insertion of 5nts (CTTGT) in Mb1246c, that produces a frameshift in the ORF resulting in a shorter protein, changing 42aa on the C-terminal of the protein on MB894. Mb1246c is a member of the PE family, composed of approximately 100 highly homologous genes unique to mycobacteria with unknown functions. This abundance suggests that the protein products of these genes may play an important role in the homeostasis of the mycobacterial cell or in generating antigenic diversity in mycobacteria [55].

Abdelaal et al. [13] utilized WGS and mice virulence assays to gain insights into polymorphism and virulence of M. bovis isolates from dairy herds circulating in the Nile Delta of Egypt. One of the isolates (MBE4) has a unique genotype compared to other isolates, including 5 SNPs and a 106 bp deletion in the PE_RGRS14 gene (Rv0834c) [13]. Other members of the PGRS subfamily of PE proteins have a potential role in mycobacterial pathogenesis [56] and the evasion of host defenses [57]; however, it is difficult to establish a virulence association without further studies.

5. Conclusions

The results obtained in this study contribute to the understanding of the host–pathogen interaction in M. bovis infection and identify putative candidates to be further evaluated as virulence factors, responsible for the virulent phenotype of this strain. The characterization of the MB894 strain as hypervirulent is supported by the findings of this work, demonstrating its ability to induce a severe form of progressive pulmonary TB in mice. The observed histopathological lesions, the higher multiplication in the evaluated organs, and the greater dissemination to the liver and spleen are consistent with the virulence pattern previously reported for hypervirulent strains of M. bovis [12,40]. The characterization of a new hypervirulent M. bovis strain isolated from pigs points out the relevance of domestic pigs as putative reservoirs and maintenance hosts of more hazardous M. bovis strains that threaten current control programs.