Mycobacterium bovis Strain Ravenel Is Attenuated in Cattle

Mycobacterium tuberculosis variant bovis (MBO) has one of the widest known mammalian host ranges, including humans. Despite the characterization of this pathogen in the 1800s and whole genome sequencing of a UK strain (AF2122) nearly two decades ago, the basis of its host specificity and pathogenicity remains poorly understood. Recent experimental calf infection studies show that MBO strain Ravenel (MBO Ravenel) is attenuated in the cattle host compared to other pathogenic strains of MBO. In the present study, experimental infections were performed to define attenuation. Whole genome sequencing was completed to identify regions of differences (RD) and single nucleotide polymorphisms (SNPs) to explain the observed attenuation. Comparative genomic analysis of MBO Ravenel against three pathogenic strains of MBO (strains AF2122-97, 10-7428, and 95-1315) was performed. Experimental infection studies on five calves each, with either MBO Ravenel or 95-1315, revealed no visible lesions in all five animals in the Ravenel group despite robust IFN-γ responses. Out of 486 polymorphisms in the present analysis, 173 were unique to MBO Ravenel among the strains compared. A high-confidence subset of nine unique SNPs were missense mutations in genes with annotated functions impacting two major MBO survival and virulence pathways: (1) Cell wall synthesis & transport [espH (A103T), mmpL8 (V888I), aftB (H484Y), eccC5 (T507M), rpfB (E263G)], and (2) Lipid metabolism & respiration [mycP1(T125I), pks5 (G455S), fadD29 (N231S), fadE29 (V360G)]. These substitutions likely contribute to the observed attenuation. Results from experimental calf infections and the functional attributions of polymorphic loci on the genome of MBO Ravenel provide new insights into the strain’s genotype-disease phenotype associations.


Introduction
Mycobacterium tuberculosis variant bovis (Mycobacterium bovis or MBO) causes significant economic hardship for livestock producers. M. bovis, a member of the Mycobacterium tuberculosis complex, is infectious to humans [1,2] and causes~150,000 cases of human disease annually [3]. However, the overall virulence of MBO is generally greater than that of M. tuberculosis [4][5][6][7], as is reflected in the extensive animal host range.
Historically, inoculation of rabbits was used to discriminate M. tuberculosis variant tuberculosis (MTB) from MBO within clinical samples; rabbits are typically resistant to MTB but highly susceptible to MBO infection. In contrast, guinea pigs and mice are susceptible to both MTB and MBO.
The MBO type strain for many comparative pathogenesis studies is M. bovis Ravenel (ATCC strain 35720). This strain was isolated from a tuberculous cow circa 1900 and deposited in the Trudeau Mycobacterial Culture Collection in 1910. At the Trudeau Institute, this strain was maintained in the non-native host rabbit before freezer stocks were made.
Experimental infection of calves with MBO strains such as 95-1315 at the National Animal Disease Center (NADC) in the US [16], WAg202 at AgResearch in New Zealand [17], and AF2122/97 at Animal Health and Veterinary Laboratory Agency (AHVLA) in the United Kingdom [18,19] consistently result in progressive disease in cattle, including typical granulomatous gross and microscopic lesions with recovery of bacilli from affected tissues. However, in a bovine TB vaccine efficacy study with cattle by Khare et al., 2007 [11], gross tuberculous lesions were not detected in vaccinated or unvaccinated groups upon necropsy 160 days after intranasal delivery of 10 6 colony forming units (CFU) of MBO Ravenel challenge, despite the recovery of MBO Ravenel from 10/10 unvaccinated and 4/10 vaccinated animals. However, histopathological changes were seen in 6/10 unvaccinated as well as 4/10 vaccinated animal's tonsils and livers. Even though all the referenced studies used different infection routes, dose rates, ages, and animal breeds, the fact remains that Ravenel failed to replicate the macroscopic granulomatous lesions seen in infection with other pathogenic strains of MBO. Because of this deficiency, pathogenesis studies no longer employ MBO Ravenel. Given these reports on apparent disease phenotype variations among different strains of MBO, the objectives of the present study were: (1) to compare the virulence of MBO Ravenel to MBO strain 95-1315 in cattle and (2) to compare the genome sequences of MBO Ravenel to three virulent strains of MBO.

Mycobacterium Tuberculosis Variant Bovis Challenge Strains
Two strains of MBO were used for the challenge inoculum: (1) MBO strain 95-1315 [USDA, Animal Plant and Health Inspection Service (APHIS) designation] originally isolated from a white-tailed deer in Michigan [20], USA and (2) MBO strain Ravenel (ATCC 35720) obtained from John Chan at Albert Einstein College of Medicine, Bronx, NY and freezer stocks were kept at NADC. Strains were prepared using standard procedures [21] in Middlebrook 7H9 liquid media (Becton Dickinson, Franklin Lakes, NJ, USA) supplemented with 10% oleic acid-albumin-dextrose complex (OADC) plus 0.05% Tween 80 and 0.5% Glycerol (strain Ravenel only).

Cattle Studies: Treatment Groups and Aerosol MBO Challenge Procedures
Holstein steers (n = 18,~1 year of age) were obtained from a TB-free herd in Sioux Center, IA, and housed in a biosafety level-3 (BSL-3) facility at the NADC, Ames, IA, according to Institutional Biosafety and Animal Care and Use Committee guidelines. In an initial study, steers (n = 3) received 10 5 CFU MBO Ravenel by aerosol. Briefly, inoculum (~10 5 CFU) was delivered to restrained calves by nebulization into a mask (Trudell Medical International, London, ON, Canada) covering the nostrils and mouth. The inoculum was inhaled through a one-way valve into the mask and directly into the lungs via the nostrils. The process continued until the inoculum, a 1 mL PBS wash of the inoculum tube, and an additional 2 mL PBS were delivered, taking~10 min. Strict BSL-3 safety protocols were followed to protect personnel from exposure to MBO.
All calves were euthanized~3.5 months after challenge by intravenous administration of sodium pentobarbital. Tissues were examined for gross lesions and collected and processed for microscopic analysis and isolation of MBO. Tissues collected included: palatine tonsil, lung, liver, and mandibular, parotid, medial retropharyngeal, mediastinal, tracheobronchial, hepatic, and mesenteric lymph nodes. Lymph nodes were sectioned at 0.5 cm intervals and examined. Each lung lobe was sectioned at 0.5-1.0 cm intervals and examined separately. Lungs and lymph nodes (mediastinal and tracheobronchial) were evaluated using a semi-quantitative gross pathology scoring system adapted from Vordermeier et al., 2002 [17]. Lung lobes (left cranial, left caudal, right cranial, right caudal, middle, and accessory) were individually scored based upon the following scoring system: 0 = no visible lesions; 1 = no external gross lesions, but lesions seen upon slicing; 2 = <5 gross lesions of <10 mm in diameter; 3 = >5 gross lesions of <10 mm in diameter; 4 = >1 distinct gross lesion of >10 mm in diameter; 5 = gross coalescing lesions. Cumulative mean scores were then calculated for each entire lung. Lymph node pathology was based on the following scoring system: 0 = no necrosis or visible lesions; 1 = small focus (1 to 2 mm in diameter); 2 = several small foci; 3 = extensive necrosis. Gross pathology data are presented as mean (±standard error) disease score for mediastinal lymph node, tracheobronchial lymph node, and lung.
Tissues collected for microscopic analysis were fixed by immersion in 10% neutral buffered formalin. For microscopic examination, formalin-fixed tissues were processed by standard paraffin-embedment techniques, cut into 5 µm sections, and stained with hematoxylin and eosin (H&E). Adjacent sections from sections containing caseous necrotic granulomas suggestive of tuberculosis were cut and stained by the Ziehl-Neelsen technique for visualization of acid-fast bacteria (AFB). Microscopic tuberculous lesions were staged (I-IV) as described by Wangoo [22]. Data are presented as the total and mean number of granulomas observed in each histologic lesion stage (i.e., I-IV) for lung and mediastinal lymph node sections (Table 1).  For quantitative assessment of mycobacterial burden, left tracheobronchial lymph nodes were removed, examined for gross lesions, weighed, and the entire lymph node (other than a small~1 g section for histology and qualitative culture) homogenized in phenol red nutrient broth using a blender (Oster, Shelton, CT, USA). Logarithmic dilutions (10 0 -10 −9 ) of homogenates in PBS were plated in 100µL aliquots plated on Middlebrook 7H11 selective agar plates (Becton Dickinson) and incubated for 8 weeks at 37 • C [23]. Data are presented as mean (±standard error) CFU per gram of tissue (Table 2).

Delayed Type Hypersensitivity (DTH) Responses (Skin Test Procedures)
Fifteen days prior to necropsy, calves received 0.

Delayed Type Hypersensitivity (DTH) Responses (Skin Test Procedures)
Fifteen days prior to necropsy, calves received 0.

Whole Genome Sequencing of MBO Ravenel
MBO Ravenel was obtained from freezer stocks stored at NADC (APHIS, USDA) and cultured on Middlebrook 7H10 slants (Hardy Diagnostics, Santa Maria, CA, USA) for 14 days. Colonies were harvested for DNA extraction. Paired-end (2 × 150 bp) library preparation using NEBNext DNA library preparation kit (New England BioLabs, Ipswich, MS, USA) was followed by NovaSeq Illumina genome sequencing (Novogene) (https://www.ncbi.nlm.nih.gov/sra/SRX10318108; accessed on 7 April 2021). All reads were quality-checked, and adapters were trimmed by Novogene's in-house custom brary preparation using NEBNext DNA library preparation kit (New England BioLabs, Ipswich, MS, USA) was followed by NovaSeq Illumina genome sequencing (Novogene) (https://www.ncbi.nlm.nih.gov/sra/SRX10318108; accessed on 7 April 2021). All reads were quality-checked, and adapters were trimmed by Novogene's in-house custom software (v1.0). Sequences were then checked for contamination using Kraken 2 with default parameters and the author-provided "Standard" database [24]. All reads were used to assemble the genome de novo, irrespective of Kraken's taxonomy assignment. ABySS v2.1.5 (k-value = 96) was used to assemble the genome [25]. RagTag v1.1.0 was used to perform MBO AF2122/97 (GenBank accession number LT708304) reference-based assembly correction followed by scaffolding [24]. QUAST v5.0.2 was used to analyze scaffolds [25] and generate Circos plots [26] and Icarus views (Figures 4 and 5-Circos-Images). Uninformative contigs (<200 bp) were removed before submission to the Prokaryotic Genome Annotation Pipeline (PGAP) v5.1 [27]. The draft genome was submitted to both PATRIC [28] and NCBI.     15), and MBO 10-7428 (isolated from cattle). MBO 10-7428 was submitted to Novogene for resequencing as described above, assembled as a draft genome (PATRIC ID 1765.618; GenBank: JAGEUC000000000.1), and used for comparative genomics and SNP extraction. Snippy [29] was used with default parameters (variant site coverage ≥ 10 reads, VCF call quality = 100, read mapping quality ≥ 60, base quality ≥ 13) to call polymorphisms and indels from paired-end reads from MBO Ravenel (SRR13938830), MBO 10-7428 (SRR13938829) and MBO 95-1315 using MBO AF2122/97 as the reference. The snippy-core script extracted polymorphisms at all positions with sufficient coverage for base calling from all four genomes. Those not unique to MBO Ravenel were removed from the analysis. Indels were placed into a separate data table. The SNPs that remained at this stage were unique to MBO Ravenel. Among these, only the missense mutations in genes with a putative function assigned were cross-checked against the annotated genome on PATRIC.
Pathogens 2022, 11, x FOR PEER REVIEW 8 of 16 Figure 5. Position of 32 high-confidence missense polymorphisms in the Ravenel assembly. The Ravenel genome assembly (GCA_018305025.1) was downloaded, and contigs longer than 1000 bp were visualized using Circleator. Working inwards, the rings represent: red, genome %GC; black, genes-fwd; black, genes-rev; blue, total unique polymorphisms (n = 486); central blue ring with labeled loci, 32 high-confidence missense polymorphisms in genes with annotated function. Of these polymorphisms, 31 are SNPs, and 1 is a multi-nucleotide polymorphism (lipN). The manuscript discusses nine SNPs in this innermost subset in greater detail.

Identification of Single Nucleotide Polymorphisms (SNPs) in MBO Ravenel
MBO Ravenel's genomic comparisons were performed against the reference strain MBO AF2122/97, MBO 95-1315 (isolated from deer, PATRIC ID 1765.15), and MBO 10-7428 (isolated from cattle). MBO 10-7428 was submitted to Novogene for resequencing as described above, assembled as a draft genome (PATRIC ID 1765.618; GenBank: JAGEUC000000000.1), and used for comparative genomics and SNP extraction. Snippy [29] was used with default parameters (variant site coverage ≥ 10 reads, VCF call quality = 100, read mapping quality ≥ 60, base quality ≥ 13) to call polymorphisms and indels from paired-end reads from MBO Ravenel (SRR13938830), MBO 10-7428 (SRR13938829) and MBO 95-1315 using MBO AF2122/97 as the reference. The snippy-core script extracted polymorphisms at all positions with sufficient coverage for base calling from all four genomes. Those not unique to MBO Ravenel were removed from the analysis. Indels were placed into a separate data table. The SNPs that remained at this stage were unique to MBO Ravenel. Among these, only the missense mutations in genes with a putative function assigned were cross-checked against the annotated genome on PATRIC. Ravenel genome assembly (GCA_018305025.1) was downloaded, and contigs longer than 1000 bp were visualized using Circleator. Working inwards, the rings represent: red, genome %GC; black, genes-fwd; black, genes-rev; blue, total unique polymorphisms (n = 486); central blue ring with labeled loci, 32 high-confidence missense polymorphisms in genes with annotated function. Of these polymorphisms, 31 are SNPs, and 1 is a multi-nucleotide polymorphism (lipN). The manuscript discusses nine SNPs in this innermost subset in greater detail.

Experimental Infection of Cattle
In an initial study with cattle (n = 3), aerosol MBO Ravenel (~10 5 CFU) elicited immune responses (i.e., DTH and IFN-γ) to MBO antigens ( Figure 1); yet, 2.5 months after challenge, tuberculous lesions were not detectable in two animals, and a single small granuloma was detected in the lung of the third animal.
To confirm this finding, an immunopathogenesis study was performed to directly compare the virulence of MBO strain Ravenel (n = 5) to that of MBO strain 95-1315 (n = 5), a strain of consistent virulence in cattle [15,16]. Significant IFN-γ responses to rESAT-6/CFP10 (E:C) or MBO PPD were elicited by MBO infection (Figure 2), regardless of strain. IFN-γ responses did not differ (p > 0.05) between the two challenge groups. Significant DTH responses to PPDs were also elicited by MBO infection (Figure 3). Responses to MBO PPD by MBO 95-1315 infected calves exceeded (p < 0.05) respective responses by MBO Ravenel infected calves.
Despite robust cell-mediated immune (CMI) responses, gross tuberculous lesions were not detectable in 4/5 MBO Ravenel-infected cattle ( Table 1). As with the initial study, one MBO Ravenel-infected steer had a single small granuloma in the left caudal lung lobe. In contrast, all 5 MBO 95-1315-infected cattle had granulomatous lesions in the lungs and lung-associated lymph nodes. Microscopic examination revealed no additional granulomas in the tissues collected.
In contrast to gross and microscopic pathology findings, MBO Ravenel was isolated from tissues of 4/5 cattle (Table 2). Tracheobronchial lymph nodes were the most common site for detectable MBO Ravenel colonization, despite no observable gross lesions in that anatomic site. As expected from prior studies, MBO 95-1315 was isolated from lungs and lung-associated lymph nodes from all five calves receiving this strain. Mean colonization of tracheobronchial lymph nodes in MBO 95-1315-infected calves exceeded that of MBO Ravenel-infected calves.

Genome Comparisons SNPs from MBO Ravenel versus Virulent MBO Strains
Comparisons between strains AF2122/97, 10-7428, Ravenel, and 95-1315 revealed no large-scale deletions in the Ravenel genome compared to the AF2122/97 genome. A total of 974 single nucleotide polymorphisms (SNPs) were extracted from the three strains against the AF2122/97 reference. Out of 974 SNPs in the set, 173 were present in and unique to MBO Ravenel. Of these, 95 were missense, and 54/95 missense mutations were in genes with a putative function assigned. Regions of the genome assembly flagged by QUAST as having potential for any degree of mis-assembly were excluded, leaving a subset of 32 highestconfidence SNPs (Figures 4 and 5, and Ravenel Supplementary Table S1). Nine highestconfidence SNPs were within "specialty genes" (a PATRIC-defined category of essential genes, virulence factors, AMR-associated genes, and others) in the functionally annotated genome available on PATRIC (1765.617) ( Table 3). These nine SNPs were cross-checked in Mycobrowser for their functional categorization [31]. Two functional categories were identified: (i) cell wall and cell processes (espH, mmpL8, aftB, eccC 5 , rpf B), and (ii) respiration or lipid metabolism (mycP 1 , pks5, fadD29, fadE29). Region of difference (RD) analysis of MBO Ravenel revealed that RD9, RD4, RD7, RD8, RD10, RD11, and RD12 were absent (classic MBO type). GenAPI reported seven total genes absent in MBO Ravenel and present in MBO AF2122/97, five of which fall into PE/PPE family proteins and may be false negatives due to the inherent difficulties in Illumina sequencing these regions, one prokka false positive for a short hypothetical protein not annotated by PGAP in the reference, and one showing a loss of the LuxR-family transcriptional regulator Mb2515c. This latter deletion was checked with 14,000 nt of sequence, including Mb2515c and flanking regions from MBO AF2122/97, and performing a blastn search against MBO Ravenel SRA reads (SRR13938830), which recapitulated the GenAPI findings showing no reads aligning over the Mb2515c gene and some flanking regions and suggests a deletion may have occurred.

Discussion
Experimental and clinical studies referred to earlier show that MBO Ravenel is attenuated and elicits a robust immune response in cattle. Several genome-wide SNP studies in M. tuberculosis have shown to have sufficient resolution to develop trait-allele interactions. In the present study, we hypothesized that drivers of attenuation are enciphered in the genome of MBO Ravenel [11] and performed comparative genomic analysis of Ravenel against three clinical MBO strains-95-1315, 10-7428, and AF2122/97. Disease phenotype was assessed by experimental infections with either Ravenel or 95-1315.
The present study utilized genomes of these select clinical strains to draw comparisons against the attenuated strain MBO Ravenel. This led to the identification of 32 highest confidence MBO Ravenel-specific missense SNPs, visualized by affected locus and position across the genome in Figure 5. Among these, a subset of 9 SNPs was selected based on functional annotation of impacted loci. These SNPs may contribute to reduced virulence and an attenuated phenotype as they affect cell wall synthesis-and transport-associated genes, pathways critical for the metabolism and intracellular survival of the pathogen.

Ravenel Elicits Robust Cell-Mediated Immune Response but Is Attenuated
Experimental calf infection with Ravenel or 95-1315 revealed that the former strain did not produce granulomata but was isolated from lymph nodes. The animals used in this study were older than the ones referenced earlier. However, the low virulence of this strain in cattle was still surprising, given the high virulence of M. bovis Ravenel in mice, rabbits, and guinea pigs [7,13,14]. Furthermore, comparative immunopathogenesis revealed significant IFN-γ responses to rESAT-6/CFP10 or M. bovis PPD were elicited by M. bovis infection, regardless of the strain used. These findings are consistent with the observations of Khare et al. [11], who demonstrated similar attenuation of Ravenel in a vaccination challenge trial. Taken together, our findings confirm that Ravenel is attenuated in the bovine host deserving a deeper explanation for the observed phenotype.
In contrast to the present study, Khare et al. observed microscopic granulomas in tonsils from 7 of 10 Ravenel-infected cattle. This may be due to the intranasal route of inoculation used rather than the aerosol route of inoculation used in our study. Aerosolization of two different strains of M. bovis, using doses similar to those presented here, also failed to induce lesions or result in the colonization of tonsils [16]. In the present study, microscopic granulomas were not observed in the lungs of Ravenel-infected cattle, in contrast to those described by Khare et al. This may also be due to a difference in the route of inoculation, or differences in the number of passages of inoculum strains.

Ravenel Carries SNPs in Genes Encoding Cell Wall Integrity
The mycobacterial membrane protein [large] (mmpL) genes encode a broad family of transmembrane-transport proteins believed to be involved in fatty acid transportation by performing the function of flippases [32][33][34][35]. These housekeeping genes are essential for survival. Members of this family, such as MmpL3, have thus been proposed as a druggable target [36], and research into small molecule inhibitors has yielded mycobactericidal compounds [37]. Williams et al. reported a list of mutations in the mmpL3-mutant strains, of which none were identifiable in MBO Ravenel [37]. Instead, MBO Ravenel had missense mutation G2662A (V888I) in mmpL8. Unlike mmpL3 s role in mycobacteria survival, evidence is limited for mmpL8 specifically. Nonetheless, being from the same family of proteins mutation in mmpL8 might affect the transportation of pathogenesis-associated compounds.
The biosynthesis of arabinogalactan by aftB is fundamental for the mycobacterial cell wall [38,39]. Raad et al. developed a Corynebacterium glutamicum ∆aftB mutant that caused outer membrane destabilization [39]. Jankute et al. demonstrated similar results in M. smegmatis [38]. Ravenel carries a mutation in aftB C1450T (H484Y), and whether this mutation results in deficient biosynthesis of arabinogalactan remains to be explored.

Ravenel Carries SNPs in Genes Associated with Respiration or Lipid Metabolism
Lipids are the primary carbon source for mycobacteria [47][48][49]. The fadD and fadE families of genes are involved in long-chain fatty acid synthesis in mycobacteria, encoding ligase, synthetase, and dehydrogenase. The enzyme encoded by fadD29 converts long-chain fatty acids to the acyl adenylates required for phenol glycolipid (PGL) production, which in turn is required for the synthesis of the outer membrane of MTB [50]. In the present study, MBO Ravenel carried a mutation in fadD29 A692G (N231S). PGL synthesis is required for mycobacterial viability, so missense changes warrant scrutiny [50,51]. MTB relies on host cholesterol import for carbon supply and survival during chronic infection through fadE29. Knockout studies by Thomas et al., and Gilbert et al., determined that fadE28 and fadE29 are essential for degrading cholesterol metabolites [52,53]. MBO Ravenel fadE29 has missense mutation T1079G (V360G). The downstream effects of missense mutations in both fadE29 and fadD29 may contribute to alterations in bacterial viability.
Polyketide synthase 5, encoded by pks5, has been identified as a virulence-associated biomarker of MTBC infection in cattle [54]. The product of pks5 is thought to be involved in multimethyl-branched fatty acid synthesis required for lipooligosaccharide (LOS) biosynthesis [55]. Loss of pks5 leads to severe MTB growth defects in animal models [56]. In MBO Ravenel, we observed a G1363A mutation (G455S) in pks5, though any effects on function require experimental testing.
The serine protease mycosin (MycP 1 ) is a conserved membrane component of ESX-1 and ESX-5 systems [57]. Ohol et al. (2010) found that inhibiting MycP 1 protease activity leads to increased ESX-1 substrate secretion [58]. C374T (T125I) in mycP 1 is, therefore, another SNP of interest in MBO Ravenel, as the destabilization of this protease is known to lead to dysregulation of the tightly controlled ESX-1 machinery. Considering that ESX-1containing RD-1 is believed to be a major contributor to the substantial attenuation of MBO BCG [15,[58][59][60], any disruption of this system by alternate means could explain virulence deficiencies despite RD-1's presence.
Out of 32 genes with functional annotations containing missense mutations unique to MBO Ravenel, nine involved in cell wall integrity and critical metabolism have been described here. A single mutation could not reasonably explain the experimental attenuation observed in MBO Ravenel compared to fully virulent strains. However, the cumulative effect of many SNPs across the genome may have contributed to the impairment of pathogenesis. Only a subset of the observed SNPs has been discussed here, and with our stringent cutoffs and filtering, some informative changes may not be included in our analysis. The remaining 23 MBO Ravenel-specific missense mutations, as well as those in loci without annotated function or those in intergenic regions, likely also play some role. While not discussed here, insertions and deletions can have profound effects on gene function, and a table of 57 called indels is provided (Supplemental Table S1). Additionally, a GenAPI comparison of MBO Af2122/97 and MBO Ravenel suggested a genomic deletion and complete loss of Mb2515c, a predicted LuxR transcriptional regulator in MTB thought to contribute to pathogenesis through unknown mechanisms [61].

Conclusions
We aimed to compare the in vivo virulence of MBO Ravenel to MBO 95-1315 in the original bovine host and discover the pathways impacted by mutations across the MBO Ravenel genome. Cattle infection experiments supported previously identified differences in MBO Ravenel pathology in the bovine host. Detection of robust immune responses across animals, despite the absence of gross lesions, also supports the Ravenel strain's deficiency in pathogenesis, causing only subclinical infection in contrast to the observed virulence from MBO 95-1315. The stringent set of identified highest-confidence missense polymorphisms in key genes involved with survival and pathogenesis of MBO provides a realistic genetic basis for attenuation. Which of these changes lead to functional impacts, if any, still requires elucidation. In summary, we confirm MBO Ravenel is attenuated and presents subclinically in the bovine host despite its broad genomic structural similarities to fully virulent MBO strains. A constellation of polymorphisms in key genes may instead explain its striking differences in disease phenotype when compared to virulent MBO 95-1315.