**3. Discussion**

In order to better understand the virulence properties of CAO strains, we focused on the clinical Rostov strain belonging to the clade. The strain was resistant to seven antituberculosis drugs and contained well-known resistance-associated mutations, as shown in Table 1. Resistance to OFX was not detected for the strain that correlates with a study of Merker et al. [21] in which the frequency of resistance to fluoroquinolones was low among CAO isolates. Besides the drug resistance-associated mutations, the strain carried a compensatory mutation in the *rpoC* gene (g764363a; G332S), which was previously described [7,24]. We sugges<sup>t</sup> that this mutation could affect the fitness and lead to an increased growth rate of the strain compared to the reference H37Rv strain shown in Figure 1, in contrast with data published for 3 strains of lineage 2, that had decreased growth rate compared to the same reference strain [25].

According to 24-locus MIRU-VNTR typing *M. tuberculosis,* Rostov belonged to the 9358-25 cluster and differed from the 94-32 cluster by two loci, as shown in Figure S1. Although this cluster was not described earlier, the phylogenetic analysis using the MIRU-VNTR-plus database revealed a clusterization with the 94-32 type and according to MIRU typing, it belongs to the M2 cluster that is specific to Central Asia population [18].

Analysis of cluster-specific SNPs revealed one significant point mutation (a2321369g; N105D) in the *Rv2063a* (MazF7) gene related to virulence, detoxification and adaptation category according to Mycobrowser database (https://mycobrowser.epfl.ch/) and Forrellad et al. [26]. The second specific SNP was identified in the *fadE29* gene resulting in an amino acid substitution Ile288Val. Such substitution did not provide the significant changes in protein structure, accordingly to BLOSUM62 Matrix [27] (Table S1). It was reported previously that the MazEF toxin–antitoxin system is very important for stress adaptation, drug tolerance, and virulence of *M. tuberculosis*, and required for persistence in vitro. The deletion of MazF reduced the pathogen virulence for guinea pigs and decreased the bacterial load in organs [28]. All other polymorphisms presented in the Table S1 are not specific for the CAO clade, but their role is likely to be important for successful spread of the Beijing genotype in the world.

Survival studies showed that mice infected with Rostov strain succumbed to death within 18–47 days p.i., whereas a large proportion of mice infected with H37Rv maintained viability up to 90 days p.i., as shown in Figure 2. Similar mortality rates were detected for Beijing *M. tuberculosis* strains; conversely, the strains belonging to other *M. tuberculosis* families—Canetti, Haarlem and Somali clades—displayed intermediate or low virulence according to Lopez et al. [9].

Analysis of the specific pulmonary lesions in mice with experimental tuberculosis on day 30 showed that both strains had characteristic pathogenic properties, i.e., were able to cause the tuberculosis process, but patterns of pathological changes in lungs and livers were different for two strains, as shown in Figure 4. Our results are in agreemen<sup>t</sup> with Ribeiro et al. [29], according to which the H37Rv strain had the least virulent properties with respect to the Beijing genotype. The obtained data indicate that infection of mice with the clinical Rostov strain of *M. tuberculosis* leads to changes in the lungs. These changes consist of a small increase in the number of macrophages in some interalveolar septums. At the same time, macrophages have a relatively narrow cytoplasm, in contrast to wide-plasma macrophages that infiltrate lung tissue when mice are infected with the H37Rv strain of *M. tuberculosis*. The absence of pulmonary infiltrates in mice infected with the Rostov strain may indicate that this strain did not activate the host defense mechanisms, compared with the response to the infection caused by the H37Rv strain, as shown in Figure 4.

In conclusion, our study showed that pre-XDR Rostov strain belonging to the CAO clade of *M. tuberculosis* Beijing genotype is characterized by high virulence for C57Bl/6 mice when compared with the laboratory H37Rv strain. We propose that characteristic alterations of the CAO clade favor the selection of highly virulent bacteria.

### **4. Materials and Methods**

### *4.1. M. tuberculosis Strains*

The Rostov strain of *M. tuberculosis* was initially isolated from a clinical sample of sputum collected from a 35-year-old man hospitalized in the South Federal District of Russia in 2013 and deposed into the State Collection of Pathogenic Microorganisms "SCPM-Obolensk" (ID B-7601). The virulent laboratory H37Rv strain of *M. tuberculosis* was obtained from the "SCPM-Obolensk" collection (ID B-4825).

Frozen stocks of bacterial cells (1 × 10<sup>5</sup> CFU) were inoculated into 30 mL Middlebrook 7H9 broth with OADC supplement (BD, Franklin Lakes, NJ, USA), and 0.05% Tween 80, in three biological replicates, incubated at 37 ◦C under static conditions (i.e., without agitation) in flask cell culture 250 mL (Greiner AG, Kremsmünster, Austria) for 30 days to estimate the growth rate. Every five days the aliquots of 0.1 mL were taken for CFU enumeration by plating the serial 10-fold dilutions in triplicates onto Middlebrook 7H11 agar (BD, Franklin Lakes, NJ, USA) enriched with OADC. Bacterial colonies were counted on the plates after incubation for three weeks at 37 ◦C. To compare the growth rate amongs<sup>t</sup> strains, we determined a growth index, calculated from the log10 of the number of CFU at each time point divided by the log10 of the number of CFU at the initial time point. Additionally, we used *C*max to compare the growth of the strains. This index means the peak point on the bacterial growth curve [25].
