Initial PCR screening of 20-2359 for RD pattern yielded atypical results. Of note, RD1 was present but RD9 did not show any amplification. RD4 and RD12 had late amplification, suggesting possible mutations in the primer or probe sites of this assay, or insertions and deletions impacting the amplicon size of the targets. WGS analysis returned atypical results for identification as well. Species identification with Kraken using a local Mycobacterium spp. database, reported 20-2359 as M. bovis, although with a low percentage of specific reads. In silico–derived spoligotype listed this strain in the most up-to-date databases as most likely M. caprae. This rare spoligotype, 0000000000000000111111111110111111111100000, had previously been reported as M. bovis or M. bovis subspecies caprae–type before M. caprae was reported as a unique species. Three other samples in our dataset isolated from primates in China (NCBI SRA nos. SRR1792164, SRR1792165, and SRR7617662) also shared this spoligotype with 20-2359 (Appendix Table 1).

An in-house lineage identification scheme using specific SNPs also failed to positively identify the isolate (Table). We found that 20-2359 lacked 1 of 2 specific mutations required to be classified as either M. bovis or M. caprae and detected none of the known lineage-specific markers. The same markers were also missing from the monkey and elephant isolates. Genomic analyses of RD confirmed that RD1 was present, but RD4 and RD9 regions were deleted in 20-2359. A more comprehensive analysis of RD in 20-2359 using RD-Analyzer (https://rdanalyzer.com) revealed a presence-absence RD pattern identical to other M. bovis–related strains and 1 M. caprae strain, NCBI SRA no. ERR1462578 (Appendix Table 2). A closer look at the RD4 sequences for the specific region Rv1496–Rv1518 in M. tuberculosis H37Rv in 20-2359 revealed a genomic deletion in 20-2359 different from all other sequences in our dataset, resulting in a unique gene cluster when compared with the other lineages (results not shown). RD4 was deleted in 20-2359, but present in the 5 closely related strains belonging to the proposed La4 lineage, which had complete RD4 and RD patterns identical to M. caprae.

SNP-based phylogeny with 100% bootstrap support using M. tuberculosis H37Rv as a reference placed 20-2359 close to isolates from 3 primates (NCBI SRA nos. SRR1792164, SRR1792165, SRR7617662) and 2 elephants kept in captivity in Japan (NCBI SRA nos. DRR120408, DRR120409) (Figure). These 6 sequences form a distinct group that branches halfway between the M. bovis La1.1 and M. caprae La2 clades. SNP distances between members of the same clade (M. caprae, M. bovis, or La1.1) all differed by <802 SNPs, whereas SNP difference across clades averaged 1,369 (range: 985–1,374 SNPs) (Figure). Within the 20-2359 cluster, the maximum SNP distance between any 2 isolates was 776. The number of SNPs between the 20-2359 cluster and any M. caprae, La1.1, or M. bovis bacillus Calmette-Guérin strain averaged 1,161; the minimum was M. caprae SRR13888754 with 1,047.

We identified Mycobacterium strains through WGS based on a combination of results from genomic database comparisons, spoligotype analysis, and detection of lineage-specific markers; each method has unique limitations. Although results generated by these methods usually agree, rare or unknown genotypes, not represented or improperly labeled in databases, can result in discordance and require a more in-depth analysis for final identification. When we first received sample 20-2359, initial presentation and culture testing did not indicate an atypical bacterium. However, when we first screened RD to confirm the strain identity, we noticed weaker amplification of some targets and the absence of RD9, indicating that the strain might belong to a less-common species or lineage within MTBC. Our attempts at identifying the strain through WGS analysis using results from Kraken, in-house lineage-specific markers, and in-silico spoligotyping all indicated it was somewhat related to M. bovis or M. caprae, but not which species or lineage.

SNP-based phylogenetic analyses using our local database, which contains >4000 clinical and nonclinical strains (data not shown), placed 20-2359 in a distinct lineage, a sister to M. caprae and more distantly related to M. bovis. A more focused phylogenetic analysis of publicly available sequences of animal-associated M. caprae, M. bovis, and other Mycobacterium spp. revealed that 20-2359 formed a well-supported cluster with 3 primate and 2 elephant isolates, distinct from M. caprae, M. bovis, and La1.1 (Figure). La1.1 is a newly classified animal-associated sublineage of M. bovis that is pyrazinamide susceptible, having branched off before acquisition of the pncA H57D mutation found in nearly all M. bovis strains worldwide, as described elsewhere (14). By comparing SNP counts between the 20-2359 cluster and the other isolates (Figure), we confirmed the distinctive nature of this cluster. The range of SNP distances (1,047–1,405) between isolates forming the 20-2359 cluster (proposed lineage La4) and isolates from other clades was lower than that between M. caprae and M. bovis and La1.1 isolates (1,203–1,463) but higher than that between M. bovis and La1.1 subclade isolates (985–1,077). Phylogenetic placement of proposed lineage La4 strains, along with the SNP distances to other clades, strongly suggests that isolates from this cluster belong to a new MTBC lineage associated with mammals from eastern and southeastern Asia.

We found the arrangement of RD4 in our clinical strain, 20-2359, unique from closely related primate and elephant isolates, which had complete RD4 gene clustering identical to M. caprae variant Allgaeu and other strains. Similarly, 20-2359 shared a spoligotype only with the 3 strains from primates, whereas the 2 strains from elephants had a spoligotype sequence with 1 extra spacer at spacer 2, identical to a spoligotype from the M. caprae clade. These differences within members of the 20-2359 cluster might reflect geographic diversity or differences in animal reservoirs. Limited information was available regarding the 3 MTBC samples from primates except that they were isolated in China. NCBI SRA nos. DRR120408 and DRR120409 samples were isolated at 2 time points from an elephant originally from the island of Borneo living in captivity in a zoo in Japan (20).

We could not establish the exact origin of clinical isolate 20-2359 based on available patient information; however, the patient grew up in Bangladesh and had potentially contracted TB through consuming raw milk. Results from a 2016 study reporting detection of M. caprae in 44 swamp buffalos from 4 farms in Thailand suggest that this strain type might have been encountered in the past (21). However, in that report, identification was based solely on spoligotype, which we have shown is conserved between some M. caprae strains and the new proposed lineage. The geographic location in that report is particularly intriguing given it is not typical for M. caprae. Although not possible to confirm with the available data, one possibility is that the swamp buffalo were infected not with M. caprae but with this newly described sister lineage. Given the distinct phylogenetic placement of this cluster, relatively long SNP distances to all M. bovis, La1.1, and M. caprae isolates in our dataset, and the case-patient’s geographic origin, which was atypical for the presence of M. caprae, we propose cluster 20-2359 belongs to a new MTBC lineage, La4, based on new nomenclature for animal-adapted MTBC lineages (14).