Mycobacterium bovis BCG-Russia Clinical Isolate with Noncanonical Spoligotyping Profile

Abstract

To date, a unique spoligotyping profile (absent spacers 3, 9, 16, and 39 to 43) has been considered a canonical feature of a Mycobacterium bovis BCG strain, regardless of its country of use. Here, we report the first case of a clinical BCG strain with an unusual, M. bovis-like spoligotyping profile with additionally deleted spacers 1 and 2. Adverse effects of the BCG vaccination, named “BCG disease,” “BCG-itis,” or “BCG-osis,” include both localized complications and disseminated infection (2). In Russia, the published rate of BCG-associated complications in 2003 was 28.1 per 100,000 newly vaccinated children, while lymphadenitis and cold abscesses were the most frequently registered complications, with 16.7 and 7.3 per 100,000, respectively (1). A stable genomic feature of all BCG strains is a large deletion of the 10-kb genomic region of difference 1 (RD1) (3, 9); this deletion played a major role in the attenuation of BCG from M. bovis (10). The other characteristic signature of the BCG strain is a structure of its direct repeat (DR) (clustered regular interspaced short palindromic repeats [CRISPR]) locus that may be determined by spacer oligonucleotide typing (spoligotyping) technique. The mutation rate of spoligotypes has previously been estimated as 0.0029 to 0.0133 per strain per year (8), which is much lower than the mutation rate of polymorphic variable-number tandem repeats (VNTRs) (6, 16) and multicopy IS6110-restriction fragment length polymorphism (RFLP) profiles (13). This relative stability of the DR locus explains the limited utility of spoligotyping as a stand-alone approach for epidemiological typing since only one spoligoprofile of a BCG strain has been described to date (www.tbovis.org). The published global spoligotype database SpolDB4 at the Pasteur Institute of Guadeloupe contains information about ∼39,000 Mycobacterium tuberculosis strains (www.pasteur-guadeloupe.fr:8081/SITVITDemo), and the unique profile for the BOVIS_BCG clade is characterized by deleted spacers 3, 9, 16, and 39 to 43 (shared-type SIT482 to SIT653 isolates from 30 countries and all continents). A total of 126 biochemically identified BCG strains isolated from HIV-negative children with clinical tuberculosis from northwestern Russia were tested in 2007 to 2010 by spoligotyping at the St. Petersburg Pasteur Institute. One hundred twenty-five strains had the aforementioned profile of BCG strain (SIT482) (Fig. ​(Fig.1).1). One strain, 1125, was found to have a somewhat different spoligotyping profile (Fig. ​(Fig.1)1) and was ascribed to SIT977, BOVIS clade, according to SpolDB4. FIG. 1. Examples of spoligotyping profiles of the M. tuberculosis complex strains recovered from tuberculosis patients in St. Petersburg, Russia. SIT number and clade name were assigned based on comparison with SpolDB4. The asterisk indicates a typical BCG profile ... Subsequent analysis of RD1 was done by in-house-developed real-time PCR (I. Mokrousov, unpublished data): RD1 was deleted in both strain 1125 and all 125 strains with the SIT482_BCG spoligotyping profile. A sequencing analysis confirmed the RD1 that was deleted in strain 1125 to be a true BCG RD1. BCG strains used in different countries may differ in their mycobacterial interspersed repetitive-unit (MIRU) profiles (15). Accordingly, strain 1125 and 15 randomly selected strains with the SIT482/BOVIS_BCG profile were subjected to VNTR typing of 12 MIRU loci (15) and some reportedly M. bovis-discriminatory loci, ETR-A, ETR-B, QUB11b, and QUB3232 (5, 12). This analysis showed that all these strains had identical profiles. In particular, the 12-MIRU VNTR profile of strain 1125 was identical to that published for strain BCG-Russia, 222324253322 (15). Analysis of the clinical and epidemiological record of the patient from whom strain 1125 was recovered revealed that the 16-month-old HIV-negative male child was admitted first to an oncology hospital and then to a tuberculosis hospital with a final diagnosis of tubercular subclavian lymphadenitis. The child underwent a successful operation, which was followed by chemotherapy (rifampin, isoniazid, and amikacin) and final recovery. This child received BCG in the maternity hospital on day 5 of life, as required by the national vaccination schedule. No evidence of previous tuberculosis infection or contact with a tuberculosis patient was found. All currently used BCG substrains originating from a single source are known to differ in regions of genomic deletions, antigen expression levels, immunogenicity, and protective efficacy. A feature specific to strain BCG-Russia is a single-nucleotide insertion in the 5′ part of the recA gene leading to an early stop codon (7). This makes BCG-Russia a natural recA mutant that fails to express RecA. It was hypothesized that recA inactivation in BCG-Russia is in part responsible for its high degree of genomic stability, resulting in a substrain that has fewer genetic alterations than other vaccine substrains with respect to M. bovis AF2122/97 wild type (7). However, these findings appear to be in some contrast with the change in the DR locus (loss of spacers 1 and 2) observed in strain 1125. Such a sudden change in the DR locus was likely a momentary event that occurred only in a small subpopulation that later become prevalent. Spoligotyping technique is based on chemiluminescence detection and is very sensitive to contaminating DNA; nevertheless, the first two signals were clearly absent in the spoligotyping profile of strain 1125 (Fig. ​(Fig.1).1). Thus, it appears that for some unclear reason this neutral change in the DR locus (loss of spacers 1 and 2) in strain 1125 was rapidly selected and retained in the progeny, likely causing the in vivo “in-patient” evolution. A recent study of the tempo and mode of molecular evolution in M. tuberculosis at the patient-to-patient level suggested that the molecular evolution of the pathogen in vivo is characterized by periods of relative genomic stability followed by bursts of mutation (14), although the underlying mechanisms remain elusive. Regarding DR locus, a sudden in vitro change of spoligotyping profile is not an impossible event: Benjamin et al. (4) described two M. tuberculosis strains isolated from the same bronchoscope within 2 days, and their spoligoprofiles differed in one signal due to IS6110 transposition. It should be mentioned that the loss of spacers 1 and 2 might also potentially occur (i) in a BCG vaccine manufacturing facility before immunization or (ii) in a clinical laboratory during the process of isolation and culturing. Indeed, BCG vaccine in Russia is produced according to international standards, and each lot is controlled by the Tarasevich State Institute of Standards (Moscow, Russia), although not by spoligotyping. Nevertheless, we detected only one such strain among those isolated in 2007 to 2010; thus, its emergence during production appears less likely. Regarding selection in the diagnostic laboratory (Laboratory of Microbiology of Tuberculosis, The Research Institute of Phthisiopulmonology, St. Petersburg, Russia), it appears unlikely, since the DNA used in this study was extracted from the first available isolate. A search in the SpolDB4 database revealed six strains with the same spoligotyping profile as that of strain 1125 (shared-type SIT977) isolated in the following countries: Argentina, Germany, and France (1 strain each); Italy (2 strains); and an unknown country (1 strain). It is unknown whether all these strains were isolated from human or animal sources, i.e., whether they represent true M. bovis or have evolved from a BCG strain. It may be noted that BCG strains representing different progeny are used in these countries (11). M. bovis SIT482 and SIT997 strains (intact RD1) were recently described in Italy (7a), suggesting the SIT482>SIT977 evolution also in animal hosts. Rapid and correct identification of a Mycobacterium bovis BCG clinical strain is both clinically relevant and ethically pertinent since (i) a BCG strain is intrinsically drug susceptible, which implies a more straightforward treatment regimen than in the case of an M. tuberculosis strain, and (ii) sensitive legal issues inevitably arise in cases of development of clinical tuberculosis as a consequence of BCG vaccination. There is no single globally standardized method to rapidly detect BCG strains; hence, it is important not to misdiagnose a clinical BCG strain as an M. bovis strain. Our findings show that the DR locus in a BCG strain may undergo an abrupt change, resulting in a noncanonical, M. bovis-like spoligotyping profile. In this view, a correct identification of a BCG strain should involve further analysis of other molecular markers, e.g., RD1.