Yersinia pestis genotyping.

To the Editor: Drancourt et al. (1) report the development of an original genotyping system for Yersinia pestis based on intergenic spacer sequencing. However, the approach appears to rely upon the characterization of polymorphisms due to tandem repeat variation. Eight spacers are used in the report, 7 of which contain tandem repeats, and the sequence variability used to produce the typing data and the strain clustering result from variation in the number of tandem repeats (and incorrect data analysis produces a dendrogram with 34 branches from only 19 different isolate types). Three of the spacers and associated polymorphisms were previously reported. Spacers YP3 and YP5 are, respectively, ms38 and ms56 (2); spacer YP10 is M61 (3). YP3 is later used to investigate ancient DNA samples, and 3 amplification products are described in detail. The sequences are compared to modern sequences by BLAST analysis, which is not relevant for tandem repeats. Instead, the Figure shows how internal variation within the array can be coded to facilitate interpretation. In this collection, Orientalis strains are "abcdeeef," whereas Antiqua strains from Africa are "abcdeef." All these different codes can be deduced one from the other by simple duplication and deletion events, with no need to invoke point mutations. The codes for all 3 ancient samples are identical to the Orientalis code "abcdeeef."

Figure

A) sequence-to-code correspondence (1 letter per 16-bp repeat unit). Differences from repeat unit "e" are shown. B) Tandem repeat arrays were coded accordingly. All sequences were obtained from Genbank (Ypseu: Yersinia pseudotuberculosis IP32953; Microtus: "Y. microtus" Chinese strain #91001).

In conclusion, the data presented by Drancourt et al. do not appear to support their claim. They did not invent a new genotyping method but used the well-known multiple locus variable analysis (MLVA) number of tandem repeats approach. The finding that the "genotype Orientalis was involved in all three pandemics" is not valid since the Orientalis type is defined by a biochemical assay, resulting in all known Orientalis strains from a 93-bp glycerol-3-phosphate dehydrogenase microdeletion (4,5), which was not investigated here.

In Response: We thank Dr. Vergnaud for his response (1). Since the time of our publication (2), 2 articles related to our paper were either submitted or published. One (3) reported identification of Yersinia pestis–specific genes in teeth from patients who died during the Justinian plague; another proposed identification of Y. pestis strains by using variable numbers of tandem repeats analysis (VNTR) (4). The authors concluded that isolates could easily be compared in their database by using 7 markers. As opposed to work with cultures where ample, high-quality DNA template is available, successful amplifications with 7 different primer sets cannot be achieved by using DNA extracted from ancient teeth (5). By comparing genome sequences, we evaluated short intergenic spacers that were more divergent. Divergences included mutations, deletions, and duplications (VNTR). Phylogenetically, an entire repeat unit has the same weight as that of a single nucleotide polymorphism. By sequencing, we have identified all events (single nucleotide polymorphism and VNTR). Sequencing is more versatile for use in strain identification (5), allows distinction at the species level, and can be applied directly on clinical and forensic samples.

The discovery of a unique sequence is critical to authenticate results in such controversial areas as paleomicrobiology (5). Fortunately, we have identified a unique sequence that contains several mutations. These mutations do not exclude this strain from being Y. pestis (see Figure). Additionally, we doubt that our conclusions would have been accepted had we simply used the VNTR, demonstrating only an amplicon of the right size on a gel.

Figure

Unrooted tree showing the phylogenetic relationships between the sequence obtained from the YP3 spacer from Justinian sample 202 and that from the genomes of Yersinia pestis strain CO92 (GenBank accession no. AJ414159), Y. pestis biovar Medievalis (AE017139), Y. pestis strain Kim (AE013993), and Y. pseudotuberculosis (BX936398). DNA sequences were aligned by using the ClustalW software, version 1.81 (2). Deletions were considered single events. A distance matrix was constructed by using the Kimura-2 parameter, and the phylogenetic tree was inferred by using the neighbor-joining method in the Mega2 software package. The scale bar represents a 0.5% nucleotide sequence divergence. Bootstrap values are indicated at the nodes (2).

In conclusion, our results have been validated by others. The sequence is original and, therefore, authentic. Dr. Vergnaud agrees that the results we presented did represent a sequence associated with the Orientalis biovar. This finding may end the controversy.