INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Plague is a disease caused by Yersinia pestis, a gram-negative bacterium in the family Enterobacteriaceae (10). There have been three major historical plague pandemics, in the 6th, 14th, and 20th centuries. The Justinian pandemic (sixth century) is reported to have killed 100 million persons in Europe, including 40% of the Constantinople (now Istanbul) population (11). The second pandemic began in the 14th century but continued well into the 17th century and resulted in perhaps 44 million European deaths in the 14th century alone. The most recent pandemic arose in China and spread quickly around the world, aided by modern transportation. Cases of plague still occur in association with established reservoirs in wild rodents in primarily rural settings.

Molecular strain discrimination is proving to be a valuable approach to the epidemiological understanding of Y. pestis (6, 9). Many commonly used PCR methods, including REP-PCR, AFLPs, and randomly amplified polymorphic DNA techniques, are usually biallelic, a fact which limits their diversity. In contrast, variable-number tandem repeats (VNTRs) are genomic regions with potentially extreme variation and, hence, great strain discrimination capacity (2, 5, 7, 8, 14). Identification of VNTRs has been the limiting factor in the development of such markers, but with an increasing number of partial and complete microbial genome sequences (3), this problem is greatly decreased.

In this study, we have examined the nearly complete Y. pestis genome sequence for simple sequence VNTRs and have discovered a tetranucleotide repeat. The repeat is in an intergenic region between two tentatively identified open reading frames (ORFs). PCR primers flanking the repeat motif have been used to characterize the repeat diversity of 35 Y. pestis strains and 1 strain of each of the closely related species Y. pseudotuberculosis and Y. enterocolitica. We have also examined the nucleotide diversity in the V antigen gene (lcrV) where, in contrast to the VNTR locus, there was a lack of nucleotide differences. Hence, the tetranucleotide repeat provides great strain discrimination potential for future epidemiological analyses.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Comparative sequencing of the V antigen gene. The complete nucleotide sequence was determined for the V antigen gene (lcrV) for most of the unique Y. pestis strains listed in Table 1, with the exception of the Indian Isolate, which is lcrV negative, and only CO92 from the U.S. isolates. PCR and DNA sequencing primers were designed on the basis of GenBank accession no. M26405. These primers were as follows: 1F, ATTAAGCGTCAGAGGGAGAG (nucleotide positions 390 to 409); 1R, CTCTTTTACTCGCTTGATGC (789 to 770); 2R, ATGGTGCCACTACTAGACAG 3' (1031 to 1012); 3F, TAGCAAGTTGCGTGAAGA (930 to 947); 4F, GGGGCGTTGGGTAATCTG (1222 to 1239); and 4R, CGTTGAGCATGGCGATAGT 3' (1561 to 1543). Initially, a 1,171-bp amplicon was generated with the 1F and 4R primers and then used as a template for the DNA sequencing reactions. The six primers described above allowed us to determine the entire nucleotide sequence on both strands of the entire amplicon. Hence, the few differences observed among Y. pestis strains were confirmed by at least two sequencing determinations.

Identification of the VNTR marker. A BLASTN (GenBank database) search was conducted on genetic sequence data for Y. pestis accessed from the Sanger Centre for Genome Research, Hinxton Hall, United Kingdom, data system. Imported sequences were analyzed for consecutive repetitive elements ranging from 2 to 4 nucleotides long. Once a VNTR locus of at least five repeat units was identified, primer pairs were designed to amplify a 200- to 250-bp region around the marker.

Strains and DNA isolation. All Yersinia strains listed in Table 1 were obtained from the U.S. Army Medical Research Institute of Infectious Diseases (USAMRIID) culture collection and were initiated from frozen stocks. Strains were first grown on sheep blood agar plates, and then a single colony was transferred into heart infusion broth. The replicate CO92 cultures 1097, 1098, 1099, 1100, 1108, and 1110 were derived from a single colony and then subcultured on either plates or in liquid media for 10 serial passages at either 28 or 37°C. CO92 variants 1101, 1102, 1103, 1104, 1107, and 1109 differ in phenotypic markers or plasmid composition from the original CO92 stock. For DNA preparation, strains were first cultured in broth and then harvested by centrifugation. DNA was released from the cells by digestion with proteinase K in the presence of sodium dodecyl sulfate. Proteins and other cellular components were removed using hexadecyltrimethylammonium bromide (CTAB; Sigma Chemical Co., St. Louis, Mo.) and then chloroform and phenol-chloroform extractions. DNA was precipitated using isopropanol, followed by an ethanol wash. All DNA samples were passed through a 0.22-µm-pore-size filter and checked for sterility prior to transfer to other laboratories.

PCR amplification. PCR amplification of the CAAA VNTR region was accomplished using two locus-specific primers: primer 1, GGTTAGGTAGGGTGTTGAAG; and primer 2, AAAGAGGCTAAGTGGCAA. PCR mixtures contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl, 0.001% gelatin, 0.2 mM each deoxynucleotide triphosphate, 1 µM R110-dUTP (Roche Molecular Systems, Inc., Branchburg, N.J.), 5 U of AmpliTaq DNA polymerase (Roche), and 20 pmol of each primer per 100 µl of total reaction volume. PCR conditions began with initial denaturation of the DNA at 94°C for 2 min followed by 35 cycles of denaturation at 94°C for 1 min, annealing at 55°C for 1 min, and extension at 72°C for 1 min. A final extension at 72°C for 5 min was followed by a 4°C soak. Because the PCR amplicons were labeled with fluorescent dUTP, analysis could be performed on an ABI377 automated DNA sequencer. This protocol labels both strands; hence, two fragments are observed in denaturing gels. Sizing of the amplicons was accomplished using Genescan software (Applied Biosystems).

Regions flanking the VNTR. ORFs flanking the VNTR region were identified using GeneQuest software (DNASTAR Inc., Madison, Wis.). BLASTN searches of GenBank were used to identify homologs of these two ORFs.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Comparative sequencing of the V antigen gene. To understand the genetic variation among Y. pestis strains and closely related species, we determined the complete lcrV gene (V antigen) sequence for 22 samples. We found that the lcrV gene had very little nucleotide variation and, with the exception of two strains, all sequences were identical to the previously reported sequence (12). The Angola strain differed at three single nucleotide positions from the others (GenBank accession no. AF167310), while the Pestoides F strain differed by a 16-nucleotide deletion (15); the Pestoides F difference involved a single deletion involving two direct repeats at the C terminus of the protein (GenBank accession no. AF167309). The lcrV gene diversity (1 minus the sum of the squared allele frequencies) based upon the three allele frequencies was low, at 0.17, for these strains. Overall, the lack of lcrV differences among these strains suggests that little evolutionary distance separates them from a common ancestor.

VNTR marker identification. In contrast to the lack of lcrV gene diversity, we have discovered a region within the Y. pestis genome that is highly polymorphic. The basis of this diversity is a tetranucleotide repeat sequence discovered by performing BLASTN searches of the partially completed Y. pestis genome sequence (Sanger Centre for Genome Research; http://www.sanger.ac.uk/Projects/Y_pestis). We found a CAAA tandem array with eight repeats in an intergenic region of CO92 (Fig. 1). The genomic regions flanking the CAAA repeats were analyzed for ORFs and regulatory sequences. ORFs were identified on both the 5' and the 3' sides of the repeats and were oriented with their 5' transcribed regions adjacent to the tandem array (i.e., both promoters must be near the tandem array). The 3' ORF (designated Orf-1) was 420 nucleotides long and separated by only 56 nucleotides from the CAAA tandem array. Orf-1 had a strong similarity (72.2%; BLAST probability, 2.9 × 10⁶⁸) to an Escherichia coli ORF of unknown function (5). The second ORF (designated Orf-2) was 560 nucleotides long and located 130 nucleotides 5' of the CAAA repeats (Fig. 1). BLASTN searches of Orf-2 against the GenBank database showed only a weak correlation (similarity, 50.8%; BLAST probability, 0.92) to the Bacillus subtilis tagO gene, encoding undecaprenyl-phosphate-N-acetylglucosaminyltransferase (GenBank accession no. AJ004803). Thus, the region 5' to the VNTR appears to be a monocistronic operon similar to genes encoding teichoic acid synthesis in gram-positive bacteria. Putative ribosome binding sequences and transcriptional promoters were identified for both of the ORFs.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Nucleotide sequence of the VNTR region. The CAAA repeat region that varies among Y. pestis strains is underlined. Orf-1 shows strong similarity to a gene encoding a 190-amino-acid hypothetical protein from E. coli. Orf-2 shows similarity to the B. subtilis gene tagO involved in teichoic acid synthesis. The putative ribosome binding sequence for Orf-1 (SD) is underlined. The PCR priming sequences are labeled and underlined.

View larger version (47K): [in this window] [in a new window]   FIG. 2.   Gel image of different alleles in the CAAA repeat region. All samples except Y. enterocolitica produced an amplicon in reactions containing the PCR primers described in the text. Strains correspond to those listed in Table 1; Pest, Pestoides. Numbers at left are in base pairs.

VNTR marker frequency. The CAAA repeat region was found to be a VNTR by PCR amplification of multiple Y. pestis strains. Amplicons from each strain were analyzed for size by electrophoresis on denaturing polyacrylamide gels. Among the 35 unique Y. pestis strains, there were nine different amplicon lengths varying with a periodicity of 4 nucleotides (Table 1). However, not all possible 4-nucleotide variants of between 1 and 12 repeats were observed. No single-repeat alleles were observed, and the only two-repeat allele was observed in Y. pseudotuberculosis. All possible alleles containing between 3 and 13 CAAA repeats were observed within the Y. pestis strains, with the exception of the 5- and 6-repeat alleles. The number of repeated CAAA sequences was calculated from the difference in size between the sequences of the observed alleles and the CO92 sequence. In addition, one allele from each size category was completely sequenced to confirm the number of repeats (data not shown). The diversity value for the CAAA repeat region was calculated to be 0.82 for the 35 unique Y. pestis strains (Table 1). This value is very high and possible only with multiple allelic loci. Alleles containing between 8 (allele H) and 11 (allele K) repeats were the most common and represented 83% (ca. 20% each) of the 35 unique Y. pestis samples (Table 1). The primers flanking the VNTR region did not successfully amplify the single Y. enterocolitica strain analyzed.

Allele stability. While the large number of alleles must result from elevated mutation rates, multiple laboratory cultures derived from one strain had identical alleles. This finding was demonstrated after multiple independent serial transfers of strain CO92 under different laboratory conditions (e.g., temperature and media). Fourteen independent cultures were examined, and all contained 10 copies of the CAAA repeats. Likewise, replicates or variants of La Paz, EV76, and Kim10 had the same VNTR alleles. Therefore, it appears that the VNTR mutation rate is not elevated to the extent that strain identity will be lost during laboratory maintenance.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have identified a unique VNTR (5'-CAAA-3') sequence from Y. pestis and characterized the repeat variation in 53 Yersinia DNAs, 35 of which are from unique Y. pestis strains. Nine alleles of the CAAA VNTR were represented within the 35 Y. pestis strains tested. This is a highly diverse genetic region (diversity value, 0.82) that may provide great discrimination in epidemiological analyses. The diversity observed in the CAAA repeat region was extremely high, in contrast to the nucleotide diversity observed across a 1,171-bp amplicon associated with the lcrV gene. Here, only three haplotypes were observed with a low diversity (diversity value, 0.17). The strong bias toward the collection of virulent strains may represent a selective force that results in a narrow definition of a pathogen and low diversity. Alternatively, all Y. pestis strains may be derived from a relatively recent ancestral strain and, thus, not have had sufficient evolutionary time to differentiate. The lcrV results are consistent with those of Achtman et al. (1), who examined five housekeeping genes in 36 Y. pestis strains without detecting even a single nucleotide difference. In contrast to the sequence homogeneity of this pathogen, the CAAA VNTR is highly diverse, with great discriminatory power as a marker among different strains.

While the biovar categorization and VNTR types did not perfectly correlate, there were definite allelic trends. Biovar antiqua (allele A) had the highest diversity of the alleles. This finding is consistent with its supposedly more ancient status. It has been suggested that the antiqua biovar is associated with the Justinian plague, the first recorded pandemic noted in history, around the year 542 A.D. (11). One can speculate that this biovar represents the earliest known strains of the bubonic plague and contains the largest number of VNTR categories. The 8 mediaevalis samples had only three VNTR alleles, while the 21 orientalis samples had five alleles. It has been suggested that the mediaevalis biovar was associated with the Black Death prevalent in 14th-century Europe. Orientalis is known from the current pandemic mainly seen today in animal reservoirs (11). In the 13 North American strains, four different alleles were observed (alleles H to K).

Identification of surrounding ORFs and the proteins that they encode may be one way of understanding the function of the VNTRs. Examples from eukaryotes and prokaryotes indicate that VNTRs can alter gene expression and, in some cases, cause genetic diseases (4, 13). The CAAA repeat region is in proximity to the putative transcriptional promoters of two ORFs. Alteration of transcriptional activity via RNA polymerase binding could be accomplished by changes in repeat length. Alternatively, selection against transcriptional changes could inhibit the formation of particular VNTR alleles. Note that the repeat numbers centered around 8 to 11 CAAA copies, with only a few observations of larger or smaller numbers. This restricted and nonrandom distribution seems to indicate constraints in the generation of VNTR diversity.