Geographic Differences in Genetic Locus Linkages for Borrelia burgdorferi

Borrelia burdorferi genotype in the northeastern United States is associated with Lyme borreliosis severity. Analysis of DNA sequences of the outer surface protein C gene and rrs-rrlA intergenic spacer from extracts of Ixodes spp. ticks in 3 US regions showed linkage disequilibrium between the 2 loci within a region but not consistently between regions.

The methods for 1) DNA extraction from ticks (11), 2) PCR amplifi cation of ospC, ospA, and IGS1 (7), 3) amplifi cation of IGS2 (8), and 4) amplifi cation of 8 chromosomal loci for MLST (12) have been described. Sequences for both strands were determined from either PCR products or cloned fragments with custom primers (7). We followed the basic nomenclature of Wang et al. (13) until, after exhausting the alphabet, we assigned both a letter and, arbitrarily, the number 3 (e.g., C3) when a new nucleotide sequence differed by >8% from known ospC alleles. We distinguished ospC variants with <1% sequence difference by adding a lowercase letter, e.g., Da and Db. Except for ospC D3 and Oa, novel polymorphisms were confi rmed in at least 1 other sample. To simplify IGS1 nomenclature, we numbered types sequentially, beginning with the original 9 types (7); ospA alleles (7) and IGS2 loci were likewise sequentially numbered. The online Appendix Table (www. cdc.gov/EID/content/16/7/1147-appT.htm) provides accession numbers for all sequences, as well as original and revised names for IGS1 sequences.
For 741 Ixodes ticks from northeastern and north-central United States or from northern California, 1 ospC allele was identifi ed and sequenced. In the remaining samples, we found a mixture of strains or evidence of >2 ospC and/or >2 IGS sequences (9). In 678 (91%) of the 741 samples with a single ospC, the allele could be matched with particular IGS1 (Table). We identifi ed 9 unique ospC sequences: Fc, Ob, Ub, A3, B3, C3, D3, E3, and F3, all from the northcentral United States. Alleles H3 and I3 of California were recently reported by Girard et al. (5). Of 32 codon-aligned ospC sequences, 6 pairs and 1 trio (Fa, Fb, and Fc) differed in sequence by <1% ( Figure, panel A). Nine novel IGS1 sequences, numbered 24-31 and 33, were discovered in samples from which ospC alleles were determined.
When we confi ned analysis to samples from northeastern states, we confi rmed linkage disequilibrium between ospC and IGS1 loci (7,10,14). However, when results from north-central states and California were included, a different picture emerged (Table, Figure, panel B). Most of the ospC alleles showed concordance with the chromosomal loci; monophyletic MLST showed either the same ospC allele or a minor variant of it. However, in several instances, the ospC alleles were linked to different IGS1 sequences, different ospA sequences, and/or different MLST with internal nodes in common. We observed this linkage for ospC alleles A, G, Hb, and N. In the case of ospC Hb, the shared internal node was deep.
We applied the Simpson index of diversity, as implemented by Hunter and Gaston (15), to the data in the Table to compare the discriminatory power (DP) of genotyping on the basis of a combination of ospC and IGS1 sequences with genotyping by 8-locus MLST (12). For double-locus typing, there were 43 types were found for 678 strains; DP value was 0.96. For MLST in this data set, 36 types were shown for 554 strains; DP was 0.95. In the study of Hoen et al. in which selection was made for geographic isolation, 37 types were distributed among 78 strains; DP was 0.97 (4).

Conclusions
Dependence on a single locus for typing may falsely identify different lineages as the same, especially when the samples come from different regions. Other loci may be as informative as ospC or IGS1, but the abundance of extant sequences for these loci justifi es their continued use. Uncertainties about the linkage of ospC and IGS1 usually can be resolved by sequencing the ospA allele (Table). IGS2 provided little additional information in this study.
One interpretation of these fi ndings is that lateral gene transfer of all or nearly all of an ospC gene has occurred between different genetic lineages. We previously had not detected recombination at the IGS1 locus on the chromo-  Table). Consensus phylograms were the output of the MrBayes version 3.1.2 algorithm (http://mrbayes.csit.fsu.edu). There were 500,000 generations with the fi rst 1,000 discarded. Nodes with posterior probabilities of >0.5 are indicated by values above the branches. Below the branches are integer values for nodes with support of >500 of 1,000 bootstrap iterations of the maximum-likelihood method, as carried out with the PhyML 3.0 algorithm (www.atgc-montpellier.fr/phyml). For both data sets and both algorithms, the models were general time reversible with empirical estimations of the proportions of invariant sites and gamma shape parameters. Scale bars indicate genetic distance. GenBank accession numbers for sequences are given in the online Appendix Table (www.cdc.gov/EID/content/16/7/1147-appT.htm). some (7), but there may be recombination at other chromosomal loci, as well as plasmid loci (6). Besides extending the understanding of the geographic structuring of the B. burgdorferi population, the results indicate that the ospC allele does not fully represent the complexity of B. burgdorferi lineages; thus, inferring phenotypes on the basis of this single locus should be made with caution.