Research Article

Phylogeography of Borrelia burgdorferi in the eastern United States reflects multiple independent Lyme disease emergence events

Anne Gatewood Hoen, Gabriele Margos, Stephen J. Bent, Maria A. Diuk-Wasser, Alan Barbour, Klaus Kurtenbach, and Durland Fish
PNAS September 1, 2009 106 (35) 15013-15018; https://doi.org/10.1073/pnas.0903810106

  1. Edited by Barry J. Beaty, Colorado State University, Fort Collins, CO, and approved July 9, 2009 (received for review April 22, 2009)

Abstract

Since its first description in coastal Connecticut in 1976, both the incidence of Lyme disease and the geographic extent of endemic areas in the US have increased dramatically. The rapid expansion of Lyme disease into its current distribution in the eastern half of the US has been due to the range expansion of the tick vector, Ixodes scapularis, upon which the causative agent, Borrelia burgdorferi is dependent for transmission to humans. In this study, we examined the phylogeographic population structure of B. burgdorferi throughout the range of I. scapularis-borne Lyme disease using multilocus sequence typing based on bacterial housekeeping genes. We show that B. burgdorferi populations from the Northeast and Midwest are genetically distinct, but phylogenetically related. Our findings provide strong evidence of prehistoric population size expansion and east-to-west radiation of descendent clones from founding sequence types in the Northeast. Estimates of the time scale of divergence of northeastern and midwestern populations suggest that B. burgdorferi was present in these regions of North America many thousands of years before European settlements. We conclude that B. burgdorferi populations have recently reemerged independently out of separate relict foci, where they have persisted since precolonial times.

Lyme disease, caused by the spirochetal bacterium Borrelia burgdorferi, was named for the town in southern Connecticut where it was described in 1976 (1, 2). However, the first described case in the US was reported from Wisconsin in 1970, before the disease was formally recognized (3). Over the past few decades, Lyme disease has spread from these initial foci to affect large areas of the Northeast and Midwest, with >27,000 cases reported from these regions in 2007 (4). The spread of Lyme disease into its current distribution in the US is primarily due to the recent range expansion of the blacklegged tick, Ixodes scapularis, the principal vector of B. burgdorferi to humans (5). A notable exception is in the Pacific coast states where B. burgdorferi is transmitted by a different tick species, Ixodes pacificus (6).

I. scapularis has a three-stage life cycle that takes at least two years to complete and requires feeding on one vertebrate host per stage (7). The two immature stages normally feed on rodents and other small and medium-sized vertebrates, while adults feed almost exclusively on the white-tailed deer, Odocoileus virginianus. Deer are not competent reservoirs for B. burgdorferi, but they serve as reproductive hosts for I. scapularis and are required for the maintenance of I. scapularis populations (8). Like other ticks of the genus Ixodes, I. scapularis is capable of off-host dispersal of only a few meters (9). However, vertebrate hosts, particularly deer, but possibly also birds, are believed to drive the dispersal of I. scapularis (10–12).

The range expansion of I. scapularis into its current distribution in the Northeast and Midwest US has been associated with the reintroduction of deer after the reforestation of much of the eastern US since the mid-20th century (13). Previous to this, much of the deciduous forest cover in the eastern part of the country was cleared for farming and for use in manufacturing during the early agricultural and industrial development of the US (14). Forest clearing, along with unregulated hunting, led to the elimination of deer populations and, in turn, I. scapularis populations throughout much of the region where both had presumably been present and widely distributed during precolonial times. Isolated deer herds were, however, continuously present on Long Island, New York (15) and smaller islands off shore of Massachusetts (14) and in remote areas of northern Wisconsin, Michigan, and Minnesota (16). These refugial herds also appear to have supported I. scapularis populations and allowed for local continuous maintenance of B. burgdorferi (17, 18).

The earliest collections of I. scapularis from the Northeast were made in the 1920s on Naushon Island, near Cape Cod, Massachusetts (19). With the return of white-tailed deer to other areas of the Northeast and Midwest, the range of I. scapularis began expanding. During the 1960s, focal I. scapularis infestations were recognized on Long Island and Nantucket Island in the Northeast and in northwestern Wisconsin (13, 20). By the mid 1970s, reports of I. scapularis indicated spread to several additional Massachusetts islands and to Cape Cod, and from eastern Long Island to the closest mainland point near the mouth of the Connecticut River at Lyme, Connecticut. By this time, I. scapularis was also known to be distributed throughout much of Wisconsin (13, 21, 22).

Although the expansion of deer populations and subsequent spread of I. scapularis is well-documented as the major cause of Lyme disease emergence in both the Northeast and Midwest US, these two foci of Lyme disease endemicity appear to be discontinuous and spreading independently of one another (23–28). However, our empirical knowledge of the origins and movement of B. burgdorferi in North America is known only from patchy entomological records of vector presence and incomplete case reports of human disease.

In this study, we use a multilocus sequence typing (MLST) scheme based on bacterial housekeeping genes that we have described (29) to examine the evolutionary past of B. burgdorferi within the context of Lyme disease emergence. We report on the processes that have shaped the population structure of B. burgdorferi in eastern North America, including an ancient east-to-west radiation of descendent clones from founding genotypes, ancient population and spatial expansion events, and recent independent emergence of B. burgdorferi out of separate refugia in the Northeast and Midwest US.

Results

Samples and Sequence Types.

We sequenced a total of 4,785 base pairs at 8 B. burgdorferi MLST loci from 78 individual tick samples, listed in Table S1. Of these, 37 were from midwestern sites and 41 were from northeastern sites, with a total of 19 different collection sites represented (Fig. 1). Due to stratified random site and sample selection procedures, the 19 sites in this study were generally representative of the range of sites in which host-seeking I. scapularis nymphs were found. However, the southern range of the distribution is poorly represented as a result of low nymphal host-seeking I. scapularis densities and absence of B. burgdorferi in the ticks found in those areas (23). The sites and samples from B. burgdorferi-positive locations are naturally divided into two regions, northeastern and midwestern, due to a paucity of host-seeking I. scapularis nymphs in and around the state of Ohio (Fig. 1). This discontinuity has been observed both in the extensive sampling associated with the present study, described in a previous report (23), and in other studies (30, 31). Of the 78 samples included in this analysis, 37 distinct sequence types (STs) were found (Table S2), of which 18 were not already present in the B. burgdorferi MLST database (http://borrelia.mlst.net).

Fig. 1.
Fig. 1.

Map of study area and sampling locations in the northeastern and midwestern US. Colored circles are the locations of the 78 B. burgdorferi samples from host-seeking I. scapularis nymphs that were analyzed by MLST in this study. Circles are shaded with hues corresponding to a longitudinal gradient: Northeastern sites are filled with warm colors and midwestern sites with cool colors. The size of the colored circles is proportional to MLST sample size. Small gray circles are sites where I. scapularis nymphs were collected but were not included in this analysis. Open circles are sites where sampling was performed but no I. scapularis nymphs were detected. Sampling sites were randomly chosen within the quadrants of a 2-degree grid (gray lines).

Sequence Diversity and Patterns of Relatedness.

The nucleotide diversity observed in the 78 samples was due to sequence variation present across all eight MLST loci (Table 1), with an average nucleotide diversity across all loci of 1.96%. All loci exhibited a higher rate of synonymous substitutions than nonsynonymous substitutions, with dN/dS ratios all well <1, indicating that the sequence variation observed was largely neutral. Of the 37 different STs observed, no single ST was found in both northeastern and midwestern sites; however, nearly all STs represented by more than one sample were found in multiple sites within the same region (Table S1). The most frequent STs, designated 3, 18, and 32, were each represented by six samples; no ST dominated the dataset. A Bayesian phylogenetic tree of concatenated sequences revealed patterns of relatedness between strains (Fig. 2). Identical STs were always confined to a single region; however, closely related STs, even those differing by only a single substitution, were frequently found in both regions.

View this table:
Table 1.

MLST sequence diversity

Fig. 2.
Fig. 2.

Outgroup-rooted Bayesian phylogenetic inference of evolutionary relationships between concatenated MLST sequences. Labels at branch tips refer to strain number, with collection site in parentheses. Clade posterior probability values are shown at tree nodes. Color-coding of taxon labels corresponds to longitude-based color gradient in Fig. 1. The branch length of the out-group, B. afzelii, is not to scale, as is indicated by slashes. (Scale bar: 1% sequence divergence.)

Spatial Dependence of Allele Frequencies.

To detect patterns in spatial structure of allele frequencies, we assessed spatial dependence using Moran's I statistic, a coefficient of spatial dependence for pairs of samples separated into distance classes. The resulting correlogram of spatial dependence >10 distance classes (Fig. 3) revealed Moran's I values that were higher than would be expected in a simulated random dataset, with 95% confidence, at the first two distance classes (≤400 km). The index of spatial autocorrelation decreased with increasing pairwise distances, and was lower than would be expected in a random dataset at two of the larger distance classes. This indicates that in neighboring sites allele frequencies tend to be more similar when they are closer together, spatial dependence of allele frequencies is limited at intermediate distance classes, and allele frequencies tend to be more different between the two regions than would be expected by chance.

Fig. 3.
Fig. 3.

Moran's correlogram of individual allele frequencies. Moran's I, an index of spatial autocorrelation, was plotted for individual allele frequencies across 10 equally sized distance classes (black line). The mean Moran's I values resulting from 10,000 random permutations of the dataset are also shown (gray line) along with a 95% confidence interval for the distribution of permuted values (gray shading). Significant values of Moran's I fall outside of the 95% confidence interval and indicate significant positive spatial dependence (positive values outside of shaded area) or significant negative spatial dependence (negative values outside of shaded area).

Clonal Complexes.

We identified clonal complexes, or groups of STs that are inferred to be clonally descended from a common founder, using the eBURST algorithm (32). In our dataset of 37 STs, we detected four distinct clonal complexes for which ancestral or founding STs could be inferred (hereafter referred to by the number assigned to the founding ST), two complexes with no inferred founder, and 13 singleton STs (STs that had six or fewer loci in common with any other ST) (Fig. 4). In all of the four rooted clonal complexes, the founding STs were obtained from sites located in the coastal Northeast US. In three cases, clonal complexes 7, 34, and 37, the founding ST was restricted to the three sites located in southern New York State and southern Connecticut. The founding ST of the fourth clonal complex, 19, was obtained from a tick from a site in southern coastal Maine. In addition, in all of the four rooted clonal complexes, the founding ST had single locus variant descendents that were from ticks collected at sites that were located inland, either in midwestern sites for clonal complexes 19, 34, and 37, or in western Pennsylvania near the western edge of the range of I. scapularis activity in the Northeast for clonal complex 7.

Fig. 4.
Fig. 4.

Geographic distributions of clonal complexes. Clonal complexes based on multilocus allelic profiles defined by the eBURST algorithm. Each colored circle represents an ST. Circle size and color correspond to sample size and location, respectively, according to the scales in Fig. 1. STs connected by a solid line are single locus variants and STs connected by a dotted line are double locus variants. Inferred founders of clonal complexes are outlined in black. For illustration purposes, the four clonal complexes with an inferred founding ST are plotted on maps of the study area; STs are plotted at approximately the centroid of all of the sampling locations where the ST was found, with adjustments made to avoid completely overlapping circles. The bottom row contains singletons and complexes with no inferred founder. Clonal complexes with inferred founders are named for their founding ST numerical assignment.

Demographic Trends.

The frequency distribution of pairwise sequence mismatches of neutrally-evolving loci can leave signatures of past demographic events (33). We examined the mismatch distributions of pairs of concatenated sequences in our dataset for evidence of past spatial and population expansions by comparing them with distributions obtained under models of constant population size, sudden population expansion, and sudden spatial expansion (34). The mismatch distribution for all pairs of sequences resembled a bell curve with a mean of 21 mismatches and a tail representing a high frequency of pairs with only a few mismatches (Fig. 5A). The observed distribution did not differ significantly from the expectations of either a population size expansion or a spatial expansion model (Harpending's raggedness index, P = 0.23 and P = 0.89, respectively) but did differ significantly from the modeled distribution for a stable, nonexpanding population (P = 0.04). This pattern of better fit of the observed data to both of the expansion models was confirmed by a sum of squared deviations an order of magnitude lower for the population and spatial expansion models (sum of squared deviations = 0.003 for both models) compared with the constant population size model (sum of squared deviations = 0.017). The mismatch distribution parameter, τ, the time in mutational steps since the modeled expansion event, was similar for both the expansion scenarios; it was estimated at 18.1 mutational steps (95% CI: 10.99, 40.05) for the population expansion and 15.7 (95% CI: 10.10, 37.25) for the spatial expansion.

Fig. 5.
Fig. 5.

Mismatch distributions. Frequency distributions of observed pairwise nucleotide differences in concatenated MLST sequences (gray bars) and mismatch distributions modeled under three different demographic scenarios (black lines, see legend). (A) All pairwise comparisons; (B) midwestern samples only; (C) northeastern samples only.

We next tested whether this pattern of expansion held up within each of the two discontinuous regions when examined individually. When the pairwise mismatch distributions were calculated for each region separately, evidence for expansion was apparent in both. The raggedness statistic for the mismatch distribution for midwestern samples was 0.026, more than twice as high as that for the northeastern samples, which had a raggedness statistic of 0.011. This is evident by the visibly more irregular shape of the midwestern distribution (Fig. 5B) compared with the northeastern distribution (Fig. 5C). In the midwestern samples, the high raggedness statistic value allowed rejection of the population expansion model (P = 0.01) but not the spatial expansion model (P = 0.149), while in northeastern sites, neither expansion model could be rejected (P = 0.18 and P = 0.84, respectively). τ was estimated at 20.7 mutational steps (95% CI: 9.5, 45.9) for the population expansion of the midwestern populations, and at 18.5 (95% CI: 11.6, 32.6) for the population expansion and 15.5 (95% CI: 10.8, 32.0) for the spatial expansion of northeastern populations. Taken together, these data provide evidence for spatial expansions of B. burgdorferi in both regions occurring at approximately the same time, assuming that the rate of evolution is similar for populations in the two regions.

Discussion

B. burgdorferi has been shown to exhibit genetic diversity throughout its range in the northeastern US based on genotyping methods relying on single loci, namely the outer surface protein C (ospC) and the 16s–23s (rrs-rrlA) rDNA intergenic spacer (35–37). However, single-locus approaches can result in distorted patterns due to the differential effects of evolutionary processes on different parts of the genome (38, 39). The multilocus approach in this study was motivated by the potential of MLST to more accurately reflect the evolutionary relationships between bacterial clones. Despite their extensive use for strain typing a large number of environmental and clinical samples, the intergenic spacer and ospC markers have not been shown to be suitable for delineating the geographical structure of fine-scale evolutionary relationships of B. burgdorferi (29, 36). Our choice of genotyping scheme allowed insights into recent and deeper evolutionary relationships among B. burgdorferi populations in the Northeast and Midwest regions of the US. This is a study of genetic diversity using neutral or nearly neutral multilocus markers in B. burgdorferi samples collected from sites systematically chosen for their locations within the range of I. scapularis-borne Lyme disease.

We observed a pattern of sequence divergence corresponding to geographic location appearing at relatively recent points in the evolutionary history of B. burgdorferi. This strongly suggests that the two foci of I. scapularis-borne B. burgdorferi in the US have a shared past and once belonged to an admixed population but are now isolated. Given the slow evolution of bacterial housekeeping genes, even a small level of sequence divergence is estimated to have occurred over long periods of time, much longer than the timescale of modern Lyme disease emergence in the US over the past three decades. If B. burgdorferi emerged in one region early in the last century and went on to seed the other region since then, we expect that we would have collected some of the same STs in both regions, that we would not have observed negative spatial dependence of allele frequencies at the scale of the entire study area, and that we would not have observed signatures of ancient population expansions in both regions. Therefore, our analysis provides phylogeographic evidence that the phenomenon of Lyme disease emergence in North America over the last few decades has been the result of two or more independent emergence events, likely as a consequence of reexpansions of multiple vector tick populations.

A genetic signature of population expansion in both the Northeast and Midwest US was reflected in the sequence mismatch distributions, which also allow for estimates of the timescale of ancient expansion events. The number of mutational steps since expansion, τ, was ≈20 for both populations under all supported expansion scenarios. This parameter is related to time, t, since expansion, according to the formula τ = 2μt, where μ is the mutation rate per nucleotide/year. Without a known mutation rate for B. burgdorferi housekeeping genes, it is difficult to accurately pinpoint the time since the inferred expansion events. However, because B. burgdorferi is a slow-growing bacterium and undergoes few genome replication events during the nine months it spends dormant in overwintering nymphal ticks, its mutation rate is likely slower than most medically-important bacterial taxa for which evolutionary rates have been estimated. Considering the time B. burgdorferi spends during its life cycle reproducing in both its tick and vertebrate hosts, approximations of its doubling time observed in the laboratory (40), and typical rates of spontaneous mutation per generation in bacteria (41), we estimated the mutation rate for B. burgdorferi to be on the order of 10−9 substitutions per site per year. A study of the evolution of a single clone of B. burgdorferi found in both the US and Europe assumed a faster evolutionary rate of 10−6 substitutions per site per year (37). Other studies have estimated prokaryotic mutation rates to be on the order of 10−10 substitutions per site per year (42–44). Even considering a wide range of mutation rates spanning several orders of magnitude (10−10–10−6), the time since expansion of B. burgdorferi in North America indicated by the mismatch distribution parameters is still at least several thousand years, and is probably closer to a million years. This indicates prehistoric spread of B. burgdorferi populations within each of the two regions long before the emergence of modern Lyme disease. This finding further supports our proposed scenario of independent Lyme disease emergence events in the Northeast and Midwest in the last century. If the modern Lyme disease epidemic in the Northeast and Midwest US was a consequence of B. burgdorferi populations spreading from one region to the other in modern times, we would not expect to have observed these signatures of ancient expansion events in populations from both regions. We speculate that the more chaotic mismatch distribution in the Midwest may be a consequence of a more stable demographic history of midwestern populations. Although the distributions we observed are consistent with expansions in both regions, the high level of raggedness of the midwestern distribution compared with the northeastern distribution may indicate a more stable demographic equilibrium in the Midwest with a less sudden expansion relative to the Northeast (45). Indeed, the expansion of B. burgdorferi since 1980 has been much more rapid and from a smaller focus in the Northeast than in the Midwest.

By grouping the 37 STs into clonal complexes with inferred founders, we uncovered clues about the origins and directionality of ancient spread of B. burgdorferi in our study region. Interestingly, all four of the clonal complexes that were identified had founding STs with distributions restricted to a few sites in coastal New England and southern New York State. Although we only identified a small number of clonal complexes, the findings suggest that this pattern is a consequence of an ancient spread of B. burgdorferi in an east-to-west direction. This evidence for an ancient east-to-west spread of B. burgdorferi in the US is of particular interest in light of recent studies that suggest a European origin for B. burgdorferi strains circulating in the US (29, 37). Our finding that all four rooted clonal complexes of B. burgdorferi had ancestral genotypes found exclusively in coastal sites of the Northeast US is compatible with the explanation that if B. burgdorferi migrated from Europe to the US, it would have first arrived in the northeast and spread westward into its current range from there. It is unknown how B. burgdorferi might have been transported across the Atlantic before the arrival of Europeans in the New World, but it has been found in birds on both continents (46, 47).

Finally, the observed B. burgdorferi population structure suggests limited contemporary migration between the two geographically discontinuous regions of activity in the US. Our sampling was extensive enough to show that most STs are widely distributed within the regions; all 17 of the STs that were represented by more than one sample were found to be restricted to either the Northeast or the Midwest, despite the fact that 15 of these were distributed between more than one site within the same region. Migration of B. burgdorferi over long distances via birds is certainly possible, and birds have been implicated in the long-distance translocation of B. burgdorferi-infected ticks (48–50). It is not unlikely that with more sampling we would have observed identical sequences occurring in both regions. Indeed, the sequence divergence that we observed between ST pairs from the Northeast and Midwest was often on the order of only one or two substitutions, much lower than the 21 mismatches that we used to infer the timescale of the major prehistoric expansion event for B. burgdorferi, suggesting that some gene flow has occurred since then. However, transport of B. burgdorferi between northeastern and midwestern I. scapularis populations appears to be relatively infrequent. Regional isolation of B. burgdorferi genotypes may have epidemiological implications due to differences in clinical outcomes associated with specific genotypes (51–55).

We have demonstrated that MLST of B. burgdorferi reveals a signature of ancient demographic processes, including spatial expansions occurring at least on the order of thousands of years ago and more recent divergence between regions. Together, these results suggest that the near-simultaneous B. burgdorferi expansion out of separate relict foci in the Northeast and Midwest US over the last several decades are independent, isolated events.

Materials and Methods

Tick Collections.

Host-seeking ticks were collected from vegetation between May and September in 2004 and 2005. Study site selection and sampling procedures were performed according to a described stratified random sampling procedure (23). Tick sampling was performed in 97 sites in 2004 and in 130 sites in 2005, with 20 sites sampled in both years for a total of 207 potential collection sites located throughout the range of I. scapularis. Each site was visited repeatedly over the summer, with a median of five visits during the season. At each visit, host-seeking ticks were collected by dragging a 1-m2 drag cloth over vegetation in five 200-m transects. The cloth was inspected for ticks at regular intervals, and ticks were preserved in barcoded vials containing 70% ethanol. The geographic coordinates of all transects relative to the map datum WGS84 were recorded using a handheld geographic position system receiver (Garmin). All nymphs were identified to species using the key by Durden and Keirans (56).

DNA Extraction and PCR.

Total DNA was extracted from individual nymphal I. scapularis ticks using ammonium hydroxide (NH4OH). All extracts were tested for the presence of B. burgdorferi DNA using real-time PCR targeting the 16S rDNA (rrs) gene using primers amplifying all known species of Borrelia described in ref. 57. Details of DNA extraction and PCR can be found in SI Materials and Methods.

Multilocus Sequence Typing.

A subset of 78 tick extracts was randomly selected for MLST analysis, described by Margos and colleagues (29). Oligonucleotide primer sequences used for MLST in this study are listed in Table S3. Mixed infections were detected in ≈20% of all samples by visual observation of superimposed chromatogram peaks. Because sequencing was performed on DNA amplified directly from tick extracts and no culture isolates were made, it was impossible to determine the multilocus sequence profiles for the individual genotypes making up a mixed infection; therefore, mixed samples were eliminated from further analyses.

All sequences were compared with described alleles in the B. burgdorferi MLST database (http://borrelia.mlst.net), and new alleles were assigned arbitrary number designations. Allelic profiles already present in the B. burgdorferi MLST database were assigned the corresponding ST number; novel combinations of alleles were assigned new arbitrary ST numbers. Details of MLST can be found in SI Materials and Methods.

Data Analysis.

Nucleotide diversity was calculated for each locus using the START version 2 software (58). A Bayesian phylogenetic tree based on concatenated sequences from the eight loci was calculated using MrBayes version 3.1. Allelic profiles were used to infer clonal complexes using the eBURST algorithm (32), which joins STs to their single and double locus variants and infers the founding ST as that with the highest number of single locus variants.

The software packages DnaSP (59) and ARLEQUIN 3.0 (34) were used to calculate frequency distributions of numbers of mismatches between pairwise sequences and to model expected distributions under the demographic scenarios of stable population size, population expansion and spatial expansion (33, 60). Model fit was examined by calculating the sum of squared deviations of the observed data relative to the model and Harpending's raggedness statistic, a measure of the irregularity of the observed distribution's shape (45). Confidence intervals for mismatch distribution parameters were obtained by performing 1,000 bootstrap replicates using ARLEQUIN (61).

The locations of all samples were mapped using a geographic information system in ArcMap Version 9 (ESRI). Locations were projected into the USA Contiguous Equidistant Conic projection and a matrix of pairwise geographic distances was calculated using the Hawth's Analysis Tools extension for ArcGIS (available at http://www.spatialecology.com/htools). To examine the spatial scale of genetic structuring, Moran's I, a measure of spatial autocorrelation (62, 63), was calculated for individual allele frequencies in 10 equally-spaced distance classes using the software package SPAGeDi (64). For significance testing, observed values of Moran's I were compared with frequency distributions obtained <10,000 random permutations of the data. Details of spatial dependence calculations can be found in SI Materials and Methods.

Acknowledgments

We gratefully acknowledge many field and lab assistants for tick collections and identifications. We thank Jonas Bunikis, Hany Mattaous, Paul Vu, and Bridgit Travinsky for DNA extraction and screening, Sarah Hamer, Roberto Cortinas, Michele Rowland, Jean Tsao, Graham Hickling, and Uriel Kitron for planning and coordinating field work, and James Childs, Gisella Caccone, Scott Glaberman, Paul Turner, Heidi Brown, and Mikaela Keller for helpful discussions. This research was funded by the Wellcome Trust (K.K.), the National Institutes of Health Grant 5R21AI065848-03 (K.K., D.F.), the G. Harold and Leila Y. Mathers Charitable Foundation (D.F.), the USDA-ARS Cooperative Agreement 58-0790-5-068 (D.F.), and the US Centers for Disease Control and Prevention Division of Vector-Borne Infectious Diseases Cooperative Agreement No. Cl00171-01 (D.F.). A.G.H. acknowledges support from the US Centers for Disease Control and Prevention Fellowship Training Program in Vector-Borne Diseases.

Footnotes

  • 3To whom correspondence should be addressed. E-mail: durland.fish{at}yale.edu

  • Author contributions: A.G.H., M.A.D.-W., A.B., K.K., and D.F. designed research; A.G.H. and G.M. performed research; G.M. and K.K. contributed new reagents/analytic tools; A.G.H., S.J.B., and K.K. analyzed data; and A.G.H., G.M., S.J.B., K.K., and D.F. wrote the paper.

  • The authors declare no conflict of interest.

  • This article is a PNAS Direct Submission.

  • This article contains supporting information online at www.pnas.org/cgi/content/full/0903810106/DCSupplemental.

  • Freely available online through the PNAS open access option.

Back to top
Article Alerts
Email Article
Citation Tools
Request Permissions
Phylogeography of Borrelia burgdorferi in the eastern United States reflects multiple independent Lyme disease emergence events
Anne Gatewood Hoen, Gabriele Margos, Stephen J. Bent, Maria A. Diuk-Wasser, Alan Barbour, Klaus Kurtenbach, Durland Fish
Proceedings of the National Academy of Sciences Sep 2009, 106 (35) 15013-15018; DOI: 10.1073/pnas.0903810106
Digg logo Reddit logo Twitter logo Facebook logo Google logo Mendeley logo
Proceedings of the National Academy of Sciences: 106 (35)
Table of Contents

Submit

Article Classifications

Jump to section

You May Also be Interested in

Multi-color molecular model
Enzymatic breakdown of PET plastic
A study demonstrates how two enzymes—MHETase and PETase—work synergistically to depolymerize the plastic pollutant PET.
Image credit: Aaron McGeehan (artist).

Similar Articles