Range Expansion and the Origin of USA300 North American Epidemic Methicillin-Resistant Staphylococcus aureus

Abstract

The USA300 North American epidemic (USA300-NAE) clone of methicillin-resistant Staphylococcus aureus has caused a wave of severe skin and soft tissue infections in the United States since it emerged in the early 2000s, but its geographic origin is obscure. Here we use the population genomic signatures expected from the serial founder effects of a geographic range expansion to infer the origin of USA300-NAE and identify polymorphisms associated with its spread. Genome sequences from 357 isolates from 22 U.S. states and territories and seven other countries are compared. We observe two significant signatures of range expansion, including decreases in genetic diversity and increases in derived allele frequency with geographic distance from the Pennsylvania region. These signatures account for approximately half of the core nucleotide variation of this clone, occur genome wide, and are robust to heterogeneity in temporal sampling of isolates, human population density, and recombination detection methods. The potential for positive selection of a gyrA fluoroquinolone resistance allele and several intergenic regions, along with a 2.4 times higher recombination rate in a resistant subclade, is noted. These results are the first to show a pattern of genetic variation that is consistent with a range expansion of an epidemic bacterial clone, and they highlight a rarely considered but potentially common mechanism by which genetic drift may profoundly influence bacterial genetic variation.IMPORTANCE The process of geographic spread of an origin population by a series of smaller populations can result in distinctive patterns of genetic variation. We detect these patterns for the first time with an epidemic bacterial clone and use them to uncover the clone's geographic origin and variants associated with its spread. We study the USA300 clone of methicillin-resistant Staphylococcus aureus, which was first noticed in the early 2000s and subsequently became the leading cause of skin and soft tissue infections in the United States. The eastern United States is the most likely origin of epidemic USA300. Relatively few variants, which include an antibiotic resistance mutation, have persisted during this clone's spread. Our study suggests that an early chapter in the genetic history of this epidemic bacterial clone was greatly influenced by random subsampling of isolates during the clone's geographic spread.

Keywords: epidemics; fluoroquinolones; founder effects; genetic drift; population genetics; range expansion.

FIG 1

ML phylogeny of USA300 with branch lengths corrected for recombination by CFML. Branch color indicates bootstrap support of the tipward node. Circle 1 indicates clades. Circle 2 indicates an FQ resistance mutation(s). The asterisk indicates an isolate with a resistance mutation in gyrA but not grlA. Circle 3 indicates the geographic sources of isolation.

FIG 2

Signatures of range expansion and the origin of USA300-NAE. Panels A to C show the regressions of θ_π, tip-to-tip distance, and Ψ, respectively, with geographic distance from Pennsylvania, when all isolates (from 2001 to 2011) were used. A solid line indicates the linear regression, and dotted lines indicate the 95% confidence intervals. The maps in panels A to C illustrate the correlations when each of the 15 populations was used as the origin, with interpolation of values between populations. Stronger evidence of origin is shown as yellow. Panels D to F show the results obtained when only recent isolates (from 2007 to 2011) were used. The asterisks on the maps in panels C and F indicate the coordinates of the selected origin when the nonlinear regression of Ψ was used.

FIG 3

Power analysis of signatures of range expansion. USA300-NAE SNPs were randomly downsampled to make 20 new data sets for each bin of downsampled SNPs and tested for a significant origin by using the signature of a decrease in θ_π (panel A) or an increase in Ψ (panel B). Yellow represents those data sets giving a significant origin in the Pennsylvania region (i.e., Pennsylvania, Ohio, New York, or Massachusetts), red represents any other significant origin, and white represents a nonsignificant origin.

FIG 4

Identification of well-sampled derived alleles of USA300-NAE. Panel A shows the relationship between the frequency of derived alleles and the number of colonized populations when all isolates (from 2001 to 2011; black dots), recent isolates (from 2007 to 2011; red dots), or all isolates with allele frequencies binned into one of five 20% bins before averaging across populations (blue dots) were used. Panels B and C show the significant positive frequency gradients for the gyrA and ssa-1 alleles, respectively. The data points, regression lines, and 95% confidence intervals are shown for all isolates (in black) along with the regression lines for recent isolates (red lines) and for all isolates with binned allele frequencies (blue lines). The different origin populations giving the best correlation with the different analyses are indicated in panel C.

FIG 5

Recombination in USA300. Panel A shows 4,109 biallelic SNPs in columns and results of recombination analysis by CFML and BNG in rows. Gray columns indicate nonrecombinant SNPs according to both methods. White columns indicate recombinant SNPs according to both methods. Black columns indicate recombinant SNPs according to one method only. Panel B shows the distribution of the number of branches for which sites are recombinant as inferred by CFML (○) and as expected under the CFML model (×). In panel B, 95% confidence intervals are shown but the intervals are sometimes too small to be visible and sometimes the symbols for observed and expected results overlap.