Introduction

The emergence of community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) is causing a worldwide public health problem. A limited variety of genetically distinct S. aureus strains is driving this epidemic in different geographic locations, such as ST8 (USA300) in the United States [1], ST80 in some European countries [2–4], or ST75 in remote indigenous Australian communities [5]. Subsequent global spreading of these clones has been detected, such as USA300 into Australia [6], Europe [7, 8], or Japan [9]. However, the pandemic spread of S. aureus is not limited to the presence of methicillin resistance: methicillin-susceptible S. aureus (MSSA) phage 80/81 was the first to sweep from hospitals in Australia to Europe, Africa, and the Americas in the 1950s [10]. A more recent example of a very successful MSSA includes ST121, which has been encountered in several European countries and parts of Russia [11–13]

Increasingly, it has been noted that returning international travelers with MRSA infections contracted strains specific to their country of vacation [14, 15]. However, it remains unclear as to what extent frequent travel and migration between regions influences the local epidemiology of S. aureus infections. We recently obtained evidence for a possible “air-bridge link” for the exchange of ST398 MSSA between the predominantly Dominican population of Northern Manhattan and the Dominican Republic (DR) [16]. This prompted further interest into the molecular epidemiology of S. aureus infections in the Caribbean, a major destination of international travel and source of agricultural export [17]. There are substantial differences in the country of origin and frequency of travelers arriving between Caribbean islands. The DR is one of the most frequently visited countries in this region, with ~4 million tourists arriving per year, including about 40% Europeans (from Germany, Italy, and the UK), 33% American, and 17% Canadian visitors, and an additional ~580,000 Dominicans returning from living overseas. In contrast, the French overseas department Martinique is the destination of ~500,000 visitors each year, with the majority (75%) arriving from France and only 3.6% from North America [17].

We, therefore, aimed to characterize S. aureus isolates from the DR, located in the Northern Caribbean and linked by travel to both North America and Europe, and contrast this with S. aureus from the South-Eastern Caribbean island Martinique, which, by air travel, is primarily connected to France. We hypothesized that these two countries with differing travel links would have a substantially distinct molecular epidemiology of S. aureus infections.

Materials and methods

Sample selection

We obtained a convenience sample of 112 clinical S. aureus isolates processed by the Referencia laboratory in Santo Domingo, DR, between February 2007 and July 2008. These strains were collected from outpatients who presented to clinics in Santo Domingo (n = 98) or across the country (n = 14), and basic demographic information was provided (Table 1). The majority of infections (n = 86, 77%) were skin and soft tissue infections (SSTIs), while the remainder (n = 26) included conjunctivitis, otitis, and urinary tract infections.

Table 1 Distribution of selected demographic and clinical variables between the Dominican Republic and Martinique
Full size table

In Martinique, we collected a convenience sample of 143 positive S. aureus isolates between August 2007 and April 2008 from outpatients presenting to Fort-de-France hospital in Fort-de-France, Martinique (Table 1). Demographic information was available for 94 of these samples. The majority of specimens were derived from SSTIs (n = 55, 59%) and the remainder was collected from bloodstream infections (n = 17, 18%), pneumonias (n = 9, 9.6%), or urinary infections (n = 10, 11.1%) or unspecified sites (n = 3).

In both sites, none of these samples were obtained from international tourists, or were specifically collected during a local outbreak of S. aureus infections.

Antimicrobial susceptibility testing

All S. aureus isolates were tested for their susceptibilities to penicillin, cefoxitin, oxacillin, erythromycin, tetracycline, levofloxacin, gentamicin, vancomycin, clindamycin, trimethoprim–sulfamethoxazole, and rifampin, using the Kirby–Bauer standard disk diffusion method according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [18].

Molecular genotyping

All isolates were characterized by S. aureus protein A (spa) typing as described previously [19] and pulsed-field gel electrophoresis (PFGE). spa polymerase chain reaction (PCR) products were sequenced and spa types automatically assigned using Ridom StaphType software (version 1.5) and compared to http://spaserver2.ridom.de. Using the integrated BURP (Based Upon Repeat Patterns) algorithm, spa types were clustered into spa clonal complexes (spa-CC) if the cost distances were <4 [20, 21]. spa types with <5 repeats were excluded, as they cannot be reliably clustered [21]. For visualization, phylogenetic trees were constructed using SplitsTree software (version 4.0) [22].

PFGE was carried out on SmaI or Clf91 digested samples [16, 23]. The resulting band patterns were analyzed by Bionumerics software (version 4.0, Applied Maths, Ghent, Belgium) to determine the relatedness between strains [24]. Profiles with >80% similarity were considered to be closely related. Multilocus sequence typing (MLST) was performed on at least one isolate of each spa-CC that also shared indistinguishable PFGE patterns as previously described [25]. Sequence types (STs) were assigned using http://saureus.mlst.net/.

Further screening for the presence of Panton–Valentine leukocidin (PVL) [26] and for the type of staphylococcal chromosomal cassette (SCC)mec for MRSA isolates was performed by PCR as previously described [27].

Statistical analysis

All statistical analyses were performed using SPSS 18 software. Chi-square tests were used for the comparison of dichotomous variables, and Fisher’s exact test with an expected cell count of <5 and independent samples t-tests to compare age. A p-value of <0.05 was considered to be statistically significant.

Results

Dominican Republic

Most of the S. aureus isolates from the DR were MSSA (90/112; 80%). Of these MSSA strains, 81/90 (90%) were resistant to penicillin, 11/90 (12%) to erythromycin, and 6/90 (7%) to tetracycline, but less than 1% of isolates were resistant to levofloxacin, gentamicin, or clindamycin. The majority of the 22 MRSA isolates harbored SCCmec IV (18/22, 80%), except for four strains with SCCmec V. Five of the 22 MRSA isolates were resistant to ≥4 antibiotics, whereas the remainder were only resistant to semi-synthetic penicillins or one additional class of drug.

Among the 112 isolates, 39 different spa types were detected, which clustered into ten distinct spa-CC by BURP analysis. One spa type was excluded (length <4 repeats) and 16 were classified as singletons (Figs. 1 and 2). Nine of the 39 different spa types had not been previously reported (t6018, t6019, t6711, t6721, t6723, t6724, t6725, t6726, and t6839).

Fig. 1
figure 1

Observed frequency of spa clonal complexes/multilocus sequence typing (MLST) for methicillin-susceptible Staphylococcus aureus (MSSA) (a) and methicillin-resistant Staphylococcus aureus (MRSA) (b) strains by geographic region (blue = Martinique, red = Dominican Republic). E = excluded from cluster analysis for short spa length. S = singleton (no cluster assigned or number <3). *More than one spa type as founder; **unknown founder

Full size image
Fig. 2
figure 2

Neighbor-joining tree based on the distance matrix produced by BURP (Based Upon Repeat Patterns) software. All observed spa types by country of origin are as follows: blue = Martinique, red = Dominican Republic, black = present in both countries. Underlined spa types were not clustered in a BURP clonal complex (singleton). The indicated MLST clonal complex corresponds to all spa types in a given branch, unless otherwise indicated

Full size image

The predominant spa type among the MSSA isolates from the DR was t665 (n = 25, 28%). These strains were further defined as ST30 by MLST typing. Using BURP cluster analysis, Spa-CC665/ST30 was the predominant CC and accounted for one-third of isolates (30/90, 33%; Fig. 1). Spa-CC002 (ST5) and t571 (ST398) each accounted for seven samples (7.8%). Notably, we only detected 3/90 (3.3%) t008 USA300 MSSA strains in this sample collection, while Spa-CC008 (ST8) accounted for 6/90 (6.6%) isolates (Fig. 1).

Among the MRSA isolates from the DR, Spa-CC665/ST30 was again the predominant CC (6/22, 27%), followed by Spa-CC148/ST72 (5/22, 23%) and Spa-CC002/ST5, (4/22, 18%) (Fig. 2). None of the MRSA samples in this study was USA300 or ST8.

Further genotyping revealed that about half of the MSSA (41/90, 46%) and MRSA (10/22, 45%) strains were PVL-positive. The majority of these PVL-positive isolates belonged to Spa-CC665/ST30 (31/51, 61%).

Martinique

Of the 143 Martinique S. aureus samples, 87 were MSSA and 56 were MRSA. About one-third of MSSA strains remained sensitive to penicillin, while only a small proportion of strains were resistant to erythromycin (13/87) or tetracycline (5/87). The majority of MRSA strains harbored resistance to multiple drugs, including erythromycin (31/56) and levofloxacin (46/56), while sensitivity to tetracyclines was mainly preserved (2/56).

We identified eight new spa types in Martinique (t5467, t5468, t5521, t5522, t5526, t5527, t6046, and t6049), and one novel MLST type (ST1793). MSSA infections were caused by a genetically heterogeneous group of strains (Fig. 1). The two predominant clusters, spa-CC1096/ST152 (13/88, 15%) and spa-CC571 (9/88, 10%), together accounted for 25% of samples (Fig. 1).

All strains in the Spa-CC571 cluster were confirmed as ST398 by MLST and shared profiles (>80% identical) by PFGE, irrespective of their country of isolation or spa type (not shown). This included t571, which we also encountered as the predominant ST398 spa type in both Northern Manhattan [16] and in the DR, as well as t1451 and t3085.

Martinique MRSA isolates clustered into ST80, ST5, and spa-CC008/ST8, which caused the majority of MRSA infections (44/55, 80%; Fig. 2). Seven of these spa-CC008 samples were t008 and USA300 by PFGE (not shown), but spa type t304 (27/55, 49%) was the predominant clone and differed substantially from USA300 by PFGE (~65% similarity). All t304 samples were SCCmec type IVc, PVL-negative, and ST8. Overall, only 5 of the 56 MRSA strains and 9 of the 87 MSSA strains were PVL-positive.

Combined assessment of S. aureus isolates from the DR and Martinique

While we attempted to obtain comparable collections of outpatient S. aureus isolates from both Caribbean countries, isolates from Martinique were derived from an older population (Table 1). In addition, Martinique patients more frequently had invasive infections (Table 1), with greater resistance to multiple antibiotics, indicating that, at least in part, samples may have been healthcare-associated. Despite these epidemiological differences, ST5 and ST80 were shared MRSA clones between the two countries, though the majority of infections were caused by ST8 in Martinique and ST30 and ST72 in the DR (Fig. 1b). When comparing the frequency of MSSA clones between the two Caribbean countries, the highly prevalent ST30 in the DR was only infrequently encountered in Martinique (Fig. 1a). The two most commonly shared CC present in both countries were spa-CC002/ST5 and spa-CC571/ST398 (Fig. 1).

Across the two islands, we identified 68 distinct and 11 shared spa types between Martinique and the DR (Fig. 2). To further elucidate the genetic relatedness of these strains, we generated a neighbor-joining tree based on the distance matrix produced by the BURP algorithm (Fig. 2). This phylogenetic tree illustrates that, while there was a high diversity in the presence of individual spa types between Martinique and the DR, usually at least one spa type per CC was shared, except for ST7, ST20, and ST152. This may suggest local diversification of a very similar pool of common strain ancestors.

Discussion

There is limited information on the molecular epidemiology of S. aureus infections in the Caribbean, a major destination of international tourism. Here, we report the presence of S. aureus clones in the DR and Martinique that previously had only been observed elsewhere. In addition, the presence of a substantial number of new spa types in both regions suggests recent local diversification of strains not otherwise encountered in other geographic areas. Furthermore, both Caribbean countries frequently shared MSSA genotypes, in particular, ST5 and ST398, but showed major differences between MRSA clones, which, in part, may be explained by disparities in the clinical sample collections tested.

Dominican MSSA and MRSA strains harbored STs commonly encountered in other parts of the world, such as ST80, in the Mediterranean, Balkan, Middle East, and Europe; ST30, in the Pacific, East Asia, and Oceania; and ST72 [2, 28, 29]. The latter strain causes CA-MRSA infections in South Korea and was also detected among immigrants from South Korea to Europe. However, the predominant South Korean spa types (t664, t324) differed from our ST72 spa types [29, 30]. This may still be consistent with the transmission of strains between regions, or, alternatively, could be explained by the much earlier dissemination of common ancestors and subsequent diversification of clones. In support of this diversification hypothesis is our observation that the presence of individual STs and spa-CC did not differ substantially between the two Caribbean countries (Fig. 2). This observation even applied to more uncommon STs, such as ST1472, which differs from ST30 in one base pair in the yqiL gene and was present as a new spa type in both Martinique (t6726) and the DR (t5521).

The high prevalence of ST30-PVL+ among Dominican MSSA and MRSA strains is consistent with recent reports indicating their global spread, including to the United States and Europe [23, 27, 28]. Interestingly, the first recognized pandemic S. aureus clone, phage 80/81 [10], also belongs to ST30. The remarkably widespread occurrence of ST30 suggests unique features in its core genome that potentially facilitate their transmission. However, the predominance of spa type t665 among Dominican ST30 MSSA and MRSA isolates suggests the emergence of MRSA from a locally successful MSSA t665/ST30 strain.

Only a very small proportion of the Martinique MSSA and none of the MRSA isolates belonged to ST30. The vast majority of MRSA from Martinique were spa t304, which is closely related to t008, the prototype of epidemic ST8/USA300 in the United States. These t304 strains differed substantially from USA300 by PFGE and none were PVL-toxin-positive, but they all belonged to ST8 and carried the SCCmec type IVc cassette. This specific spa type has infrequently been described as CA-MRSA in South Korea and in parts of Europe [30]. Spa-t304 was only rarely detected among MSSA isolates and suggests that MRSA t304/ST8 evolved independently of a successful MSSA counterpart. Alternatively, this could also be consistent with a local outbreak of t304 MRSA strains.

In contrast to the DR, a substantial number of MSSA infections in Martinique were due to ST152, which has been associated with colonization in Mali and infections in Nigeria [31, 32]. However, in both Caribbean countries, ST398 MSSA accounted for a substantial number of infections (~10%). ST398 has been linked to MRSA outbreaks among Dutch pig farmers, though transmission beyond the immediate animal contacts or their families appears rare [33]. While we cannot exclude animal contact of infected patients in our sample, it is notable that the Dominican and Martinique ST398 isolates were collected at two urban medical centers (population of ~2 million in Santo Domingo and ~90,000 in Fort-de-Frances). This adds to our earlier observation in Northern Manhattan, where ST398 MSSA was detected in individuals without direct animal exposure [16].

Spa-t571 was the most frequently encountered ST398 type in our samples from New York [16], the DR, and Martinique, and was also attributed in a case of fatal necrotizing pneumonia in France [34]. Both additional ST398 spa types from Martinique, namely, t3085 and t1451, have almost exclusively been detected in France. This raises the possibility that ST398 strains have either been imported from France into Martinique or vice versa. Detailed sequence comparisons may aid in gaining further understanding into the origin and dissemination of these strains.

There are several limitations to our study. First, samples are mainly representative of single centers in both countries, with the exception of some isolates (14%) collected across the DR. Second, while we aimed to obtain a representative snapshot of community-associated S. aureus infections in both sites, we were only able to collect a convenience sample with limited clinical or epidemiological information, such as co-morbidities, or recent hospitalizations or travel of patients. In particular, the higher age of patients from Martinique, the relatively high numbers of invasive isolates, and fluoroquinolone resistance suggests that a number of infections were related to healthcare exposure. Third, while the samples were not collected during known outbreaks of S. aureus infections at the study sites, we cannot fully exclude the possibility of clusters of infections accounting for some of the predominant clones. Fourth, we only obtained samples from one time period and were, therefore, unable to study changes in predominant strain patterns over time. Nevertheless, the results provide a profile of strain diversity in this previously unexamined geographic region.

It has been suggested that MSSA are less restricted to a particular geographic region than MRSA strains [35]. However, in the current study, we primarily observed the presence of a diversity of the international strains ST8, ST30, ST80, and ST72 in MRSA samples from the Caribbean. In light of reports of travelers returning to their home countries with S. aureus skin infections endemic to their travel destination [14], a tourist region visited by a diversity of travelers may also serve as a melting pot for MSSA and MRSA exchange and transmission.

Further prospective studies are warranted in order to more specifically study the molecular epidemiology of MSSA and MRSA infections in relation to “air-bridges” between countries with frequent travel or migration, such as in the Caribbean.