MRSA Clonal Complex 22 Strains Harboring Toxic Shock Syndrome Toxin (TSST-1) Are Endemic in the Primary Hospital in Gaza, Palestine

Background Methicillin-resistant Staphylococcus aureus (MRSA) is an important pathogen in both community and healthcare-related settings worldwide. Current knowledge regarding the epidemiology of S. aureus and MRSA in Gaza is based on a single community-based carriage study. Here we describe a cross-sectional analysis of 215 clinical isolates collected from Al-Shifa Hospital in Gaza during 2008 and 2012. Methods All isolates were characterized by spa typing, SCCmec typing, and detection of genes encoding Panton-Valentine leukocidin (PVL) and toxic shock syndrome toxin (TSST-1). Representative genotypes were also subjected to multilocus sequence typing (MLST). Antibiotic susceptibility testing was performed using VITEK2 and MicroScan. Results MRSA represented 56.3% of all S. aureus strains, and increased in frequency from 2008 (54.8%) to 2012 (58.4%). Aside from beta-lactams, resistance was observed to tetracycline, erythromycin, clindamycin, gentamicin, and fluoroquinolones. Molecular typing identified 35 spa types representing 17 MLST clonal complexes (CC), with spa 998 (Ridom t223, CC22) and spa 70 (Ridom t044, CC80) being the most prevalent. SCCmec types I, III, IV, V and VI were identified among MRSA isolates, while type II was not detected. PVL genes (lukF/S-PV) were detected in 40.0% of all isolates, while the TSST-1 gene (tst) was detected in 27.4% of all isolates, with surprisingly high frequency within CC22 (70.4%). Both PVL and TSST-1 genes were found in several isolates from 2012. Conclusions Molecular typing of clinical isolates from Gaza hospitals revealed unusually high prevalence of TSST-1 genes among CC22 MRSA, which is noteworthy given a recent community study describing widespread carriage of a CC22 MRSA clone known as the ‘Gaza strain’. While the latter did not address TSST-1, tst-positive spa 998 (Ridom t223) has been detected in several neighboring countries, and described as endemic in an Italian NICU, suggesting international spread of a ‘Middle Eastern variant’ of pandemic CC22 strain EMRSA-15.


Introduction
Staphylococcus aureus is one of the most prevalent human pathogens isolated from hospitalized patients worldwide, and its importance in community settings continues to increase. S. aureus causes a broad variety of diseases ranging from skin and soft-tissue infections to bacteremia, osteomyelitis, infective endocarditis, and necrotizing pneumonia [1][2][3]. In recent decades, MRSA has emerged as the most frequently identified antibiotic-resistant pathogen in many parts of the world, including North Africa and the Middle East [4]. While hospital-acquired MRSA (HA-MRSA) strains remain endemic in most of these regions, in recent years community-acquired (CA-MRSA) strains have emerged as a cause of invasive and life-threatening infections in young, healthy patients with no significant healthcare exposure [2,[5][6][7][8]. Within the past few years, livestock-associated MRSA (LA-MRSA) have posed an additional threat [9][10][11]. Fortunately, whereas HA-MRSA isolates are generally multi-drug resistant, CA-MRSA and LA-MRSA tend to be resistant primarily to beta-lactam antibiotics and, in the case of LA-MRSA CC398, to tetracycline as well.
Molecular epidemiological data about S. aureus and MRSA in the Middle East, including the Palestinian Territories, are generally scarce and insufficient [4,[33][34][35], with current knowledge regarding the epidemiology of S. aureus in Gaza based on a single recent communitybased carriage study [36]. The Gaza Strip (geographic coordinates 31°25 0 N, 34°20 0 E) is a narrow territory (41 km long and 6-12 km wide) along the eastern Mediterranean coast, with tightly-controlled borders abutting Israel and the Sinai Peninsula of Egypt. It is considered one of the most densely populated areas in the world, with a population of about 1.7 million inhabitants (37.5% of the total estimated Palestinian population), and a population density (4,073/km 2 ) nearly ten-fold greater than that of the West Bank (433/km 2 ) [37]. Over the last decade, the socioeconomic situation in Gaza has declined steadily, leaving nearly 80% of the population dependent on international assistance. The healthcare infrastructure within Gaza suffers accordingly from the long-term effects of war, economic isolation, and border closures, with marked scarcity of medical equipment and instrumentation. Recent events such as war, border fighting and regional political unrest have only served to exacerbate the situation [38].
Here we describe the molecular characteristics and antibiotic susceptibilities of clinical MRSA and MSSA isolates collected from the largest medical complex and primary hospital in Gaza (Al-Shifa Hospital) in 2008 and 2012. To the best of our knowledge, this is the first report describing clinical S. aureus infection within the Gaza Strip. We note the high prevalence of a tst-positive CC22 MRSA clone, which is likely related to the 'Gaza strain' described in the aforementioned community-based carriage study [36], as well as the 'Middle Eastern variant' of EMRSA-15 from a recent NICU outbreak in Sicily, Italy [28].

Bacterial isolates
For the present study, 215 unique S. aureus isolates associated with diverse clinical infections were collected from the largest public tertiary referral hospital (Al-Shifa Hospital) in Gaza City. Of the 215 unique S. aureus isolates, 126 were collected in 2008, out of a total unique 1091 bacterial isolates (March 1-July 31); while 89 were collected in 2012, out of a total unique 1121 bacterial isolates (March 1-August 11). Isolates were obtained directly from the clinical laboratory of Al-Shifa Hospital, and represent complete capture of all S. aureus isolates during the stated collection periods. Although a defined sampling strategy was not employed, the collected strains likely reflect the clinical epidemiology of S. aureus in Gaza, since Al-Shifa Hospital is the primary referral hospital for patients from all areas of the Gaza Strip.

Identification of S. aureus and MRSA
S. aureus isolates were identified phenotypically using one or more of the following methods: catalase test, tube coagulase test, Pastorex Staph Plus latex agglutination (Bio-Rad, Hercules, California), and the Staph ID 32 API system and VITEK 2 Compact Gram Positive Card (bio-Mérieux, Paris, France). Multiplex PCR was also used for detection of the nuc (S. aureus species-specific) and mecA (methicillin-resistance) genes [39,40]. DNA for molecular typing was isolated as described previously, with different methods used for the 2008 [18] and 2012 [41] isolates.

Molecular typing
All 215 S. aureus isolates were characterized by staphylococcal protein A (spa) typing [12]; 2008 isolates were spa typed using the Ridom StaphType software [13,42], while 2012 isolates were spa typed using eGenomics software [12,41], with Ridom assignments made using the Spa Server website (http://spa.ridom.de/). In order to avoid confusion, both designations are used throughout this manuscript, with eGenomics spa types prefaced by the term "spa", and Ridom spa types shown in parentheses. All MRSA isolates were subjected to staphylococcal cassette chromosome (SCC)mec typing; 2012 isolates were typed using multiplex real-time PCR [43], while 2008 isolates were typed using both multiplex conventional [44] and real-time [43] PCR. Multilocus sequence typing (MLST) was performed as described previously [15] on a representative subset of 42 isolates, with clonal complexes inferred via eBURST analysis [45]; all other clonal complexes were inferred from spa typing data as described previously [41], using both the Ridom Spa Server website and our own internal eGenomics database. Clonal complex sub-groups with distinct genotypic signatures were classified as individual CCs [8], e.g. ST239 strains were classified as CC239 rather than CC8; ST34 strains were classified as CC34 rather than CC30; and ST291 strains were classified as CC291 rather than CC398.
Production of TSST-1 toxin by 17 representative tst-positive strains was assayed via reversed passive latex agglutination (RPLA), using the Oxoid TST-RPLA Toxin Detection Kit from Thermo Fisher Scientific (Oxoid Ltd., United Kingdom), according to the manufacturer's instructions.

Sequencing of TSST-1 and SaPI element
Long-range PCR assays were employed to partially amplify the SaPI region of three tst-positive isolates. Known SaPI sequences were downloaded from GenBank and used as templates for primer design. Forward primers were designed to target the SaPI integrase gene, while the reverse primer is located within tst and is universally conserved among all known SaPIs. Primer oligonucleotide sequences were as follows: a) Int-F1, GCGTGATGTTTAGCGTCTGA (targets the SaPI integrase in strains N315, Mu50, and RN3984); b) Int-F2, TTTGAAGTTTTGCCT-CAGCA (targets the SaPI integrase in strains RF122 and ED133); c) Int-F3, TTCCCTACAA-GAGATGCAACA (targets the integrases in SaPI68111 and SaPIj50); d) Int-F4, CCACTCATCACTTGCAGCAT) (targets the integrase sequence in Tn557); and e) reverse primer Tst-Rv, GTAAGCCCTTTGTTGCTTGC (conserved in all tst-harboring SaPI elements). Genomic bacterial DNA was isolated using a previously-described method [48], and long-range PCR was performed using the Advantage GC Genomic LA PCR Polymerase Kit (Clontech Laboratories, Inc., Mountain View, CA), according to the manufacturer's protocol. The resulting amplicons were then sequenced by primer walking as described previously [48]. One of the three identical SaPI sequences was submitted to GenBank (NCBI) as accession number KP334121.

Statistical analysis
Comparisons of selected data from 2008 and 2012 were performed by Chi-square analysis using GraphPad Software (San Diego, California). Chi-square tests (two-tailed) with p-values 0.05 were considered statistically significant.

Ethics Statement
The study protocol and data handling were approved by the ethical committee at the laboratory medicine department of Al Azhar University-Gaza, which does not have an Institutional Review Board (IRB) at this time. The study was performed in accordance with the ethical standards established in the 1964 and 1975 Declaration of Helsinki, and the modifications thereafter.

Antimicrobial resistance
Resistance to multiple classes of antimicrobials was observed among both MRSA and MSSA, including aminoglycosides, beta-lactams, cephalosporins, fluoroquinolones, fusidic acid, macrolides, and tetracycline. Since different susceptibility testing methods were used in 2008 (VITEK 2) and 2012 (MicroScan), direct comparisons were only possible for 15 of the total 35 antimicrobials tested. Fig. 1
As mentioned, the overall frequency of CC22 decreased slightly from 35.7% (45/126) in 2008 to 29.2% (26/89) in 2012, but there were notable changes in the strains mapped to this clonal complex over the four-year period. Among 79 strains classified as CC22, there were a total of eight different spa types, all identified as ST22 by MLST (with the exception of a single strain that typed as ST737, a single locus variant). Within the 2008 CC22 isolates, four different spa types were observed, with spa 113 (t005) and spa 998 (t223) accounting for 31.1% and 42.2% of these, respectively. Within the 2012 CC22 isolates, four different spa types were also observed, with spa 998 (t223) accounting for 64.7% of these; by contrast, the frequency of spa 113 (t005) was much lower in 2012 (14.7%), while spa 1599 (t7063) was not detected. All of the A total of 17 tst-positive strains were selected in order to assay for TSST-1 toxin production, including 13 CC22 strains encompassing all observed spa, SCCmec, and PVL genotypic combinations; two representatives of each tst-positive CC30 spa type; and each of the singleton CC80 and CC239 tst-positive strains. Production of TSST-1 was determined via reverse passive latex agglutination (RPLA) as per the manufacturer's instructions. All of the strains tested were positive for TSST-1 production, including those which also harbored genes for PVL.

SaPI characterization
In order to better characterize the tst-positive'Gaza strain', the SaPI elements from three tstpositive spa 998 (t223) strains were sequenced, including two MRSA (from 2012) and one MSSA (from 2008). Long PCR yielded a 10.5 kb amplicon using primer set intF1 and tst-Rv, and the resulting tst-bearing SaPI elements were completely sequenced and compared against the NCBI genome database. All three SaPI elements were identical, while BLAST results indicated >99.9% similarity to the sequence of SaPI2 from S. aureus "Harrisburg" strain RN3984 [49], with only 3 nucleotide differences (SNPs) within 10,578 bp. As described in the Methods, one of the three SaPI element sequences (from strain BK 38088) was submitted to GenBank (accession number KP334121).

Discussion
The Palestinian Territories consist of two discontinuous areas: a) the West Bank, which shares a large border with Israel and Jordan; and b) the Gaza Strip, a small territory on the Mediterranean coast which shares most of its border with Israel, except for an 11-km southern border with Egypt. Since 2006, Gaza has been a semi-autonomous region, subject to border and travel restrictions enforced by Israel and Egypt. Although residents of Gaza occasionally travel to Israel for advanced medical care, Egypt is the primary portal for travel to the outside world, consisting mainly of students and foreign workers. The Egyptian border crossing in Rafah often serves as a portal for movement of people and goods, but has been increasingly subject to closure recently. As a result, the population of Gaza is somewhat more isolated than that of the West Bank, and may therefore be considered a "semi-closed" population [50].
Data regarding S. aureus and MRSA within the Palestinian Territories are generally scant [4,35]. Previous reports have mostly focused on Northern Palestine [34,51] and the West Bank [33], with relatively little detail regarding molecular epidemiology. Recently, a joint Israeli-Palestinian community-based nasal carriage study addressed the epidemiology of S. aureus in Gaza for the first time, highlighting the widespread presence of a CC22 clone dubbed the 'Gaza strain' [36], which was subsequently described in two reports from Sicily, Italy [28,29]. In this study, we describe the molecular characteristics of clinical isolates collected from the primary tertiary-care hospital in Gaza in 2008 and 2012. We observed a similarly high prevalence of CC22, including a clone with the same spa type (998/t223) as the aforementioned Gaza strain. Moreover, we detected unusually high frequencies of CC22 isolates that harbor the gene (tst) coding for toxic shock syndrome toxin (TSST-1). To the best of our knowledge, this is the first report describing the molecular epidemiology of S. aureus clinical infection in the Gaza Strip.
The overall MRSA prevalence in this study was 56.3%, representing an increase of 3.6% from 2008 (54.8%) to 2012 (58.4%). Nosocomial MRSA prevalence in the Eastern Mediterranean was previously described as the highest in the Mediterranean region, with overall frequencies ranging as high as 52% (Egypt), 55% (Cyprus), and 56% (Jordan) [52]. Recent studies in Lebanon [53], Israel [54], Saudi Arabia [27] and Jordan [55] have noted nosocomial MRSA rates of 30%, 35%, 50%, and 62% respectively, while the only nosocomial report to date from Palestine (West Bank) described a MRSA prevalence of only 8.7% [51]. All other reports of MRSA prevalence in the Palestinian Territories have involved nasal carriage rather than infection [33,34,36], although notably, 45% of the S. aureus isolates from the aforementioned Gaza carriage study were methicillin-resistant [36]. Therefore, the MRSA frequencies described in this report may be considered among the highest described in the region thus far.
Molecular analysis revealed a surprising degree of genotypic diversity. A total of 35 different spa types corresponding to 17 MLST clonal complexes were identified, with spa type diversity increasing from 2008 to 2012. Consistent with the S. aureus literature, MRSA strains showed less genetic diversity than MSSA, and as observed in the carriage study from Gaza [36], CC22 was the predominant lineage, with spa 998 (t223) accounting for half (50.7%) of the CC22 isolates. Interestingly, much more diversity was found in the current hospital-based study than in the carriage study, possibly due to the longer sampling period in our study. Whereas spa 998 (t223) was the only CC22 spa type found in the carriage study, accounting for 64% of all MRSA isolates, in this study a total of six different CC22 spa types were identified, with spa 998 (t223) accounting for only 27.3% of all MRSA isolates. Moreover, whereas the carriage study only identified three additional MRSA clonal backgrounds (CC5, 78, and 80) and four MSSA clonal backgrounds (CC15, 22, 45, and 291), the present study identified nine MRSA-associated clonal backgrounds (CC5, 6, 8, 22, 30, 80, 121, 239, and 1153) and fifteen MSSA-associated clonal backgrounds (CC1, 5, 6, 8, 15, 22, 30, 34, 88, 97, 121, 133, 291, 398, and 1153). Nevertheless, the high prevalence of the Gaza strain in the carriage study argues in favor of a community-based origin [36], and suggests that the high levels of MRSA carriage (12-13%) observed in the community may serve as a reservoir for the nosocomial CC22 lineages described in our study [25].
Notably, the predominant clonal backgrounds and spa types observed in both Gaza studies differ markedly from those described in the West Bank and Israel. Previous reports from the West Bank have identified CC1 and CC5 [33], as well as CC239 [56], while reports from Israel have described CC5 [54,57], CC8 [54,57], and CC45 [57,58] as the dominant clonal backgrounds. Unfortunately, molecular epidemiologic data from Egypt are exceedingly rare, consisting of two studies featuring small numbers of PVL-positive CA-MRSA isolates from lineages also found in this study [59,60], and, more recently, a nasal carriage study involving outpatients in Egypt and Saudi Arabia [61]. By contrast, more detailed studies from Lebanon [2,32,53] and Jordan [24,25,32,55,62] reveal a number of spa types and clonal backgrounds also detected in this study. The contrast between the molecular epidemiology of S. aureus in Gaza and Israel is noteworthy, given that residents of Gaza frequently seek advanced medical care within Israel. Interestingly, whereas SCCmec types I, II, and V are common in Israeli hospitals [7,54], types I and V are relatively uncommon in Gaza, while type II was not detected in this study.
Within the Middle Eastern region, 2/16 (12.5%) EMRSA-15 isolates from Kuwaiti hospitals [26]; 4/21 (19.0%) CC22 isolates from a hospital in Abu Dhabi [23]; 6/30 (20.0%) CC22 isolates from a hospital in Riyadh (Saudi Arabia) [27]; and 5/5 (100.0%) spa 998 (t223) isolates from a community-based carriage study in Jordan [24] were all identified as tst-positive, while a more recent study in Jordan identified tst-positive spa 998 (t223) as the predominant MRSA strain among nasal isolates from healthcare workers and community-based individuals [25]. As of this writing, a carriage study in Egypt and Saudi Arabia also identified spa 998 (t223) in nasal isolates from outpatients, but presence of tst was not addressed [61]. Meanwhile, a large study in southern India uncovered 6 clinical and 7 nasal EMRSA-15 isolates which were all positive for both PVL and TSST-1 [3]. Similarly, isolates from 2005-2007 referred to the national Staphylococcal Reference Laboratory in the United Kingdom included five CC22 strains harboring both TSST-1 and PVL [30]. Lastly, the first report of an EMRSA-15 variant in the United States involved a tst-positive isolate from a 10-month old infant which was likewise described as spa 998 (t223), SCCmec IV, and PVL-negative [31].
In this study, TSST-1 genes were also detected in single isolates of CC80 and CC239, suggesting that CC22 strains may be serving as a TSST-1 reservoir for other clonal backgrounds in Gaza. The CC80 isolate also harbored PVL, and the presence of both genes in an isolate corresponding to the Gaza strain suggests the potential for greater numbers of isolates harboring genes for both toxins. These observations are supported by the detection of both PVL and TSST-1 genes in 3/5 spa 998 (t223) isolates obtained from Shohada Al-Aqsa Hospital in Gaza during the same collection period in 2012 (data not shown). Moreover, in a recent regional survey which identified TSST-1 genes in 18/31 (58.1%) CC80 strains from Jordan, 7 also harbored PVL genes [32]. As the authors of that study note, while carriage of both PVL and TSST-1 is rare in S. aureus, a previous study from Jordan also identified two CC80 isolates harboring the genes for both toxins [24]. Likewise, the aforementioned studies from India [3] and the U.K. [30] also observed simultaneous carriage within several clonal backgrounds, including CC22. Although the clinical implications of harboring both toxins are unclear at this time, a recent study suggests they may not be expressed at similar levels during infection [64].
Our study has several limitations. As mentioned earlier, very little clinical and demographic data were available, thereby limiting the possibility of epidemiological inference, including a clear distinction between healthcare-associated (HA) vs. community-associated (CA) infections. Likewise, morbidity and mortality data was not available, leaving the clinical implications of PVL-and TSST-1 bearing strains unresolved. Future studies should aim to further characterize TSST-1 harboring CC22 clones, in Gaza as well as other countries where they have been reported, in order to discern whether such strains are spreading internationally, or whether CC22 strains are acquiring TSST-1 independently. A recent phylogenomic analysis of 193 spatio-temporally diverse strains detailed the population structure of CC22, clearly distinguishing EMRSA-15 from various other lineages. However, strains from the Middle East were not included in the study, while only two MSSA strains were observed to harbor TSST-1, highlighting the unique nature of the Gaza strain among known CC22 lineages [63].
Given the pandemic spread of EMRSA-15, the potential for acquisition of TSST-1 from CC22 strains such as the ones described in this study is of significant clinical concern. Moreover, in light of recent reports from Sicily [28,29], surveillance programs should endeavor to monitor spread of the Gaza strain to other regions [25], as well as the spread of TSST-1 to other clonal backgrounds, and further occurrence of both PVL and TSST-1 within the same strain. Lastly, clinical epidemiology studies should be undertaken to determine risk factors, transmission patterns, and clinical outcomes for CC22 strains harboring TSST-1 [25], given the high rate of endemicity observed in multiple studies including this report [25,28,29] Author Contributions