Carriage frequency, phenotypic, and genotypic characteristics of methicillin-resistant Staphylococcus aureus isolated from dental health-care personnel, patients, and environment

There is limited data on methicillin-resistant Staphylococcus aureus (MRSA) carriage in dental clinics. 1300 specimens from patients, health personnel, and environmental surfaces of a dental clinic in Egypt were tested for MRSA. Antibiotic susceptibility, biofilm formation, Staphylococcal protein A (spa) typing, SCCmec typing, and PCR-based assays were used to detect mecA, mecC, vanA, Panton-Valentine Leukocidin toxin (PVL), and toxic shock syndrome toxin-1 (tst) genes. Among 34 mecA-positive MRSA isolates, five (14.7%) were PVL-positive, seventeen (50%) were tst-positive, ten (29.4%) were vanA-positive, while none harboured mecC. MRSA hand carriage rates in patients, nurses, and dentists were 9.8%, 6.6%, and 5%. The respective nasal colonization rates were 11.1%, 6.7%, and 9.7%. 1.3% of the environmental isolates were MRSA-positive. Strong and moderate biofilm-forming isolates represented 23.5% and 29.4% of MRSA isolates. 24 MRSA isolates (70.6%) were multi-resistant and 18 (52.9%) harboured SCCmec IV. Among eight spa types, t223 (26.5%), t267 (23.5%), and t14339 (23.5%) were predominant. We noted an alarming genetic relatedness between 7 (20.6%) MRSA isolates and the epidemic EMRSA-15 clone, as well as a combined occurrence of tst and PVL in 3 (8.8%) isolates. Results suggest high MRSA pathogenicity in dental wards highlighting the need for more efficient surveillance/infection control strategies.


Results
Prevalence of MRSA and other staphylococci. In this study, a total of 1300 swab specimens were collected from six different wards within a dental clinic in Egypt, including: 1030 (79.2%) specimens from environmental surfaces and 270 (20.8%) specimens from hands (n = 182) and anterior nares (n = 88) of both patients and DHCPs. These 1030 specimens from environmental surfaces fall into two categories, those from clinical-contact surfaces (n = 602) and those from housekeeping surfaces (n = 428) ( Table 1).
The isolates recovered from housekeeping surfaces demonstrated a lower prevalence of S. aureus as compared to those recovered from clinical-contact surfaces; however, this difference was statistically non-significant [4.3% versus 8.4%; P = 0.053 by Fisher's exact test]. For clinical contact surfaces, the dentists' chairs had the highest prevalence of S. aureus (8%), followed by dentists' drills (7.1%) and patients' faucet sinks (6.2%). For housekeeping surfaces, door knobs had the highest prevalence of S. aureus (2.68%), followed by disinfectant containers, floors, and light switches, which had equal prevalence rates of 7.1% each.

Characterization of the MRSA isolates.
Genetic groups based on spa typing and SCCmec typing. The spa typing analysis revealed 8 distinct spa types within the tested MRSA isolates. The spa type attribution of each isolate is reported in Table 2. The frequencies, geographical spread, and repeat successions for each identified spa type are reported in Table 4. The spa type including the largest number of isolates was t223 (n = 9, 26.5% of all tested MRSA isolates). This was immediately followed by t14339 and t267 (each of which contained 8 isolates, 23.5%). The other spa types were less frequent, including t084 (n = 3), t3689 (n = 2), t380 (n = 2), t8506 (n = 1), and t1339 (n = 1).
Two of the most frequent spa types in our study (t223 and t267) were disseminated in different wards, since they were recovered from five out of the six tested wards (Table 5). Conversely, the spa type 14339 was mainly predominant in the dental surgery ward (62.5%, 5/ 8).
Only four spa types (t223, t14339, t267, and t084) coexisted in both personnel and environmental surface specimens, with higher frequencies of t267 and t084 in personnel specimens, an equal distribution of t14339 in both specimen categories, and higher frequency of t223 in environmental surface specimens (Table 5). On the other hand, the other four spa types (t3689, t8506, t1339, and t380) were only detected among isolates from personnel ( Table 5).
The majority of the tested MRSA isolates (n = 24, 70.6%) were multidrug resistant (non-susceptible to at least one agent in three or more of the tested antimicrobial classes, other than β-lactams). In this regard, the non-susceptibility rates, which include both intermediate and resistant isolates, to two, three, four, five, six, seven, and eight antimicrobials were 5.9%, 8.8%, 29.4%, 26.5%, 17.6, 5.9%, and 2.9%, respectively. There was no significant difference in the prevalence of multidrug resistance between the MRSA isolates recovered from environmental surfaces and those isolated from personnel (76.9% versus 66.67%, P = 0.704 by Fisher's exact test).
The full resistance rates were generally higher for personnel isolates than for environmental surface isolates for all tested antimicrobial agents. For cefoxitin, no difference was found between the two specimen categories. For chloramphenicol, doxycycline, erythromycin, and vancomycin, full resistance rates for the environmental surface isolates were higher than their personnel counterparts.
A comparison of the occurrence of antimicrobial resistance among the investigated MRSA isolates in relation to different clonal lineages is presented in Table 7. A significantly higher frequency of ciprofloxacin resistance was recorded among isolates from spa type t3689 than other types (χ 2 = 14.62, df = 7, P = 0.0412). Similarly, resistance to clindamycin occurred at significantly higher frequencies in MRSA with spa types t3689 or t380 as compared to other types (χ 2 = 15.8, df = 7, P = 0.027). For the remaining antimicrobial agents, non-significant differences in resistance were observed between the spa types identified in the current study ( Table 7).
Ten (29.4%) of the tested MRSA isolates (distributed equally between environmental surfaces and personnel; n = 5 for each group) were positive for the vanA gene. Four (40%) of these isolates were associated with spa-CC223, four (40%) were associated with multiple spa types, and two (20%) had a spa type (t380) that was excluded from clustering (Tables 2 and 4). The vanA gene was detected only in isolates harbouring SCCmec type IV (n = 7, 70%) and SCCmec type I (n = 3, 30%) ( Table 6).
Five (14.7%) out of the 34 MRSA isolates contained the PVL gene, which seemed to be more associated with environmental surfaces than with personnel isolates (15.38% versus 14.29%), but this difference was non-significant (P > 0.9999 by Fisher's exact test). While all of these PVL-positive MRSA isolates carried SCCmec type IV (Table 6), the majority (80%, 4/5) were associated with spa-CC223, while only one isolate (20%) had a spa type (t380) with non-predictable CC (Tables 2 and 4). Additionally, three of the PVL-positive strains were also positive for the tst gene, one of which also harboured the vanA gene (  Biofilm formation. From the 34 MRSA isolates tested for biofilm formation, 8 (23.5%) isolates were classified as strong biofilm producers, 10 (29.4%) were moderate, 7 (20.6%) were weak, and 9 (26.5%) were non-biofilm producers (Fig. 2). This classification was based on the criteria established by Stepanovic and colleagues 34 . All isolates from personnel were shown to be moderate, weak, or non-biofilm producers, except two isolates with strong biofilm-forming ability ( Table 2, IDs: 7 and 93), which were derived from hand and nasal swabs of two different patients attending the prosthetic dentistry and prosthodontic wards, respectively. For isolates recovered from enviromental surfaces, only 7.7% (1/13) of the isolates were non-biofilm formers, while the rest were biofilm formers [46.2% (6/13) strong, 23% (3 /13) moderate, and 23% (3/13) weak] (Fig. 2).
There was no statistically significant difference in biofilm-forming abilities between MRSA isolates recovered from personnel and those recovered from environmental surfaces (χ 2 = 7.733, df = 3, P = 0.0519). Similarly, no significant differences were found in biofilm production between isolates recovered from clinical contact surfaces and those recovered from housekeeping surfaces (χ 2 = 2.829, df = 3, P = 0.4188).

Discussion
A very limited number of studies, none of which was performed in Egypt, investigated carriage frequency, antibiotic resistance, virulence properties, and genetic diversity of MRSA strains isolated from personnel and environmental surfaces from dental health-care personnel (DHCPs), dental patients, and dental environment. In an effort to fill this knowledge gap, we phenotypically and genotypically characterized MRSA isolates from six  different wards at a university outpatient dental clinic in Egypt. We elected to restrict screening of personnel samples to hand and nasal swab specimens, while the environmental surfaces chosen included those commonly encountered within and outside of the patient care area. The recovered isolates were initially identified based on resistance to cefoxitin and oxacillin, and were further confirmed by mecA gene detection. Cefoxitin demonstrated a 100% sensitivity for MRSA detection in our hands, as compared to 73.5% in the case of oxacillin, which missed the detection of 10 (26.47%) of the mecA-positive isolates ( Table 2, IDs: 2 , 11, 15, 20, 25, 35, 68, 70, 71, 93). The superiority of cefoxitin for MRSA identification has been reported by a number of authors [35][36][37] .
In the current study, MRSA represented 3.9% of all recovered bacterial isolates (34/863) and 30.4% (34/112) of all recovered S. aureus strains. A 6.6% (12/182) MRSA prevalence was observed among DHCPs, which is in line with the 6.1% prevalence reported among health care workers in the Middle East 38 . This rate is higher than the average global prevalence of MRSA carriage by DHCPs of 4.6% 38 . The MRSA nasal carriage rate among the outpatients screened in the current study (10.2%, 9/88) was lower than the 32% rate reported in a sample of Egyptian outpatients attending primary health care centers [48]. Similarly, the MRSA hand carriage rate among our outpatients (9.8%, 6/61) was much lower than the 47.4% rate obsereved among those attending a dermatology clinic 39 . We detected higher nasal carriage rates of MRSA in outpatients (11.1%, 3/27) than DHCPs (8.2%, 5/61). This is in agreement with the results of a previous study that showed a higher prevalence of nasal MRSA colonization in patients compared with health care workers (5.1% vs. 4.8%) 40 . Overall, our results related to the hand and nasal carriage of MRSA among medical and non-medical personnel, as well as MRSA colonization on surfaces in the clinic environment are consistent with results reported in a number of studies [41][42][43][44][45] . The differences between prevalence rates in our study and others could be attributed to differences in study design, sample size, patient characteristics, and specimen types tested.
Excluding that of prosthodontics, all the investigated wards had one or more of their surfaces positive for MRSA. The absence of MRSA in the tested surfaces of the prosthodontics ward may be attributed to the nature of dental procedures performed in this particular ward, being mostly non-invasive 46 . Supporting this suggestion is the finding that MRSA-positivity of patients' sink faucets in our study occurred exclusively in the prosthetic dentistry and periodontics wards, where bleeding of patients is common, due to the invasive dental procedures performed in both wards 46 . As might be expected, surfaces with more patient contact (clinical contact surfaces) had higher rates of MRSA colonization than surfaces with less patient contact (housekeeping surfaces). Door knobs were the most contaminated among the investigated housekeeping surfaces. Given the absence of MRSA

Weak (n = 7)
Moderate (n = 10) Cluster 1 spa-CC 223  91 ; the spa types shown are based on the Ridom StaphType software. b All the MRSA isolates investigated in the current study were mecA-positive and mecCnegative. C The isolates were classified as biofilm non-producers, weak, moderate, and strong biofilm producers based on previously published criteria 34 .

Hand
in housekeeping surfaces that are mainly touched by DHCPs (disinfectant containers, dentist/nurse hand washing sink, and nurses' desks), it may be reasonable to assume that door knobs in the present study were mainly contaminated by patients' hand contact. Looking at the possibility of MRSA transmission among the various dental wards and/or among the various specimen categories, only three pairs of isolates showed the same antibiogram ( Table 2, IDs: 26 and 44; 63 and 65; and 30 and 95). However, differences were found in the SCCmec types and spa types carried by the two isolates within each of these pairs. The majority (n = 24; 70.6%) of our MRSA isolates showed multidrug resistance pattern to most of the antimicrobials used. While this may be a reflection of the excessive, unjustified use of broad-spectrum antibiotics in Egypt 47 , this pattern is usually associated with HA-MRSA, because antibiotic resistance in CA-MRSA strains is often limited to β-lactams 48,49 . Surprisingly, 14 (51.9%) of the multidrug-resistant isolates in our study carried SCCmec type IV, which is commonly found in CA-MRSA. The emergence and spread of these multidrug-resistant CA-MRSA isolates could also be the result of the selective pressure of excessive and inappropriate antibiotic usage in our community. In the case of antibiotics known for their potent anti-MRSA activities, the isolates showed a relatively high rate of resistance to vancomycin and linezolid (29.4% and 26.4%, respectively), while there was virtually no resistance to imipenem. Overall, nineteen and thirteen antibiotic resistance profiles were observed among MRSA isolates from personnel and environmental surfaces, respectively. This difference might reflect the presence of strong selective pressure from antibiotic usage in the personnel group.
The investigated isolates were characterized using SCCmec and spa molecular typing tools. Based on the former typing method, the classical nosocomial SCCmec types I and II represented 38.2% and 2.9%, respectively, whereas SCCmec type III was completely absent. On the other hand, SCCmec type IV (which is usually considered a CA-MRSA marker) was the most predominant type. However, the multidrug resistance profiles and the relatively low prevalence of the PVL gene (14.7%) seen in our isolates are also common in HA-MRSA strains. The predominance of SCCmec type IV in this study is in agreement with other studies conducted on community-derived and hospital-derived MRSA isolates in the neighbouring territories of Jordan and Gaza [50][51][52] . Taken collectively, these findings suggest that in Egypt, and probably other neighbouring regions, the population Figure 1. Population structure of the tested MRSA isolates (n = 34) based on BURP analysis. This analysis was performed using the Based Upon Repeat Pattern (BURP) algorithm of the Ridom StaphType software (Ridom GmbH, Würzburg, Germany) at a cost setting of ≤ 5 and excluding spa-types with 5 or fewer repeats. Each dot represents a different spa type, with the diameter of the dot being proportional to the quantity of the corresponding spa type. Clusters of linked spa types correspond to spa clonal complexes (spa-CCs). The predict founder of a cluster (which was used for defining the cluster) is shown in blue, while the others in black. Near the lines of connection, the mutations involved in the transition from a spa type to the next one are reported in detail. All DNA changes are meant to occur from the founder to the periphery. Legend: numbers along the lines refer to the repeat sequence involved in the mutation; +indicates the acquisition of a repeat sequence; -indicates the loss of a repeat sequence; within circles the numbers of the strains of each CC appear between brackets. In summary, the analysis identified a single clonal complex (spa-CC223) comprising spa types t223, t14339, t3689, and t8506; n = 20 isolates, and accounted for 58.8% of all tested MRSA isolates, as well as 3 singletons (t267, t084, and t1339; n = 12 isolates, 35.2%), while excluded 2 isolates (t380, 5.8%) from the clustering, as they consisted of four repeat units only.
SCIEntIFIC REPORTS | 7: 7390 | DOI:10.1038/s41598-017-07713-8 structure of MRSA in the community is starting to mirror that found in the hospital setting, making the boundaries between these two categories so blurred. A similar observation has been made in other countries, which may be attributed to increased MRSA colonization rates in the community, or increased prevalence of nosocomial MRSA [53][54][55][56][57][58][59] .
In the current study, spa-CC223 was the main spa clonal complex (58.8% of the total MRSA isolates). Interestingly, spa-CC223 has been reported as the second most predominant spa-CC in a study conducted in Kuwait 60 . Additionally, we identified 8 different spa types among the tested MRSA isolates, 6 of which (t223,  The gene encoding the Panton-Valentine leucocidin toxin; vanA: the gene encoding an enzyme that causes a structural change in the terminal amino acid of the pentapeptide chain of peptidoglycan, thus conferring vancomycin resistance; tst: the gene encoding the toxic shock syndrome toxin; spa: staphylococcal protein A; a minus sign (−) denotes the absence of an attribute. a All the MRSA isolates investigated in the current study were mecA-positive and mecC-negative. b The isolates were classified as biofilm non-producers, weak, moderate, and strong biofilm producers based on previously published criteria 34 .   t267, t084, t380, t8506, and t1339) have been previously reported in other Arab countries ( Table 4). The limited diversity and the high frequency of the spa type t223 are in agreement with previous data from general population studies as well as from hospital-based studies in the Arab region 50,61,62 . This may suggest that certain MRSA clones are more successful than others at surviving, colonizing, and spreading in this geographical region, which is consistent with what has been reported in Europe 63 . The other two remaining spa types identified in the current study, namely t14339 and t3689, have been previously reported in isolates from Ireland and Denmark, respectively. Herein, the detection of these two types may be due to the accquision of their respective clones during international travel, or may be a function of the study location in a clinic within a private university, where Arab and non-Arab students from different nationalities are enrolled, some of which have received their secondary education in European countries. Interestingly, the spa types from environmental surface isolates were far more clonally conserved than the spa types from personnel (4 and 8 different spa types, respectively; Table 5). We have also observed differences in the antimicrobial resistance profiles of the strains recovered from the two specimen categories. Taken together, these observations may reflect the different reservoirs of strains to which the two categories are exposed. Contrary to other studies with similar sample sizes, MRSA isolates in this study showed a limited genetic diversity, with CC80-MRSA-IV-t267 (17.7% of all tested MRSA isolates) and CC22-MRSA-IV-t223 (14.7% of all tested MRSA isolates) being the most predominant clones. While this limited genetic diversity may be attributed to the monocentric nature of the study, the predominance of CC80 and CC22 among the identified clonal complexes is consistent with what has been described previously in studies from Egypt and other neighbouring countries 50,52,64,65 . Additionally, five (14.7%) of the current MRSA isolates ( Table 2, IDs: 5, 11, 13, 111, and 112) were identified as tst-positive CC22-IV isolates, which are expected to belong to the 'Middle Eastern variant' of EMRSA-15. Isolates with similar characteristics have been also reported in studies from Egypt, Italy, the Gaza Strip, Saudi Arabia, Jordan, Kuwait, and the United Arab Emirates 50,61,62,[66][67][68][69] . Interestingly, the strains reported in Italy and Gaza appeared to be endemic in the tested health care units 66 , and both showed the spa type t223, as with the majority of our isolates, suggesting the dissemination of this clone into hospital settings in the Arab and Mediterranean regions. It is noteworthy that the Italian strain exhibited an antimicrobial profile that was different from most of our isolates, because the former was a non-multiresistant MRSA strain. A similar diversity has been shown in a recent study, in which six variants of CC22-MRSA-IV have been detected in the Gulf region 70 . Therefore, the two tst-negative CC22-MRSA-IV isolates recovered in the current study ( Table 2, IDs: 65, and 103) might be either tst deletion mutants of the 'Middle Eastern variant' strain or derived from imported European UK-EMRSA-15/Barnim epidemic strain. The latter possibility is supported by the spa type (t3689) possessed by of one of the two isolates ( Table 2, ID: 65), since it is a common type in Denmark, but not in the Arab region (Table 4). Further investigations are necessary to track the origin of the seven CC22-MRSA-IV isolates described herein, as well as to assess whether they represent a true HA-MRSA clone, such as the classic EMRSA-15, or, alternatively, a CA-MRSA clone that might have spread into the tested dental clinic via the DHCPs or the patients. Future studies should determine risk factors, geographical abundance, transmission patterns, population dynamics, and clinical implications for CC22 strains harbouring the tst gene, given the high rate of endemicity characterizing these strains 50,62,66,71 .
An interesting aspect of the current study is the genotypic characterization of the MRSA isolates, which were examined for the presence or absence of five genes (mecA, mecC, vanA, tst, and PVL) with antimicrobial or virulence-related functions. Given the correlation between antimicrobial resistance and antibiotic consumption 72,73 , and given the widespread use of non-prescription antimicrobial agents in Egypt 74 , it may not be surprising that mecA-positive MRSA strains were recovered from hand and nasal specimens of the outpatients examined in the current study. Additionally, the mecA-positive MRSA isolates included strains recovered from DHCPs and environmental surfaces. Both specimen categories are exposed to high antibiotic pressure, caused by being in daily contact with patients receiving antibiotics, with ample opportunity to acquire antibiotic-resistant bacteria, and thus antibiotic resistance genes 43 . Contrary to mecA, the mecC gene was absent from all isolates. This finding supports previous findings, in which mecC-positive MRSA strains have been reported almost exclusively in Europe [75][76][77] .
When the prevalence of toxin-encoding genes was investigated, five of the tested MRSA isolates were found positive for the PVL gene, four of which (80%) were recovered from the hand swabs of DHCPs (including two nurses and two dentists, Table 2). Similarly, seventeen of the tested MRSA isolates were positive for the tst gene, five of which (29.4%) were recovered from both nasal and hand swabs of DHCPs (including four nurses and one dentist, Table 2). The hand and nasal colonization of DHCPs with MRSA isolates expressing these toxins is of public health interest, since DHCPs can serve as sources of transmission of these isolates in the community, especially among patients. The clinical implication of this becomes partciulary apparent when considering the carriage of more than one toxin-encoding gene, since three of the investigated isolates carried the gene for PVL in combination with the gene for tst. One of these three also harboured the vanA gene ( Table 2, IDs: 11, 111, and 112). While the carriage of both PVL and tst genes has been reported in a limited number of studies, the majority of these studies 50,51,64 , as well as the current one, have been conducted on isolates recovered from the Arab region. This is an alarming observation that needs to be prioritized in the formulation of national and regional health care policies, especially considering the large population exchange between these countries.
The prevalence rate of PVL-positive isolates in the present study (14.7%) is comparable to two other studies that have indicated prevalence rates of 15% and 19% among hospital-isolated MRSA strains from Egypt 78,79 . Similar to previous studies that have shown an association between PVL-producing genes and specific MLST CCs 80-82 , the majority (80%, 4/5) of the PVL-positive isolates in this study were associated with CC22.
In the case of the tst gene, the MLST lineage that showed the strongest association with tst-positive strains was also CC22 (70.6%, 12/17) and, to a lesser extent, CC80 (23.5%, 4/17). This fits well with previous studies that have reported the occurrence of tst-positive CC22 MRSA strains in Jordan 51, 62 , Kuwait 61 , Saudi Arabia 68 , and the United Arab Emirates 69 , as well as tst-positive strains belonging to CC80 in Jordan 51, 64 .
Some MRSA strains are able to produce biofilm on both mucosal and inanimate surfaces, making them difficult to eradicate 83 . Therefore, one of the aims of the present study was to evaluate the biofilm-forming ability of the tested isolates. The results showed that among the eight MRSA isolates recovered from nasal swabs, one (12.5%) isolate was classified as strong biofilm producer, three (37.5%) were moderate, one (12.5%) was weak, and three (37.5%) were non-biofilm producers. This is consistent with the general idea that a dispersed mode of growth is favoured over a biofilm mode during S. aureus nasal colonization 84 . Interestingly, three of the five isolates that are likely to be related to the 'Middle Eastern variant' of EMRSA-15 were strong biofilm producers ( Table 2, IDs: 5, 11, and 13). Collectively, the detection of multiple virulence and antimicrobial resistance genes suggests the pathogenic potential of the MRSA isolates in the current study, especially when combined with their ability to form biofilms, and thus their potential to resist disinfectants or sanitizers. This is the first study to provide an overview of MRSA clones currently circulating among patients, DHCPs, and environmental surfaces in dental clinics in Egypt. The main findings of this study include: (i) the limited genetic diversity of MRSA isolates within dental clinics in Egypt (ii) the detection of five tst-positive and two tst-negative CC22-IV isolates that are likely to be linked to the epidemic EMRSA-15 clone; (iii) the combined occurrence of tst and PVL in three of the isolates; (iv) the high level of multidrug resistance in the isolates; (v) the predominance of SCCmec type IV-harbouring MRSA isolates in the population; (vi) the blurring of traditional distinctions between CA-MRSA and HA-MRSA based on SCCmec type and antibiotic resistance in the community, which may suggest the infiltration of CA-MRSA into the hospitals in the area; and (vii) the detection of isolates with spa types (t14339 and t3689) that have never been reported before in any Arab country. In conclusion, the results suggest that personnel and dental clinic surfaces may serve as sources for transmission of MRSA. They can also act as important reservoirs for antibiotic resistance genes. Results reinforce the need for continuous national and regional MRSA surveillance programmes in order to keep track of the emerging clones. Strict antibiotic policy and infection control measures should be implemented to reduce the incidence of MRSA infection in dental clinics and other health care settings.

Materials and Methods
Study design and sample collection. In the present cross-sectional, monocentric study, a total of 1300 samples were collected from six different dental wards of a university outpatient dental clinic in Egypt, between January and May 2013. Samples collected were obtained from: (i) the hands and anterior nares of patients and DHCPs; and (ii) environmental surfaces within the clinic. The six dental wards included the endodontics, operative dentistry, periodontics, prosthetic dentistry, prosthodontics, and dental surgery wards 19 .
SCIEntIFIC REPORTS | 7: 7390 | DOI:10.1038/s41598-017-07713-8 All samples collected in this study (whether from personnel or environmental surfaces) were obtained during the working hours of the clinic. Participants were chosen randomly to differentiate CA-infections from HA-infections. A written informed consent was obtained from each subject. The study protocol was approved from the Ethics Committee of the Faculty of Pharmacy, October University for Modern Sciences and Arts (MSA). All methods were performed in accordance with the relevant guidelines and regulations. Hand swabs were collected during working days, immediately after removing the gloves (if applicable) and before washing. For sampling, the palms and periungual areas were vigorously rubbed with sterile saline-moistened cotton swabs. Paired nasal swabs were collected from each participant according to a previously described method 85 .
The environmental surfaces investigated in this study were categorized into two groups: (i) the clinical-contact surfaces (that is, surfaces that are touched frequently during dental procedures), which included: dental light arms, dentists' chairs, dentists' drills, dentists' tool racks, patients' sink faucets, and X-ray switches; and (ii) the housekeeping surfaces (surfaces outside of the patient care area), which included: dentist/nurse hand-washing sinks, disinfectant containers, door knobs, floors away from the dentists' chairs, light switches, and nurses' desks.
One set of environmental surface samples from each of the investigated wards included all the above-mentioned areas from both clinical contact and housekeeping surfaces. These samples were collected at the end of the clinic hours (before general cleaning for the next day) and following the CDC guidelines for environmental infection control 86 . Briefly, each sample was collected by applying a sterile water-moistened swab firmly over an approximate area of 5 × 20 cm of the specific object. In wards with two to twelve chairs, two chairs were randomly chosen, and samples were taken from both chairs, while in wards with two dental chairs, samples were collected from both chairs.
Identification of S. aureus and screening for methicillin resistance. Swabs (collected from hands, anterior nares, or environmental surfaces) were inoculated into 2 ml of double strength brain heart infusion broth (BHI; Difco, USA), and incubated at 37 °C for 18-24 h. A volume of 100 μl was withdrawn from cultures showing growth, plated onto mannitol salt agar (Difco, USA), and the plates were incubated aerobically at 37 °C for 24 h. The yellow-colored colonies on mannitol salt agar were collected for further identification using standard microbiological methods. These methods included colony morphology on blood agar, Gram stain, in addition to catalase and coagulase tests. Strains with ambiguous biochemical results were analyzed by 16 S rRNA gene sequencing as described elsewhere 87 .
All S. aureus isolates were screened for methicillin resistance using oxacillin (1 µg) and cefoxitin (30 µg) disc diffusion tests. Briefly, bacterial cultures were adjusted to the 0.5 McFarland turbidity standard, which is equivalent to 1.5 × 10 8 CFU/ml, and inoculated (using a sterile cotton swab) on the surface of a Mueller Hinton Agar (MHA) (Oxoid, UK) in the case of cefoxitin, or MHA supplemented with 2% sodium chloride in the case of oxacillin. Zone diameters were measured and interpreted according to the guidelines of the Clinical Laboratory  Table 8. PCR primers used in this study. * F: forward primer; R: reverse primer.
Standard Institute (CLSI) 88 . All MRSA isolates were stored at −20 °C in BHI containing 15% glycerol for further characterization.

Molecular typing methods. SCCmec typing.
A multiplex PCR with five primer-pairs (Table 8) was used as previously described 89 to discriminate between SCCmec types I, II, III, IV, and V. The SCCmec type was determined on the basis of the band pattern obtained. Isolates with no visible bands, or with a band pattern that was not in agreement with one of the five predicted band patterns, were classified as non-typeable.
spa typing and Based Upon Repeat Patterns (BURP) analysis. Using the primers listed in Table 8, all the investigated MRSA isolates (n = 34) were subjected to a PCR assay for amplification of the polymorphic repeat region (X region) of the spa gene as described elsewhere 90 . The spa amplicons were then purified using a Qiagen DNA purification kit (Qiagen GmbH, Hilden, Germany) and sequenced by Macrogen ® (Seoul, South Korea) using capillary electrophoresis. spa typing was conducted as described by Harmsen et al. 91 , and the resulting spa types were then clustered into related spa clonal complexes (spa-CCs) using the BURP algorithm implemented in the Ridom StaphType software version 2.2.1 (Ridom GmbH, Würzburg, Germany). The default parameters of the BURP algorithm (exclusion of spa types shorter than 5 repeats and clustering of spa types if cost is less or equal to 5) were applied in this analysis, in order to prevent the formation of spa clusters that are too large or non-specific 92 .
The spa type that could not be assigned to a cluster was considered as a singleton. Due to the high concordance between spa typing and MLST 24 , the MLST clonal complexes (CC) corresponding to the respective spa-CCs were deduced from the data on the Ridom SpaServer (http://spaserver.ridom.de, last accessed on 15 November 2016) and by literature search.
Antimicrobial susceptibility testing. Excluding vancomycin and linezolid, the antimicrobial susceptibility of the isolates to a panel of commonly used antibiotics was determined using the Kirby Bauer disc diffusion method on MHA plates according to CLSI guidelines and breakpoints 88  The susceptibilities of the isolates to vancomycin and linezolid (both from Sigma Aldrich, Germany) were determined using the agar dilution method following the CLSI guidelines and interpretative criteria 88 . Throughout the study, the antimicrobial susceptibility tests were quality controlled using S. aureus ATCC 43300 (methicillin-resistant strain) and S. aureus ATCC 29213 (methicillin-sensitive strain).
Detection of mecA, mecC, vanA, tst, and PVL-encoding genes. Presumptive MRSA isolates were confirmed by polymerase chain reaction (PCR) using primers targeting the mecA and the mecC genes. Additionally, all the MRSA isolates were subjected to a PCR assay for detecting the lukF/S-PV genes encoding the PVL toxin and the tst gene encoding the TSST-1 toxin, while only those found to be phenotypically resistant or intermediately resistant to vancomycin were tested for the presence of vanA gene. The primers used in these assays (Table 8) were synthesized by Eurofins MWG Operon (Ebersberg, Germany). Genomic DNA extraction and purification was done using the GeneJET ® Genomic DNA purification kit (Thermo Scientific, USA) according to the instructions of the manufacturer. Each PCR amplification cycle consisted of an initial denaturation step at 95 °C for 10 minutes, followed by denaturation at 95 °C for 30 s, annealing at 47-52 °C (depending on primers used, Table 8) for 30 s, and extension at 72 °C for 1 min for each kb of DNA amplified. This cycle was repeated 35 times followed by a final extension step at 72 °C for 10 minutes. The final volume of the reaction mixture for each PCR assay was 25 µl, and amplifications were performed using the Biometra TAdvanced thermal cycler (Biometra, Göttingen, Germany). All PCR-based assays employed known positive and negative controls. After amplification, 10 µl of each PCR reaction was separated on a 2% (w/v) agarose gel, stained with ethidium bromide (0.5 mg/ml), and visualized under a Gel Doc EZ Imager (Bio-Rad Laboratories, USA).
Biofilm formation assay. The ability of the MRSA isolates to form biofilm onto polystyrene microtiter plates was evaluated as described previously with slight modifications 93 . Briefly, overnight bacterial cultures in trypticase soy broth (TSB, Difco, USA) were diluted in the same medium to match the 0.5 McFarland turbidity standard, followed by further dilution (1:100) in TSB supplemented with 2% (w/v) glucose and 2% (w/v) NaCl. A volume of 200 µl of this diluted bacterial suspension was cultured in triplicates in microtiter wells (96 wells; Nunc, Denmark), while negative control wells contained uninoculated medium. The plates were incubated at 37 °C for 24 h. Following incubation, the plates were washed carefully three times with 200 μl of tryptone water (Difco, USA) to remove nonadherent planktonic cells, and the plates were subsequently dried at room temperature. The established biofilm was stained with 100 μl/well of 0.1% membrane filtered crystal violet solution (Sigma Aldrich, Germany) at room temperature for 2 min. The wells of the microtiter plates were then washed twice with sterile pyrogen-free water, and finally a mixture of ethanol:acetone (4:1, v/v) was used to elute bound crystal violet. The eluted crystal violet was diluted 1:10 with the same mixture of solvents, and the optical density was determined spectrophotometrically at λ = 545 nm using microplate ELISA reader (Stat Fax ® 2100). The isolates were classified as biofilm non-producers, weak, moderate, and strong biofilm producers based on previously published criteria 34 .
Statistical analysis. Categorical variables were compared using the Chi-square test (χ2) or Fisher's exact two-tailed test, as appropriate, with P values of < 0.05 as the level of significance. These statistical analyses were carried out using the GraphPad Prism (version 6; GraphPad Software Inc.; USA).