Species and drug susceptibility profiles of staphylococci isolated from healthy children in Eastern Uganda

Staphylococci are a key component of the human microbiota, and they mainly colonize the skin and anterior nares. However, they can cause infection in hospitalized patients and healthy individuals in the community. Although majority of the Staphylococcus aureus strains are coagulase-positive, some do not produce coagulase, and the isolation of coagulase-positive non-S. aureus isolates in humans is increasingly being reported. Therefore, sound knowledge of the species and characteristics of staphylococci in a given setting is important, especially isolates from children and immunocompromised individuals. The spectrum of Staphylococcus species colonizing children in Uganda is poorly understood; here, we aimed to determine the species and characteristics of staphylococci isolated from children in Eastern Uganda. Seven hundred and sixty four healthy children less than 5 years residing in Iganga and Mayuge districts in Eastern Uganda were enrolled. A total of 513 staphylococci belonging to 13 species were isolated from 485 children (63.5%, 485/764), with S. aureus being the dominant species (37.6%, 193/513) followed by S. epidermidis (25.5%, 131/513), S. haemolyticus (2.3%, 12/513), S. hominis (0.8%, 4/513) and S. haemolyticus/lugdunensis (0.58%, 3/513). Twenty four (4.95%, 24/485) children were co-colonized by two or more Staphylococcus species. With the exception of penicillin, antimicrobial resistance (AMR) rates were low; all isolates were susceptible to vancomycin, teicoplanin, linezolid and daptomycin. The prevalence of methicillin resistance was 23.8% (122/513) and it was highest in S. haemolyticus (66.7%, 8/12) followed by S. aureus (28.5%, 55/193) and S. epidermidis (23.7%, 31/131). The prevalence of multidrug resistance was 20.3% (104/513), and 59% (72/122) of methicillin resistant staphylococci were multidrug resistant. Four methicillin susceptible S. aureus isolates and a methicillin resistant S. scuiri isolate were mupirocin resistant (high-level). The most frequent AMR genes were mecA, vanA, ant(4')-Ia, and aac(6')-Ie- aph(2'')-Ia, pointing to presence of AMR drivers in the community.


Introduction
Staphylococci are a key component of the human microbiota, and around 47 species have been identified [1,2]. They mainly colonize the skin, anterior nares, the nasopharynx, and gut [2][3][4]. Staphylococci are generally nonpathogenic however, when there is breach of the body barriers (e.g. damage to the skin and mucous membranes) or when the immune system is weakened, they become opportunistic pathogens and cause a variety of mild to severe community-and hospital-acquired infections [2][3][4].
One of the most useful characteristics used to differentiate staphylococci is their ability to produce coagulase. In this regard, staphylococcal isolates are broadly classified into two groups-the coagulase-positive staphylococci (CoPS) comprising of mainly Staphylococcus aureus, which is more pathogenic, and the coagulase-negative staphylococci (CoNS) that are less pathogenic but include majority of the species [2]. While majority of S. aureus strains are overwhelmingly coagulase-positive, some are atypical in that they do not produce coagulase [5][6][7][8][9][10][11]. Among the CoNS, the medically relevant species include S. epidermidis, S. haemolyticus, S. hominis and S. saprophyticus [1,2]. Interestingly, one species, Staphylococcus schleiferi, includes both CoNS and CoPS subspecies i.e. S. schleiferi subsp. schleiferi and S. schleiferi subsp. coagulans, respectively [2]. Furthermore, there are five additional CoPS species i.e. S. delphini, S. hyicus, S. intermedius, S. intrae and S. pseudintermedius, which are typically animal-associated [2] but they have also been isolated from human samples. Although these non-S. aureus CoPS species are characteristically animal-associated [2], they are likely to be prevalent in human specimens in the developing countries [5] where rural, agrarian populations predominate.
Staphylococcal colonization is a known risk factor for staphylococcal disease [12]. Globally, there is an increase in prevalence of nosocomial infections caused by CoNS and atypical S. aureus, and this is attributed to the increase in use of procedure-related devices in human medicine [2]. CoNS and atypical S. aureus infections are difficult to treat as they are more likely to be methicillin-resistant and multidrug-resistant; moreover, CoNS species like S. lugdenensis and S. saprophyticus are associated with severe infections i.e. infective endocarditis and urinary tract infects, respectively [2]. It is therefore important to unambiguously identify staphylococci to species-level, particularly isolates from vulnerable groups such as children and immunocompromised individuals. Additionally, an understanding of the dynamics of staphylococcal colonization in a given setting, especially methicillin resistant CoNS (MRCoNS), is critical as they are considered to be the reservoir of antibiotic resistance genes for the more pathogenic species (S. aureus) [3,13]. Several studies have reported transfer of the staphylococcal cassette chromosome mec (SCCmec), a transposon that bears the molecular determinant of methicillin resistance (mecA gene), between S. epidermidis and S. aureus [14]. While the role of CoNS as reservoirs of the SCCmec element is well-established, the spectrum of species potentially involved in SCCmec exchange is not well documented, particularly in the developing countries.
Here we aimed to determine the species and characteristics of staphylococci isolated from children in Eastern Uganda. As virulence and antimicrobial resistance (AMR) tend to coexist in staphylococci [15,16], we also determined the frequency of genes associated with these phenotypes among community-associated staphylococci in Eastern Uganda. 2011-183; SBS194), and by the Uganda National Council for Science and Technology (Ref HS 1080) [17][18][19][20]. Written informed consent was obtained from the parents/guardians of the children, and the consent procedure included consent for storage of samples for further studies. Since the study was conducted on stored isolates, the Research Ethics Committees waived the need for re-consenting of the participants.

Study setting, samples and isolates
This cross-sectional study was nested in previous studies/projects that investigated the pneumococcal nasopharyngeal colonization and integrated community case management of malaria and pneumonia in children less than 5 years of age at the Iganga/Mayuge Health and Demographic Surveillance Site (IMHDSS) in rural Eastern Uganda [20,21]. The study population was described before [20][21][22]; briefly, 764 healthy children less 5 years at the IMHDSS were enrolled and sampled. Following consent, a study nurse collected two nasopharyngeal samples (swabs) from each child. Swabs were transported in small batches to the Clinical Microbiology laboratory in Kampala for culturing. Samples were processed for isolation of staphylococci by following standard microbiology procedures described previously [5,20] (and summarized below).

Species identification and antimicrobial susceptibility testing
Following overnight culturing of samples on non-selective medium (blood agar plates), Grampositive and catalase-positive isolates were sub-cultured again on solid Brain Heart Infusion (BHI) medium and confirmed to species level by using the Phoenix 100 Identification (ID) / Antimicrobial Susceptibility Testing (AST) Automated Microbiology System from the Becton, Dickinson and Company (Franklin Lakes, New Jersey), as described previously [23,24]. Morphologically distinct colonies with characteristic appearance of staphylococci were selected for automated identification. Antibiotic susceptibility testing by minimum inhibitory concentrations (MICs) to a panel of 16 antibiotics (cefoxitin, penicillin, ampicillin, tetracycline, trimethoprim/sulfamethoxazole, erythromycin, chloramphenicol, gentamicin, ciprofloxacin, rifampicin, clindamycin, mupirocin [high-level], teicoplanin, linezolid, vancomycin and daptomycin) was determined with the Phoenix Automated Microbiology System as previously described [23,24] [25]. Multidrug resistance (MDR) was defined as isolates resistant to three or more classes of antimicrobials. To validate automated species'-level identification by the Phoenix ID/AST Expert System, we randomly selected 24 isolates representing presumptively identified isolates of S. aureus, S. epidermidis and S. haemolyticus, and molecularly re-identified them by polymerase chain reaction (PCR) as described previously [26]. To determine the SCCmec types among the methicillin resistant staphylococci (MRS), SCCmec typing was done and interpreted as described previously [17,22,27]. For quality control, S. aureus ATCC™ 29213 and Enterococcus faecalis ATCC™ 29212 were included in the Phoenix ID/AST panels and molecular assays.

Virulence and antimicrobial resistance genes
In this study, all isolates were investigated for carriage of AMR and virulence genes in addition to the mecA gene described above. The most frequent AMR genes in S. aureus isolates were vanA, ant(4')-Ia, and aac(6')-Ie-aph(2'')-Ia while the most frequent virulence genes were hld and hla, Table 4. Although phenotypic resistance to vancomycin was not detected, 20 isolates carried the vanA gene while 6 carried the vanB gene. Eleven (42%, 11/26) of the isolates with vanA/vanB also carried the mecA gene. Furthermore, 62 isolates carried the AMEs genes (any of aac (

6')-Ie-aph(2'')-Ia, ant(4')-Ia and aph(3')-IIIa), of which five (8%)
were phenotypically resistant to gentamicin (an aminoglycoside). Generally, the hld, hla, sea and LukS-PV / LukF-PV genes were more frequent in S. aureus compared to CoNS species, Table 4. Additionally, in S. aureus the bhp gene was more frequent in MRSA isolates compared to MSSA while in CoNS, the sea, sdrE, bhp, hla, and hld genes were more frequent in MRSCoNS, Table 4. Among the CoNS, the AMR genes were more frequent in S. epidermidis compared to other species. The hlg gene was detected in only four isolates while hlb and ica/ D genes were not detected. Overall, the virulence and AMR genes were generally more frequent in S. aureus compared to CoNS species (for example, the LukS-PV / LukF-PV genes were only detected in S. aureus).

Discussion
In this study, we used a rigorous approach to determine the species and characteristics of staphylococci colonizing children in Eastern Uganda. The Automated system we used i.e. the Phoenix AST/ID Expert System, is more accurate than the manual methods at identifying Gram-positive bacteria, as well as AMR testing [30]. However, in the developing countries, manual methods are the mainstay for identification of staphylococci, and in most cases the tube coagulase test is the confirmatory test for identification of S. aureus [5]. The tube coagulase test is considered to be more reliable at identifying S. aureus when a firm clot that doesn't move on tilting the tube occurs; therefore, the subjectivity in interpreting the tube coagulase test probably leads to misidentification of CoNS as S. aureus [31] In this study, a high staphylococcal nasopharyngeal colonization rate (63.5%, 485/764) was reported for healthy children in Eastern Uganda. However, our rate is comparable to the rates reported by other investigators e.g. �80% by Budri et al [32] and Faria et al [1]. Furthermore, the number of Staphylococcus species (i.e. 13) in this study is also comparable to the 9 species reported by Faria et al in Denmark [1], but less than the 19 species reported by Xu et al in a study of staphylococcal colonization in healthy individuals and the environment in London  (1) [14]. S. aureus was the most prevalent species followed by S. epidermidis, which contradicts reports of S. epidermidis as the most prevalent Staphylococcus species among staphylococcal isolates from humans [1,14,32]. However, as the overall prevalence of CoNS in this study (i.e. 42%, 320/764) was higher than that of S. aureus (37.6%, 193/513), the prevalence of S. epidermidis might be higher than what we have reported had we succeeded in identifying all the CoNS to species level. Still, several factors could have affected our recovery of S. epidermidis; for instance, unlike S. aureus, CoNS especially S. epidermidis mainly colonize the skin [33] yet we sampled the nasopharynx which is mainly colonized by S. aureus. Additionally, the enrichment methods employed in isolation of staphylococci generally favor recovery of S. aureus [32]. Interestingly, while majority of the CoNS species identified are human-associated (i.e. the S. epidermidis-like group-S. epidermidis, S. haemolyticus, S. capitis, S. hominis, etc. [2]), animal-associated CoNS species (i.e. S. caprae) and animal-associated CoPS species (i.e. S. intermedius) were also detected, which perhaps reflects a rural, agrarian population typical of Eastern Uganda. The detection of animal-associated staphylococci in this setting is a cause for concern as it may reflect (1) contamination of human samples by animal-associated strains, (2) occurrence of animal-associated CoPS in human samples, which could yield false-positive results on coagulase testing and overestimation of S. aureus/MRSA rates [30]. Around 5% of the children in this study were colonized by two or three Staphylococcus species, fewer than reports by investigators from Europe (i.e. 7.5% to 30%) [1,3,32]. In agreement with previous reports [1,3,32], the commonest co-colonizing species were S. aureus and S. epidermidis. Only one of the co-colonized children harbored S. aureus and S. haemolyticus/lugdunensis, which is not surprising as S. lugdnensis produces lugdunin, an antibacterial agent that reduces the probability of S. aureus colonization [3,34]. Further, 66.7% (16/24) of the co-colonized children harbored a multidrug resistant Staphylococcus while 58.3% (14/24) harbored a MRS. However, only three of the co-colonized children harbored different species of MRS, and this is in agreement with the previously reported negative correlation of co-colonization by distinct MRS species [1,3,32].
Antimicrobial resistance rates were generally low across species. However, in sharp contrast to reports from the Nordic countries [35], penicillin resistance and MRS nasopharyngeal colonization rates were quiet high, which is consistent with global reports of increasing MRS prevalence in the community [36]. The detection of SCCmec types I+II+III combined and SCCmec types IV+V combined suggests coexistence of hospital-associated and community-associated MRS that seems to result from dissemination of hospital-associated strains into the community [36]. Further, there were five isolates with high-level mupirocin resistance (HLMup r ), four of which were MSSA while one was a methicillin resistant S. scuiri (MRSS). This is consistent with increasing levels of mupirocin resistance staphylococci, especially in MRCoNS [3] [37], which is worrying as mupirocin is the drug of choice for eradication of Staphylococcus colonization. The occurrence of HLMup r in Africa before widespread use of mupirocin for MRSA decolonization requires continuous monitoring.
With the exception of mecA-positive isolates, majority of the isolates harboring AMR genes i.e. vanA/vanB and AMEs (any of aac(6')-Ie-aph(2'')-Ia, ant(4')-Ia and aph(3')-IIIa) were phenotypically susceptible to drug(s) targeted by the gene products. For example, phenotypic resistance to vancomycin was not detected; however, we found 26 isolates with the vanA/vanB genes, and 42% (11/26) of them carried mecA. vanA and vanB genes encode high-and lowlevel resistance to vancomycin, respectively. vanA type of resistance is widely spread in enterococci and it is transferrable between enterococci and staphylococci [29,38,39]. While phenotypic resistance to vancomycin among enterococci and staphylococci is rare in Uganda [24,40], vanA/vanB PCR-positive vancomycin-susceptible isolates of enterococci and staphylococci have been reported in Uganda [28,40]. While it is puzzling, vanA-positive-vancomycin-susceptible enterococci do occur, and they have also been reported from the developed countries [39,41,42]. These isolates have been found to lack key genes e.g. vanR and vanS, which are required for activation of transcription of the vancomycin resistance genes. Additionally, the vanA-positive-vancomycin-susceptible isolates can possess insertion sequences e.g. the ISL3-family element that silence transcription of the vanA operon [39,41,42]. The fact that horizontally transferrable silenced-vanA successfully reverted into resistance during vancomycin treatment [39], genotypic testing of invasive vancomycin-susceptible isolates has been recommended [33]. Relatedly, while phenotypic resistance to aminoglycosides correlates well with the detection of AMEs genes [43,44], in this study, few isolates with AMEs genes (i.e. 8%, 5/62) were phenotypically resistant to gentamicin (an aminoglycoside). However, gentamicinsusceptible isolates positive for AMEs genes have been reported before [37,38]. Overall, the carriage of AMR genes in absence of significant antibiotic selective pressure points to presence of AMR drivers in the community that should be investigated [32,36] with robust approaches like whole genome sequencing [42] to better understand the mechanisms underlying altered AMR susceptibility in our setting.
In this study, the virulence and AMR genes were generally more frequent among S. aureus isolates compared to CoNS species. For example, the LukS-PV / LukF-PV genes were detected only in S. aureus but not CoNS. The hlg gene was detected in only four S. aureus isolates, while hlb and ica/D were not detected at all. The low frequency of virulence genes in CoNS is consistent with reports that CoNS generally possess a smaller repertoire of virulence genes compared to S. aureus [2], and that human-associated S. aureus strains do not express beta toxins (hlb) [33]. Furthermore, the bhp gene was prevalent in S. aureus and it was significantly associated with MRSA. Interestingly, bhp encodes a protein homologue of Bap (biofilm associated protein) and it is generally absent in human-associated S. aureus strains. However, the Bap homologue termed bhp, is present in S. epidermidis [2]. Investigators in the developed countries found an association between presence of bhp and aacA [aac(6)-aph (2)] genes with the "not cured" clinical outcome among patients who underwent surgery and got infected by staphylococci [45].
The fact that staphylococci are opportunistic pathogens, identification of virulence/invasive strains is critically important in guiding diagnosis and treatment of staphylococcal infections [46]. To this end, the detection of staphylococcal virulence factors, especially genes involved in biofilm formation, is frequently cited as a means through which we can rapidly identify invasive strains [19,28,46]. Compared to previously characterized staphylococcal isolates in Uganda [19,28], we noted that the virulence and AMR genes are significantly more prevalent in S. aureus and generally in hospital isolates. In fact, the detection of icaA/D and hlg genes have been reported in hospital isolates but not community isolates [19,28], and current study conforms to this notion.
The polysaccharide intercellular adhesin/poly-N-acetylglucosamine (PIA/PNAG) is associated with biofilm formation in staphylococci especially S. epidermidis. PIA/PNAG is encoded by the ica gene cluster, which comprises the icaA, icaD, icaB and icaR genes. Deletion of these genes is associated with biofilm absence [2,46]. Biofilm production and presence of the ica operon correlate with disease causing clinical isolates and in mouse models, the PIA/PNAG-negative mutants were significantly less pathogenic [2,28]. The bhp/Bap genes are also associated with the biofilm phenotype and may be useful as markers in distinguishing between pathogenic strains and normal flora [46] [2]. Overall, the implication of detecting these genes in flora of children at the IMHDSS, albeit at lower frequencies relative to hospital-associated isolates, is that (a) strains bearing these genes could be potentially virulent and, (b) the genes could be useful as markers for screening potentially invasive strains.

Conclusions
The staphylococcal nasopharyngeal colonization rate in children in Eastern Uganda is high but comparable to rates from other countries. Also, the Staphylococcus species' distribution in the children mirrors what has been described from other settings but with S. aureus as the dominant species. Approximately 5% of the children were colonized by two or three Staphylococcus species but co-colonization by distinct MRS species is rare. Although the AMR rates were low, nasopharyngeal colonization by MRS is quiet high. The detection of high-level mupirocin resistance requires further investigation. Furthermore, while majority of CoNS and CoPS were human-associated, the detection of animal-associated CoNS and CoPS implies that individuals at the IMHDSS could be infected by zoonotic staphylococci. Importantly, with the occurrence of animal-associated CoPS species in human samples, tube-coagulase testing probably overestimates S. aureus/MRSA rates in this setting. Therefore, species-level identification of staphylococci with robust approaches is advised, especially clinically relevant isolates. The carriage of AMR genes in community-associated isolates, especially the vanA-positive-vancomycin-susceptible bacteria, points to existence of AMR drivers in the community that should be investigated for a better understanding of genetic determinants of altered antibiotic susceptibility. Overall, virulence and AMR genes, especially ica and bhp/Bap, were significantly more prevalent in S. aureus, could be useful as markers for screening potentially invasive strains.
Supporting information S1 Table. General characteristics of community-associated staphylococci isolated from children less than 5 years in rural Eastern Uganda.