Characterization of a community-acquired-MRSA USA300 isolate from a river sample in Austria and whole genome sequence based comparison to a diverse collection of USA300 isolates

The increasing emergence of multi-resistant bacteria in healthcare settings, in the community and in the environment represents a major health threat worldwide. In 2016, we started a pilot project to investigate antimicrobial resistance in surface water. Bacteria were enriched, cultivated on selective chromogenic media and species identification was carried out by MALDI-TOF analysis. From a river in southern Austria a methicillin resistant Staphylococcus aureus (MRSA) was isolated. Whole genome sequence analysis identified the isolate as ST8, spa type t008, SCCmecIV, PVL and ACME positive, which are main features of CA-MRSA USA300. Whole genome based cgMLST of the water isolate and comparison to 18 clinical MRSA USA300 isolates from the Austrian national reference laboratory for coagulase positive staphylococci originating from 2004, 2005 and 2016 and sequences of 146 USA300 isolates arbitrarily retrieved from the Sequence Read Archive revealed a close relatedness to a clinical isolate from Austria. The presence of a CA-MRSA USA300 isolate in an aquatic environment might pose a public health risk by serving as a potential source of infection or a source for emergence of new pathogenic MRSA clones.


Genetic comparison of ST8 isolates.
To assess relationship of the river water isolate W1 to clinical and food associated isolates, 18 clinical isolates from the Austrian national reference laboratory for coagulase positive staphylococci were characterized by WGS and sequences of 146 arbitrarily chosen USA300 isolates were retrieved from the Sequence Read Archive (SRA).
Whole genome based cgMLST phylogenetic analysis including S. aureus reference strain COL (accession no. , and 143 genomes from isolates retrieved from SRA was performed and a minimum spanning tree was calculated (Fig. 1). Distance calculation between all 164 samples revealed a maximum allelic distance between samples of 351 and an average allelic distance of 93.5 across the MST. Clinical Austrian isolates differed among each other in minimum of two, a maximum of 327, and an average distance of 100 alleles. Based on the defined complex threshold (CT) of 24 allelic differences 18 25 different complexes were obtained. Six out of 18 clinical Austrian isolates (H1/04-H3/04; H4/05-H6/05) showed close relatedness with a maximum allelic difference of 21 and were all located in complex 2 ( Fig. 1) respectively in complex 1 (Fig. 2 Phenotypic susceptibility testing revealed that the water isolate W1 was resistant to beta-lactams (benzylpenicillin, amoxicillin-clavulanic acid, cefoxitin), fluoroquinolones (ciprofloxacin, moxifloxacin) and erythromycin among the seventeen antibiotics tested (Table 1).
The presence of antimicrobial resistance against penicillin, methicillin, erythromycin and ciprofloxacin in isolate W1 was confirmed by in silico analysis using Mykrobe predictor.
Comparative analysis of isolate W1 to the closest relative isolate H5/16 via OrthoFinder revealed following differences in presence and absence of protein coding genes: additional presence of two domain-containing proteins (DUF443) and one hypothetical protein in isolate W1 and the additional presence of ten hypothetical proteins, four transposases, two family proteins (TIGR01741), one domain-containing protein (DUF5079), one immunoglobulin G-binding protein A and one transcriptional regulator in isolate H5/16.
Additionally 15 genes in isolate W1 were identified for which no orthologue could be assigned in the other genomes in this comparison: two ATP-binding proteins, two family proteins (TIGR01741), two hypothetical proteins, one acyltransferase, one LysR family transcriptional regulator, one AI-2E family transporter, one homoserine dehydrogenase, one recombination protein RecJ and one DUF5079 domain-containing protein. Three genes revealed no significant result and could not be assigned to any known group of protein coding genes (Supplementary Table S3). The circular map (CGView Server V 1.0 (2007)) depicts sequence similarity between the reference strain FPR3757, isolates H5/16 and W1 and sequence similarity between COL (NC_002951.2), FPR3757 and isolate W1 in Fig. 3.

Discussion
The discharge, persistence and dissemination of AMR in nature are considered a major public health threat worldwide 19 . Therefore a strategy to better control the release of antibiotics into the environment and subsequently to prevent contact of bacteria from human and animal sources with environmental organisms 20 is of utmost importance. Staphylococcus aureus was particularly successful in developing resistance to antibiotics. MRSA is listed as of high priority on WHOs antibiotic-resistant pathogens list (http://www.who.int/medicines/publications/ global-priority-list-antibiotic-resistant-bacteria/en/). Since the first detection of MRSA with association to healthcare (HA-MRSA) settings in 1961 21 , new epidemic MRSA clones have emerged affecting the community (CA-MRSA) or being associated with animals (LA-MRSA) [22][23][24] . Surveillance and characterisation of clinical, animal and environmental isolates is a public health imperative.
The MRSA isolate described was obtained from a water sample from a small river in southern Austria 24 . Characterisation of this isolate by WGS analysis revealed that the isolate has all characteristic features of CA-MRSA USA300 7 . CgMLST and wgMLST based gene-by-gene comparison 18 to eighteen clinical isolates (USA300) available from the National Reference Laboratory for Staphylococci including Staphylococcus aureus revealed a very close relatedness to one clinical USA300 isolate (H5/16) collected 2016 in Vienna, differing only by two alleles in their core-and by four alleles in their whole-genome targets. The water and patient isolates shared 60 identical virulence and 14 antibiotic resistance alleles. We were unable to find an epidemiological link between these two isolates. Four isolates collected in 2016 from a hospital located in one of the settlements next to this river, showed no close relatedness to the USA300 water isolate.
An interesting additional outcome of this WGS based typing study was that clinical USA300 isolates from Austria are a diverse population indicating that USA300 clones were introduced several times in Austria and are not all descendants from the first Austrian USA300 isolate from 2004. According to the WGS patterns all eight clinical Austrian USA300 isolates from 2016 are unrelated to each other. Thus in comparison to previous typing methods that did not allow a differentiation (same MLST, same spa type), WGS based typing is an important improvement representing an added information value for hospitals allowing a clear discrimination of outbreak and transmission events from unrelated isolates. In due consideration of staphylococcal mutation rates of one SNP per 15 weeks 25 we can assume that the water isolate W1 and several clinical isolates from Austria are descendants of the complex containing the hypervirulent USA300-C2406 strain from a patient with necrotizing pneumoniae isolated in Canada in 2004 26 and reference strain FPR3757 1 . In contrast to the aforementioned Austrian USA300 isolates from 2016, the first USA300 isolates detected in Austria in 2004 to 2005 10 belong to four different complex types. One complex contained isolates collected from two patients' , stationary in the same hospital in 2005, whereof the initial one was derived from a postal worker processing post from the United States, Target ND ND  ND  ND  ND ND ND  1  ND  ND  ND  aminoglycosides  BA000017.4   aphA3  ND 1  1  1  1  1  1  1  1  ND  1  ND  ND  ND  1  ND ND  ND  1  ND  ND  aminoglycosides  CP009681.1   blaI  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  ND  1  7  ND  ND  beta-lactam  SRR016154   blaR  15  15  15  15  15  15  15  15  15  15  15  15  15  15  15  15  ND  ND  15  6  ND  beta-lactam  SRR016154   blaZ  10  10  10  10  10  10  10  10  10  10  10  10  10  10  10  10  ND  1  10  6  ND  beta-lactam  SRR016154   ccrA2  6  NAT 6  6  6  6  6  6  6  NAT 6  6  6  6  6  6  ND  17  6  1  6  methicillin  NC010079   ccrB2  6  NAT 6  6  6  6  6  6  6  ND  6  6  15  6  6  6  ND  1  6  1  6 3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  4  3  3  1  3 27 and the detection in a river water sample with close relatedness of the water isolate to a clinical isolate might indicate a possible risk for public health.

aadD ND ND ND ND ND ND ND ND ND ND
In conclusion, this is to our best knowledge the first genome based characterisation of an environmental USA300 isolate confirming the emerging global health threat caused by environmental AMR. While numerous environmental AMR studies have reported on Gram negative bacteria our example shows that also Gram positive For appropriate actions further investigations are essential to determine the occurrence, risk, environmental impact of AMR in the environment as well as use of optimal wastewater treatment technologies to alleviate this public health threat 28 . Samples were collected in sterile 500 ml sodium thiosulfate containing bottles (Corning ® Gosselin ™ , NY, USA), 20 to 30 cm below the river surface and 50 to 100 cm apart from the river bank. For MRSA screening 100 ml aliquots of samples were filtrated and bacterial isolates were enriched by incubating the filter in tryptone soya broth (containing 3.5 mg/ml cefoxitin and 75 mg/ml aztreonam) at 37 °C overnight according to the standard protocol for isolating MRSA from dust samples (http://www.eurl-ar.eu/233-protocols.html, Isolation of MRSA from dust samples). To minimize the risk of contamination water sample filtration and cultivation were separated spatially, locally and temporally from cultivation of clinical samples. Laboratories were weekly screened for contaminations using air settlement plates and surface swabbing according to a standard operation protocol (SOP) following international guidelines 29  Whole genome sequencing and data analysis. For whole genome sequencing (WGS) high quality genomic DNA (gDNA) was isolated using the MagAttract HMW DNA Kit (Qiagen, Hilden, Germany) and quantified with a Qubit ® 2.0 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) using the dsDNA BR Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Libraries of bacterial genomes were prepared using Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA) according to the manufacturer's protocol. A desired coverage of at least 50-fold was calculated with Sequencing Coverage Calculator (www.illumina.com/ CoverageCalculator) and bacterial isolates were paired-end sequenced with a read length of 2 × 300 basepairs on a MiSeq instrument (Illumina, San Diego, CA, USA). SPAdes version 3.9.0 30 was used for read assembly and SeqSphere + version 4.1.9 (Ridom, Münster, Germany) for strain characterisation using a recently published cgMLST scheme 18 and calculation of minimum spanning trees (MST). Phylogenetic relatedness between water and related clinical isolates was further visualized using CGView Server V 1.0(2007) 31 .

Analysis of gene presence and absence.
Detailed analysis of additional genomic information was performed for 10 genomes: Reference strains C2406 (CP019590.1) and FPR3757 (CP000255.1) and isolates H1/04, H2/04, H3/04, H4/04, H5/04, H6/04, H05/16 and W1. The contigs of each assembly was filtered for a minimum length of 1,000 nucleotides. Genes were predicted using prodigal v. 2.6.3 42 with default parameters, orthologous groups where calculated with OrthoFinder v. 1.1.4 43 . Orthologous groups with differences in presence/absence were selected. Annotation of the orthologous groups and genes without group assignment was performed using NCBI-BLASTp v. 2.6.0+ with e-vaule cutoff of 0.01 44 and the RefSeq nr-protein database release 84 45 . The most frequent annotation in an orthologous groups was selected as description for the group.