A comparative genomic analysis between methicillin-resistant Staphylococcus aureus strains of hospital acquired and community infections in Yunnan province of China

A total of 60 MRSA strains were selected from 2013 to 2016 in both patients and food isolates in Yunnan province, southwest China. All strains were collected consecutively during the same period in both hospital and food surveillance. CA-MRSA strains of patients were isolated from The First People’s Hospital of Yunnan province (hospital A, HA). In contrast, HA-MRSA were isolated from People’s Hospital of Kunming City (hospital B, HB). CA-MRSA case was defined as MRSA isolated within 48 h after admission without the following risk factors: history of hospitalization, surgery, dialysis or stay in a long-term care unit within 1 year; dependence on an indwelling catheter, intravenous line or a percutaneous device when cultured; or previous isolation of MRSA. HA-MRSA cases were defined as MRSA isolation within 48 h with at least one risk factor, as mentioned above, or beyond 48 h, regardless of risk factors [15]. Hospitals A and B are tertiary care academic medical centers in southwest China and cover the entire Kunming area. The basic clinical information of patients from two hospitals were collected. For all these patients’ isolates, we have performed pulsed field gel electrophoresis (PFGE), multilocus sequence typing (MLST) and spa typing and selected them from previous systemic surveillance results (Additional files 1 and 2). The selection criterions of strains in this study were shown as follows: a. three major clones of patients’ MRSA, including ST239-t030, ST59-t437 and ST6-t701 were involved in the study, also contained some highly similarity strains with food derived strains (ST188-t189 and ST965-t062 etc); b. ST239-t030, ST59-t437, ST6-t701 and ST188-t189 generated four cluster groups based on PFGE patterns (Additional file 1), and we selected different PFGE pattern strains for sequencing in each genotype profile, except some highly similar isolates with food strains, such as seven ST59-t437, four ST6-t701 and three ST965-t062 strains.

Minimum inhibitory concentrations (MICs) for the 13 antibiotics were determined through the broth microdilution method using customized microtiter plates (Sensititre, UK) according to the manufacturers’ instructions. The antibiotics tested were penicillin (PCN), oxacillin (OXA), gentamicin (GEN), ciprofloxacin (CIP), levofloxacin (LEV), moxifloxacin (MXF), erythromycin (EM), clindamycin (CM), linezolid (LZD), vancomycin (VAN), tetracycline (TET), rifampicin (RIF) and trimethoprim/sulfamethoxazole (SXT). The tests were performed and interpreted in accordance with the Clinical and Laboratory Standards Institute (CLSI) guidelines (M100, 2018) [16], and S. aureus ATCC 29213 was used as a quality control. The mec gene was detected by PCR as described by previous study [17].

All MRSA isolates were recovered on Brain Heart Agar (BHI) (Luqiao, Beijing) at 37 °C for 24 h. Total genomic DNA of the isolated bacteria was extracted using a bacterial total genomic DNA extraction kit (Tiangen, Beijing) following the manufacturer’s instructions. All DNA samples were stored at − 20 °C for complete genome analysis.

Bacterial genome sequencing of all isolates was performed by our laboratory on the Illumina HiSeq platform using 2 × 150 bp paired-end reads. The libraries were built using a Nextera XT DNA Library Prep Kit following the manufacturer’s reference guidelines. Generally, 1 ng genomic DNA of each strain was used. After segment and purification, index PCR was performed to add the Illumina Nextera barcodes using i5 and i7 primers, and then the purification was executed again to remove non-target fragments. Finally, the libraries were normalized, pooled and sequenced using an Illumina HiSeq sequencing system (Illumina, San Diego, USA).

The raw data were trimmed for quality control, and low-quality (<Q40) reads were filtered by Trimmomatic (version 0.38) [18]. Draft genomes were assembled using SOAPdenovo (version 2.04), with k-mer values (25, 31, 37, 47, 59, 71, 83 and 95) optimized to the best assembly results [19, 20]. Gapcloser (version 1.12r6) was used to fill the genomic gaps. GenemarkS software (version 4.28) [21] with default parameters was used to predict the open reading frame (ORF) of each genome, and the predicted amino acid sequences were aligned and annotated by DIAMOND [22] (E-value: 1e-5, top 5) to NCBI non-redundant nucleotide database (NR), SwissProt, TrEMBL, KEGG, GO, COG, CDD, Pfam, Pathogen-Host Interaction database (PHI), Antibiotic Resistance Genes Database (ARDB), and Virulence Factor Database (VFDB).

We performed core- and pan-genome analysis using PGAP [23], which employed an all-versus-all BLAST search to group all CDSs into homologous clusters with default settings. All of the 60 strains sharing CDSs were considered to be core genes, and CDSs without orthologs were considered to be unique genes. Orthologs Clusters were annotated in COG functional distributions. Twenty-four MRSA references from other global strains in publically available datasets were used for phylogenetic analysis to determine the dissemination and evolution of these clonal lineages in southwest China (Additional file 3). Pairwise colinear comparisons of all genome sequences were performed using Mummer3 [24]. The pan-based tree was generated based on the presence or absence of genes in the pan-genome using the neighbor-joining method, and the SNP-based tree was generated based on the total genomic sequence results by using neighbor-joining method as well. The phylogenetic trees were visualized and ordered with FigTree (version 1.4.3) and iTol.

The sequence data have been deposited into the National Center for Biotechnology Information (NCBI), https://​www.​ncbi.​nlm.​nih.​gov/​ with accession numbers VCEL00000000 to VCGT00000000 (PRJNA543691).

For all the MRSA used in this study, forty strains (66.7%) were isolated from patients, and 20 (33.3%) isolated from food. Among the patients’ strains, sixteen (27%) were recognized as CA-MRSA, compared with 24 (40%) for HA-MRSA (Fig. 1a). ST6-t701 (18, 30%), ST59-t437 (16, 26.7%) and ST239-t030 (14, 23.3%) were the three major genotype profiles in this study. Other genotypes contained ST965-t062, ST188-t189 and ST5-t14723, as Fig. 1b shown. ST6-t701 was predominated in food strains, while ST59-t437 and ST239-t030 were the primary clones in patients (Fig. 1b). The comparison of clinical features between CA and HA-MRSA from patients revealed that except for sex, the factors age, clinical manifestations, and sample types were statistically different (P < 0.05), as Table 1 shown. Most of the CA-MRSA cases were children from pediatrics under the 5 years. The genotype profiles of CA-MRSA strains showed highly heterogeneity (Simpson’s index = 0.750) compared with HA-MRSA (Simpson’s index = 0.487). All the genotypes could be found in CA-MRSA isolates, except ST239-t030. However, the HA-MRSA cases were mostly from old patients, and the patients had different primary diseases, such as hypertension, cardiovascular disease, trauma, tumor or other infections. 58.3% of HA-MRSA were ST239-t030, and 41.7% were ST59-t437 genotype. All the food strains were isolated from eight cities of Yunnan province, and the majority (21.7%) was isolated from grain products. The details of MRSA strains information in this study were shown in additional file 4.

All the MRSA in this study were resistant to OXA and PCN, but sensitive to VAN. The mec gene was positive for all the isolates. In general, 55% of all the strains were resistant to MXF, followed by SXT (41.7%) and EM (36.7%). Compared the antibiotic resistant results between patients and food indicated that higher antibiotic resistant rates were found in patients’ strains, such as LZD, CIP, LEV, RIF, TET and MXF (Table 2). The antibiotic resistant results between genotypes of strains were also different. ST59-t437 genotype strains showed higher antibiotic resistant rates in CM, CIP, EM, SXT and TET, while ST239-t030 had higher resistant rate in MXF (P < 0.05). ST6-t701 clone had the lowest resistant rate for most of antibiotics, as Table 2 shown.

The molecular determinants based on ARDB annotations revealed several resistant genes were commonly found between these strains, such as liaR, mprF, pgsA and rpoC resistant to daptomycin; UhpT, murA and GlpT resistant to fosfomycin; gyrA and parC resistant to fluoroquinolones; gyrB and parE resistant to aminocoumarin; rpoB resistant to rifampicin; rpsL resistant to Streptomycin. The majorities of ErmC, ErmT, ErmY, ErmG, ErmA, ErmB resistant to erythromycin and tetM, tetO, tetS, tet32, tetW, tetQ resistant to tetracycline were found in ST59-t437 isolates.

Totally, 682,961,826 raw reads were obtained for all the MRSA strains, 643,281,744 valid reads were retained after quality control and the effective rate was 94.19%. The average genome sizes of MRSA were 2.79 ± 0.05 Mbp, with GC content 33% (Additional file 5). The average predicted genes were 2693.35 ± 65.19, with average length 875.05 ± 7.04 (bp), and the coding rate of all predicted genes were 84.50 ± 0.20% for all MRSA in this study. In addition, 50.72 ± 8.38 average tRNA-coding genes and 7.27 ± 0.92 average rRNA operons were identified. The average sequencing depth was 548.48 ± 92.51, and the average coverage of genome was 93.79 ± 1.28%, as Additional file 5 shown. In general, 2572.03 ± 60.90, 2122.35 ± 34.99, 2037.05 ± 16.12, and 2571.48 ± 60.81 average genes were annotated in NR, PFAM, Swissprot and TrEMBL database respectively. 1896.53 ± 17.79 and 1484.57 ± 8.98 average genes were annotated in GO and KEGG pathways, while 1963.68 ± 24.11 average genes were annotated in COG categories.

All CDSs from all MRSA sequenced in this study were used for core and pan-genome analyses. Core genes were considered as the most conserved and shared by all strains, whereas dispensable genes were shared by unique genes identified in only one of the genomes. The most of the CDSs were highly conserved among MRSA strains, and the core genomes of these isolates were 1593 genes (Fig. 2a). Dispensable genes were distributed ranging from 16 to 50 genes in different strains (Fig. 2a). According to homologous whole clusters COG expansions, 1040 genes were assigned to metabolism, e.g., E (amino acid transport and metabolism), G (carbohydrate transport and metabolism) and P (inorganic ion transport and metabolism). Five hundred sixty-four genes were assigned to information storage and processing category, e.g., K (transcription), L (replication, recombination and repair) and J (translation, ribosomal structure and biogenesis). Three hundred ninety-six genes were assigned to cellular processes and signaling, e.g., M (cell wall/membrane/envelope biogenesis), O (posttranslational modification, protein turnover, chaperones) and V (defense mechanisms), as Fig. 2b shown. Generally, the size of the core and pan-genome depended on the number of strains were considered in the analyses. It was evident that the pan-genome curve converges rapidly, indicating the variation in MRSA strains. However, the core genome size was stable, decreasing initially and reached a plateau at 2000 genes (Fig. 2c). This revealed that MRSA has an open pan-genome in this study.

Phylogenetic analysis based on pan-genome (core and dispensable genes) of MRSA showed that five clustering groups were generated with brown, yellow, green, blue and gray color (Fig. 3a). Interestingly, the different genotype or clones of strains clustered into different groups, thus, clustering group brown was ST6-t701 strains, green was ST59-t437 isolates, blue was ST239-t030, yellow was ST188-t189, gray was ST965-t062 and ST5-t14723 clone. The SNP based phylogenetic tree also generated identical five clustering groups showing with different color in this study (Fig. 3b). Pan-based phylogenetic tree of 84 MRSA strains (60 isolates in this study and 24 references) revealed that different clusters were generated according to the genotype profiles as mentioned above (Fig. 3c). All the ST239-t030 strains in this study were more closely related to T0131 isolate from Tianjin, China, belonged to ‘Turkish clade’ from Eastern Europe, but far away from TW20 of ‘Asian clade’ or other global reference isolates. Two groups of ST59-t437 clones of MRSA in Yunnan province were generated with some reference strains. YNSA7 to YNSA167 group was more closely related to SA40TW isolate, belonged to the ‘Asian-Pacific’ clone (AP), whereas YNSA154 to YNSA210, YNSA53 and YNSA11 group were closely related to SA957, M013 and SA268 strains, belonged to the ‘Taiwan’ clone (TW). Most of the CC5 clonal MRSA (ST5 and ST965 clone; YNSA34 to YNSA31) were separated from reference strains except YNSA453, as Fig. 3c shown.

Clustering ST239-t030 clone comprised all the HA-MRSA cases in this study; it was the primary clone for hospital acquired infection in southwest China. Colinear comparison of strains YNSA73 and YNSA94 in this genotype showed that the chromosome of YNSA73 was collinear with YNSA94 with 99.73% identity (Fig. 4a). ST6-t701 clone (clustering brown) mainly referred to food related diseases in community, since the patients and food strains clustered into identical group, and close similarities could be found among these strains. In this group, YNSA21 was a patient isolates, and YNSA365 was a food strain, the chromosome of YNSA21 was collinear with YNSA365 showing 99.71% identity (Fig. 4b). In addition, YNSA311 and YNSA480 were both isolated from food strains of different sources, their genomes were also collinear with each other showing 99.76% identity (Fig. 4c). ST59-t437 clone represented the most heterogeneous cluster in this study. The CA and HA-MRSA cases both contained ST59-t437 genotype, and some isolates from patients were related with food. Colinear comparison between YNSA115 (food) and YNSA279 (patient) revealed the 99.79% identity of two isolates (Fig. 4d). Therefore, ST59-t437 clone had a high degree of diversity for provoking different clinical spectrum of diseases in both community and hospital.