Group B Streptococcus Infections Caused by Improper Sourcing and Handling of Fish for Raw Consumption, Singapore, 2015–2016

Policies and guidelines regarding sale of ready-to-eat raw fish dishes have been updated.


GBS Analysis
Fish and fish tank water samples collected from ports were analyzed by Agri-Food and Veterinary Authority of Singapore by using a proprietary method that involved screening for GBS and GBS serotype III with a multiplex PCR method and isolation of viable GBS ST283 by culture method. GBS analyses on all other samples were conducted at EHI by using a modified protocol (1). Each fish surface swab sample was prepared by applying a premoistened sponge swab (3M Company, Maplewood. MN, USA) onto the entire surface area of each whole fish or fish part sample. The sponge swab was then suspended in 9 mL of Todd Hewitt broth (Acumedia, Lansing, MI, USA) containing colistin sulfate (7.5 g/mL) and nalidixic acid (5 g/mL) and homogenized by using a Stomacher (model 400; Seward Medical, Worthing, UK) with a paddle speed of 230 rpm for 30 s. A total of 25 g each of fish muscle sample, fish organ composite sample, and sliced fish sample was homogenized with 225 mL of Todd Hewitt broth containing antimicrobial drugs by using a Stomacher.
Each 500-mL fish tank water sample was concentrated by using membrane filtration with sterile 0.45-m pore size filters made of mixed cellulose esters (Merck Millipore, Darnstadt, Germany). Membranes were then suspended in 20 mL of Todd Hewitt broth containing antimicrobial drugs. Homogenized aliquots and membranes were incubated at 37°C for 24 h. Secondary enrichment was performed by transferring 1 mL of the enriched Todd Hewitt broth to 5 mL brain heart infusion broth (Acumedia) and incubating at 37°C for 24 h. Nucleic acids were extracted from enriched brain heart infusion broth and selected preenriched sample suspensions in Todd Hewitt broth containing antimicrobial drugs by using a method described (2). GBS was detected by using a singleplex PCR specific for the dltR gene (3). All GBS PCR-positive samples were cultured to obtain viable isolates. Culture involved plating of 100 L of serially diluted (0-10 5 dilutions) enriched brain heart infusion broth onto Brilliance GBS Agar (Thermo Fisher Scientific, Darmstadt, Germany) and incubating the agar at 37°C for 24 h.
Presumptive mauve colonies were confirmed as GBS by serologic analysis using the Streptococcal Grouping Kit (Oxoid, Basingstoke, UK) according to the manufacturer's instructions.
Although 103 of 522 enriched samples tested by EHI were positive for GBS by PCR, only 35 of these samples yielded GBS isolates when plated on chromogenic agar; GBS was not recovered from the other 68 PCR-positive samples. A high background flora could have masked GBS in these 68 enriched samples. This masking is supported by the observation that of the 68 isolate-negative samples, 57 initially showed PCR-negative results, but then showed PCRpositive results after a 2-day enrichment. This change in results indicated low amounts of GBS in the original samples. Presampling manipulation, such as freezing, could have also contributed to the low yield.
Because recovery of viable isolates was not always successful for GBS PCR-positive samples, the enriched broth of these samples was further screened for GBS serotype III by using published primers (4). This screening was performed to estimate the highest possible contamination rate of GBS serotype III ST283 in fish.
A total of 84 GBS isolates were characterized by using serotyping, multilocus sequence typing (MLST), and whole-genome sequencing methods. Serotyping of GBS isolates was performed by using the IMMULEX STREP-B Antisera Set (Staten Serum Institut Diagnostica, Hillerød, Denmark) according to the manufacturer's instructions. Results were confirmed by using a published multiplex PCR protocol (5). MLST was performed by using published primers and Sanger sequencing (6). Aligned nucleotide sequences of amplified products were compared with sequences in the PubMLST Streptococcus agalactiae Database (http://pubmlst.org/sagalactiae/) for identification of sequence types.
Whole-genome sequencing was performed by the Genome Institute of Singapore and the Agency for Science, Technology and Research by using a NextSeq 500 sequencer (Illumina, San Diego, CA, USA) according to protocols described (7). A total of 1 mL of each GBS overnight culture in brain heart infusion broth was centrifuged, and the pellet of bacterial cells was lysed by using enzymatic lysis buffer at 37°C for 45 min, followed by extraction using the DNeasy Blood and Tissue Kit (QIAGEN, Valencia, CA, USA) according to the manufacturer's instructions. Genomic DNA shearing was performed by using an M220 Focused Ultrasonicator (Covaris, Woburn, MA, USA), and library preparation was performed by using the TruSeq Nano DNA LT Library Prep Kit Pooled libraries (Illumina). Samples were then sequenced by using a NextSeq 500 sequencer with 2 × 151-bp reads. All sequencing data are available in GenBank under BioProject PRJNA293392.

Sequence Analysis
All primary sequence analysis was performed by the Genome Institute of Singapore with the Efficient Rapid Microbial Sequencing Platform. Reference-based analyses were performed by using the SG-M1 genome (8) as a reference. FASTQ files were mapped by using Burrows-Wheeler Aligner version 0.7.10 software (9). Indel realignment and single-nucleotide polymorphism (SNP) calling was performed by using Lofreq* version 2.1.2 with default parameters (10). Maximum-likelihood SNP trees (ignoring indels) of GBS ST283 strains were created by using SNPhylo (11). Neighbor-joining trees of all GBS strains were created by using APE version 3.5 (12). All phylogenetic trees were visualized with GGTREE version 3.2 (13)

SPCs and E. coli and S. aureus Counts
The protocols used by EHI are described below. Each fish surface swab sample was prepared by applying a premoistened sponge swab onto a 10 cm × 10 cm surface area of each fish part or whole fish sample. The sponge swab was then suspended in 9 mL Butterfield buffer and homogenized by using a Stomacher. A total of 25 g of each fish muscle and sliced fish sample was homogenized with 225 mL of universal preenrichment broth (Acumedia) by using a Stomacher. Each 500-mL fish tank water sample was concentrated by membrane filtration using 0.45-m pore size filters and suspended in 20 mL of universal preenrichment broth. SPCs and E. coli and S. aureus counts were determined performed according to described methods (18). The presence of the methicillin resistance gene and enterotoxin-producing genes (sea, seb, sec, sed, see, seg, seh, sei, sej, and sel) of 18 S. aureus isolates from 17 fish was determined by using published primers (19-21).

Detection of Aeromonas spp., Listeria monocytogenes, Salmonella spp., Vibrio cholerae, and V. parahaemolyticus
Each fish surface swab sample was prepared by applying a premoistened sponge swab onto the entire surface area of each fish part or whole fish sample. The sponge swab was then suspended in 9 mL of universal preenrichment broth and homogenized by using a Stomacher. All stomached aliquots (surface swab specimens, fish muscle, fish organs, sliced fish) resuspended in universal preenrichment broth were incubated at 37°C for 24 h for detection of Aeromonas

SPCs in Fish Pieces before and after Rinsing
We analyzed 21 sets of saltwater fish muscle samples to determine if SPC could be reduced by rinsing with water. Each set of samples consisted of 2 muscle pieces (each 5 cm × 5 cm × 1 cm). One piece served as a control (before rinsing), and the other piece was rinsed with 500 mL of sterile water. We analyzed fish muscle samples for SPCs according a described method (18).  Year festive dish consisting of raw fish slices served with raw vegetables and condiments; yusheng, Chinese style sliced raw fish dish served separately with porridge. †Food stalls refer to stalls housed within larger eating establishments that include hawker centers, coffee shops, and eating houses. Supermarkets refer to fresh produce sections of supermarkets and exclude sashimi and sushi counters of supermarkets. ‡-, no genes encoding resistance from the ARGannot database (including those acting on aminoglycosides, β-lactams, colistin, fosfomycin, fluoroquinolones, glycopeptides, macrolides, lincosamide, streptogramines, phenicols, rifampin, sulfonamides, tetracyclines, and trimethoprim) were detected by SRST2 from the genome sequencing data. §Isolates also described in a study (7) on the analysis of clinical, epidemiologic, and bacterial sequencing data obtained during investigation of group B Streptococcus infections. ¶Fish identified as tilapia (Oreochromis spp.) by using the DNA barcoding method (29). #Gene encoding resistance to tetracycline (tetO) was detected in 3 of the 5 isolates.