Healthcare-associated infections caused by chlorhexidine-tolerant Serratia marcescens carrying a promiscuous IncHI2 multi-drug resistance plasmid in a veterinary hospital

Abstract

The bacterium Serratia marcescens can cause opportunistic infections in humans and in animals. In veterinary settings, the diversity, reservoirs and modes of transmission of this pathogen are poorly understood. The phenotypes and genotypes of Serratia spp. isolated from dogs, cats, horses, a bird and a rabbit examined at an Australian veterinary hospital between 2008 and 2019 were characterised. The isolates were identified as S. marcescens (n = 15) or S. ureilytica (n = 3) and were placed into four distinct phylogenetic groups. Nine quasi-clonal isolates associated with post-surgical complications in different patients displayed high levels of resistance to the antimicrobials fluoroquinolones, cephalosporins, aminoglycosides, and to the disinfectant chlorhexidine. A Serratia sp. with a similar resistance profile was also isolated from chlorhexidine solutions used across the Hospital, suggesting that these infections had a nosocomial origin. A genomic island encoding a homolog of the Pseudomonas MexCD-OprJ biocide efflux system was detected in the chlorhexidine-tolerant Serratia. The nine multi-drug resistant Serratia isolates also possessed a Ser-83-Ile mutation in GyrA conferring fluoroquinolone resistance, and carried a large IncHI2 conjugative plasmid encoding antimicrobial and heavy metal resistances. This replicon was highly similar to a plasmid previously detected in a strain of Enterobacter hormaechei recovered from the Hospital environment. IncHI2 plasmids are commonly found in Enterobacteriaceae, but are rarely present in Serratia spp., suggesting that this plasmid was acquired from another organism. A chlorhexidine-tolerant Serratia isolate which lacked the IncHI2 plasmid was used in mating experiments to demonstrate the transfer of multi-drug resistance from a E. hormaechei donor. This study illustrates the importance of environmental surveillance of biocide-resistance in veterinary hospitals.

Fig 1. Heatmap of all-against-all ANI values for 133 complete Serratia spp. genomes and 18 veterinary isolates, showing distinct intra-specific subgroups within S. marcescens sensu lato.

The top dendrogram was generated from a distance matrix calculated from the ANI values using the “dist” function (euclidean method) followed by hierarchical clustering using the “hclust” function (complete method) in R. The ANI values are represented by a scale of grey. The darkest shade represents ANI values >97.5%, which is used as the lowest limit to define groups A-F within the S. marcescens sensu lato group. The strain names and ANI groups are indicated on the right side of the map and their corresponding accession numbers are listed in the S1 Table. The GenBank files headers were used to extract the reported taxonomic classification (top sidebar) and the host or origin (left sidebar) of each strain, and this information was color-coded accordingly. The 18 animal isolates from the U-vet are indicated by a pink square on the right side of the map. The number of ARGs per genome was calculated from individual ABRicate reports and color-coded on the far-left sidebar.

Fig 2. Core-genome phylogeny inferred from SNP analysis of the S. marcescens “sensu-lato” group, suggesting a quasi-clonal relation amongst the MDR isolates.

An unrooted tree was generated from the multiple alignment containing high-quality SNP, Indels, and structural variations within the core genome, by Parsnp, from 81 complete genomes using strain ATCC_13880 as a reference and excluding potential recombination sites. The resulting tree was retrieved from the Parsnp output files and decorated with iTOL. The species, strain names, host and accession numbers extracted from GenBank are specified for each node. The 18 animal isolates from the U-Vet are indicated by a pink square.

Fig 3. Effect of chlorhexidine exposure on the growth of Serratia isolates from the U-Vet, indicating the ability of the resistant strains to recover from routine disinfection protocols.

Duplicate cultures of each isolate were incubated in a 96-well plate. The average of the two OD620nm values after blank subtraction are plotted against time on a log-transformed scale. The single asterisk * indicates S. ureilytica isolates. A: Bacterial suspensions exposed to sterile water for 5 minutes, then diluted 1:10 in Letheen Broth. B: Bacterial suspensions exposed to 0.5% chlorhexidine for 5 minutes, then diluted 1:10 in Letheen broth. Red lines, filled circles: Chlorhexidine-resistant Serratia isolates including all 9 MDR isolates and the non-MDR strain CM2017_569 (double asterisk **). Blue lines, filled squares: Chlorhexidine-susceptible Serratia strains, E. hormachei strain CM2018_216, and control strain E. coli K12. Purple line, filled squares: S. ureilytica strain CM2008_163, showing intermediate resistance to Chlorhexidine. Insets indicate the OD620nm values shortly after the dilution step in Letheen broth (time 0).

Fig 4. Phylogenetic analysis of Serratia spp. genomes, demonstrating the distribution of unusual antimicrobial resistances patterns and plasmid incompatibility groups in phylogenetically related strains.

A maximum likelihood tree was generated from 147 concatenated multiple alignments of 13 house-keeping genes representing 19074 positions with MEGA using the General Time Reversible model with discrete Gamma distribution and Invariable sites (GTR+G+I), selected by the lowest Bayesian Information Criterion (BIC) score. A bootstrap consensus tree was inferred from 100 replicates. Bootstrap values are represented by branch thickness and circles with proportional diameter. The presence of ARGs and/or plasmids in the genomes were predicted by ABRicate using the databases resfinder and plasmidfinder, respectively. The unrooted tree was decorated with iTOL. Nodes indicate the species, strain and origin of each isolate as reported in the corresponding GenBank file. Bars represent the number of ARGs with importance rating. Circles or stars represent the incompatibility group of plasmid(s). Triangles indicate the presence of a Quinolone Resistance Determining Region (QRDR) in the GyrA protein sequence.

Fig 5. Comparison of IncHI2 plasmids from MDR Serratia spp., E. hormachei, Klebsiella pneumoniae and E. coli, showing the high similarity between the two replicons found at the U-Vet, and suggesting a common origin for this mobile genetic element within the hospital.

A) Mauve sequence alignments, showing local blocs of co-linearity and structural conservation between plasmids pCM2015_854 from Serratia (top), pCM2018_216 from E. hormachei (second top) and other most-closely related plasmids. B) Comparative alignments of plasmid sequences by BRIG showing the conservation of antimicrobial resistance loci. The four outermost rings (red-orange) correspond to the plasmid carried by the representative MDR Serratia strain CM2015_854, used as a reference. The inner rings (shades of blue) represent highly similar plasmids from other Enterobacteriaceae. From outer to inner rings: 1- plasmid pCM2015_854 annotations (red), 2- ARG loci (orange), 3- map of annotated features (red), 4- nucleotide sequence (red), 5- plasmid pCM2018_216 sequence (dark blue), 6- to 8- other plasmids sequences (medium to light blue). C) Local genetic maps of ARG loci of plasmids pCM2015_854 from Serratia and pCM2018_216 from E. hormachei.

Fig 6. Progressive Mauve alignment of the chromosomes from representative U-vet Serratia isolates, showing the presence of an Integrative Conjugative Element in two chlorhexidine-tolerant strains.

The position of the predicted ICE region in CM2015_854 and CM2017_569 is indicated by the red box. Locally colinear blocks are indicated in solid colors.