WGS of a cluster of MDR Shigella sonnei utilizing Oxford Nanopore R10.4.1 long-read sequencing

Abstract

Objectives

To utilize long-read nanopore sequencing (R10.4.1 flowcells) for WGS of a cluster of MDR Shigella sonnei, specifically characterizing genetic predictors of antimicrobial resistance (AMR).

Methods

WGS was performed on S. sonnei isolates identified from stool and blood between September 2021 and October 2022. Bacterial DNA from clinical isolates was extracted on the MagNA Pure 24 and sequenced on the GridION utilizing R10.4.1 flowcells. Phenotypic antimicrobial susceptibility testing was interpreted based on CLSI breakpoints. Sequencing data were processed with BugSeq, and AMR was assessed with BugSplit and ResFinder.

Results

Fifty-six isolates were sequenced, including 53 related to the cluster of cases. All cluster isolates were identified as S. sonnei by sequencing, with global genotype 3.6.1.1.2 (CipR.MSM5), MLST 152 and PopPUNK cluster 3. Core genome MLST (cgMLST, examining 2513 loci) and reference-based MLST (refMLST, examining 4091 loci) both confirmed the clonality of the isolates. Cluster isolates were resistant to ampicillin (bla_TEM-1), trimethoprim/sulfamethoxazole (dfA1, dfrA17; sul1, sul2), azithromycin (ermB, mphA) and ciprofloxacin (gyrA S83L, gyrA D87G, parC S80I). No genomic predictors of resistance to carbapenems were identified.

Conclusions

WGS with R10.4.1 enabled rapid sequencing and identification of an MDR S. sonnei community cluster. Genetic predictors of AMR were concordant with phenotypic antimicrobial susceptibility testing.

Introduction

MDR and XDR shigellosis are emerging public health issues, particularly given the ease of transmission combined with limited clinical data to guide empirical management of XDR Shigella.¹ Although infection is generally self-limited, the emergence of antimicrobial resistance limits the effectiveness of empirical antibiotic treatment, which has been utilized to decrease bacterial burden to mitigate the potential for transmission, particularly in the community. In Vancouver, British Columbia, we have observed a cluster of MDR Shigella sonnei in patients presenting to hospital for medical care and/or diagnostic testing.² Cases occurred predominantly among people experiencing homelessness (PEH) in the downtown eastside community in Vancouver, where the probable mode of transmission was faecal-oral.^3,4 Notably, the MDR S. sonnei cluster was associated with significant morbidity including bloodstream infections. Among the invasive isolates of S. sonnei, there was evidence of clonality with genotype 3.6.1.1.2 identified through WGS.⁴

Recent descriptions of outbreaks due to S. sonnei have been described and highlighted the utility of genomic sequencing to support outbreak investigations and interventions.^5–7 WGS in these outbreaks was performed utilizing Illumina sequencing, which enables high-throughput short-read sequencing. Although this technology has established itself as the gold standard for accuracy, it has proven to be relatively slow (days). This is especially true when compared with nanopore long-read sequencing, which has historically had low per-read and consensus accuracies but can be conducted rapidly (hours). The speed of nanopore sequencing has been leveraged in clinical settings for viral amplicon analysis, metagenomics and antimicrobial resistance (AMR) gene detection; however, the low accuracy has hampered application to bacterial outbreaks, which requires near perfect accuracy to resolve closely related isolates.

Recently, advancements in reagents, flowcells, basecalling software and bioinformatic tools have enabled drastic improvements in the quality of nanopore sequencing, enabling near-finished bacterial genomes in real time.⁸ Here, we leverage recent advances in nanopore sequencing to rapidly investigate a cluster of MDR S. sonnei cases in PEH in order to genetically characterize AMR genes and to support potential interventions.

Methods

WGS was performed on S. sonnei isolates identified from stool and blood at our microbiology laboratory servicing the downtown Vancouver community between September 2021 and October 2022. Isolates were extracted on the MagNA Pure 24 (Roche Diagnostics) and sequenced on the GridION (Oxford Nanopore Technologies; ONT) utilizing R10.4.1 flowcells. Cluster-related MDR S. sonnei isolates were from patients with no travel history,⁴ whereas outlier isolates were from patients with a recent history of travel outside of Canada. Phenotypic antimicrobial susceptibility testing with disc diffusion (ampicillin, ceftriaxone, ciprofloxacin, trimethoprim/sulfamethoxazole, tetracycline) and E-test (azithromycin) were performed as per CLSI M100.⁹ For this study, MDR S. sonnei was defined as resistance to the recommended first-line reportable antibiotics for Shigella spp. in the CLSI M100 32nd edition (ampicillin, a fluoroquinolone and trimethoprim/sulfamethoxazole), as well as azithromycin.

Raw data were basecalled with Guppy (v6.4.6) super accuracy (SUP) model on the ONT GridION. Reads were demultiplexed enforcing barcodes at both ends to avoid barcode crosstalk. Raw FASTQ files were submitted to BugSeq (version 20 April 2023) for further processing.¹⁰ In short, reads underwent preprocessing including filtering of short (<100 bp), low-quality (mean Phred <8) and low-complexity (DUST score >7) reads. Remaining reads were assembled with metaFlye (v2.9.2). Assemblies were binned by taxonomy, assessed for taxonomic purity and analysed for AMR with BugSplit.^11,12 Bin completeness was assessed with BUSCO (v5).¹³ AMR predictors were additionally assessed with ResFinder 4.1.¹⁴ Plasmids were detected with MOB-Suite, using BugSeq’s curated database, which preserves cluster identifiers but improves plasmid host prediction (e.g. by reassigning Escherichia plasmids to Enterobacterales, as they may transfect Shigella spp.).¹⁵ Plasmid host range was predicted using replicon type and relaxase type, along with observed host range of similar plasmids in the literature.¹⁶ Strain typing was performed with MLST, core genome MLST (cgMSLT), reference-based MLST (refMLST), PopPUNK and global genotype (identified with Mykrobe).^17–21 The phylogenetic tree (Figure 1) was constructed from refMLST allele distances using iTOL v6.²² The tree was rooted at the reference genome for S. sonnei (NCBI accession GCF_013374815.1) and default iTOL settings were otherwise used. Every ‘cluster’ isolate had another isolate within 10 alleles using the cgMLST method and 20 alleles using the refMLST method. refMSLT also assigned all ‘cluster’ isolates to the same 20-allele stable cluster code.^21,23

Figure 1.

Phylogenetic tree of Shigella sonnei isolated in our laboratory during the study period, including genomic predictors of resistance, core cluster plasmids and days since the first study case (TMP-SMX = trimethoprim-sulfamethoxazole). The tree is rooted at GCF_013374815.1, the reference genome for S. sonnei from NCBI. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Ethics approval was obtained through the University of British Columbia/Providence Health Care Research Ethics Board (H22-02183).

Results

During the study period, there were 100 S. sonnei identified, including 91 that were culture positive and 9 that were PCR-positive culture negative. Of the culture-positive samples, we sequenced 56 isolates, including 3 from blood cultures and 53 from stool cultures, to a median depth of 73-fold using ONT’s new R10.4.1 chemistry, achieving complete (>95% single-copy orthologues) and pure (<5% duplicated single-copy orthologues) assemblies for all isolates sequenced. Mean N50 read length for the ONT sequencing runs was 11 kbp.

Fifty-three ‘cluster’ isolates were identified as S. sonnei by sequencing, with global genotype 3.6.1.1.2 (CipR.MSM5), MLST 152 and PopPUNK cluster 3. cgMLST (examining 2513 loci) and refMLST (examining 4091 loci) both confirmed the clonality of the cluster isolates (Figure 1). There were three outlier isolates with travel history, which are summarized in Table 1. Sequence data have been uploaded to GenBank (BioProject number PRJNA1019784).

Table 1.

Characteristics of Shigella sonnei isolates during the study period

	Travel history	Global genotype	Global genotype name	MLST	Antimicrobial resistance genes
Cluster	N/A	3.6.1.1.2	CipR.MSM5	152	blaTEM-1, dfA1, dfrA17, sul1, sul2, gyrA S83L, gyrA D87G, parC S80I, ermB, mphA
Outlier 1	Belize	3.7.18	Global III	152	blaCMY, dfrA1, sul2, qnrB19, tet(A)
Outlier 2	Ghana	3.7.16	Global III	152	dfrA1, sul2, tet(A)
Outlier 3	Denmark	3.6.1	CipR parent	152	dfrA1, gyrA S83L

Plasmid content was highly conserved across cluster isolates, including a core set of 10 plasmids detected in 516/530 plasmid-isolate combinations (Table 2). In contrast, these plasmids were only detected in 13/30 outlier plasmid-isolate combinations. An additional two plasmids, AB685 and AB272, were detected in 31 and 25 out of 53 cluster isolates, respectively. Of the 10 core plasmids, 2 were predicted to be conjugative, 6 mobilizable and 2 non-mobilizable (Table 2) using the criteria of MOB-Suite, which incorporates mate-pair formation, relaxase and origin of replication markers.¹⁵ The two conjugative plasmids were also the plasmids harbouring all plasmid-borne AMR genes.

Table 2.

Details of 10 conserved plasmids relating to the cluster

Plasmid cluster IDa	Mate-pair formation type	Origin of replication type	Median size across samples (bp)	Median coverage across samples (n = 53)	Predicted host range	AMR genes found	Nearest NCBI accession	Replicon type(s)b	Relaxase type(s)	Mobility
AA282	MPFI	MOBP	94 230	84	Enterobacteriaceae	aadA1; aph(3′′)-Ib; aph(6)-Id; dfrA1; floR; sat2; sul2	KU932021	K2/Z type	MOBP	Conjugative
AA336	MPFF	MOBF	80 246	49	Enterobacterales	aadA5; blaTEM-1; dfrA17; erm(B); mph(A); sul1	CP044156	FIA type	MOBF	Conjugative
AA579	—	MOBQ	8032	113	Enterobacteriaceae		CP030114	rep_cluster_1778	MOBQ	Non-mobilizable
AA670	—	MOBunknown	2689	108	Bacillota/Pseudomonadota/Actinomycetota/Bacteroidota		CP038000	rep_cluster_2350	—	Non-mobilizable
AA747	—	—	109 443	20	Enterobacteriaceae		NC_018651	rep_cluster_488	—	Mobilizable
AA971	—	—	11 410	99	Enterobacteriaceae		CP043022	rep_cluster_2141	MOBQ	Non-mobilizable
AA974	—	—	9712	93	Enterobacteriaceae		CP019693	Col156	MOBQ	Non-mobilizable
AA979	—	—	11 369	99	Enterobacteriaceae		KP970685	Col156	MOBQ	Non-mobilizable
AB446	—	MOBP	10 319	135	Bacillota/Pseudomonadota/Actinomycetota/Bacteroidota		CP034069	ColRNAI_rep_cluster_1857; rep_cluster_2350	MOBP	Non-mobilizable
AC748	—	—	2438	96	Enterobacteriaceae		CP029581	Col(BS512)	—	Mobilizable

^aPlasmid cluster ID derives from MOB-Suite and reflects unique taxonomic identifiers for each plasmid.

^bIf a detected replicon cannot be assigned to a known incompatibility group, it is assigned to a replicon cluster (‘rep_cluster_*’).

Cluster isolates (n = 53) were universally resistant to ampicillin, trimethoprim/sulfamethoxazole, azithromycin and ciprofloxacin, and susceptible to ceftriaxone. BugSeq’s predicted phenotype matched the observed phenotype for all isolates: bla_TEM-1 (ampicillin); dfA1 and dfrA17 (trimethoprim); sul1 and sul2 (sulfamethoxazole); gyrA S83L, gyrA D87G and parC S80I (ciprofloxacin); and ermB and mphA (azithromycin) were present in all isolates. A subset of cluster isolates (n = 11) from October 2021 to mid-February 2022 harboured the tet(A) gene on plasmid AB595, which corresponded to phenotypic susceptibility testing (tetracycline resistant). All isolates recovered after mid-February 2022 no longer carried plasmid AB595 and the tet(A) gene, and of those available for testing (11/42), all 11 were phenotypically susceptible (Figure 1). No genomic predictors of resistance to carbapenems were identified. Resistance predictors were consistently detected on plasmids AA282 and AA336, with the exception of the chromosomal quinolone resistance mutation (Table 2).

Discussion

There is a pressing need for rapid and accurate sequencing to assist in the identification and containment of infectious disease clusters and outbreaks. Sequencing and identification of a clonal strain can provide critically important information in identifying the relevant interventions to stem transmission.^3–6 Traditionally, utilization of WGS for outbreak analysis has been limited, as testing is often only available at reference laboratories with long turnaround times. With technological advancements in nanopore sequencing and bioinformatics, we have been able to implement WGS in our clinical laboratory. Although accuracy has been a historical issue with long-read sequencing, the R10.4 flowcell was recently evaluated against the MiSeq 2 × 200 bp, showing comparability of bacterial genome sequencing at 40-fold coverage.⁸

We demonstrated across a range of typing methods—from the least granular method PopPUNK, to traditional MLST, cgMLST, global genotyping, and the most granular method refMLST—clonality of a single strain of S. sonnei causing disease in our population. Although all methods use different thresholds to establish epidemiological relatedness, interpretation for all methods in the context of our cluster was achieved with long-read sequencing including other methods beyond MLST that are sensitive to sequencing errors when examining larger proportions of the genome. PopPUNK, MLST, global genotype and refMLST all enabled stable cluster nomenclature in the context of a growing number of cases; global genotype and refMLST cluster names additionally offered the convenience of conveying relatedness to other clusters in a manner similar to SNP addresses.²⁴ The identification of a clonal cluster of S. sonnei contributed to informing potential interventions to mitigate community transmission. Shigellosis has been previously associated with sexual transmission among MSM patients,^4,5 but the local cluster of cases suggested the potential for faecal-oral transmission, shifting the focus to environmental measures, such as sanitation and hygiene.^4,25

Although shigellosis primarily affects individuals residing within a community, cases also impact acute care facilities, resulting in visits to the emergency department, hospital admissions and referrals to specialists. When severe infections are caused by MDR Shigella, there may be limited oral treatment options available on discharge from hospital.⁴ Supplementary benefits of sequencing relate to the availability of genotypic evidence of AMR genes and mutations. In response to the increasing number of cases, our microbiology laboratory added PCR testing of stool (in addition to routine enteric cultures) to improve the turnaround time for detecting Shigella. Although not performed in this study, theoretically, for cases that are PCR positive but culture negative, WGS can potentially provide genotypic prediction of AMR, which would not otherwise be available. Comparison of phenotypic antimicrobial susceptibility interpretation was concordant with the genotypic prediction using nanopore sequencing, including the detection of SNPs conferring quinolone resistance, which again demonstrates the base-level accuracy of our sequencing. In our clonal strain, three mutations associated with quinolone resistance (gyrA S83L, gyrA D87F and parC S80I) correlated with phenotypic results using the disc diffusion method, in keeping with previously reported quinolone AMR markers.^7,20 In outlier 1 (Table 1) sequenced during this study, only one genetic marker (qnrB19) was identified, which tested susceptible to ciprofloxacin by disc diffusion. However, outlier 3 carried gyrA S83L and tested intermediate by disc diffusion. The identification of only one genetic marker of resistance to quinolones has been associated with ‘decreased susceptibility’ rather than resistance,⁷ but could not be confirmed in our study as we did not have MIC results available. The clinical effectiveness of a quinolone with ‘decreased susceptibility’ remains unclear because that interpretation is extrapolated from Salmonella breakpoints in CLSI, though clinical failures have been reported.²⁶ Further study is needed to assess the relationship between the identification of AMR genetic markers, in vitro antimicrobial susceptibility testing and clinical effectiveness.

In addition, WGS enabled interrogation of AMR predictors for antimicrobials that may not be first-line antibiotics routinely tested in microbiology laboratories for Shigella spp., such as tetracycline and carbapenems, which are considered Tier 4 agents according to CLSI.²⁷ With increasing reports of MDR and XDR Shigella, WGS can provide clinicians with supplementary AMR data to inform management of difficult cases where there are limited or no effective antimicrobials except those with limited clinical data, such as carbapenems or pivmecillinam.^1,5

This study has some limitations. It was completed at a single microbiology laboratory in response to a cluster of MDR S. sonnei cases. Due to resource constraints, we were unable to sequence every isolate associated with the cluster during the study period, but were able to include a representative proportion of isolates throughout the study (approximately 60% of the total isolates). The only notable evolution of the clonal strain over time was the loss of tet(A) and tetracycline resistance, though we are limited in the ability to determine if the clonal lineage lost the plasmid or whether this was another similar lineage that never had the plasmid.

In response to a cluster of MDR S. sonnei infections, nanopore sequencing identified a clonal strain of S. sonnei with genotypic resistance to ampicillin, trimethoprim/sulfamethoxazole, ciprofloxacin and azithromycin. As sequencing technology continues to evolve, WGS for outbreak investigations and AMR testing will be more accessible to clinical laboratories, enabling the potential for WGS data to be used beyond surveillance to inform antimicrobial treatment and other interventions to interrupt transmission.

Funding

This study was supported by internal funding.

Transparency declarations

S.D.C. is a shareholder in BugSeq Bioinformatics Inc. The other authors report no relevant conflicts of interest.