A genomic survey of Clostridioides difficile isolates from hospitalized patients in Melbourne, Australia

ABSTRACT There has been a decrease in healthcare-associated Clostridioides difficile infection (CDI) in Australia, coupled with an increase in the genetic diversity of strains isolated in these settings, and an increase in community-associated cases. To explore this changing epidemiology, we studied the genetic relatedness of C. difficile isolated from patients at a major hospital in Melbourne, Australia. Whole-genome sequencing of C. difficile isolates from symptomatic (n = 61) and asymptomatic (n = 10) hospital patients was performed. Genomic comparisons were made using single-nucleotide polymorphism (SNP) analysis, ribotyping, and toxin, resistome, and mobilome profiling. C. difficle clade 1 strains were found to be predominant (64/71), with most strains (63/71) encoding both toxins A and B (A+B+). Despite these similarities, only two isolates were genetically related (≤2 SNPs) and a diverse range of ribotypes was detected, with those predominating including ribotypes commonly found in community-associated cases. Five non-toxigenic (A−B−CDT−) clade 1 strains were identified, all in asymptomatic patients. Three clade 4 (A−B+CDT−) and four clade 5 (A+B+CDT+) strains were detected also, with these strains more likely to carry antimicrobial resistance determinants, many of which were associated with mobile genetic elements. Overall, within a single hospital, C. difficile-associated disease was caused by a diverse range of strains, including many strain types associated with community and environmental sources. While strains carried asymptomatically were more likely to be non-toxigenic, toxigenic strains were isolated also from asymptomatic patients, which together suggest the presence of diverse sources of transmission, potentially including asymptomatic patients. IMPORTANCE There has been a decrease in healthcare-associated Clostridioides difficile infection in Australia, but an increase in the genetic diversity of infecting strains, and an increase in community-associated cases. Here, we studied the genetic relatedness of C. difficile isolated from patients at a major hospital in Melbourne, Australia. Diverse ribotypes were detected, including those associated with community and environmental sources. Some types of isolates were more likely to carry antimicrobial resistance determinants, and many of these were associated with mobile genetic elements. These results correlate with those of other recent investigations, supporting the observed increase in genetic diversity and prevalence of community-associated C. difficile, and consequently the importance of sources of transmission other than symptomatic patients. Thus, they reinforce the importance of surveillance for in both hospital and community settings, including asymptomatic carriage, food, animals, and other environmental sources to identify and circumvent important sources of C. difficile transmission.

C lostridioides difficile is the primary aetiological agent of infectious diarrhea in hospitalized patients (1), with infection remaining a significant global problem in healthcare facilities worldwide (2,3).However, both the clinical and molecular epidemi ology of C. difficile infection (CDI) have changed over time and, while community-associ ated cases are increasing, there is now a decrease in healthcare-associated cases.These cases are associated with increasingly genetically diverse strains (4).
C. difficile can transiently (or longer in hospital patients) colonize the gut asymptomat ically or cause a spectrum of diarrheal diseases collectively known as CDI, mediated by two major toxins TcdA (toxin A) and TcdB (toxin B).Some strains also encode an additional binary toxin known as (C.difficile transferase (CDT) (5).Variant strains carrying different combinations of these three toxins or containing variations within the toxin genes are seen also (6,7).The spread of C. difficile is facilitated by the production of highly resistant spores, which are shed in the feces of infected individ uals and are thought to be the major source of transmission (8).However, infection prevention and control measures, and antimicrobial stewardship in healthcare facilities have been successful in preventing transmission of C. difficile between symptomatic patients, and the genetic diversity of strains isolated in these settings has increased (4,(9)(10)(11)(12)(13).This genetic diversity mirrors that seen in community-associated CDI, supporting the importance of infectious sources other than symptomatic patients.Asymptomatic carriers of toxigenic C. difficile strains have been suggested as an alternative source of infection, and transmission between community and healthcare settings has been proposed also, with a recent study showing that 79% of C. difficile ribotypes detected in Australian hospitals were detected also in the community (14).Therefore, the identifica tion and molecular typing of C. difficile strains carried by both asymptomatic carriers and symptomatic CDI patients within healthcare facilities may provide important information about the epidemiology of CDI and inform infection prevention and control practices.
During a recent study at a Melbourne hospital, 71 C. difficile isolates were recovered from patients who presented with CDI symptoms (n = 61), or those who did not (n = 10).Subsequently, we investigated the genetic relatedness of these strains with PCR ribotyping and analysis of whole-genome sequence data using in silico multi-locus sequence typing (MLST), and further characterisation based on antimicrobial resistance (AMR) determinants and mobile genetic elements (MGEs).This investigation provides important information on circulating strains in a large Australian hospital.The findings support previous work suggesting that strains causing healthcare-associated CDI within a single facility are genetically distinct outside of an outbreak setting, and propose the presence of important sources of infection other than symptomatic patients.

Study design and participants
From 2014 to 2015, a C. difficile surveillance study was conducted at a >500-bed public hospital in Melbourne, Australia.Patients were included if they had C. difficile isolated from stool; current clinical guidelines recommend testing all patients with diarrhea who are inpatients.Standard laboratory criteria rejected non-liquid stool, except where specifically requested by the clinician due to a high clinical suspicion of CDI.For this study, faecal samples were collected in sterile containers and tested for C. difficile toxin using C. difficile Tox A/B II (TechLab).Isolates described as asymptomatic were recovered from formed stools from patients without CDI across multiple wards and floors from the same public hospital.C. difficile was isolated from enzyme immunoassay positive samples using ChromID C. difficile Agar (bioMérieux) and cultured for storage in cooked meat medium.

Bacterial culture
For the extraction of genomic DNA, C. difficile from cooked meat medium stocks was cultured anaerobically at 37°C in supplemented Heart Infusion (HIS) broth or on agar plates (Oxoid), supplemented with 0.375% glucose, 0.1% cysteine, and 0.1% sodium taurocholate.

Isolation of genomic DNA and sequencing
Genomic DNA was isolated from C. difficile as described previously (15).Whole-genome sequencing was performed on the Illumina MiSeq Platform, as described previously (16).

PCR ribotyping
Purified genomic DNA, as used for sequencing, was used as the PCR template.PCR ribotyping was performed as previously described (26,27).PCR ribotyping reaction products were concentrated using the Qiagen MinElute PCR purification kit (Qiagen Sciences, Germantown, MD, USA) and run on the QIAxcel capillary electrophoresis platform (Qiagen Sciences).The analysis of PCR ribotyping products was performed using the BioNumerics software package, v.7.6.3 (Applied Maths, Sint-Martens-Latem, Belgium).PCR RTs were identified by comparison with banding patterns in a reference library, consisting of a collection of 80 reference strains from the European Centre for Disease Prevention and Control and the most prevalent RTs currently circulating in Australia and Asia (T.V. Riley, unpublished data).Isolates that could not be identified with the reference library were designated with internal nomenclature, prefixed with QX, if there were more than two isolates with the same banding pattern.A single isolate with a new banding pattern was designated "novel."

Isolation and whole-genome sequencing of C. difficile from hospitalized patients
During the sample period, 71 C. difficile isolates were recovered using faecal swabs from patients with (n = 61) or without (n = 10) symptomatic CDI.Following the recovery of C. difficile from patient samples, genomic DNA was extracted for PCR ribotyping and whole-genome sequencing.In silico analyses of draft genome sequences included single-nucleotide polymorphism (SNP) analysis and core genome phylogeny, determin ing multi-locus sequence type (MLST), toxin gene profiling, pan-genome phylogeny, and resistome and mobilome analysis (Table 1).

Toxin gene profiles of hospital-associated C. difficile isolates
Genetic markers of virulence may be useful predictors of disease severity, thus all genome sequences were analyzed for the presence of genes encoding toxins A and B, and CDT.This analysis revealed that 63 of the 71 isolates (89%) encoded both major toxins (A+B+), with 3 (4%) encoding toxin B only (A−B+), and the remaining 5 (7%) being non-toxigenic (A−B−) (Fig. 1).All three A−B+ isolates belonged to clade 4 (STs 37 and 38) and were recovered from the 61 patients with symptomatic CDI (5%).There were also Total asymptomatic strains = 10 a The detection of mobile genetic elements including transposons, plasmids, and bacteriophage was performed using NCBI BLASTn against reference gene sequences, and the presence of antibiotic resistance genes was predicted using the Nullarbor v2.0 pipeline (https://github.com/tseemann/nullarbor).MLST STs were determined using the Nullarbor pipeline according to the pubMLST database (https://pubmlst.org/organisms/clostridioidesdifficileresamples and the Shimodaira-Hasegawa test which was performed with FastTree (18).The midpoint root for the tree was estimated using iTol and tree annotation was also performed in iTol (28).The isolates detected in symptomatic CDI patients are denoted in black, and those detected in asymptomatic patients are denoted in gray.Scale bar = substitutions per site.
four A+B+ isolates that were CDT+, with all four belonging to clade 5 (ST11).Three of the A+B+CDT+ isolates were isolated from patients with symptomatic CDI (5%) (Fig. 1).The remaining A+B+CDT+ isolate was isolated from 1 of the 10 asymptomatic patients as were the five non-toxigenic isolates all belonging to clade 1.The remaining four isolates in this patient group were A+B+CDT− and belonged to clade 1 (Fig. 1).

The mobilome of hospital-associated C. difficile isolates
MGEs are an important component of the C. difficile genome, with up to 11% of the genome originating from exogenous elements, and the associated insertions and recombination events driving evolution of the C. difficile genome (30,31).Therefore, the genome sequence of each isolate was probed for MGEs previously identified in C. difficile, finding that MGEs were carried by approximately 52% (37/71) of isolates in this study, including approximately 47% (30/64) of clade 1 isolates, and all clade 4 (3/3) and clade 5 (4/4) isolates (Table 1).The detection of transposons primarily focussed on those known to encode AMR determinants; however, all identified resistance genes were also analyzed for transposon-related genes up-and downstream.Known transposons and novel transposon-like elements were detected in approximately 31% (22/71) of isolates in this study, with the presence of each being successfully correlated with resistance genes predicted in silico (Table 1).Known and predicted transposons were detected in 26% (16/61) of isolates from symptomatic patients and 60% (6/10) of isolates from asymptomatic patients (Table 1).Ten isolates carrying Tn916-like elements, and three isolates carrying Tn5397 were predicted to encode tet(M), one isolate carrying Tn4453 was predicted to encode catD, and five isolates carrying Tn6194 were predicted to encode erm(B) (Table 1).Four isolates were predicted to encode the aadE-sat4-aphA-3 aminoglycoside resistance locus on a novel Tn6189-like element, whereas it appears that the region usually encoding ermB had been replaced with a region encoding aadE-sat4-aphA-3 (Table 1), with conserved flanking regions showing up to 96% nucleotide identity (data not shown).Similarly, two clade 5 isolates carrying a Tn4453a/b-like transposon were predicted to encode the aminoglycoside resistance locus, aph2-aadA-aac-aacA/ aphD, which also appears to have replaced a region usually encoding catD (32).
This analysis also included examination for C. difficile bacteriophages with a predicted role in virulence, including ϕCD38-2, ϕCD27, and ϕCD119, which have been implicated in the regulation of TcdA and TcdB (34).Only two isolates (2/71) were predicted to carry a bacteriophage, with both encoding the ϕCD38-2 prophage (Table 1).

DISCUSSION
An epidemiological survey for the presence of C. difficile in patients admitted to a Melbourne hospital was conducted, in which 71 isolates of C. difficile were recovered from symptomatic (n = 61) and asymptomatic (n = 10) patients.All isolates were whole-genome sequenced, with phylogenetic analysis showing that the circulating strains in this single facility were genetically diverse, and thus unlikely to have been transmitted between symptomatic inpatients.Most of the isolates (64/71) fell within the clade 1 phylogenetic lineage, a highly heterogeneous group, with the remaining isolates belonging to clade 4 (3/71) and clade 5 (4/71) (Fig. 1).This result is in line with recent studies showing an epidemiological shift in healthcare-associated CDI, with high levels of genetic diversity detected within hospitals (4,13,14), rather than single lineages dominating the hospital environment as seen in outbreaks.
The genetic diversity of isolates detected in this study also supports the likelihood of a multitude of infection sources other than symptomatic patients, including asymp tomatic carriers, food, and the environment (4, 9-11, 13, 14, 35-40).Similarly, high levels of genetic diversity between C. difficile strains isolated from symptomatic CDI patients have previously been reported (41), though the findings of this study indicate that this diversity is recapitulated even in patients not suffering from symptomatic CDI.Further supporting the divergence in infection sources are the increasingly blurred lines between strain types that cause healthcare-acquired versus community-associated CDI.Of note, the most prevalent ribotypes detected in hospitalized patients in this study were commonly associated with community-associated infection and environmental contamination, including the clade 1 (A+B+CDT−) ribotypes 014/020 (19/71) and 002-like (7/71), two of the most frequently isolated ribotypes in Australia (14,35) (Fig. 1).C. difficile ribotype 014/020 strains are commonly isolated from both hospital-acquired and community-associated CDI cases in Australia, as well as being detected in asympto matic patients (14,36), and environmental sources such as public lawns (39), soil and mulch from hospital grounds (40), and retail root vegetables (38).C. difficile 014/020 strains are also commonly isolated from pigs in Australia (37,42), with high levels of genetic relatedness between pig and human isolates, suggesting that transmission occurs between species (37).C. difficile ribotype 002 strains are often associated with community-associated CDI in Australia (14,36), and can also be detected in environmen tal sources such as public lawns (39) and root vegetables (38).Similar trends have been seen in the UK, where one study showed that C. difficile isolated from patients with CDI within a single hospital tends to be genetically diverse, however highly related strain types can be isolated from environmental sources in the same geographic region (13).Therefore, it is possible that sources such as animals, food, and the environment have contributed to the spread of these lineages throughout Australia, leading to their inevitable prevalence in hospitalized patients.
Epidemic ribotypes that are commonly isolated in the Northern Hemisphere such as the clade 2 ribotype 027 and the clade 5 ribotype 078 are not endemic in Australia and were not detected in this study (Fig. 1).However, similar clade 5 ribotypes such as ribotype 126-which is closely related to ribotype 078-and ribotypes 127 and 033 are found in Australian livestock (27).This was reflected in this study, although a novel ST11 strain, ES1368 (A+ B+CDT+) made up 5.6% (4/71) of the strains detected (Fig. 1), suggesting further genetic diversity within this clinically important clade.These clade 5 strains were also the only strains found in this study to encode the C. difficile binary toxin (CDT) (Fig. 1) which is associated with "hypervirulent" lineages.Unsurprisingly, all isolates recovered from symptomatic CDI patients in this study encoded toxin B (Fig. 1), consistent with previous findings that toxin B is essential for virulence in C. difficile (43).Further supporting the role of toxin B in CDI pathogenesis is the increasing prevalence of strains encoding toxin B but not toxin A (7,44,45), three of which were identified in this study in symptomatic CDI patients (Fig. 1).These toxin-variant clade 4 (A−B+CDT−) strains made up 4.2% (3/71) of the total isolates detected in this study, with two of these clade 4 isolates belonging to ribotype 017 (Fig. 1).Clade 4 ribotype 017 strains are predominant in Asia (46), and their low prevalence in Australia, as seen in this and previous studies, suggests that these cases may be caused by imported, rather than circulating strains (46).While asymptomatic patients had a higher rate of non-toxigenic C. difficile carriage in this study (Fig. 1), the presence of toxigenic C. difficile in this patient group is consistent with a recent meta-analysis showing toxigenic strains present upon admission in asymptomatic patients ranging in prevalence from 4.1% to 15% (12), suggesting that asymptomatic carriage of toxigenic C. difficile may represent an important source of transmission within hospitals and the community.However, the relatively few C. difficile detected in asymptomatic patients (n = 10) compared to symptomatic patients (n = 61) in this study limits our ability to make comparisons between these patient groups.This limitation was also evident in the analysis of MGE carriage in this study, which found a higher prevalence in the smaller asymptomatic patient group (60% vs 26% in symptomatic patients).
C. difficile is known as one of the most urgent threats to public health due to its antimicrobial resistance (47), and the carriage of MGEs encoding AMR genes in this study sees C. difficile acting as a potential reservoir of these resistance genes in the gut (Table 1).While there is an established link between AMR and CDI outbreaks (1), the isolates recovered in this study were determined to be genetically distinct, with no indication of a C. difficile outbreak being responsible for infections.However, the abundance of AMR determinants carried by the isolates in this study, particularly in clades 4 and 5 (Table 1), suggests that antimicrobial use contributes to the persistence of these lineages.Determination of antibiotic resistance susceptibility in these clades 4 and 5 isolates may provide further insight into the functional role of the detected genes in the persistence of these isolates.Of note, there were seven unique amino acid substitutions detected in GyrB in this study, four of which have been previously associated with fluoroquinolone resistance in C. difficile (S366A, S366V, S416A, and E466V) (48)(49)(50); however, Australia has largely avoided the clonal spread of C. difficile associated with fluoroquinolone use in the Northern Hemisphere, likely because of the conservative use of these antimicrobials in Australia (51).
Overall, the results of this study correlate with those of other recent investigations (9-14), supporting the observed increase in genetic diversity and prevalence of communityassociated strain types amongst C. difficile within a single hospital setting in Australia, and consequently, the apparent importance of sources of transmission other than symptomatic patients.Further surveillance is required for C. difficile in both hospital and community settings, including asymptomatic carriage, food, animals, and other environmental sources, to identify and circumvent important sources of C. difficile transmission.

FIG 1 7 FIG 2
FIG 1Core phylogenetic analysis, MLST, ribotyping, and toxin profiling of hospital-associated C. difficile isolates.Using the Nullabor v2.0 pipeline (https:// github.com/tseemann/nullarbor),a core genome phylogeny of 99,370 SNPs was inferred with a maximum-likelihood tree using a GTR + G4 model, and multi-locus sequence types and clades were determined according to the C. difficile pubMLST database (https://pubmlst.org/organisms/clostridioidesdifficile).Ribotypes were determined via PCR ribotying, and toxins were detected in silico using NCBI BLASTn against reference gene sequences.The phylogenetic analysis was performed in comparison to the reference strain, C. difficile 630, denoted here in bold.The isolates detected in symptomatic CDI patients are denoted in black, and those detected in asymptomatic patients are denoted in gray.Isolates marked with an asterisk (*) could not be ribotyped.Each isolate is represented on the phylogenetic tree in relation to their clade, multi-locus sequence type (ST), ribotype (RT), and encoded toxins (toxins A, B, and CDT) with the + andsymbols denoting the presence or absence, respectively, of these toxin genes in silico.Each color represents the different groups detected in each typing scheme as a visual representation of diversity.Scale bar = substitutions per site.

TABLE 1
The mobilome and resistome of hospital-associated C. difficile isolates a

Isolate type/strain no. Mobile genetic elements Antibiotic resistance determinants Classification Plasmids Bacteriophage Transposons Resistance genes GyrA mutations GyrB mutations RT ST
(Continued on next page) Research Article Microbiology Spectrum November/December 2023 Volume 11 Issue 6 10.1128/spectrum.01352-234

TABLE 1
The mobilome and resistome of hospital-associated C. difficile isolates a (Continued)

Isolate type/strain no. Mobile genetic elements Antibiotic resistance determinants Classification Plasmids Bacteriophage Transposons Resistance genes GyrA mutations GyrB mutations RT ST
cMCD80Tn5397 tet(M) 039(Continued on next page)

TABLE 1
The mobilome and resistome of hospital-associated C. difficile isolates a (Continued)

Isolate type/strain no. Mobile genetic elements Antibiotic resistance determinants Classification Plasmids Bacteriophage Transposons Resistance genes GyrA mutations GyrB mutations RT ST
). Antibiotic resistance genes associated with transposons are listed in the same row.Amino acid substitutions in GyrAB proteins were detected by multiple sequence alignments were performed using EBI-EMBL Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/), in comparison to the fluoroquinolonesusceptible reference strain C. difficile 630.GyrAB mutations previously reported to confer fluoroquinolone resistance are in bold.