Clostridioides difficile from Fecally Contaminated Environmental Sources: Resistance and Genetic Relatedness from a Molecular Epidemiological Perspective

Clostridioides difficile is the most important pathogen causing antimicrobial-associated diarrhea and has recently been recognized as a cause of community-associated C. difficile infection (CA-CDI). This study aimed to characterize virulence factors, antimicrobial resistance (AMR), ribotype (RT) distribution and genetic relationship of C. difficile isolates from diverse fecally contaminated environmental sources. C. difficile isolates were recovered from different environmental samples in Northern Germany. Antimicrobial susceptibility testing was determined by E-test or disk diffusion method. Toxin genes (tcdA and tcdB), genes coding for binary toxins (cdtAB) and ribotyping were determined by PCR. Furthermore, 166 isolates were subjected to whole genome sequencing (WGS) for core genome multi-locus sequence typing (cgMLST) and extraction of AMR and virulence-encoding genes. Eighty-nine percent (148/166) of isolates were toxigenic, and 51% (76/148) were positive for cdtAB. Eighteen isolates (11%) were non-toxigenic. Thirty distinct RTs were identified. The most common RTs were RT127, RT126, RT001, RT078, and RT014. MLST identified 32 different sequence types (ST). The dominant STs were ST11, followed by ST2, ST3, and ST109. All isolates were susceptible to vancomycin and metronidazole and displayed a variable rate of resistance to moxifloxacin (14%), clarithromycin (26%) and rifampicin (2%). AMR genes, such as gyrA/B, blaCDD-1/2, aph(3′)-llla-sat-4-ant(6)-la cassette, ermB, tet(M), tet(40), and tetA/B(P), conferring resistance toward fluoroquinolone, beta-lactam, aminoglycoside, macrolide and tetracycline antimicrobials, were found in 166, 137, 29, 32, 21, 72, 17, and 9 isolates, respectively. Eleven “hypervirulent” RT078 strains were detected, and several isolates belonged to RTs (i.e., RT127, RT126, RT023, RT017, RT001, RT014, RT020, and RT106) associated with CA-CDI, indicating possible transmission between humans and environmental sources pointing out to a zoonotic potential.


Introduction
Clostridioides difficile (formerly Clostridium difficile) is a Gram-positive, anaerobic, sporeforming, toxin-producing, rod-shaped bacterium, which can cause diarrhea but also more severe disease, such as pseudomembranous colitis and even toxic megacolon [1,2].CDI usually occurs after antibiotic exposure when the normal gut microbiota is disrupted, giving vegetative and spores of C. difficile the ability to thrive.Treatment with antimicrobials, including penicillins, cephalosporins, fluoroquinolones and the macrolide-lincosamidestreptogramin B (MLS B ) antimicrobials, is considered a high risk factor for CDI development [3][4][5].The pathogenicity of C. difficile strains is predominately dependent on the release of two toxins; toxin A (tcdA) and toxin B (tcdB), which contribute to CDI and the respective genes, are encoded on a 19.6 kb pathogenicity locus (PaLoc) together with the regulatory components, TcdR, TcdC and TcdE [6].Additionally, binary toxin (CDT) encoded by cdtAB is associated with so called "hypervirulent" strains [7].Besides CDT, these "hypervirulent" strains might harbor mutations in the toxin repressor gene tcdC, leading to a higher toxin production [8].
C. difficile can be characterized by PCR ribotyping on a molecular level, and several ribotypes (RTs) are of epidemiologic importance.For instance, nosocomial CDI is often associated with "hypervirulent" RT027, which has been frequently found in hospital settings and outbreaks, especially in Europe, North America and to some extent in Asian countries [9][10][11].Furthermore, other "hypervirulent" RTs, such as RT023, RT078, RT126, RT127, and RT176, are known [12][13][14][15].Of note, RT078 is more commonly associated with community associated (CA)-CDI.In previous years, the zoonotic potential of C. difficile has been under scientific investigation.Several studies have reported that the environment, including animals and food, can be considered as a potential source of CA-CDI [7,[16][17][18].However, up to this date, these reservoirs and C. difficile transmission outside the hospital environment are not fully understood.
Animal manure and sewage sludge often contains C. difficile spores after being treated by digestion or composting in digesters or biogas plants [22,24,25].Subsequently, the disposal of animal manure and feces, manure-, biogas plant-and thermophilic digesterderived materials or digested sewage sludge as agricultural fertilizers might contribute to environmental contamination with C. difficile.
Exemplified for RT078, strains from both humans and animals are genetically related based on subtyping techniques, such as whole genome sequencing (WGS) following by subsequent phylogenetic analysis [13,[26][27][28], which demonstrates evidence for zoonotic transmission of C. difficile between humans and animals.In particular, WGS provides more-in-depth information about genetic diversity and relatedness resulting in a better understanding of the source and the evolution of C. difficile contributing to the current molecular CDI epidemiology [29].
Furthermore, the rapid resistance formation in C. difficile strains poses a significant threat to global health, driven by the increased use of antimicrobials as a treatment against other intestinal pathogens [3], and is known to promote CDI.Several recent studies have reported the emergence of virulent-resistant bacterial pathogens from a variety of sources, increasing the need for the appropriate use of antimicrobial agents.In C. difficile, accessory antimicrobial resistance (AMR) genes are often located on mobile genetic elements (MGEs) (i.e., conjugative and mobilizable transposons, plasmids, and prophages).They can be transferred via horizontal gene transfer (HGT), within toxigenic and non-toxigenic C. difficile strains [30] as well as other bacterial species (i.e., Bacillus subtilis and Enterococcus faecalis) [31,32].In this study, the strain composition and corresponding phenotypic and genotypic antimicrobial resistance and virulence-associated factors were evaluated giving insight into the molecular epidemiology of C. difficile of environmental origin from Northern Germany.In a second step, the genetic relationship between C. difficile isolates was determined by using core genome multi-locus sequence typing (cgMLST) based on WGS to show possible epidemiologic intersections.

PCR-Ribotyping and Toxin Genotyping
PCR ribotyping was conducted as described previously [34].In short, a standardized ESCMID (European Society of Clinical Microbiology and Infectious Diseases) protocol was utilized together with capillary gel electrophoresis.The obtained C. difficile isolates were characterized for toxin A (tcdA), toxin B (tcdB) and binary toxins (CDT, cdtA B) by conventional PCR [35], and results were confirmed by analyzing the genome of C. difficile strains (see below in Section 2.4).

Whole Genome Sequencing and Data Analysis
To determine the genetic relationship of the C. difficile isolates, 166 isolates were subjected to WGS using the Pacific Biosciences long-read platform Sequel IIe (Pacific Biosciences Inc., Menlo Park, CA, USA) and were subsequently de novo-assembled using the SMRT Link software versions 10 and 11 (Pacific Biosciences Inc.) as described recently [36].For molecular subtyping and to determine the genetic relationship of the different isolates, the cgMLST approach as described elsewhere was applied [37].Using the Ridom SeqSphere + software version 9 (Ridom GmbH, Münster, Germany), the cgMLST genes were extracted, and a minimum-spanning tree was constructed to display the genotypic clustering.For backwards compatibility, the "classical" MLST Sequence Types (STs) were extracted in accordance to the C. difficile MLST database of the PubMLST website (https://pubmlst.org/organisms/clostridioides-difficile/.Accessed 15 November 2022).In addition to the minimum-spanning tree analysis, all single nucleotide polymorphisms (SNPs) were extracted from the cgMLST target genes that were present in all strains investigated, and a phylogenetic tree (neighbor-joining tree) was constructed using the SeqSphere + software.Subsequent graphical representation was done using the iTOL tool version 5 [38].For further in-depth analysis, the WGS datasets were annotated using the RAST server (the rapid annotation using subsystem technology) version 2.0 (https://rast.nmpdr.org/.Accessed 15 November 2022) [39].AMR genes were identified by screening contigs with the CARD version 2 (the comprehensive antibiotic resistance databases) using resistance gene identifier (RGI) (https://card.mcmaster.ca/.Accessed 11 April 2023), BacAnt [40], ResFinder 4.1 (https://cge.food.dtu.dk/services/ResFinder/. Accessed 11 April 2023) [41], ARG-ANNOT [42] and Vrprofile2 [43].The genomes were further analyzed for the presence of known point mutations associated with resistance to fluoroquinolones (e.g., substitution in GyrA and GyrB subunit of the gyrase enzyme) and rifampicin (substitution in RpoB enzyme) using CARD and Snippy v.4.6.0 (https://github.com/tseemann/snippy. Accessed 25 November 2022), respectively.
The toxin genes were identified by using the virulence factors database from Ba-cAnt [40] as well as by annotation provided by the RAST server (https://rast.nmpdr.org/.Accessed 15 November 2022).
All contig sequences generated were submitted to NCBI GenBank under BioProject number (PRJNA1011814).

Results
The collection of environmental C. difficile isolates, which were characterized phenotypically and genotypically in the current study, was obtained from different environmental sources in the Northern region of Germany as described previously [22].The isolates were characterized for antimicrobial susceptibility patterns, and the genomic characterization was assessed for the RT diversity and the prevalence of virulence-encoding genes and AMR genes.In addition, the genetic relatedness among C. difficile isolates was performed using cgMLST based on WGS.

Antimicrobial Susceptibility
The antimicrobial susceptibility of 166 C. difficile isolates to five tested antibiotics and their corresponding RTs and STs is shown in Table 2 and Table S1.All C. difficile strains were susceptible to metronidazole and vancomycin.Overall resistance towards clarithromycin, moxifloxacin and rifampicin was encountered in these strains as follows: 26% ( 43 2).

Antimicrobial Resistance (AMR) Genes
All 166 C. difficile strains harbored at least four accessory AMR genes (Table S1).The most common accessory AMR genes were gyrA and gyrB, conferring fluoroquinolone resistance and found in all strains, caused via mutations in the quinolone resistance determining regions (QRDRs) of DNA gyrase subunits A (gyrA) and/or B (gyrB) (not separately shown in Figure 7).The blaCDD-1 encoding beta-lactamase could be detected in 137 strains (83%) whereas the blaCDD-2 gene was found only in 29 strains (17%).The second most abundant resistance gene is tet(M) detected in 72 strains (43%) and conferring tetracycline resistance by protecting the ribosomal protection protein.The aph(3 )-IIIa gene encoding aminoglycoside resistance was found in 64 strains (39%) whereas ant(6)-la gene conferring also aminoglycoside resistance was found in 36 strains (22%).The sat-4 gene encoding streptothricin resistance was found in 32 strains (19%), and ermB encoding a methylase enzyme that protects the 23S rRNA from the binding of the MLS B group antimicrobials was found in 21 strains (13%) (Figure 7A).

Discussion
The impact of environmental sources for CDI development is still poorly understood.The presence of toxigenic or non-toxigenic C. difficile has been documented in different environmental sources outside healthcare institutions, such as animal feces, manure, soil, food, and municipal WWTPs [17, 21,22,24,44], which could be served as potential sources of CA-CDI.
In the present study, a large strain diversity was evident with several strains being of higher epidemiologic importance.In particular, RT014 and RT020 as one of the most often encountered RTs in human disease could be detected together with RT001 which is considered to be a nosocomially associated strain [12].Furthermore, RT001 and RT014 were one of the most frequently detected in isolates from poultry meat in Germany [19].RT014 was also detected in soil samples being located next to a dairy farm [45].RT014 and RT020 were the predominant RT among soil isolates obtained from home gardens in Western Australia [46] and poultry feces [20].
On the other hand, strains that harbor the binary toxin, such as RT126, RT127 and RT078, were present as well.Of note, RT127 was a major clinical strain in Northwestern Taiwan for the years 2009-2015 [14] and was the most numerous RT detected in this study.Moreover, this RT was most frequently found in toxigenic isolates (50.2%) with CDT among obtained RTs from a calf farm in Australia [47].
A similar situation is given for RT126.RT126 was predominately detected in the feces of calves.RT126 has already been described in cattle [21,44] and pigs [44,48].Furthermore, RT126 has been observed as one of the predominant RTs in a veal calf farm in Belgium [49].In Spain, RT126 is one of the most common RTs among clinical isolates [48], and RT126 was also identified in clinical isolates in Southern Taiwan [50].In a study carried out by Primavilla et al. [51] in hospital food in central Italy, RT126 was also the second most frequently detected RT in CDI cases.
RT023 was identified in 2% of isolates being obtained from RS, RSS, and TDB samples.RT023 prevalence, isolated from humans in Europe, was ~3% [12].Interestingly, RT018 was found in three isolates recovered from municipal WWTP samples (RS and DSS).In the past, RT018 has been associated with a C. difficile outbreak in Southern Germany [57].More importantly, RT018 is considered to be the most predominant RT in Northern Italy with prevalence rates exceeding 40% [58].
In summary, concerning molecular epidemiology: RTs being frequently encountered in humans, such as RT001, RT014, and RT020 were present in the collected environmental samples.This might indicate that digested sewage sludge, untreated sewage, raw sewage sludge, biogas plant derived materials and thermophilic digesters treating biowaste or sewage sludge could pose a reservoir of toxigenic C. difficile RTs.
In addition to the classical differentiation of C. difficile isolates by ribotyping, the genome sequences were determined as well.This enabled us to further subgroup the isolates.Initially, the grouping was performed based on the cgMLST allelic profiles.This analysis revealed 20 clusters and 47 singletons.Many clusters corroborated with ribotyping results.However, in some instances, cgMLST was unable to group the isolates in accordance with their RTs, e.g., isolates sharing RT078 and RT126 or RT014 and RT020, where the allelic profiles only differed in up to five alleles.This is, however, in agreement with recent studies, which observed clustering of several RTs (e.g., RT078/RT126, RT014/RT020) [61,62].Here, the current study could demonstrate that the distribution of virulence genes, coding for i.e., the toxins A and B and the binary toxins, is concordant with the phylogenetic branching.This indicates that the different branches, which also represent to some extent the different clades, are stable lineages, and acquisition of the mentioned toxins was an early process during the evolution of these lineages, which goes in line with the clonal population structure [63].
For backwards compatibility, "classical" MLST STs (with seven loci) were also extracted from the genomic data set.Here, 32 distinct STs were determined that showed a good correlation to cgMLST typing results.In contrast, the comparison to ribotyping was not always concordant.For example, isolates of ST11 exhibited different RTs (RT127, RT126 and RT078), which were also separated in most instances using cgMLST.In summary, these results go in line with previous results, where RTs could be correlated with STs only to some extent [63].
C. difficile has been known to be resistant to multiple antimicrobials, such as tetracyclines, fluoroquinolones, lincomycin, erythromycin, aminoglycosides, macrolides, and beta-lactam antimicrobials, that are commonly used against bacterial infections in clinical settings [3,5] and continue to be associated with the highest risk for CDI [3].In the present study, resistance to MXF was frequently detected in ST11 (RT126 and RT078) isolates from the feces of calves and digested sewage sludge-amended soils.Many of RT126 isolates were additionally resistant to CLR, which belongs to the macrolide antibiotic class.These findings are in accordance with what have been reported in calf farms in Italy [21].Rates of antimicrobial resistance in C. difficile differ in diverse geographic regions [4].In particular, resistance to fluoroquinolones, macrolides, lincosamides, and tetracyclines has been associated with the spread of ST11 sublineages [64].In addition, C. difficile has evolved multiple AMR mechanisms that contribute to the development of AMR in C. difficile: (a) harboring of resistance-associated genes in the bacterial chromosome that could be transferred via HGT, including conjugation, transduction or transformation, (b) selection pressure leading to gene mutations, (c) alterations in the antibiotic targets and/or in metabolic pathways in C. difficile and (d) biofilm formation [3,65].
In the current study, six different tetracycline resistance genes in 51% of isolates were identified, including tet(M), tet (40), tetA(P), tetB(P), tet(O), and tet(L).The tet(M) was the predominant gene of the tet class in C. difficile strains (43%) and the majority of C. difficile RT126 and RT127 isolates were positive for tet(M), confirming that tetracycline resistance is widespread among ST11 isolates from a cattle farm.This finding supports the hypothesis of a zoonotic origin of these infections caused by large amounts of tetracyclines used in animal husbandries resulting in a high load released into the agro-ecosystem via organic fertilizers [21,66].Also, tet(M) gene was identified in non-toxigenic C. difficile RT140 and RT031 strains.It has been reported that all non-toxigenic tet(M)-positive strains from Indonesia and Thailand carried Tn916 or Tn5397 transposons [65].In C. difficile, acquired accessory AMR genes are often located on MGEs, and the most common element associated with tet(M) mediated tetracycline resistance is Tn5397 and Tn916-like transposons [3,5].These elements play a crucial role in HGT between distinct toxigenic and non-toxigenic C. difficile strains and between C. difficile strains and other intestinal pathogens.For instance, Tn5397 carrying tet(M) gene was shown to be transferred from C. difficile to Bacillus subtilis [31] and Enterococcus faecalis [32].The tet(40) gene, which encodes tetracycline efflux, was identified only in RT126 and RT078 isolates which represent 10% from 166 isolates.In a recent study, in 2.1% of 10,330 publicly available C. difficile genomes, tet(40) gene could be identified [65].Intriguingly, other tet resistance genes, such as tetA(P) and tetB(P) were found in nontoxigenic RT073 and toxigenic RT001 strains.The tetA(P) gene, which mediates active efflux of tetracycline, and tetB(P) gene related to ribosomal protection protein family and were first described in anaerobic bacteria, such as Clostridium perfringens [67].Therefore, it is proposed that tetA(P) and tetB(P) genes are acquired by the conjugative transfer into C. difficile from some other pathogenic bacteria.Non-toxigenic strains can act as a reservoir for many AMR genes that could be transferred horizontally to toxigenic strains, as well as to other zoonotic pathogenic bacteria.
Resistance to fluoroquinolones was mediated by the presence of chromosomal mutations in the QRDRs of the gyrA and gyrB genes.The presence of the mutations in gyrA and gyrB genes was highly associated with high-risk clones, such as ST11 and ST3, being the most prevalent in the current study.Interestingly, most of obtained amino acid substations patterns in QRDRs of gyrA and gyrB genes have been previously identified among fluoroquinolone-resistant C. difficile strains, belonging to different genotypes, such as RT001, RT018, RT176, and RT046 [68].
Obtained environmental isolates harbored an aminoglycoside-streptothricin resistance cassette (aph(3 )-IIIa-sat-4-ant(6)-la) and were assigned to ST11 (RT126 and RT078), which is similar to the cassette found in Erysipelothrix rhusiopathiae, a species commonly found in pig gut [65] and was also detected in Enterococcus faecium [69].The sat-4 gene was previously detected in Campylobacter coli and Enterococcus faecium [69,70] and the cassette of resistance genes is found in many bacterial species, indicating the possibility of interspecies transmission.In general, ST11 strains (RT126, RT127, and RT078) show a high proportion of antimicrobial resistance determinates.
For MLS B resistance, the ermB gene was identified in 13% of total isolates, which has been associated with CDI outbreaks in Europe [71].The ermB gene is mostly found in the conjugative and mobilizable transposons, Tn5398, Tn6194, Tn6218, and Tn6215 [3,4].
For vancomycin resistance, multiple van gene clusters were identified in obtained C. difficile isolates, which were analyzed in this study.However, a complete van resistance operon was not detected in these isolates.Several van gene clusters, including vanA, vanB, vanG, vanW, and vanZ1, have been identified in C. difficile and associated with high vancomycin minimum inhibitory concentrations (MICs) [72].The expression of these clusters is controlled by two-component regulatory systems, vanS (membrane sensor kinase) and vanR (cytoplasmic response regulator) [72,73], suggesting that these clusters were described to be phenotypically silent.Therefore, the presence of van resistance clusters in environmental C. difficile strains does not always result in their expression in vitro resistance to vancomycin.These strains could be considered susceptible to vancomycin.
For beta-lactam resistance, blaCDD-1 or blaCDD-2 genes were detected in all isolates analyzed, which confer resistance against various beta-lactam antibiotics.These enzymes previously identified in C. difficile strains allowing to have intrinsic resistance to antimicrobials, such as penicillins and cephalosporins [74], which is highly conserved among those C. difficile genomes.

Conclusions
This study demonstrated a large genetic overlap between RTs being isolated from environmental samples and humans that may represent a reservoir for CA-CDI.Although RT027 was absent, "hypervirulent" RT078 was found in digested sludge-amended soils, which could possess the ability for zoonotic transmission between humans and environmental sources.Furthermore, a broad variety of AMR genes were predominantly present in the ST11 sublineages.Although resistance to antimicrobials used to treat CDI is rare, this study provides evidence to support the role of AMR in the spread of C. difficile.Future studies need to address the question to which extent HGT, e.g., via MGEs (i.e., transposons, prophages, or plasmids), is present-and further triggered by antimicrobial selection pressure-e.g., for the development and emergence of new epidemic strains.

Figure 3 .
Figure 3. Distribution of C. difficile STs (A) and in diverse environmental samples (B).CF: calf feces, BP: biogas plant digestate, TDS/TDB: thermophilic digester for treating sewage sludge or biowaste, S: soil, ARC/E: anaerobic lab-scale bioreactors treating sewage sludge (control and experiment), DSS-S: digested sewage sludge-amended soils.Others indicate STs with fewer than three assigned strains or samples.

Figure 3 .
Figure 3. Distribution of C. difficile STs (A) and in diverse environmental samples (B).CF: calf feces, BP: biogas plant digestate, TDS/TDB: thermophilic digester for treating sewage sludge or biowaste, S: soil, ARC/E: anaerobic lab-scale bioreactors treating sewage sludge (control and experiment), DSS-S: digested sewage sludge-amended soils.Others indicate STs with fewer than three assigned strains or samples.

Figure 4 .
Figure 4. Minimum-spanning tree based on allelic profiles of 166 C. difficile isolates.Each circle represents a separate genotype, and distances between two genotypes are based on the allelic profiles of up to 2147 target genes, pairwise ignoring missing targets.The values on the connecting lines indicate the number of allelic differences between the connected isolates.Circle sizes are proportional to the numbers of isolates per genotype (i.e., the allelic profile).Related genotypes (≤6 alleles distance) are shaded in gray, and the isolates are colored according to their RT.RSS: raw sewage sludge, RS: raw sewage, ASS: activated sewage sludge, DSS: digested sewage sludge, CF: calf feces, BP: biogas plant digestate, ARC/E: anaerobic lab-scale bioreactors treating sewage sludge (control and experiment), DS: digested sewage sludge-amended soils, TDS: thermophilic digester for treating sewage sludge, TDB: thermophilic digester for treating biowaste, S: soil.

Figure 4 .
Figure 4. Minimum-spanning tree based on allelic profiles of 166 C. difficile isolates.Each circle represents a separate genotype, and distances between two genotypes are based on the allelic profiles of up to 2147 target genes, pairwise ignoring missing targets.The values on the connecting lines indicate the number of allelic differences between the connected isolates.Circle sizes are proportional to the numbers of isolates per genotype (i.e., the allelic profile).Related genotypes (≤6 alleles distance) are shaded in gray, and the isolates are colored according to their RT.RSS: raw sewage sludge, RS: raw sewage, ASS: activated sewage sludge, DSS: digested sewage sludge, CF: calf feces, BP: biogas plant digestate, ARC/E: anaerobic lab-scale bioreactors treating sewage sludge (control and experiment), DS: digested sewage sludge-amended soils, TDS: thermophilic digester for treating sewage sludge, TDB: thermophilic digester for treating biowaste, S: soil.

Table 1 .
The ribotypes (RTs) of C. difficile linked to STs and MLST clades.
(*) Human CA-CDI.STs correspond to more than two RTs marked with bold.