A publicly accessible database for Clostridioides difficile genome sequences supports tracing of transmission chains and epidemics

SaLTy: a novel Staphylococcus aureus Lineage Typer

Staphylococcus aureus asymptomatically colonises 30 % of humans but can also cause a range of diseases, which can be fatal. In 2017 S. aureus was associated with 20 000 deaths in the USA alone. Dividing S. aureus isolates into smaller sub-groups can reveal the emergence of distinct sub-populations with varying potential to cause infections. Despite multiple molecular typing methods categorising such sub-groups, they do not take full advantage of S. aureus genome sequences when describing the fundamental population structure of the species. In this study, we developed Staphylococcus aureus Lineage Typing (SaLTy), which rapidly divides the species into 61 phylogenetically congruent lineages. Alleles of three core genes were identified that uniquely define the 61 lineages and were used for SaLTy typing. SaLTy was validated on 5000 genomes and 99.12 % (4956/5000) of isolates were assigned the correct lineage. We compared SaLTy lineages to previously calculated clonal complexes (CCs) from BIGSdb (n=21 173). SALTy improves on CCs by grouping isolates congruently with phylogenetic structure. SaLTy lineages were further used to describe the carriage of Staphylococcal chromosomal cassette containing mecA (SCCmec) which is carried by methicillin-resistant S. aureus (MRSA). Most lineages had isolates lacking SCCmec and the four largest lineages varied in SCCmec over time. Classifying isolates into SaLTy lineages, which were further SCCmec typed, allowed SaLTy to describe high-level MRSA epidemiology. We provide SaLTy as a simple typing method that defines phylogenetic lineages (https://github.com/LanLab/SaLTy). SaLTy is highly accurate and can quickly analyse large amounts of S. aureus genome data. SaLTy will aid the characterisation of S. aureus populations and ongoing surveillance of sub-groups that threaten human health.

Comparative Genomics of Clostridioides difficile

Clostridioides difficile, a Gram-positive spore-forming anaerobic bacterium, has rapidly emerged as the leading cause of nosocomial diarrhoea in hospitals. The availability of large numbers of genome sequences, mainly due to the use of next-generation sequencing methods, has undoubtedly shown their immense advantages in the determination of C. difficile population structure. The implementation of fine-scale comparative genomic approaches has paved the way for global transmission and recurrence studies, as well as more targeted studies, such as the PaLoc or CRISPR/Cas systems. In this chapter, we provide an overview of recent and significant findings on C. difficile using comparative genomic studies with implications for epidemiology, infection control and understanding of the evolution of C. difficile.

Core genome multilocus sequence typing of Clostridioides difficile to investigate transmission in the hospital setting

PURPOSE

Traditional epidemiological investigations of healthcare-associated Clostridioides difficile infection (HA-CDI) are often insufficient. This study aimed to evaluate a procedure that includes secondary isolation and genomic typing of single toxigenic colonies using core genome multilocus sequence typing (cgMLST) for the investigation of C. difficile transmission.

METHODS

We analyzed retrospectively all toxigenic C. difficile-positive stool samples stored at the Lausanne University Hospital over 6 consecutive months. All isolates were initially typed and classified using a modified double-locus sequence typing (DLST) method. Genome comparison of isolates with the same DLST and clustering were subsequently performed using cgMLST. The electronic administrative records of patients with CDI were investigated for spatiotemporal epidemiological links supporting hospital transmission. A comparative descriptive analysis between genomic and epidemiological data was then performed.

RESULTS

From January to June 2021, 86 C. difficile isolates were recovered from thawed samples of 71 patients. Thirteen different DLST types were shared by > 1 patient, and 13 were observed in single patients. A genomic cluster was defined as a set of isolates from different patients with ≤ 3 locus differences, determined by cgMLST. Seven genomic clusters were identified, among which plausible epidemiological links were identified in only 4/7 clusters.

CONCLUSION

Among clusters determined by cgMLST analysis, roughly 40% included unexplained HA-CDI acquisitions, which may be explained by unidentified epidemiological links, asymptomatic colonization, and/or shared common community reservoirs. The use of DLST, followed by whole genome sequencing analysis, is a promising and cost-effective stepwise approach for the investigation of CDI transmission in the hospital setting.

A local-scale One Health genomic surveillance of Clostridioides difficile demonstrates highly related strains from humans, canines, and the environment

Although infections caused by Clostridioides difficile have historically been attributed to hospital acquisition, growing evidence supports the role of community acquisition in C. difficile infection (CDI). Symptoms of CDI can range from mild, self-resolving diarrhoea to toxic megacolon, pseudomembranous colitis, and death. In this study, we sampled C. difficile from clinical, environmental, and canine reservoirs in Flagstaff, Arizona, USA, to understand the distribution and transmission of the pathogen in a One Health framework; Flagstaff is a medium-sized, geographically isolated city with a single hospital system, making it an ideal site to characterize genomic overlap between sequenced C. difficile isolates across reservoirs. An analysis of 562 genomes from Flagstaff isolates identified 65 sequence types (STs), with eight STs being found across all three reservoirs and another nine found across two reservoirs. A screen of toxin genes in the pathogenicity locus identified nine STs where all isolates lost the toxin genes needed for CDI manifestation (tcdB, tcdA), demonstrating the widespread distribution of non-toxigenic C. difficile (NTCD) isolates in all three reservoirs; 15 NTCD genomes were sequenced from symptomatic, clinical samples, including two from mixed infections that contained both tcdB+ and tcdB- isolates. A comparative single nucleotide polymorphism (SNP) analysis of clinically derived isolates identified 78 genomes falling within clusters separated by ≤2 SNPs, indicating that ~19 % of clinical isolates are associated with potential healthcare-associated transmission clusters; only symptomatic cases were sampled in this study, and we did not sample asymptomatic transmission. Using this same SNP threshold, we identified genomic overlap between canine and soil isolates, as well as putative transmission between environmental and human reservoirs. The core genome of isolates sequenced in this study plus a representative set of public C. difficile genomes (n=136), was 2690 coding region sequences, which constitutes ~70 % of an individual C. difficile genome; this number is significantly higher than has been published in some other studies, suggesting that genome data quality is important in understanding the minimal number of genes needed by C. difficile. This study demonstrates the close genomic overlap among isolates sampled across reservoirs, which was facilitated by maximizing the genomic search space used for comprehensive identification of potential transmission events. Understanding the distribution of toxigenic and non-toxigenic C. difficile across reservoirs has implications for surveillance sampling strategies, characterizing routes of infections, and implementing mitigation measures to limit human infection.

Recombinant Fusion Protein Vaccine Containing Clostridioides difficile FliC and FliD Protects Mice against C. difficile Infection

Bacterial flagella are involved in infection through their roles in host cell adhesion, cell invasion, auto-agglutination, colonization, the formation of biofilms, and the regulation and secretion of nonflagellar bacterial proteins that are involved in the virulence process. In this study, we constructed a fusion protein vaccine (FliCD) containing the Clostridioides difficile flagellar proteins FliC and FliD. The immunization of mice with FliCD induced potent IgG and IgA antibody responses against FliCD, protected mice against C. difficile infection (CDI), and decreased the C. difficile spore and toxin levels in the feces after infection. Additionally, the anti-FliCD serum inhibited the binding of C. difficile vegetative cells to HCT8 cells. These results suggest that FliCD may represent an effective vaccine candidate against CDI.

Molecular epidemiology of Clostridioides difficile in companion animals: Genetic overlap with human strains and public health concerns

Introduction

The changing epidemiology of Clostridioides difficile reflects a well-established and intricate community transmission network. With rising numbers of reported community-acquired infections, recent studies tried to identify the role played by non-human reservoirs in the pathogen's transmission chain. This study aimed at describing the C. difficile strains circulating in canine and feline populations, and to evaluate their genetic overlap with human strains to assess the possibility of interspecies transmission.

Methods

Fecal samples from dogs (n = 335) and cats (n = 140) were collected from two populations (group A and group B) in Portugal. C. difficile isolates were characterized for toxigenic profile and PCR-ribotyping. The presence of genetic determinants of antimicrobial resistance was assessed in all phenotypically resistant isolates. To evaluate the genetic overlap between companion animals and human isolates from Portugal, RT106 (n = 42) and RT014/020 (n = 41) strains from both sources were subjected to whole genome sequencing and integrated with previously sequenced RT106 (n = 43) and RT014/020 (n = 142) genomes from different countries. The genetic overlap was assessed based on core-single nucleotide polymorphism (SNP) using a threshold of 2 SNP.

Results

The overall positivity rate for C. difficile was 26% (76/292) in group A and 18.6% (34/183) in group B. Toxigenic strains accounted for 50% (38/76) and 52.9% (18/34) of animal carriage rates, respectively. The most prevalent ribotypes (RT) were the toxigenic RT106 and RT014/020, and the non-toxigenic RT010 and RT009. Antimicrobial resistance was found for clindamycin (27.9%), metronidazole (17.1%) and moxifloxacin (12.4%), associated with the presence of the ermB gene, the pCD-METRO plasmid and point mutations in the gyrA gene, respectively. Both RT106 and RT014/020 genetic analysis revealed several clusters integrating isolates from animal and human sources, supporting the possibility of clonal interspecies transmission or a shared environmental contamination source.

Discussion

This study shows that companion animals may constitute a source of infection of toxigenic and antimicrobial resistant human associated C. difficile isolates. Additionally, it contributes with important data on the genetic proximity between C. difficile isolates from both sources, adding new information to guide future work on the role of animal reservoirs in the establishment of community associated transmission networks and alerting for potential public health risk.

Dogs as carriers of virulent and resistant genotypes of Clostridioides difficile

While previous research on zoonotic transmission of community-acquired Clostridioides difficile infection (CA-CDI) focused on food-producing animals, the present study aimed to investigate whether dogs are carriers of resistant and/or virulent C. difficile strains. Rectal swabs were collected from 323 dogs and 38 C. difficile isolates (11.8%) were obtained. Isolates were characterized by antimicrobial susceptibility testing, whole-genome sequencing (WGS) and a DNA hybridization assay. Multilocus sequence typing (MLST), core genome MLST (cgMLST) and screening for virulence and antimicrobial resistance genes were performed based on WGS. Minimum inhibitory concentrations for erythromycin, clindamycin, tetracycline, vancomycin and metronidazole were determined by E-test. Out of 38 C. difficile isolates, 28 (73.7%) carried genes for toxins. The majority of isolates belonged to MLST sequence types (STs) of clade I and one to clade V. Several isolates belonged to STs previously associated with human CA-CDI. However, cgMLST showed low genetic relatedness between the isolates of this study and C. difficile strains isolated from humans in Austria for which genome sequences were publicly available. Four isolates (10.5%) displayed resistance to three of the tested antimicrobial agents. Isolates exhibited resistance to erythromycin, clindamycin, tetracycline and metronidazole. These phenotypic resistances were supported by the presence of the resistance genes erm(B), cfr(C) and tet(M). All isolates were susceptible to vancomycin. Our results indicate that dogs may carry virulent and antimicrobial-resistant C. difficile strains.

K-mer based prediction of Clostridioides difficile relatedness and ribotypes

Comparative analysis of Clostridioides difficile whole-genome sequencing (WGS) data enables fine scaled investigation of transmission and is increasingly becoming part of routine surveillance. However, these analyses are constrained by the computational requirements of the large volumes of data involved. By decomposing WGS reads or assemblies into k-mers and using the dimensionality reduction technique MinHash, it is possible to rapidly approximate genomic distances without alignment. Here we assessed the performance of MinHash, as implemented by sourmash, in predicting single nucleotide differences between genomes (SNPs) and C. difficile ribotypes (RTs). For a set of 1905 diverse C. difficile genomes (differing by 0–168 519 SNPs), using sourmash to screen for closely related genomes, at a sensitivity of 100 % for pairs ≤10 SNPs, sourmash reduced the number of pairs from 1 813 560 overall to 161 934, i.e. by 91 %, with a positive predictive value of 32 % to correctly identify pairs ≤10 SNPs (maximum SNP distance 4144). At a sensitivity of 95 %, pairs were reduced by 94 % to 108 266 and PPV increased to 45 % (maximum SNP distance 1009). Increasing the MinHash sketch size above 2000 produced minimal performance improvement. We also explored a MinHash similarity-based ribotype prediction method. Genomes with known ribotypes (n=3937) were split into a training set (2937) and test set (1000) randomly. The training set was used to construct a sourmash index against which genomes from the test set were compared. If the closest five genomes in the index had the same ribotype this was taken to predict the searched genome’s ribotype. Using our MinHash ribotype index, predicted ribotypes were correct in 780/1000 (78 %) genomes, incorrect in 20 (2 %), and indeterminant in 200 (20 %). Relaxing the classifier to 4/5 closest matches with the same RT improved the correct predictions to 87 %. Using MinHash it is possible to subsample C. difficile genome k-mer hashes and use them to approximate small genomic differences within minutes, significantly reducing the search space for further analysis.

Haem is crucial for medium-dependent metronidazole resistance in clinical isolates of Clostridioides difficile

Background

Until recently, metronidazole was the first-line treatment for Clostridioides difficile infection and it is still commonly used. Though resistance has been reported due to the plasmid pCD-METRO, this does not explain all cases.

Objectives

To identify factors that contribute to plasmid-independent metronidazole resistance of C. difficile.

Methods

Here, we investigate resistance to metronidazole in a collection of clinical isolates of C. difficile using a combination of antimicrobial susceptibility testing on different solid agar media and WGS of selected isolates.

Results

We find that nearly all isolates demonstrate a haem-dependent increase in the MIC of metronidazole, which in some cases leads to isolates qualifying as resistant (MIC >2 mg/L). Moreover, we find an SNP in the haem-responsive gene hsmA, which defines a metronidazole-resistant lineage of PCR ribotype 010/MLST ST15 isolates that also includes pCD-METRO-containing strains.

Conclusions

Our data demonstrate that haem is crucial for medium-dependent metronidazole resistance in C. difficile.

Transition From PCR-Ribotyping to Whole Genome Sequencing Based Typing of Clostridioides difficile

Clostridioides difficile causes nosocomial outbreaks which can lead to severe and even life-threatening colitis. Rapid molecular diagnostic tests allow the identification of toxin-producing, potentially hypervirulent strains, which is critical for patient management and infection control. PCR-ribotyping has been used for decades as the reference standard to investigate transmission in suspected outbreaks. However, the introduction of whole genome sequencing (WGS) for molecular epidemiology provides a realistic alternative to PCR-ribotyping. In this transition phase it is crucial to understand the strengths and weaknesses of the two technologies, and to assess their correlation. We aimed to investigate ribotype prediction from WGS data, and options for analysis at different levels of analytical granularity. Ribotypes cannot be directly determined from short read Illumina sequence data as the rRNA operons including the ribotype-defining ISR fragments collapse in genome assemblies, and comparison with traditional PCR-ribotyping results becomes impossible. Ribotype extraction from long read Oxford nanopore data also requires optimization. We have compared WGS-based typing with PCR-ribotyping in nearly 300 clinical and environmental isolates from Switzerland, and in addition from the Enterobase database (n=1778). Our results show that while multi-locus sequence type (MLST) often correlates with a specific ribotype, the agreement is not complete, and for some ribotypes the resolution is insufficient. Using core genome MLST (cgMLST) analysis, there is an improved resolution and ribotypes can often be predicted within clusters, using cutoffs of 30-50 allele differences. The exceptions are ribotypes within known ribotype complexes such as RT078/RT106, where the genome differences in cgMLST do not reflect the ribotype segregation. We show that different ribotype clusters display different degrees of diversity, which could be important for the definition of ribotype cluster specific cutoffs. WGS-based analysis offers the ultimate resolution to the SNP level, enabling exploration of patient-to-patient transmission. PCR-ribotyping does not sufficiently discriminate to prove nosocomial transmission with certainty. We discuss the associated challenges and opportunities in a switch to WGS from conventional ribotyping for C. difficile.

Agricultural fertilization with poultry manure results in persistent environmental contamination with the pathogen Clostridioides difficile

During a field experiment applying broiler manure for fertilization of agricultural land, we detected viable Clostridioides (also known as Clostridium) difficile in broiler faeces, manure, dust and fertilized soil. A large diversity of toxigenic C. difficile isolates was recovered, including PCR ribotypes common from human disease. Genomic relatedness of C. difficile isolates from dust and from soil, recovered more than 2 years after fertilization, traced their origins to the specific chicken farm that had delivered the manure. We present evidence of long-term contamination of agricultural soil with manure-derived C. difficile and demonstrate the potential for airborne dispersal of C. difficile through dust emissions during manure application. Clostridioides genome sequences virtually identical to those from manure had been recovered from chicken meat and from human infections in previous studies, suggesting broiler-associated C. difficile are capable of zoonotic transmission.

Molecular Epidemiology and Antimicrobial Resistance of Clostridioides difficile in Hospitalized Patients From Mexico

Clostridioides difficile is a global public health problem, which is a primary cause of antibiotic-associated diarrhea in humans. The emergence of hypervirulent and antibiotic-resistant strains is associated with the increased incidence and severity of the disease. There are limited studies on genomic characterization of C. difficile in Latin America. We aimed to learn about the molecular epidemiology and antimicrobial resistance in C. difficile strains from adults and children in hospitals of México. We studied 94 C. difficile isolates from seven hospitals in Mexico City from 2014 to 2018. Whole-genome sequencing (WGS) was used to determine the genotype and examine the toxigenic profiles. Susceptibility to antibiotics was determined by E-test. Multilocus sequence typing (MLST) was used to determine allelic profiles. Results identified 20 different sequence types (ST) in the 94 isolates, mostly clade 2 and clade 1. ST1 was predominant in isolates from adult and children. Toxigenic strains comprised 87.2% of the isolates that were combinations of tcdAB and cdtAB (tcdA+/tcdB+/cdtA+/cdtB+, followed by tcdA+/tcdB+/cdtA-/cdtB-, tcdA-/tcdB+/cdtA-/ cdtB-, and tcdA-/tcdB-/cdtA+/cdtB+). Toxin profiles were more diverse in isolates from children. All 94 isolates were susceptible to metronidazole and vancomycin, whereas a considerable number of isolates were resistant to clindamycin, fluroquinolones, rifampicin, meropenem, and linezolid. Multidrug-resistant isolates (≥3 antibiotics) comprised 65% of the isolates. The correlation between resistant genotypes and phenotypes was evaluated by the kappa test. Mutations in rpoB and rpoC showed moderate concordance with resistance to rifampicin and mutations in fusA substantial concordance with fusidic acid resistance. cfrE, a gene recently described in one Mexican isolate, was present in 65% of strains linezolid resistant, all ST1 organisms. WGS is a powerful tool to genotype and characterize virulence and antibiotic susceptibility patterns.

Prevalence and phylogenetic relationship of Clostridioides difficile strains in fresh poultry meat samples processed in different cutting plants

Clostridioides difficile is one of the most frequent causes of nosocomial infections in humans leading to (antibiotic-associated) diarrhea and severe pseudomembranous colitis. With an increasing frequency, C. difficile infections (CDI) are also observed independently of hospitalization and the age of the patients in an ambulant setting. One potential source of so-called community-acquired CDI is a zoonotic transmission to humans based on direct contact with animals or the consumption of food. To estimate the exposure of humans with C. difficile via food, we screened 364 different retail fresh poultry meat products purchased in Berlin and Brandenburg, Germany and further characterized the isolates. None of the 42 turkey or chicken meat samples without skin was contaminated. However, 51 (15.8%) of 322 tested fresh chicken meat samples with skin were C. difficile-positive. The vast majority (84.3%) of all isolates exhibited toxin genes tcdA and tcdB, whereas the binary toxin cdtA/B was absent. Most of the isolates (50/51) were susceptible to all six investigated antimicrobials. However, one non-toxigenic strain was multidrug resistant to the antimicrobials clindamycin and erythromycin. The isolates were mainly represented by PCR-ribotypes (RT) 001, RT002, RT005, and RT014, which were already associated with human CDI cases in Germany and were partially detected in poultry. The relatively high contamination rate of fresh retail chicken meat with skin purchased in Germany indicates chicken meat as a potential source of human infections. Moreover, we identified cutting plants with a higher rate of a C. difficile-contamination (21.4-32.8%). To compare the phylogenetic relationship of the isolated strains from certain cutting plants over several months in 2018 and 2019, we analyzed them using NGS followed by core genome MLST. Interestingly, highly related strains (0-3 alleles distance) of common clinical RT001 and RT002 isolates, as well as of the non-toxigenic RT205 isolates were detectable in same cutting plants over a period of three and 16 months, respectively.The continuous contamination with the same strain could be explained by the longterm persistence of this strain within the cutting plant (e.g., within the scalder), or with a recurring entry e.g. from the same fattening farm.

Retrospective Definition of Clostridioides difficile PCR Ribotypes on the Basis of Whole Genome Polymorphisms: A Proof of Principle Study

Clostridioides difficile is a cause of health care-associated infections. The epidemiological study of C. difficile infection (CDI) traditionally involves PCR ribotyping. However, ribotyping will be increasingly replaced by whole genome sequencing (WGS). This implies that WGS types need correlation with classical ribotypes (RTs) in order to perform retrospective clinical studies. Here, we selected genomes of hyper-virulent C. difficile strains of RT001, RT017, RT027, RT078, and RT106 to try and identify new discriminatory markers using in silico ribotyping PCR and De Bruijn graph-based Genome Wide Association Studies (DBGWAS). First, in silico ribotyping PCR was performed using reference primer sequences and 30 C. difficile genomes of the five different RTs identified above. Second, discriminatory genomic markers were sought with DBGWAS using a set of 160 independent C. difficile genomes (14 ribotypes). RT-specific genetic polymorphisms were annotated and validated for their specificity and sensitivity against a larger dataset of 2425 C. difficile genomes covering 132 different RTs. In silico PCR ribotyping was unsuccessful due to non-specific or missing theoretical RT PCR fragments. More successfully, DBGWAS discovered a total of 47 new markers (13 in RT017, 12 in RT078, 9 in RT106, 7 in RT027, and 6 in RT001) with minimum q-values of 0 to 7.40 × 10^-5, indicating excellent marker selectivity. The specificity and sensitivity of individual markers ranged between 0.92 and 1.0 but increased to 1 by combining two markers, hence providing undisputed RT identification based on a single genome sequence. Markers were scattered throughout the C. difficile genome in intra- and intergenic regions. We propose here a set of new genomic polymorphisms that efficiently identify five hyper-virulent RTs utilizing WGS data only. Further studies need to show whether this initial proof-of-principle observation can be extended to all 600 existing RTs.

Accurate reconstruction of bacterial pan- and core genomes with PEPPAN

Bacterial genomes can contain traces of a complex evolutionary history, including extensive homologous recombination, gene loss, gene duplications, and horizontal gene transfer. To reconstruct the phylogenetic and population history of a set of multiple bacteria, it is necessary to examine their pangenome, the composite of all the genes in the set. Here we introduce PEPPAN, a novel pipeline that can reliably construct pangenomes from thousands of genetically diverse bacterial genomes that represent the diversity of an entire genus. PEPPAN outperforms existing pangenome methods by providing consistent gene and pseudogene annotations extended by similarity-based gene predictions, and identifying and excluding paralogs by combining tree- and synteny-based approaches. The PEPPAN package additionally includes PEPPAN_parser, which implements additional downstream analyses, including the calculation of trees based on accessory gene content or allelic differences between core genes. To test the accuracy of PEPPAN, we implemented SimPan, a novel pipeline for simulating the evolution of bacterial pangenomes. We compared the accuracy and speed of PEPPAN with four state-of-the-art pangenome pipelines using both empirical and simulated data sets. PEPPAN was more accurate and more specific than any of the other pipelines and was almost as fast as any of them. As a case study, we used PEPPAN to construct a pangenome of approximately 40,000 genes from 3052 representative genomes spanning at least 80 species of Streptococcus The resulting gene and allelic trees provide an unprecedented overview of the genomic diversity of the entire Streptococcus genus.