From colonization to infection: Genomic evolution of Clostridioides difficile pathogenesis

SUMMARY

Clostridioides difficile is a spore-forming, toxin-producing anaerobe that is a leading cause of healthcare-associated infections. Its success as a pathogen reflects a complex interplay between bacterial evolution, virulence regulation, ecological adaptation, environmental selection, and host susceptibility. Comparative genomics has revealed deep C. difficile lineage diversification, driven by mobile genetic elements and selective pressures from antibiotics and host environments. These events affect strain-specific virulence by shaping the organization and regulation of the pathogenicity toxin loci, metabolic adaptations for nutrient utilization, and enhanced spore resilience. This review integrates evolutionary and genomic perspectives to illustrate how adaptive diversification has sculpted C. difficile pathogenesis and epidemic success.

INTRODUCTION

C. difficile is a gram-positive, spore-forming, anaerobic bacterium and the leading cause of healthcare-associated gastrointestinal infection and antibiotic-associated diarrhea worldwide.^1,2 Classified as an urgent threat antibiotic-resistant pathogen by the US Centers for Disease Control and Prevention since 2013 (Figure 1), C. difficile infection (CDI) currently imposes a significant clinical and economic burden, affecting approximately 500,000 individuals and costing $5.4 billion annually in the US alone.^3,4 Major risk factors for CDI include age, hospitalization, weakened immune system (e.g., by chemotherapy or immunosuppressants), and, most predominantly, prior antibiotic exposure.⁵ In healthy individuals, C. difficile spore germination is suppressed by the protective antagonism of the commensal gut microbiota.⁶ Antibiotic-induced disruption of the gut microbiota and metabolism promotes spores to germinate into toxin-producing vegetative cells, leading to intestinal epithelial damage and inflammation.^7,8 This presents a major therapeutic challenge, as conventional antibiotics could further disrupt the microbiota but failed to eliminate the spores, which persist and reinitiate infection. This feedback loop results in recurrent CDI (rCDI), affecting approximately 20%–30% of patients following the initial treatment episode.⁹

Figure 1. Milestones in the evolutionary history, epidemic events, and genomic study of C. difficile.

mya, million years ago; PMC, pseudomembranous colitis; NGS, next-generation sequencing; C. diff, C. difficile. Created using Biorender.com.

The clinical spectrum of C. difficile ranges from asymptomatic carriage to mild diarrhea, severe colitis with pseudomembrane formation, and in some cases fatal colitis.^1,10 One critical determinant of C. difficile clinical outcomes is the virulence of the colonizing strain.^11,12 Like other bacterial pathogens, C. difficile continuously evolves and adapts under selective pressures.^13,14 Through evolutionary mechanisms such as horizontal gene transfer (HGT), homologous recombination, and mutation accumulation, C. difficile acquires genetic diversity that underlies strain-specific differences in virulence, colonization, and, eventually, epidemiological success.^11,15,16 Additionally, diversification of virulence genes can cause sequencing-based diagnostic failure, such as failing to detect target genes using PCR typing.^17,18 Therefore, it is critical to identify genomic features that correlate with CDI risk and understand the evolutionary mechanisms that shaped these features to better understand, predict, and diagnose CDI. In addition to the major toxins, other genomic features such as sporulation-related genes, cell-surface genes, and mobile genetic elements (MGEs) also contribute to C. difficile colonization and pathogenesis by enhancing sporulation, biofilm formation, antibiotic resistance, and toxin production.^19–22 In this review, we synthesize the current knowledge on (1) diversity and typing methods of C. difficile lineages, (2) evolution of C. difficile virulence and putative virulence genes and their resulting effects on pathogenesis, and (3) epidemiology of representative C. difficile virulent strains.

Diversity and typing methods in C. difficile

Since the first complete C. difficile genome (reference strain 630) was sequenced in 2006 using shotgun Sanger sequencing (Figure 1),¹⁴ advances in high-throughput sequencing, assembly, and gene annotation have generated over 20,000 C. difficile genomes, greatly improving understanding of C. difficile genome diversity and evolution.^11,23 C. difficile has a large, diverse genome comprising a 4.1–4.3 Mb circular chromosome with ~3,600 protein-coding sequences.^14,24 The C. difficile core genome, which is the set of genes present in all C. difficile strains, only accounts for less than 20% of the total genome on average and primarily encodes essential cellular functions including DNA replication machinery and central energy metabolism.^25,26 However, genomic diversity appears considerably more constrained in clinically dominant lineages.^25,27 The C. difficile accessory genome includes a wide array of MGEs, such as conjugative transposons (CTns), prophages, and plasmids,^14,23 which can encode or facilitate the transfer of virulence, colonization-related, and antibiotic-resistance genes (ARGs). This combination of a small core genome with a large accessory genome pool supports C. difficile to adapt to different selective pressures and facilitates its success as both a persistent gut colonizer and an opportunistic pathogen.

A variety of molecular typing methods that target either the entire genome, several loci, or a single region have been applied to estimate and investigate C. difficile genetic diversity and to trace outbreaks.^28,29 Early approaches included pulsed-field gel electrophoresis, which resolves large chromosomal fragments to compare strain-specific restriction patterns, and toxinotyping, which differentiates isolates based on sequence variation in the pathogenicity locus (PaLoc). The current standard C. difficile typing approaches are multilocus sequence typing (MLST) and PCR ribotyping,^30–32 combined with whole-genome sequencing (WGS). In fact, reclassification of C. difficile from the Clostridium to the Clostridioides genus is based on the advanced resolution from PCR ribotyping and WGS in 2015 (Figure 1).³³ Based on allelic profile variation across seven housekeeping genes, MLST has identified over 600 sequence types (STs) and breaks C. difficile into five major clades (C1 to C5) and at least three cryptic clades (C-I to C-III).^13,25 By contrast, PCR ribotyping distinguishes strains by amplifying the 16S–23S rRNA intergenic spacer region, from which length polymorphisms yield over 800 defined ribotypes (RTs).³⁴ The exact numbers of defined RTs and STs have varied with the adaptation of new sequencing technologies and gene annotation,^13,29 while both numbers are expected to increase as WGS methods advance and new C. difficile lineages are identified.

Currently, the most prevalent clade in both hospital- and community-associated CDI is C1. With over 100 STs, C1 is also the most genetically diverse clade (Table 1). C2 contains the epidemic hypervirulent strain RT027 which has caused multiple disease outbreaks in North America and Europe,^46,47 though recent evidence indicates it is no longer epidemiologically predominant.⁴⁸ The C3 and C4 are less clinically relevant than C1 and C2 but both contain pathogenic strains (e.g., RT023 and RT017) with unique PaLoc configurations.^39,42 C5 isolates are more prevalent in agriculture samples and contribute to community-associated CDI.^23,43 Though containing some pathogenic isolates, cryptic clades are clinically rare and are the most divergent lineages.²⁷ Bayesian molecular dating estimates the most recent common ancestor of C1–5 existed 1.1–6.7 million years ago, with the cryptic clades diverging several million years earlier (Figure 1).²⁵ Whole-genome average nucleotide identity (ANI) analysis also shows that cryptic clades fall below the species demarcation threshold of 95%, indicating they could be distinct Clostridioides species (Table 1).²⁵

Table 1.

Genomic, pathogenic, and epidemiologic features of major C. difficile clades and cryptic clades

Clade	Genomic/phenotypic features	Pathogenicity/epidemiology	Representative STs	Representative RTs	% ANI	16S rRNA % similarity	References
C1	most genetically diverse; contains over 200 STs; most PaLoc has typical TcdA+TcdB+ configuration	globally dominant lineage; broad host range; moderate virulence; includes RT106, a rising epidemic strain in the US	ST2, ST3, ST6, ST8, ST14, ST35, ST42, ST44, ST54	RT001, RT002, RT012, RT014, RT046, RT106	100	100	Knight et al.25; Roxas et al.35; Liao et al.36
C2	carries fluoroquinolone resistance mutations (gyrA, gyrB); most PaLoc has typical TcdA+TcdB+ configuration; often CDT+; high toxin production and sporulation rate	includes historically hypervirulent strain RT027; causes major 2000s hospital outbreaks in NA/EU but now declining	ST1, ST41, ST67, ST48	RT027, RT036, RT244	98.7	99.9	Knight et al.25; Midani et al.37; Lim et al.38
C3	PaLoc contains Tn6218 transposon; often TcdA+TcdB+ and CDT+; unique surface structure for niche adaptation	emerging hypervirulent lineage; limited spread but associated with severe CDI and recurrence	ST5, ST22, ST96, ST285	RT023, RT058	97.9	100	Knight et al.25; Shaw et al.39; Chen et al.40; Shaw et al.41
C4	contains TcdA−TcdB+ PaLoc; usually CDT−; resistant to clindamycin and fluoroquinolones	causes major HA-CDI outbreaks in Asia; moderate virulence	ST37, ST39	RT017, RT369	97.8	100	Knight et al.25; Imwattana et al.42
C5	contains TcdA−TcdB+ PaLoc; often CDT+; carries multiple tetracycline resistance genes	common in livestock (pigs, cattle) and could be zoonotic originated; major cause of emerging CA-CDI	ST11	RT033, RT078, RT126	96.1	99.7	Knight et al.25; Elliott et al.43; Alves et al.44
C-I	contains divergent PaLoc (TcdA+TcdB−) and distinct GC content	rare and has limited human cases; contains potential environmental lineage	ST205	RT151, RT177, RT178, RT179, RT181	91.1	99.9	Ramírez-Vargas et al.17; Monot et al.18; Knight et al.25
C-II	distinct metabolic profile; contains TcdA−TcdB+ PaLoc	rare and has limited human cases; potentially environmental or animal origin	ST205	RT151	93.5	99.8	Knight et al.25; Ducarmon et al.45
C-III	lacks typical PaLoc; large genomic divergence from rest of the clades	rare and no known human infections; likely ancestral environmental lineage	–	–	89.2	99.7	Knight et al.25

Both ANI and 16S rRNA gene % similarity are relative to C1. ST, sequence type; RT, ribotype; ANI, average nucleotide identity.

C. difficile virulence and putative virulence genes evolution

While the pathogenesis of C. difficile involves a complex interplay that requires host immune responses and commensal gut microbiota, these mechanisms have been comprehensively reviewed elsewhere.^49,50 Here, we focus on how strain-specific virulence genes (toxin loci) and putative virulence genes (sporulation genes, cell wall genes, and MGEs) evolve and impact pathogenesis and colonization.

Toxin loci

PaLoc

The virulence of C. difficile is primarily determined by the expression of two major exotoxins, TcdA and TcdB, encoded by the tcdA and tcdB genes, respectively.¹⁹ These two toxin genes are colocalized with three other regulatory genes (tcdR, tcdC, and tcdE) within a ~19 kb PaLoc locus on the chromosome (Figure 2A).⁵¹ Toxigenic C. difficile strains are defined by the presence of the PaLoc locus, whereas in non-toxigenic strains this locus is replaced by a highly conserved 115-bp non-coding sequence.^19,52 Expression of TcdA and TcdB is regulated by the sigma factor, TcdR, encoded by tcdR (Figure 2A).⁵³ TcdR autoregulates its own expression and is modulated by changes in temperature and nutrient availability. tcdC encodes an anti-sigma factor, TcdC, which antagonizes TcdR-mediated toxin transcription.⁵⁴ tcdE encodes a holin-like protein TcdE that mediates non-lytic toxin secretion, while lytic toxin release can also be triggered by physiological cues.^55,56

Figure 2. The evolutionary mechanisms that shape the diversity of PaLoc and TcdB.

(A) The PaLoc encodes TcdA, TcdB, the positive regulator TcdR, the negative regulator TcdC, and holin TcdE. TcdA/B has glucosyltransferase (GTD), cysteine protease (CPD), delivery and receptor binding domain (DRBD), and C-terminal combined repetitive oligopeptide sequences (CROPS) domain.

(B) Major mechanisms of TcdB evolution include HGT, homologous recombination, and mutation accumulation. Examples of TcdB and other putative virulence genes’ evolution and outcome are included. Created using Biorender.com.

TcdA (~308 kDa) and TcdB (~270 kDa) are large single chain, multi-domain glycosylating toxins with four conserved functional domains: (1) an N-terminal glucosyltransferase domain (GTD) that glucosylates and inactivates host Rho GTPases; (2) an autoprotease domain (APD), which is responsible for the autoproteolytic cleavage of the toxin and release of GTD; (3) a delivery and receptor binding domain (DRBD) that enables receptor binding, pore formation, and cargo translocation; and (4) a C-terminal combined repetitive oligopeptide sequences (CROPS) domain that mediates glycan recognition^57,58 (Figure 2A). Recent structural studies have refined our understanding of TcdB and its host-receptor binding: frizzled (FZD) receptors bind the DRBD domain,⁵⁹ LRP-1 binds CROPS,⁶⁰ and Nectin-3 binding is independent of CROPS,⁶¹ while chondroitin sulfate proteoglycan 4 (CSPG4) forms a multidomain interface spanning APD, DRBD, and the CROPS hinge region.^62,63 TcdA primarily binds to glycan structures via its CROPS domain.⁶⁴ Sulfated glycosaminoglycans (sGAGs) and low-density lipoprotein receptors (LDLR) have been implicated as additional receptor targets for TcdA,⁶⁵ while glycoprotein 96 (gp96) has also been proposed, but further studies are needed to confirm their physiological relevance.⁷ After binding to their cognate receptors, TcdB undergoes clathrin mediated endocytosis, while TcdA utilizes protein kinase C and PACSIN2-dependent mechanisms for endocytosis.⁶⁶ Endosomal acidification then triggers a conformational change in the toxins, allowing GTD translocation into the cytosol and inositol hexakisphosphate-dependent autoproteolytic activation.⁶⁷ Once delivered, GTD catalyzes the transfer of a glucose moiety from UDP-glucose to Rho-family GTPases (Cdc42, Rho, and Rac1).⁶⁸ While both TcdA and TcdB target multiple Rho GTPases, differences in substrate preference and efficiency have been reported.^69,70

Inactivation of the Rho-family GTPases disrupts actin cytoskeleton regulation, causing G-actin accumulation, F-actin depolymerization, cell rounding, and disassembly of focal adhesions and tight junctions.⁷ Beyond cytoskeleton disruption, TcdA and TcdB trigger cell death through multiple mechanisms depending on cell type and toxin concentration. In epithelial cells, TcdA triggers mitochondrial apoptotic pathways via activation of caspases and mitochondrial damage.^71,72 TcdB shows concentration-dependent cytotoxicity of epithelial cell death: at low toxin concentrations, it induces apoptosis; while at high toxin concentrations, it causes pyknosis, a necrotic cell death phenotype.⁷³ In addition to the cytotoxic effects, both toxins elicit robust inflammatory responses by disrupting epithelial barrier integrity and stimulating immune cells to secrete pro-inflammatory cytokines (e.g., IL-1β, IL-8, and TNF-α), further driving recruitment of neutrophils, monocytes, and macrophages into the gut mucosa.⁷⁴ This contributes to pseudomembrane formation, tissue damage, and diarrhea characteristic of CDI.⁷⁵

The relative contribution of TcdA and TcdB to CDI remains context-dependent and debated. Using isogenic mutants in a virulent C. difficile strain background, Lyras et al. demonstrated that TcdB is essential for virulence in an infection model, while TcdA is dispensable for virulence.⁷⁶ On the contrary, Kuehne et al. showed that C. difficile producing either toxin TcdA or TcdB caused cytotoxicity in vitro and led to disease in a hamster model of infection.¹⁹ Another study examining GTD-disrupted mutants indicated that TcdA retains the ability to contribute to pathology in the presence of TcdB with disrupted GTD.⁷⁷ Clinical isolates producing high levels of TcdA further underscore its pathogenic potential. Overall, TcdB appears more potent in cultured cells,⁷⁰ while both toxins exhibit comparable effects in vivo suggesting that observed differences reflect receptor tropism and substrate specificity in particular cell types rather than intrinsic toxin potency.

Comparative genomics has revealed that PaLoc is highly dynamic, shaped by HGT and intralocus recombination. In particular, tcdB has mosaic regions across its entire coding sequence that were likely generated during ancestral intragenic recombination and thus displaying greater toxin subtype diversity (B1–12) compared with tcdA (A1–7), which mainly varies in the number and length of repeats in its C-terminal CROPS domain (Figure 2B).¹⁶ However, only a subset of these toxin variants, found in major clonal lineages, are associated with CDI.¹⁶ The two dominant clinical subtypes are TcdB1, encoded by lineages like VPI10463 and reference strain 630, and TcdB2, carried by RT027.⁷⁸ These subtypes differ in receptor tropism; TcdB1 uses the FZD receptor on colonocytes, while TcdB2 primarily utilizes TFP1 expressed on colonic crypt cells, with CSPG4 providing additional binding, reflecting potential adaptation to interact with distinct host cell populations.⁷⁹

While the precise evolutionary origin and trajectory of PaLoc remains debated, evidence suggests it was acquired through clade-specific long-homologous recombination events after clade divergence.^43,80 For example, C3 PaLoc is uniquely associated with the transposon Tn6218, which was gained after clade divergence.^40,80 C5 has two separate PaLoc acquisition events for TcdA⁺TcdB⁺ and TcdA⁻ TcdB⁺ PaLoc variants, respectively (Table 1). Additionally, the prophage hypothesis suggests that PaLoc recombination was facilitated by an ancient phage integration event, which is supported by the presence of the phage-like holin gene tcdE within the locus.^18,81 This hypothesis also proposes an evolutionary trajectory from two “mono-toxin PaLoc” to one canonical “bi-toxin PaLoc,” consistent with sequence similarities between intergenic regions of the toxin genes.¹⁸

TcdA and TcdB share dynamic evolutionary history with other large clostridia toxins (LCT), found across species like Paraclostridium sordelli (TcsL and TcsH), Clostridium novyi (TcnA), and Clostridium perfringens (TpeL).⁸² While Clostridium sordelli possesses a dual-toxin system like C. difficile, Clostridium novyi and Clostridium perfringens encode single LCT homologs, suggesting lineage-specific gain or loss of toxin genes. At the protein level, the glucosyltransferase activity is highly conserved among LCTs indicating that core enzymatic pathogenic mechanisms are ancient and preserved. In contrast, receptor-binding regions and CROPS are variable in homologs, reflecting fine-tuning to different host receptors, tissues, and infection strategies. Notably, Schorch et al. were the first to demonstrate that TpeL had a CROPS independent receptor-binding mechanism, challenging the assumption that CROPS are mandatory for host cell engagement.⁸³ Recent genomic analysis has revealed the presence of an LCT-like translocase outside classical clostridia lineages, suggesting an ancient origin with potentially diverse ecological roles.⁸⁴ Receptor specificity is therefore a key evolutionary driver as highlighted by TcsL and TcdB, where a few mutations in the receptor-binding region are sufficient to switch binding specificity of TcsL to TcdB, providing a molecular basis for differential tissue tropism and disease progression (Figure 2B).⁸⁵ LCT diversity is further shaped by HGT. The PaLoc locus has been shown to transfer between C. difficile strains under lab conditions by a conjugation-like mechanism,¹⁵ converting non-toxigenic strains into toxigenic variants, however the transfer appears to be highly strain dependent.⁸⁶ Similarly, toxin genes in Paraclostridium sordelli are carried on pCS1 family plasmids, and transfer of these plasmids between strains has been found to be mediated by a conserved clostridial conjugation locus.⁸⁷ Together, these observations support a model in which toxin evolution is driven by modular diversification, receptor adaptation, and tropism, with HGT and recombination potentially contributing to the emergence of related toxin variants.

CdtLoc

In addition to the PaLoc virulence locus, C. difficile can also harbor a 6.2 kb binary toxin locus, or the C. difficile transferase locus (CdtLoc), which carries the binary toxin genes cdtA and cdtB, along with their positive regulator gene cdtR.^88,89 CDT is comprised of two components: CDTa (~53 kDa), the enzymatic component and CDTb (~99 kDa), responsible for receptor binding and translocation.⁹⁰ CDTa undergoes proteolytic cleavage to a 48 kDa active form, with its C terminal domain carrying ADP-ribosyltransferase activity. CDTb contains four functional domains mediating receptor binding, membrane insertion, pore formation, and oligomerization.⁹¹ It binds the lipolysis-stimulated lipoprotein receptor (LSR), triggering endocytosis.⁹² Host serine proteases activate CDTb, enabling heptamer formation and under acidic endosomal conditions, β-barrel pore formation through which CDTa is translocated to cytosol^91,93,94

In the cytosol, chaperones HSP90 and cyclophilin A facilitate CDTa refolding, which then ADP-ribosylates G-actin at Arg-177, blocking polymerization and causing cytoskeletal collapse and tight junction disruption between intestinal epithelial cells.^95,96 Additionally, CDT leads to the formation of protrusion bodies composed of microtubules that increase C. difficile adherence to host cells.⁹⁷ CDT has been reported to activate TLR-2/NF-kB signaling pathway, leading to the production of pro-inflammatory cytokines and suppression of protective host eosinophil response. The TLR-2 signaling has also been proposed to prime inflammasome activation.^98,99 More recently, CDT has been shown to directly activate the NLRP3 inflammasome in vitro through its pore forming component, CDTb, via a K⁺ efflux mechanism.¹⁰⁰ However, this inflammasome activation was not recapitulated in murine model, suggesting that while CDT clearly enhances inflammatory responses, the precise mechanisms remain incompletely understood.

The binary toxin is produced by approximately 20% of C. difficile clinical isolates, including RT027 and RT078, and is considered an accessory virulence factor, although its precise role in CDI pathogenesis remains unresolved.⁹⁰ Many clinical isolates have been shown to carry truncated, non-functional forms of the CDT locus without virulence attenuation¹⁰¹ and TcdA⁻ TcdB⁻CDT⁺ strains demonstrate no typical symptoms of CDI in vivo.¹⁰² While a 69 bp deletion in the cdtR gene significantly reduced virulence in an ST1 strain,¹¹ CdtR has also been shown to influence regulation of TcdA and TcdB, making it difficult to disentangle CDT’s independent contribution to virulence. Though the evolutionary trajectory of CDT remains understudied, the discovery of complete and functional CDT loci in phiSemix9P1 phage suggests a possible phage origin, similar to the PaLoc phage acquisition theory.¹⁰³ CDT belongs to a broader family of binary toxins distributed across phylogenetically diverse bacteria including Clostridium perfringens iota toxin, Clostridium spiroforme CST toxin, and insecticidal proteins from Bacillus thuringiensis,^88,104 suggesting an evolutionarily conserved mechanism whose functional significance in CDI remains unclear.

Sporulation-related genes

A key feature of C. difficile pathogenesis is its ability to form metabolically dormant and resilient endospores that can survive hostile environmental conditions, facilitating transmission, environmental persistence, and disease recurrence.⁸ Successful colonization requires spores to germinate into vegetative cells in the intestinal environment in response to primary bile acid germinants and amino acid or divalent cation co-germinants^6,105. Unlike most spore-forming bacteria, C. difficile lacks the canonical transmembrane nutrient-germinant receptor and instead utilizes a pseudoprotease CspC:CspA heterodimer as signaling node for both germinants and co-germinants.¹⁰⁶ In contrast, key taxa in the gut microbiota deconjugate these primary bile salts into secondary bile salts, thereby inhibiting germination.¹⁰⁷

To transit infection, C. difficile activates sporulation pathways prior to exiting the host. The core sporulation machinery, including the master sporulation regulator, Spo0A, and the sporulation-specific sigma factors, is conserved across diverse C. difficile lineages and is largely encoded within the species core genome.¹⁰⁸ Spo0A is conserved across most endospore formers and is typically activated by a multistep phosphorelay. However, C. difficile has diverged significantly and employs orphan histidine kinases for the same process.²⁰ This simplification likely reflects adaptation to the intestinal niche, favoring rapid commitment to sporulation over complex signaling cascades. Further, the Spo0A regulon in C. difficile also regulates toxin production, directly coupling virulence gene expression to sporulation.²⁰ The sporulation specific sigma factors are conserved across C. difficile strains and show similarity to model spore formers like Bacillus subtilis; however, their activation order differs significantly between the two species. Likewise, spore coat formation in C. difficile employs conserved components, such as SpoIVA, alongside more variable coat associated factors like SipL, CotL, and CdeM, indicating divergence from canonical spore forming systems like that of Bacillus subtilis.^6,20,109 Collectively, the sporulation gene repertoire in C. difficile reflects homology with other bacteria indicating a layered evolutionary trajectory.¹¹⁰

Cell-surface genes

During the vegetative growth phase, C. difficile is covered by a dense, protective, procrystalline surface layer (S-layer) composed primarily of the surface layer protein A (SlpA) and 29 cell wall proteins (Cwps).^111,112 SlpA is synthesized as a precursor encoded by the slpA gene and subsequently cleaved by the cysteine protease Cwp84 into high- and low-molecular-weight subunits,^111,113 which bind and form the compact S-layer 2D array. Since binding to the intestinal epithelial cell surface is a primary step in C. difficile colonization and pathogenesis, SlpA promotes virulence by increasing C. difficile persistence in the gut environment and inducing pro-inflammatory cytokines by binding with cell-surface Toll-like receptors (TLR).^21,114 Notably, purified SlpA from hypervirulent strains (e.g., RT126, RT001, and RT084) has been shown to downregulate tight junction proteins and increase TLR4, IL-6, and TNF-α expression levels compared with nonvirulent strains in vitro, suggesting strain-specific SlpA can indirectly enhance virulence by mediating cellular junction disruption and host immune response.²¹ Additionally, SlpA has been suggested to function as a receptor for C. difficile specific bacteriophages from the Myoviridae and Siphoviridae families, with phage specificity largely determined by SlpA isoform variation,^113,115,116 positioning it as a potential target for cocktail phage therapy against CDI.

Cwps, which constitute approximately 5%–20% of the S-layer, also play a critical role in C. difficile pathogenesis.^117–119 While 12 out of 29 cwp genes are clustered within the slpA locus, 17 genes are distributed in the rest of the genome.^112,120 Some (e.g., Cwp16, Cwp18, and Cwp25) are highly conserved across clinical isolates, while some have high heterogeneity that results in different phenotypes.¹¹² For example, mutations in Cwp84 could result in accumulation of uncleaved SlpA, which impair biofilm formation and surface hydrophobicity.¹¹⁷ CwpV promotes C. difficile aggregation through the C-terminal repetitive domain, which is distinct between strains and can influence antigenicity and anti-phage protection.^118,119 Additionally, surface appendages such as flagella and pili also contribute to immune response.^121,122 Flagellin (encoded by fliC gene), the major flagella protein component, is recognized by surface TLR5 and activates the MAPK and NF-κB signaling cascades, leading to pro-inflammatory mucosal response.¹²² However, the role of flagella in adherence and colonization is strain-dependent; loss of motility from the fliC mutant reduced adherence and colonization in hypervirulent RT027 strain but showed the opposite phenotypes in reference strain 630, suggesting the genetic background could modulate flagella-dependent functions.^123,124

Whole-genome analyses have revealed that cell-surface genes are among the most recombination-enriched loci and contribute to interclade virulence variation.^120,125 For example, SlpA has been suggested to co-evolve with secA2 and cwp66 genes to form a 10-kb cassette in the cwp cluster, which are driven by homologous recombination and has resulted in 12 distinct isoforms that correlated with different antigenicity phenotypes.^113,120 Flagellar assembly and cell envelopment biogenesis functional pathways are also shown enriched for recombination,¹²⁵ and flagellar expression gene heterogeneity has been shown to enhance virulence in hypervirulent strain RT027, suggesting evolution in the flagellar switch contributes to disease outcome.¹²⁶

MGEs (phages, transposable elements, and plasmids)

MGEs, including bacteriophages, transposons, and plasmids, act as major drivers of C. difficile genome plasticity by allowing gene acquisition from different C. difficile lineages and different bacteria species.^13,127 These elements facilitate horizontal transfer of virulence genes, antimicrobial resistance genes, and genes for gut environment adaptation as a major mechanism of C. difficile evolution.

Bacteriophages are bacterial viruses that can infect and replicate in their hosts, either through lytic replication that results in host cell lysis or through lysogeny, in which the phage genome gets integrated into the host chromosome or plasmid.⁸¹ Prophages are highly prevalent across C. difficile pangenome, with individual isolates often carrying one to three prophages.⁸¹ Currently, more than 20 prophages have been identified in C. difficile and all belong to the Myoviridae and Siphoviridae families.^81,128 Nearly all the C. difficile phages are lysogenic and some can directly enhance or attenuate virulence during lysogeny.^129,130 For example, ϕCD38–2 replicates as a circular plasmid in lysogeny and can increase tcdA and tcdB expression in RT027¹³⁰ up to 2.1-fold. The deletion of ϕCDKH02 has been shown to decrease TcdB protein levels, lower sporulation rate, and lower adhesion in RT027 and RT176 strains.¹³¹ Conversely, ϕCD119 has been shown to reduce tcdA and tcdB expression, likely through the phage repressor RepR binding to the upstream of PaLoc region.¹³² Beyond virulence regulation, phages also regulate other HGT, such as transposon integration into the C. difficile genomes.¹²⁹ For example, ϕC2 mitigates the incorporation of CTn Tn6215, which encodes erythromycin resistance gene erm(B) in different C. difficile strains.¹²⁹ While the influence of individual phages likely depends on strain specific genetic backgrounds, genomic analysis of C. difficile clinical isolates showed that nontoxigenic isolates generally have an increased abundance and diversity of prophages compared with toxigenic isolates.¹³³

Transposable elements, particularly CTns, are also highly prevalent in C. difficile genome and are major drivers of evolution. These elements can excise, transfer, and reintegrate ARGs and virulence genes into the genome through the activity of encoded integrases, excisionases, and recombinases.^134–136 For instance, Tn5398, one of the earliest described C. difficile CTns, carries two copies of erm(B), which encodes a 23S rRNA methylase that confers resistance to macrolide-lincosamide-streptogramin B (MLS) antibiotics, a phenotype now widespread among clinical C. difficile isolates.^135,137 Similarly, Tn6194 is frequently found in RT027 and also harbors erm(B) that can be transferred into Enterococcus faecalis, highlighting C. difficile as a potential ARG reservoir for other gut bacterial species.¹³⁴ Besides erm(B), transposons also carry other ARGs, such as resistance genes against tetracyclines (tet(M) on Tn916/Tn5397; tet(W) on TnB1230), chloramphenicol (catD on Tn4453a/b), and linezolid (cfr on Tn6218).^14,138–140 Additionally, CTns can also influence virulence by carrying cell-surface genes that affect adherence and colonization, including sortase and sortase substrates.^41,141

In addition to CTns, plasmids are major carriers of ARGs and virulence genes in C. difficile, with 13% of isolates predicted to harbor at least one potential plasmid.¹⁴² Both previous and current clinical CDI antibiotic treatments, metronidazole and vancomycin, have plasmid-mediated resistance identified: metronidazole resistance is correlated with plasmid pCD-METRO and plasmid pX18–498 carries N-acetylmuramoyl-L-alanine amidase gene, which encodes vancomycin resistance.^22,143 Though rare, plasmids can also carry toxin genes.^17,144 Three C-I isolates have been shown to escape PCR-based diagnosis and carry a monotoxin TcdB next to a full CdtLoc in putative plasmids.¹⁷ Additionally, five more isolates with similar toxin-encoded plasmids were identified from C2 and C4, suggesting potential plasmid exchange between clades that increased toxin locus diversity.¹⁴⁴

Dynamic C. difficile epidemiology

Since the first identification of C. difficile as a cause of pseudomembranous colitis in the 1970s, C. difficile epidemiology has substantially changed, particularly in the past two decades (Figure 1). The shifts are characterized by differences in dominant strains, incidence rates, mortality rates, and acquisition routes.¹⁴⁵ Here we synthesize genetic features of representative C. difficile epidemic strains in North America and their contributions to epidemic success. A marked rise in CDI outbreaks occurred in the early 2000s (Figure 1), mostly in healthcare-associated CDI (HA-CDI). From 2000 to 2008, the CDI diagnosis incidence in US hospitals rose from 3.82 to 8.75 cases per 1,000 discharges,¹⁴⁶ while CDI-attributed deaths increased nearly 10-fold, from 793 to 7,483 cases. Notably, 93% of CDI-associated deaths in 2008 occurred in patients over 65 years of age, underscoring age as a major risk factor of CDI susceptibility.³ During this period, the majority of HA-CDI cases in the United States and Canada were caused by the RT027 strain, a hypervirulent C2 lineage with high rates of mortality, recurrence, and fast transmission between hospitalized patients.^1,147,148 Though the major genes that contribute to RT027’s epidemic success remain debatable, fluoroquinolone resistance, through mutations in gyrA and gyrB, is considered one of the most important drivers of its expansion from North America to Europe and beyond.^149,150 Additional genetic features have been suggested to enhance RT027 virulence, including truncation at tcdC and the presence of binary toxin.^90,151 A single mutation at trehalose repressor gene has also been hypothesized to contribute to enhanced virulence through increased trehalose metabolism but was not supported by subsequent studies.^152,153 Overall, the epidemic success of RT027 appears to reflect selective amplification of a pre-adapted lineage under antibiotic selection with additional virulence enhancements.

With the implementation of targeted prevention, including infection-control practices and antimicrobial stewardship programs,^154,155 the incidence of HA-CDI in North America started declining in the 2010s after the 2000s outbreaks. In the US, the estimated number of HA-CDI cases decreased from 306,500 in 2011 to 235,700 in 2017,¹ while in Canada, the rates fell from 4.74 to 3.61 cases per 10,000 patient-days between 2015 and 2019.¹⁵⁶ Alongside the decline in HA-CDI cases, the incidence of RT027 has also decreased, likely due to restricted fluoroquinolone usage¹⁵⁷ and competitive displacement by emerging virulent strains such as RT106 (Figure 1).^27,158 Unlike RT027, RT106 belongs to C1 and is characterized by a lower mortality rate and the absence of binary toxin.¹⁵⁸ RT106 also harbors a unique 48-kb genomic island that encodes vanZ-like genes conferring glycopeptide resistance and a putative srtB-anchored collagen-binding adhesin, which enhances collagen-dependent biofilm formation.³⁵ Other emerging lineages also have the potential to become epidemic strains by carrying distinct genetic features and phenotypes. For instance, RT078 is a C5 strain that is frequently associated with community-acquired CDI (CA-CDI), characterized by its potential zoonotic origin and the tetM tetracycline ARG in MGEs.¹⁵⁹ Whereas RT023, a strain from C3, carries a unique slpA variant that enhances host adhesion and immune evasion and a large genetic insertion encoding a putative siderophore biosynthetic cluster that may facilitate iron acquisition to better colonize distinct environmental niches.^39,41 Together, diverse genetic traits affecting virulence, transmission, and colonization have been independently acquired across lineages, highlighting how genomic plasticity of C. difficile can shift its epidemiology.

In addition to the emergence of new predominant C. difficile strains, another potential epidemiological shift is the rise of CA-CDI. Though C. difficile ubiquitously exists in the environment, distinct reservoirs for toxigenic strains have been identified, such as asymptomatic carriers shedding toxigenic spores to the surrounding environment and contributing to CA-CDI.^160,161 Toxigenic strains such as RT078 and RT126 have been identified in livestock and have also been isolated from wastewater and soil samples, suggesting potential zoonotic transmission and the role of environmental contamination in spreading CDI in the community.^23,162,163

Conclusion

Advances in molecular typing methods and WGS have been fundamental in defining the lineage and evolutionary landscape of C. difficile. These approaches have resolved its population structure into distinct clades with varying genetic features that shape the virulence potential, transmission route, and epidemiological impact. Among the evolution of virulence and putative virulence genes, evolution of PaLoc is a central determinant of virulence outcome across lineages. Comparative genomic analysis with related clostridial species, which have diverged to target different receptors/tissues, suggests that HGT and receptor adaptation have driven toxin evolution. Beyond toxin genes, sporulation-related genes, cell wall genes, and MGEs have further contributed to C. difficile virulence by enhancing gut colonization, immune response, and antibiotic resistance across lineages. The historic emergence of lineages such as RT027 shows how antibiotic pressure, particularly fluoroquinolone use, selectively amplified a pre-existing lineage carrying advantageous traits rather than the emergence of novel virulence factors.

Looking forward, the distribution and diversity of putative virulence genes across clinical C. difficile lineages and their contribution to disease-associated phenotypes could be further defined through advanced WGS, combined with comparative genomics and genome-wide association analysis. As the epidemiology of C. difficile continues to shift, with changes in predominant strains and acquisition route, continued genomic surveillance is essential for tracking emerging lineages, identifying CDI risk factor genes, and anticipating future outbreaks. Finally, defining the evolutionary origins, mechanisms, and trajectories of virulence and putative virulence genes could help identify variants with the potential to enhance virulence or prevalence, enabling early detection and informing strategies to limit their dissemination. Together, these approaches will improve our ability to understand, predict, and diagnose CDI.