Clostridium difficile ribotype 017 – characterization, evolution and epidemiology of the dominant strain in Asia

ABSTRACT Clostridium difficile ribotype (RT) 017 is an important toxigenic C. difficile RT which, due to a deletion in the repetitive region of the tcdA gene, only produces functional toxin B. Strains belonging to this RT were initially dismissed as nonpathogenic and circulated largely undetected for almost two decades until they rose to prominence following a series of outbreaks in the early 2000s. Despite lacking a functional toxin A, C. difficile RT 017 strains have been shown subsequently to be capable of causing disease as severe as that caused by strains producing both toxins A and B. While C. difficile RT 017 strains can be found in almost every continent today, epidemiological studies suggest that the RT is endemic in Asia and that the global spread of this MLST clade 4 lineage member is a relatively recent event. C. difficile RT 017 transmission appears to be mostly from human to human with only a handful of reports of isolations from animals. An important feature of C. difficile RT 017 strains is their resistance to several antimicrobials and this has been documented as a possible factor driving multiple outbreaks in different parts of the world. This review summarizes what is currently known regarding the emergence and evolution of strains belonging to C. difficile RT 017 as well as features that have allowed it to become an RT of global importance.


Introduction
Clostridium difficile is an important cause of antimicrobial-associated diarrhoea (AAD) in both humans and animals [1]. In humans, the disease can progress from watery diarrhoea to life-threatening pseudomembranous colitis (PMC) and toxic megacolon [2]. C. difficile infection (CDI) is a toxin-mediated disease and major virulence factors include toxin A (TcdA, 308 kDa) and toxin B (TcdB, 270 kDa) [3]. An additional binary toxin (C. difficile transferase, CDT) is produced by some strains only. CDT-producing strains of C. difficile account for an increasing proportion of human infections in some parts of the world (currently ca. 20% of CDI cases in non-outbreak situations) but are common in animals [4,5]. C. difficile can be classified into different PCR ribotypes (RTs) using banding patterns of the amplified intergenic spacer region between the 16S and 23S rRNA genes [6]. Currently, over 600 RTs exist in the United Kingdom-based C. difficile Ribotyping Network (CDRN) database [7].
Brief history of C. difficile infection and the emergence of C. difficile RT 017 were no reports of human gastrointestinal infections associated with C. difficile until 1978 when, after a period of intense trans-Atlantic competition between researchers, C. difficile was identified in faecal specimens from patients with PMC [25].
Not all strains of C. difficile produce toxins and cause disease. Initially, it was thought that all toxigenic strains of C. difficile produced both major toxins [26]. For two decades after the association between C. difficile and PMC was shown, it was believed that TcdA was required to cause initial damage to the intestinal mucosa before TcdB could exert its potent cytotoxic effect [27], and the significance of TcdAnegative, TcdB-positive (A-B+) stains was not apparent [17]. To further support this belief, the first few strains of C. difficile isolated with an A-B+ phenotype were associated only with asymptomatic carriage [28]. During this same period, there was a move away from using the faecal TcdB cytotoxicity assay and/or culture of C. difficile for diagnostic purposes due to the time and expense involved in maintaining and using cell lines, and the long turnaround time of culture. Concomitantly, there was an emphasis on developing rapid immunoassays for the detection of TcdA [29]. TcdA was chosen because of the continued mistaken belief that C. difficile produced either both TcdA and TcdB, or no toxins, because it was easier to manufacture antibodies against TcdA, and because detection of TcdA had greater sensitivity compared to detection of TcdB [30]. These tools made the detection of C. difficile easier, but with far less overall sensitivity, and further obscured the significance of A-B+ C. difficile strains.
The importance of A-B+ strains of C. difficile was finally appreciated at the end of the twentieth century when 16 patients in a Canadian tertiary-care hospital developed PMC with an A-B+ strain. Stool samples from these patients tested negative for C. difficile TcdA but were later shown via a cytotoxin assay to contain C. difficile that produced a functional TcdB only [17]. Similar findings were published from other countries [13,16] and further studies confirmed these strains as A-B+ C. difficile RT 017 [8]. At the same time, a study reported that not only could TcdB exert its cytotoxic effect in the absence of TcdA, but also that human intestinal mucosa was around 10 times more sensitive to TcdB than TcdA [31]. This was the first time that the clinical significance of A-B+ C. difficile became evident [32]. Over the last 20 years, C. difficile RT 017 has been isolated from many parts of the world, however, it is likely that C. difficile RT 017 originated from a single geographical region and its global dispersal has been a relatively recent event [33].

Characteristics of C. difficile RT 017
Epidemiological typing of C. difficile RT 017 Currently, PCR ribotyping is a method of typing C. difficile that is widely used in many parts of the world due to its relative simplicity and high discriminatory power [34]. However, ribotyping requires comparison of banding patterns with those of standard strains present in a library of patterns that was established in 1999 [6]. Thus, reports of C. difficile before or around that time classified C. difficile by various other methods [17,35]. Table 1 summarizes these different methods used when referring to C. difficile RT 017. Early ribotyping studies in Japan used their own nomenclature and assigned "fr" to RT 017 [36].
Before genotype-based methods, C. difficile was classified using phenotypic methods that, in general, had poor reproducibility, low typeability, and lacked sufficient discriminatory power to be applied to epidemiological studies [42]. However, serogrouping was widely used early and showed a good correlation with toxigenicity [43]. Serogrouping classified C. difficile RT 017 as either serogroup F or X [37].
Many genotypic methods, including ribotyping, use unique banding patterns of different PCR products to classify C. difficile strains. Toxinotyping detects differences in the Pathogenicity Locus (PaLoc) and classifies C. difficile RT 017 as toxinotype VIII [38]. Pulsed-field gel electrophoresis is more commonly used in North America and classifies C. difficile RT 017 as North American pulsed-field gel electrophoresis type 9 (NAP 9) [39]. Restriction endonuclease analysis (REA) typing has greater discriminatory power than ribotyping and divides C. difficile RT 017 into several REA types which are grouped as REA groups CF and CG [37].
Multi-locus sequence typing (MLST) is another genotype-based method involving 7 housekeeping genes. However, it is not based on banding patterns but rather the unique sequences of these genes and thus has been used mainly in evolutionary studies. This method classifies C. difficile RT 017 as sequence type (ST) 37 belonging to evolutionary clade 4 [40]. MLST has good discriminatory power, however, it is relatively more complicated to perform [34]. The advent of nextgeneration sequencing makes in silico MLST now more accessible [44]. A recent study in China reported that RT 017 can also be identified using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) with high sensitivity and specificity [45]. However, this study did not include other C. difficile strains from clade 4 and another Chinese study suggested that different clade 4 strains may not be distinguishable by this method [46].
C. difficile RT 017 toxin C. difficile RT 017 is classified as A-B+ C. difficile as it produces only a functional TcdB [8]. Its TcdB also gives a different cytopathic effect (CPE) in cell cytotoxin assays using various cell lines compared to other strains that is often referred to as a variant CPE [16,47]. Studies on the tcdA gene of C. difficile RT 017 revealed a 1.8 kb deletion in the repeating region (3 ′ end) ( Figure 1) and a point mutation in the 5 ′ end which results in a premature stop codon [49,50]. The 1.8 kb deletion corresponds to a deletion of the carboxy repetitive oligopeptide (CROP) region of TcdA, which is the recognition site of many TcdA enzyme immunoassays (EIAs), making the toxin undetectable by these EIAs [47]. The nonsense mutation at 5 ′ end corresponds with a loss of catalytic action of the TcdA, thus making the toxin non-functional [47,49].
Notably, despite lacking a functional TcdA, most of the tcdA gene in C. difficile RT 017 remains intact and can be detected by PCR if primers specific to the nonrepeating region of the tcdA gene are used. In such cases, C. difficile RT 017 could be incorrectly detected as both tcdA-and tcdB-positive C. difficile [51].
While these primers are efficient for detection of toxigenic strains in clinical practice, the results may appear confusing in an epidemiological study. An additional primer set is needed to identify the deletion in the repeating region of tcdA gene and differentiate C. difficile RT 017 from true A+B+ C. difficile strains [28,52].
Interestingly, the TcdB of RT 017 (TcdB-F) is different from the TcdB commonly found in most C. difficile strains. TcdB-F behaves as a "functional hybrid," combining characteristics of both TcdB and the Clostridium sordellii lethal toxin, TclS. While TcdB-F binds to the same cellular receptors as TcdB, the two proteins display differences in their target specificity, with TcdB primarily glucosylating Rho, Rac and Cdc42 targets, and TcdB-F glucosylating Rac and Ras targets ( Figure 2) [48]. The difference in cellular targets is thought to be responsible for the different CPE observed for the two toxins [50].
Infection due to C. difficile RT 017 Despite producing toxin B only, several studies suggest that C. difficile RT 017 causes clinical disease that is indistinguishable from that caused by other C. difficile RTs [9,12]. In addition, C. difficile RT 017 causes disease as severe as that caused by "hypervirulent" C. difficile RT 027 [10]. In an outbreak setting, mortality due to C. difficile RT 017 can be as high as 37.5% [47], but this high mortality rate may have been due to the exclusion of mild cases. There have been no clinical studies of C. difficile RT 017 infection in South East Asia, where there is a high prevalence of RT 017 [41,53,54]. Given that CDI in this region was, in general, associated with low mortality and recurrence [55], it will be interesting to see whether the less severe CDI in this region is specifically associated with C. difficile RT 017 or if there are other unknown protective factors in the population or region, such as a high prevalence of carriage of non-toxigenic strains, which may occupy the same niche and competitively exclude toxigenic strains from the gut [53,56,57].

Evolution and transmission of C. difficile RT 017
Based on MLST and Bayesian evolutionary model analysis (Figure 3), C. difficile has evolved into at least five clades and three cryptic clades. This clade divergence occurred more than a million years ago [34]. C. difficile RT 017 (ST 37; red arrowhead in Figure 3) is a member of C. difficile clade 4 along with many non-toxigenic, and some similar toxigenic, strains [46,[58][59][60][61]. Despite limited data, it is clear that both A-B+CDT-and non-toxigenic strains of C. difficile (orange and green, respectively, in Figure 3) are equally distributed throughout clade 4, indicating  [48]. The nonsense mutation near the 5 ′ terminal of the tcdA gene is responsible for the loss of function of TcdA [49]. The 1.8 kb deletion near the 3 ′ terminal of the tcdA gene makes TcdA undetectable by many toxin EIAs [47].
that the clade 4 ancestor could either be a toxigenic (A-B+CDT-) or non-toxigenic strain. A recent study analyzed time-scaled core-genome phylogenies and suggested that the clade 4 ancestor was a nontoxigenic strain of C. difficile, and that acquisition of the PaLoc in C. difficile RT 017 occurred around 500 years ago [59].
To date, the genomes of two C. difficile RT 017 strains (CF5, isolated in Belgium in 1995, and M68, isolated in Ireland in 2006) have been completely sequenced, providing important reference chromosomes for whole genome sequencing (WGS) studies of this lineage [62]. Figure 4 shows the genome of C. difficile strain M68. Using WGS, Cairns et al. showed that 23 of 24 of C. difficile RT 017 strains from one hospital were closely related and formed a single cluster. The only unrelated C. difficile RT 017 strain was isolated from a patient with community-acquired CDI and this belonged to a cluster from outer London hospitals. These findings suggested that C. difficile RT 017 was mostly transmitted between patients in the same ward and between wards in the same hospital. The study further found that environmental contamination with clinical isolates was possible and that RT 017 could withstand decontamination with hydrogen peroxide vapour [22].
Another WGS study of 277 different C. difficile RT 017 strains isolated from around the world, including 24 from animals (cattle, dogs, and horses) showed that C. difficile RT 017 could be transmitted between humans and animals, and also reported that deletions and insertions found in RT 017 genomes were distributed throughout all geographical areas [33]. The finding of little genetic diversity implies that C. difficile RT 017 originated in a single geographical area and that global spread occurred relatively recently, however, it remained unclear where that single geographical area was. Cairns et al. [33] concluded that C. difficile RT 017 originated in North America and then spread to Europe, Asia and other parts of the world [33]. This conclusion contradicts many epidemiological studies (see below) that, taken collectively, suggest that the origin of C. difficile RT 017 is in Asia. The Cairns et al. study included only a limited number of historic C. difficile RT 017 isolates from Asia (2 strains from Korea and 1 strain from Japan, all isolated in 1995) and a greater number of C. difficile RT 017 strains from North America (9 strains from the United States isolated from 1990 to 1996).
Reports on C. difficile RT 017 infection started in the late 1990s with a series of outbreaks in Poland [13], Japan [14,15], the Netherlands [16], Canada [17], and Argentina [18]. During the early 2000s, there were outbreaks of so-called "hypervirulent" C. difficile RT 027 in Europe and North America [82], and the number of RT 017 reports appeared to subside   [19][20][21], and in Sweden in 2008 [23]. The most recent documented outbreak of RT 017 infection started in 2009 in England and persisted for at least 3 years [22].
Among these C. difficile RT 017 outbreaks, clinical characteristics of the patients were described only in reports from the outbreak in Canada, with 18.8% (3/ 16) of cases having PMC, 31.3% (5/16) of cases being recurrent and a 37.5% (6/16) mortality rate [17,47]. Outbreaks of C. difficile RT 017 infection have been linked to the use of clindamycin [16] and fluoroquinolones [21]. In both outbreaks, discontinuation of the offending agent resulted in a rapid decline in the number of CDI cases due to C. difficile RT 017 [16,21]. This suggests that these outbreaks were associated with the use of specific antimicrobials and that antimicrobial stewardship helped to control spread. Besides many outbreaks, there have also been nonoutbreak reports of C. difficile RT 017 throughout the world. The majority of these reports with high prevalence figures were from Asia, while reports from non-Asian countries mostly recorded low prevalence figures. Data summarizing the prevalence of C. difficile RT 017 in Asia and non-Asian countries can be found in Tables S1 and S2 in the supplementary document.

C. difficile RT 017 in Asia
It is likely that C. difficile RT 017 is endemic in Asia and has been resident in this region for a long time, for three different reasons. First, in contrast to non-Asian countries, RT 017 appeared mainly in non-outbreakrelated prevalence studies [41,53,54,69,[71][72][73][74][75][86][87][88]. Second, there have been reports of A-B+CDT-C. difficile RTs in the region other than C. difficile RT 017 with similar deletions in the tcdA gene, some of which have also been classified in MLST clade 4 [46,58,60,61,71,89]. Third, the earliest Asian isolates of RT 017 in humans can be dated back to 1993 in Indonesia, where five strains of RT 017 were isolated from healthy infants [15]. The high prevalence and diversity of A-B+CDT-C. difficile in Asia and the evidence of old C. difficile RT 017 isolates suggest that the origin of this RT is in Asia. While Asia is a very large continent, current information suggests that C. difficile RT 017 is endemic in at least two different regions of the continent: parts of East Asia, and South East Asia [90].
In contrast to these reports, Japan saw an outbreak of C. difficile RT 017 infection in 1996 [14,15], perhaps coincidentally, around the same time as RT 017 outbreaks in Poland, the Netherlands and Canada [13,16,17,67]. However, there have been no major reports of C. difficile RT 017 infection in Japan since. Interestingly, in 2001, there was an outbreak of CDI caused by an A-B+ strain of C. difficile with an RT pattern resembled C. difficile RT 017 [91]. This strain was later identified as the novel C. difficile RT 369, a strain that is closely related to C. difficile RT 017 [36], and that was recently identified in China as ST 81, a single loci variant of ST 37 [92]. To date, RT 369 remains among the most common toxigenic strains isolated in Japan while only a small number of C. difficile strains belonging to RT 017 have been detected [93].

South East Asia
Most epidemiological studies in South East Asia have been conducted in Thailand [41,53,89] with additional reports from Indonesia [54], Laos [86], Malaysia [56,87] and Singapore [88]. Although the information is limited, based on these publications, and some publications from Thailand that detected a high prevalence of A-B+ C. difficile [94][95][96], it is likely that RT 017 is endemic throughout this region.
Despite isolating C. difficile RT 017 strains as early as 1993 [15], there were no epidemiological studies in the region until 2006 [41]. All studies thereafter reported similar results. In Thailand, three studies confirmed that C. difficile RT 017 ranks among the most common toxigenic strains present (ca. 30.8% -41.5%) [41,53,89]. In Indonesia, C. difficile RT 017 was the most prevalent RT isolated from patients [54]. C. difficile RT 017 has been isolated in Laos [86], although only five patients were included in this report. The most recent report from South East Asia came from Malaysia where the prevalence of C. difficile RT 017 was 20.0% [56]. In contrast to other South East Asian countries, a study in Singapore reported a low prevalence of RT 017 of 4.9% (3/61), and an RT distribution more like European countries. The comment was made that this possibly reflected the international population of Singapore, both resident and passing through [88].

Europe
Apart from obvious outbreaks, reports of RT 017 in Europe were scarce. During the late 1990s, the prevalence of RT 017 was 2.5% (9/364) in France [67]. During 2008-2009, RT 017 was responsible for 4.9% (2/41) of severe CDI cases in Germany [84]. In 2012, only one out of 171 (0.6%) C. difficile isolates from Austria was classified as C. difficile RT 017 [68]. A pan-European study reported an overall prevalence of Figure 5. Timeline of C. difficile RT 017 reports around the world. Outbreaks refer to an increase in the regional prevalence of RT 017, which is confirmed either to be clonal or with evidence suggesting that isolates came from the same source. Endemic presence refers to prevalence reports that were not associated with outbreaks.

Australia
Several epidemiological studies conducted in various regions of Australia with C. difficile RT 017 being found at a much lower prevalence compared to Asia. The prevalence of C. difficile RT 017 infection was ca. 3% [78][79][80][81] suggesting that those cases are are more likely to be imported rather than caused by endemic strains.

Africa
The number of studies on CDI in Africa is very limited. To date, the only country with reported C. difficile RT 017 infection is South Africa, where a very high prevalence of RT 017 among diarrhoeal patients in tuberculosis hospitals was seen [77,98,99]. Historically, Cape Town in South Africa has been an important port city where ships coming from and going to Asia, Australia and Europe stopped during their voyages. The introduction of C. difficile RT 017 may merely reflect travel between these regions, however, it appears that C. difficile RT 017 has now become established within the hospital system in South Africa. Patients testing positive for C. difficile are at high risk of mortality, and tuberculosis is an additional risk factor for CDI in populations with HIV [100].

C. difficile RT 017 in animals
Recently, many C. difficile strains associated with CDI in humans have also been isolated from animals or animal products suggesting that CDI may be transmitted from animals [101]. Despite its high prevalence in the Asian human population [102], there have never been any reports of C. difficile RT 017 in animals in this region [103,104], and it has rarely been reported in animals elsewhere. C. difficile RT 017 has been isolated from calves in Canada [105] and rabbits in Italy [106]. The WGS study undertaken in the United Kingdom by Cairns et al. involving 277 C. difficile RT 017 strains only included 24 strains of animal origin [33]. The reasons why RT 017 is apparently not prevalent in animals have not been elucidated.
Role of antimicrobial resistance in the outbreaks of C. difficile RT 017 AMR plays an important role in the dissemination of many C. difficile RTs. Being resistant to antimicrobials while the intestinal microbiota is disrupted allows C. difficile to survive, produce toxins and eventually cause disease [2]. Furthermore, being intrinsically resistant to alcohol and desiccation, C. difficile as a spore can survive within the hospital environment and spread to patients. Antimicrobial resistance has been associated with CDI outbreaks in the past; in particular, the outbreaks of "epidemic" C. difficile RT 027 in North America and Europe were associated with fluoroquinolone and rifampicin resistance.
Outbreaks of infection with C. difficile RT 017 have been linked with clindamycin-and fluoroquinoloneresistant strains [13,16,18,21]. Besides these antimicrobials, C. difficile RT 017 also has higher rates of resistance to tetracyclines and rifaximin [107][108][109]. Tetracycline resistance was associated with an outbreak of C. difficile RT 078 [110,111]. Rifaximin is a derivative of rifampicin which was also associated with the outbreak of C. difficile RT 027 [112,113]. There is no doubt that misuse of these antimicrobials may lead to the future outbreaks of C. difficile RT 017, given that it is endemic in East and South East Asia, where tetracycline and rifampicin are commonly prescribed for many tropical infections and tuberculosis, respectively.

Conclusions
C. difficile RT 017 is one of the most successful RTs of C. difficile in the world. It was the first A-B+ C. difficile shown to cause CDI following several outbreaks. This discovery led to a better understanding of the pathogenesis of CDI in general, together with the roles of TcdA and TcdB, and eventually lead to changes in the way the laboratory diagnosis of CDI was made. The high rate of resistance to many antimicrobial agents provides hints as to how C. difficile RT 017 spread throughout the globe. It also gives us a warning that antimicrobial stewardship is needed to prevent further outbreaks.
The ancestral home of C. difficile RT 017 remains controversial, however, the weight of epidemiological evidence suggests that this strain originated in Asia and spread to other regions of the world long before the much-publicised spread of RT 027. Particular clinical characteristics of C. difficile RT 017 infection have yet to be determined. Why C. difficile RT 017 is not found more commonly in animals despite successful human spread also remains unclear, however, this may just reflect a lack of animal studies in Asia. Also, there has been no study comparing phenotypic characteristics of C. difficile RT 017, such as sporulation, germination and motility, with other epidemic strains. Since these properties are related to the spread of C. difficile, such studies may uncover important factors that help in the control of C. difficile RT 017 spread and prevent further outbreaks.

Disclosure statement
No potential conflict of interest was reported by the authors.