International Clostridium difficile animal strain collection and large diversity of animal associated strains

Background Clostridium difficile is an important cause of intestinal infections in some animal species and animals might be a reservoir for community associated human infections. Here we describe a collection of animal associated C. difficile strains from 12 countries based on inclusion criteria of one strain (PCR ribotype) per animal species per laboratory. Results Altogether 112 isolates were collected and distributed into 38 PCR ribotypes with agarose based approach and 50 PCR ribotypes with sequencer based approach. Four PCR ribotypes were most prevalent in terms of number of isolates as well as in terms of number of different host species: 078 (14.3% of isolates; 4 hosts), 014/020 (11.6%; 8 hosts); 002 (5.4%; 4 hosts) and 012 (5.4%; 5 hosts). Two animal hosts were best represented; cattle with 31 isolates (20 PCR ribotypes; 7 countries) and pigs with 31 isolates (16 PCR ribotypes; 10 countries). Conclusions This results show that although PCR ribotype 078 is often reported as the major animal C. difficile type, especially in pigs, the variability of strains in pigs and other animal hosts is substantial. Most common human PCR ribotypes (014/020 and 002) are also among most prevalent animal associated C. difficile strains worldwide. The widespread dissemination of toxigenic C. difficile and the considerable overlap in strain distribution between species furthers concerns about interspecies, including zoonotic, transmission of this critically important pathogen.


Background
Clostridium difficile, an anaerobic sporogenic bacterium, is recognized as the major pathogen in healthcare associated intestinal infections in humans and also as an important animal pathogen. In addition to the potential for serious (including fatal) infections in animals, are companion and food animals considered as an important potential source for human community-acquired infections [1][2][3][4]. This indicates the importance of preventive measures targeting animals and food [5]. Several studies have looked at similarity between strains isolated from humans and animals [1,[6][7][8][9][10], but typically focusing on limited number of different species and restricted to a narrow geographic region.
PCR ribotyping is currently the method of choice for differentiation of C. difficile strains. In humans more than 300 PCR ribotypes are recognized while the number of reported animal associated PCR ribotypes is much lower [1,8,11]. It could be expected that variety of animal associated C. difficile will increase with the increased number of typed animal isolates. To date piglets and pig farms are hosts and environments where C. difficile has been most extensively studied [3,[12][13][14][15][16][17][18]. For this reservoir the modes of transmission and potential association with human infections are also best understood [14,16].
C. difficile strains can also be differentiated into toxinotypes according to the differences in the toxin A and toxin B encoding region (PaLoc) [19]. Toxinotypes V and XI are particularly often isolated from animals [19], but the reason for this association is not known. Many published studies report C. difficile PCR ribotypes, but not many do report the toxinotypes [9,13].
The aim of this study was to collect C. difficile isolates from different countries and different animals and to compare them with classical agarose gel-based and sequencerbased PCR ribotyping and determine the toxinotypes.

Results
Altogether 112 C. difficile isolates from 13 animal species were obtained from 14 laboratories from 12 different countries (Austria, Belgium, Canada, Czech Republic, Denmark, Germany, Italy, Scotland, Slovenia, Spain, Switzerland and USA) ( Table 1, Additional file 1). Inclusion criteria were one strain (PCR ribotype) per animal species per laboratory. Each participating country contributed 1 to 24 isolates (Additional file 1). Strains were isolated between 1998 and 2012. Only five isolates were from 1998 to 2002, with the majority of isolates (n = 34) from 2011.
Only 38 isolates were from animals with clinical signs, 56 were from clinically normal animals and for 18 isolates disease status of the animal was not known.

Molecular characterization of strains with two PCR ribotyping approaches
Results of agarose gel-based and sequencer-based PCR ribotyping for 112 C. difficile isolates are presented in Table 1 and Additional file 1.

Distribution of PCR ribotypes and toxinotypes in different animal hosts and in different countries
PCR ribotype 014/020 was found in the majority of the animal species included in the collection (pig, cattle, horse, poultry, cat, dog, rabbit and a goat) across 5 different countries ( Figure 2). PCR ribotypes that were found only in particular animal species, but in several different countries, were PCR ribotype 150, which was found only in pigs but in four different countries, and PCR ribotype 033, which was found in cattle in three different countries and only in a single case in horse.
Toxinotype V strains of different PCR ribotypes were mainly associated with cattle, horses and pigs and were rarely found in other animals. Also toxinotype XI strains (a and b) were mainly associated with cattle and horses.  *Btb + − presence of binary toxin CDT; **including racoons, wild hare, rabbits, goats, partridges, goose, and crow; ***12 countries participated; tox--nontoxigenic strain (lacking the PaLoc and genes coding for binary toxin CDT); A "CE" prefix, eg. (CE)039, indicates that the PCR ribotype assigment was made in reference laboratory (CDRN Leeds) using the newer capillary electrophoresis-based approach. PCR-ribotypes in the table are ordered according to the number of countries in which the strains were found and then according to the number of strains belonging to that ribotype. Figure 1 Dendrogram showing similarities of banding patterns generated with classical agarose gel electrophoresis based PCR ribotyping for all 38 different PCR ribotypes included in the collection. For each "gel-based" PCR ribotype all profiles generated with capillary electrophoresis PCR ribotyping are added for comparison. A "CE" prefix, eg. (CE) 039, indicates that the ribotype assigment was made in reference laboratory (CDRN Leeds) using the newer capillary electrophoresis-based approach.

Discussion
Several publications are available about isolation and characterization of C. difficile from animals, but they are usually limited to a specific geographical region and data of the different studies are sometimes difficult to compare due to nomenclature of PCR ribotypes that is not unified. The advantage of the international animal C. difficile strain collection is in performance of strain typing in the single setting, hereby minimizing the ambiguities in PCR ribotype designations. Additionally, two PCR ribotyping approaches have been used; the standard one using 'Cardiff ' nomenclature and WEBRIBO based analysis (giving inter-laboratory comparable results independent of Cardiff/Leeds reference strains). This is not a prevalence study with collection of samples from defined number of farms or hosts as was done for hospitals and human isolates in Europe [20]. The isolates in this study were collected from different labs in different countries and the criterion was one PCR ribotype per animal species per lab. Therefore the collected strains are not reflecting the prevalence of PCR ribotypes but are giving a good basis to assess the diversity of animal associated C. difficile.
Our results show that variability of PCR ribotypes present in different animal species is large. Isolates from pigs and cattle were most common, perhaps reflecting the importance of C. difficile in pigs and concerns about zoonotic transmission from cattle (or more specifically, meat) ( Table 1, Table 2). In contrast to cattle and pigs, are cats, dogs, rabbits and poultry clearly understudied and are probably associated with a broader variety of C. difficile PCR ribotypes than currently recognized. Only two countries have contributed poultry isolates to the collection, but they represent 10 different PCR ribotypes. Strains from captive rabbits were contributed only by one country, but represent 7 different PCR ribotypes (Additional file 1).
Some PCR ribotypes seem to be more often associated with a particular animal host. Many publications report PCR ribotype 078 in pigs [7,10,15] and type 033 in cattle   [22,24] (Table 2). Results presented here confirm this observation (Figure 2), but also suggest that the wellknown animal-associated PCR ribotype 078 may not be currently present in animals in all countries. In addition to types 078 and 033 some other PCR ribotypes are typically associated with pigs, such as PCR ribotypes 150, 002, 045 and 081 (Table 1). Recent reports from Australia describe a new genotype in terms of PCR ribotype (237) and toxin genes (tcdB+, tcdA-, cdtA + and cdtB+) in pigs [18]. PCR ribotype 027 that was prevalent in humans in many countries within the past ten years was in this collection of animal strains found only in USA and Canada (Additional file 1). Toxinotype V and XI and binary toxin positive strains were initially reported to be the prevalent population of strains isolated from animals (70-100%) [19,29]. However, the results of this study show that nonvariant strains of toxinotype 0 are widespread in animals and that the proportion of binary toxin positive strains can be as low as 35.7%.
All isolates in the collection were distributed into 38 PCR ribotypes, but only a few of those contained five or more isolates, and the majority was represented only by a few or a single isolate (Table 1). This resembles the situation with human strain collections with many different PCR ribotypes, but only a few of them having a large number of isolates [20]. However, as the inclusion criterion for the collection was one PCR ribotype per animal species per laboratory, the number of isolates of a given PCR ribotype does not reflect the actual prevalence of this PCR ribotype in animal host. But the high number of isolates of a given PCR ribotype in the collection does reflect its broader geographic presence and possibly broader spectrum of animal species from which it was isolated.
All animal-associated PCR ribotypes with four or more isolates reported here (Table 1) are among the most prevalent in diverse studies of human isolates [8,11,20,30,31]. In particular PCR ribotype 014/020 is currently the most prevalent type isolated in Europe and in some USA studies and is the only type that was present here in the majority of animal hosts (Table 1). Although type 014/020 is not recognized as hypervirulent and is not associated with outbreaks or severe disease in humans, its ability to colonize many diverse hosts and its ubiquitous presence indicates considerable endemic potential of this particular PCR ribotype [8].

Conclusions
PCR ribotype 078 is the most prevalent C. difficile type associated with this collection of animal isolates, but only with some hosts and in some countries. The second most prevalent PCR ribotype is 014/020 which, in contrast, has broader range of animal hosts. Variability of animal associated PCR ribotypes is substantial and is likely to increase with the number of typed animal isolates. Large overlap of animal associated C. difficile PCR ribotypes with human strains furthers concerns about interspecies, including zoonotic, transmission of this important pathogen. Moreover, strains that are prevalent in humans are also prevalent in different animals from different geographic areas, emphasizing that certain strains have a large potential for global dissemination.

Inclusion criteria and requested isolate information
Laboratories from different countries with publications on C. difficile in animals were invited to contribute the isolates. Participating laboratories were asked to provide only a single representative PCR ribotype per animal species and laboratory/country. No formal structure was used to guide isolate submission.
Participating laboratories were asked to provide additional information about the individual isolate and the animal host: date and country/city of C. difficile isolation or specimen collection, molecular characterization (i.e. PCR ribotype, toxin genes), animal species, age, status (farm, domestic or wild animal), clinical signs and antimicrobial exposure history (when available).
All isolates were stored in Microbank cryogenic vials (Pro-lab Diagnostics) at −80°C.

PCR ribotyping and toxinotyping
All isolates were characterized by toxinotyping, agarose gelbased PCR ribotyping and capillary gel electrophoresisbased PCR ribotyping.
Agarose gel-based PCR ribotyping was performed as described elsewhere [33]. PCR ribotypes were determined by comparison of banding patterns with the internal library using the BioNumerics software v7.1 (Applied Maths, Belgium). Banding patterns were compared with a reference library of 48 Cardiff type strains. Strains that were not consistent with those in the library were sent to reference laboratory (CDRN Leeds) for confirmation and are named with prefix (CE), indicating that the assignment was made with the newer capillary electrophoresis-based approach. Three strains generated new ribotyping profiles (not previously recognized in Leeds/Cardiff collection) and are designated with the internal nomenclature (SLO and 3-digit code).
Dendrograms were constructed using the Dice coefficient and the unweighted pair group method with arithmetic means (UPGMA). Position tolerance and optimization were set to 1.1%.
Capillary gel electrophoresis-based PCR ribotyping was performed as described previously [34]. Primers described by Bidet and colleagues [35] were used for amplification of intergenic spacer regions (ISRs), with forward primer labelled with a WellRED dye D4-PA (Sigma-Aldrich, Germany). PCR products were analyzed with CEQ™ 8000 Genetic Analysis System (Beckman Coulter) using a 33 cm capillary and gel LPAI. CEQ 600-bp DNA size standard (Beckman-Coulter) was used to determine fragment lengths. Capillary separation conditions were as follows: samples injection voltage of 2 kV over 60 s, separation voltage of 4.8 kV with a total running time of 75 min and a capillary temperature of 50°C.
PCR ribotypes were determined with WEBRIBO database (https://webribo.ages.at/) [34]. Fragment sizes were also imported into BioNumerics software for comparison of banding patterns generated with classical agarose gel electrophoresis and sequencer-based capillary gel electrophoresis.

Additional file
Additional file 1: Overview of C. difficile strains, animal hosts and countries represented in the collection.