Introduction
The main causative agent of nosocomial diarrhea associated with the use and abuse of antibiotics is
Clostridium difficile; an anaerobic, Gram-positive, spore-forming bacterium.
C. difficile infection (CDI) symptoms include watery diarrhea, anorexia and leukocytosis and in severe disease presentations these features could be accompanied by dehydration, hypoalbuminemia, acute kidney injury, and hypotension or death [
1]. Different genotypes of this bacterium have been isolated worldwide and the prevalence rates varied between the countries [
2].
Most
C. difficile strains produce toxin A (TcdA) and/or toxin B (TcdB), which display cytotoxic activity through glycosylation of small, monomeric GTPases. In addition, a subset of strains synthesize an actin-ADP-ribosylating toxin known as binary toxin (CDT) that targets the actin cytoskeleton [
3]. Genes encoding TcdA and TcdB are located on a so-called Pathogenicity Locus (PaLoc). The genes for CDT, by contrast, are typically located elsewhere on the chromosome [
4] or on extrachromosomal molecules [
5,
6].
TcdA and TcdB induce the secretion of several proinflammatory cytokines (e.g. IL-1β, IL-12, and TNF-α) and lead to neutrophil and macrophage infiltration to intestinal tissues that can result in host damage, as seen in intestinal histopathology specimens [
7,
8].
The global population of
C. difficile is distributed in eight multilocus sequence typing (MLST) clades, of which five contain strains of common detection in human hosts [
9]. In the last decade, many countries have reported outbreaks of CDI, partly due to the emergence and rapid spread of hypervirulent or epidemic strains, such as one classified as NAP1 by PFGE, RT027 by ribotyping, and ST01 by MLST [
10‐
13].
Besides its epidemic potential, the NAP1/RT027/ST01 strain has received particular attention due to its link to CDI cases of increased severity and mortality [
14]. This strain overproduces TcdA and TcdB, secretes binary toxin (CDT), harbors mutations in genes encoding S-layer proteins (SLPs) that increase its adherence to the gut epithelium and overproduce spores [
15]. Some reports have highlighted these characteristics as likely contributors to its increased virulence [
10]. However, the magnitude of these phenotypes is not equal in all isolates from this genotype [
16], and their relative contribution to its epidemicity remains to be elucidated.
The NAP1/RT027/ST01 strain is classified into the so-called hypervirulent MLST Clade 2 [
17,
18] along with 66 additional STs [
19,
20]. It is yet unknown whether “hypervirulence” is a universal trait of the members of this clade. To answer this open question and to widen our current knowledge on the pathogenesis of
C. difficile Clade 2 strains, we performed a comprehensive in vivo analysis of the pathogenicity and lethality of clinical
C. difficile strains typed as ST01, ST41, ST67, and ST252 in two animal models and interpreted the results on the basis of their phylogenomic relatedness.
Discussion
The use of MLST technique has facilitated the classification of isolates into several clades. Clade 2 includes ST01 strains [
17,
18], which have caused outbreaks worldwide [
21,
22]. This clade has been divided into subgroups 2i (ST01, ST67, ST41, ST62 or ST140) and 2ii (ST32, ST47, ST61, ST95, ST97, ST123 or ST252) [
23]. This work revealed a broad spectrum of virulence and pathogenic potential among five clinical isolates from different STs within the
C. difficile “hypervirulent” Clade 2.
Strain 5758-ST01 from subgroup 2i was the most virulent strain in all in vivo experiments, as it induced higher cytokine secretion in intestinal tissue, tissue damage, multiorgan alterations, and mortality. This could be due a combination of already reported characteristics of the NAP1/RT027/ST01 strain, such as toxin overproduction [
24], binary toxin secretion [
25], and increased sporulation [
26]. However, other unexplored mechanisms such as polymorphisms and/or regulation patterns of colonization determinants like S-layer protein, flagellin, β-glucosidases and other adhesins [
27‐
30] could also have played a role in the pathogenesis. A detailed studied of those mechanisms is beyond the scope of this study.
An even wider pathogenicity spectrum could have been observed if more isolates from other STs had been studied. Nonetheless, our results suggest that the relative virulence of the non-ST01 strains goes in line with their phylogenomic distance to the well-characterized NAP1/RT027/ST01 genotype.
Strain 5757-ST67 was second in terms of the damage inflicted in animal experiments. This is somewhat expectable, as ST-67 strains also belong to subgroup 2i and have been reported to cause severe CDI [
31]. Despite having a variant TcdB, 5757-ST67 produced an elevated inflammatory response and alteration of biochemical markers that could be explained by toxin overproduction [
32]. Moreover, mutations in
treR (trehalose repressor), CpG depleted genomes [
31] and a PaLoc rather similar to that of strain R202091-ST01 [
33] have been described in this ST, all of which are characteristics of hypervirulent
C. difficile strains.
The proinflammatory response and mortality levels recorded for strain 2811-ST41, also belonging to subgroup 2i, did not reach 5758-ST01 and 5757-ST67 levels. Furthermore, strain ICC45-ST41 produced weaker proinflammatory responses than 2811-ST41. Among multiple explanations, this difference could be tracked to the synthesis of a variant TcdB in the former [
34]. However, information on the amount of secreted TcdA and CDT by strains from this ST is still lacking, though it could clarify the observed differences in proinflammatory potential.
Strain 5809-ST252, from subgroup 2ii, presented the weakest proinflammatory response and lethal potential of the studied strains. We did not find reports in the literature on the pathogenesis of this type of C. difficile strains and therefore cannot provide explanations for its behavior.
Our results could be explained by undescribed differences in the strains´ exoproteomes and attributed to structural particularities on the bacterial surface that alter human intestinal homeostasis [
29]. Other factors, unexplored in this work, such as germination potential, colonization efficiency, the number of metabolically active
C. difficile cells in the intestine, and bile acids metabolism could explain the differences in the damage observed in the infection models [
35‐
37].
Contradictory observations on in vitro phenotypes expressed by isolates from the same ribotype, such as RT027, have been reported, particularly with regard to sporulation and toxin production [
16,
38,
39]. Therefore, in vivo studies, such as the work herein presented, have the potential to clarify pending issues and should be carried out with two or more animal models and including others STs from the “hypervirulent” Clade 2.
We observed indications of extraintestinal organic damage, mainly in renal and hepatic functions, as well as an increase in leukocytes, these alterations were recorded after 1 or 2 days of
C. difficile detection on feces. The animals infected with 5758-ST01 showed altered biochemical parameters sooner than those infected with non-ST01 strains (exposed for 3 additional days). As such, in an early infection, the general health condition is altered by 5758-ST01 infection but not by non-ST01 strains. These findings have been described in the severe outcome of CDI in humans [
40,
41], which suggest a severe pathology in animals infected by the 5758-ST01 strain.
We conclude that “hypervirulence” is not a widespread feature in C. difficile Clade 2 strains. This result lay the foundations for interpretation of future in vitro and genomic comparisons of these and other Clade 2 strains. Moreover, it highlights the importance of C. difficile typing and of targeted diagnosis of NAP1/027/ST01 strains.
Materials and methods
C. difficile strains
This study was done with five
C. difficile MLST Clade 2 strains that were isolated from stool samples of symptomatic patients. Briefly, samples were treated with 96% ethanol and inoculated onto Cefoxitin-Cycloserine-Fructose agar plates (Oxoid) that were later incubated in anaerobic chamber (90% N
2, 5% CO
2, 5% H
2). Colonies were identified phenotypically with the RapID 32A system (bioMerieux) and by PCR-amplification of the
tpi gene. These bacteria were previously analyzed by PFGE, an end-point PCR targeting different PaLoc fragments, and toxigenic culture following the previously reported protocols [
42]. All strains were cryopreserved at − 80° C and recovered through culturing on Brucella agar plates supplemented with vitamin K and laked horse blood under anaerobic conditions (Table
2).
Table 2
C. difficile MLST Clade 2 strains used in this study
5758 | 01 | NAP1 | tcdA+/tcdB+/cdtB+/tcdC+, del | COv |
2811 | 41 | New type | tcdA+/tcdB+/cdtB+/tcdC+ | C |
ICC45 | 41 | New type | tcdA+/tcdB+/cdtB+/tcdC+ | V |
5757 | 67 | NAP1 | tcdA+/tcdB+/cdtB+/tcdC+, del | VOv |
5809 | 252 | New type | tcdA+/tcdB+/cdtB+/tcdC+ | C |
Preparation of bacterial cell-free supernatants
All strains were grown in TYT-broth (3% tryptose, 2% yeast extract, 0.1% thioglycolate, and pH 6.8) for 24 h under anaerobic conditions. At this time point, viable bacterial cells counts for all strains were in the 10
7 CFU/mL order (Additional file
1: Fig. S3). Bacterial cells were then pelleted by centrifugation for 30 min at 20,000 g and the resulting supernatants were filtered using 0.2 µm pore membranes [
32]. These bacterial cell-free supernatants were used in the ileal loop assay in mice (see below).
Murine ileal loop model
Four to 5 weeks old, male, Hsd:ICR mice with a body weight of 18–25 g were used. These animals were grouped in polycarbonate cages and maintained under controlled conditions of temperature (19.9 ± 0.7 °C and 23.5 ± 1.0 °C), relative moisture (75 ± 5% and 89 ± 3%), noise (81.4 ± 2.2 dB); and photoperiod (12 h). Water and feed were available ad libitum. All animal proceedings were in compliance with local legislation (Ley de Bienestar de los Animales N° 7451) and approved by the Comité Institucional de Cuido y Uso de Animales (CICUA) from the Universidad de Costa Rica (CICUA 52-15). We used the maximum amount of animals authorized by CICUA based on the 3R’s principles.
Mice were fasted overnight and then anesthetized with ketamine (60 mg/kg) and xylazine (5 mg/kg). Through a midline laparotomy, a 4 cm ileal loop was ligated and injected with 0.3 ml of bacterial cell-free supernatants or control solutions (PBS, TYT-broth or non-toxigenic
C. difficile ATCC
® 700057 supernatant). Four hours after inoculation mice were euthanized and the length and weight of the intestinal loops were recorded [
43,
44].
Myeloperoxidase (MPO) assay
Neutrophils degranulation in homogenized ileal tissue was evaluated using a colorimetric MPO activity assay [
45]. Briefly, ileal tissue (100 mg) was homogenized in hexadecyltrimethylammonium bromide (HTAB) (Sigma) buffer (PBS, HTAB 50% w/v, and H
2O
2 0.1% v/v) and cleared by centrifugation at 4500 g for 7 min at 4 °C. The resulting supernatants were incubated with a 0.017%
o-dianisidine solution (Sigma) and after 5 min the absorbance was determined at 450 nm. Results were reported as MPO/100 mg of ileal tissue.
Detection of TNF-α, IL-1β, IL-6, and IL-10
The concentration of proinflammatory interleukins (IL-1β, IL-6, IL-10) and tumor necrosis factor alpha (TNF-α) in ileal tissue homogenates was determined with commercial enzyme-linked immunosorbent assays (ELISA) following the instructions of the manufacturer (R&D Systems).
Hamster infection model
Six to eight weeks old, male, Golden Syrian hamsters (
Mesocricetus auratus) with a body weight of 85–120 g were used. Similarly to previous studies [
46,
47], the hamsters were separated into groups of 5 animals.
The animals were maintained in polycarbonate cages under the conditions mentioned above for mice. Two clindamycin doses were administered to each individual through the oral route 10 (30 mg/kg) and 5 days (50 mg/kg) prior to the beginning of the experiment. On day 0, each animal was inoculated orally with 10
4 spores that were prepared as previously described [
48]. Seven groups of animals were used, including five treatments (five strains) and two negative controls (PBS or DMEM).
Animals were monitored for 15 days for infection signs, such as weight loss and wet tail (diarrhea), and to record their mortality. Stool samples were collected every 2 days throughout the experiment for isolation of
C. difficile following published protocols [
49]. When an experimental subject died, or at day 15 (when euthanasia was performed), a chirurgical resection of the ascendant colon and the cecum was performed to monitor colonization by
C. difficile.
Analysis of blood biomarkers
Additional groups of 5 Golden Syrian hamsters were inoculated with spores of each bacterial strain and negative controls (PBS or DMEM) as described above. All animals infected with 5758-ST01 strain were euthanized on day 3 (due to its rapid lethality rate associated with this strain) and all other groups on day 6 post-spore inoculation. The criteria in this matter was that euthanasia (and subsequent blood collection) should be performed 1 or 2 days after successful colonization or infection by C. difficile (as demonstrated by a positive fecal C. difficile culture).
Then, a portion of whole blood was collected after decapitation to measure serum levels of albumin, creatinine, blood urea nitrogen (BUN), sodium, chloride, lactate dehydrogenase (LDH), alanine amino-transferase (ALT), and aspartate amino-transferase (AST) levels with an automated analyzer (Roche Cobas® 3c11). Another portion of total blood was collected in microtubes with EDTA to obtain white blood cells (WBC) differential counts through visualization of Wright-stained blood smears.
Histopathologic assessment of tissue sections
Murine ileal and hamsters’ cecum samples were fixed in 10% buffered formalin, processed, and stained with hematoxylin and eosin (H&E) for histopathological evaluation. These preparations were evaluated for epithelial damage, edema, and neutrophil infiltration using a histopathological scoring (HS) system that ranged from 0 (absence of alterations) to 4 (severe) [
50]. All histopathological assays were performed by an expert DVM pathologist in a single blind setting.
Phylogenomic analyses
Whole genomes were reconstructed from 2 × 250 bp paired-end Illumina reads from the NCBI BioProject PRJNA293889. The genomic relatedness of the non-ST01 isolates to 5757-ST01 strain was determined using Mash [
51], which is based on a MinHash dimensionality-reduction technique, and through calculation of average nucleotide identity (ANI) values with an alignment-free method [
52].
Statistical analyses
Data from the animal models are presented as mean ± standard error (SEM) or as medians. Means and median were compared using one-way ANOVA tests with Bonferroni correction or Kruskall-Wallis tests followed by Dunn’s multiple comparison tests with Benjamini–Hochberg adjustment, respectively. Mortality was evaluated using Kaplan–Meier curves. P values < 0.05 were considered statistically significant.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.