Abstract
Clostridium perfringens is an anaerobic toxin-producing bacterium associated with intestinal diseases, particularly in neonatal humans and animals. Infant gut microbiome studies have recently indicated a link between C. perfringens and the preterm infant disease necrotizing enterocolitis (NEC), with specific NEC cases associated with overabundant C. perfringens termed C. perfringens-associated NEC (CPA-NEC). In the present study, we carried out whole-genome sequencing of 272 C. perfringens isolates from 70 infants across 5 hospitals in the United Kingdom. In this retrospective analysis, we performed in-depth genomic analyses (virulence profiling, strain tracking and plasmid analysis) and experimentally characterized pathogenic traits of 31 strains, including 4 from CPA-NEC patients. We found that the gene encoding toxin perfringolysin O, pfoA, was largely deficient in a human-derived hypovirulent lineage, as well as certain colonization factors, in contrast to typical pfoA-encoding virulent lineages. We determined that infant-associated pfoA+ strains caused significantly more cellular damage than pfoA− strains in vitro, and further confirmed this virulence trait in vivo using an oral-challenge C57BL/6 murine model. These findings suggest both the importance of pfoA+ C. perfringens as a gut pathogen in preterm infants and areas for further investigation, including potential intervention and therapeutic strategies.
Similar content being viewed by others
Main
C. perfringens has been frequently linked to various intestinal diseases in both humans and animals, particularly in neonates1. It is known to secrete >20 toxins, including several pore-forming toxins, for example, β-toxin, perfringolysin O (PFO), NetB and C. perfringens enterotoxin (CPE), alongside encoding other virulence factors associated with different facets of pathophysiology2,3,4,5,6. Although whole-genome sequencing (WGS) has advanced our understanding of adult-associated C. perfringens, gastroenteritis-associated strains, association studies with other intestinal diseases in younger age groups, including carriage in healthy individuals, are very limited7,8,9.
NEC, the most severe and lethal neonatal gastrointestinal (GI) emergency worldwide, is an acquired inflammatory gut disease that manifests as tissue necrosis in the GI tract10,11. NEC typically affects 5–15% of very-low-birthweight (<1,500 g) preterm infants, leading to high mortality rates (~40%) and severe longer-term complications12,13,14,15. Clinical observations support a microbial role in NEC, including gas cyst formation in the gut wall (pneumatosis intestinalis), and antimicrobial treatment often resolves symptoms16,17,18. Several pathogens including Klebsiella spp. and C. perfringens have been linked to clinical NEC19,20, with microbiota-based, preterm infant cohort studies implicating overabundance before NEC21,22. C. perfringens has been consistently reported as a potential pathogenic agent associated with NEC1,23,24, with CPA-NEC symptoms often more fulminant and severe compared with ‘classic’ NEC19,25,26.
In the present study, we describe an extensive genomic study of C. perfringens clinical preterm isolates from non-CPA-NEC (95%) and CPA-NEC patients (5%). We probed virulence factors and plasmids associated with infant/CPA-NEC C. perfringens strains and traced putative hospital transmission. Additional experiments on a selection of C. perfringens strains and an oral-challenge infection model allowed us to assay key virulence and pathology readouts.
Results
Eight distinct phylogenetic lineages of C. perfringens
We isolated and performed WGS on 272 C. perfringens isolates from infant stool samples collected longitudinally from 70 individuals (including 4 CPA-NEC infants), obtained from 5 UK neonatal intensive care units (NICUs) (and at-home collections) between 2011 and 2016 (Supplementary Table 1), which indicated an ~24% incidence. A strain-level, core-genome, global phylogenetic tree (isolates and metagenome-assembled genomes; Fig. 1a) clustered into eight major lineages (Fig. 1b,c). The relatively distant lineage VIII consisted of mostly typical human gastroenteritis isolates, whereas lineage VII included mixed human and animal sources, indicating a potential zoonotic signature. Lineage V comprised exclusively human-derived strains (Fig. 1d), with more than half obtained from infants and hospital linked.
a, Mid-point rooted, maximum-likelihood phylogenetic tree of 302 representative C. perfringens strains (derived from a collection of 673 C. perfringens genomes) inferred from 8,506 SNPs (from 858 single-copy core genes), aligned with metadata including genome types, sources, sample hosts, hospitals and lineages assigned using hierBAPS. Black dashed lines, type strain ATCC13124 and several representative strains were labelled in this phylogenetic tree. Branch lengths are indicative of the estimated nucleotide substitution per site (SNPs). MAG, metagenome-assembled genomes (from MGnify study, human gut microbiome-derived C. perfringens genomes only). b, Unrooted phylogenetic tree coloured by assigned lineages. c, Pairwise SNP distances among strains located within lineages I (n = 8), II (n = 7), III (n = 11), IV (n = 28), V (n = 57), VI (n = 66), VII (n = 98) and VIII (n = 27) and between lineages. The box in the boxplot represents 50% of the central data, in between the lower and upper quartiles, with a central line representing the median, whereas the whiskers show the most extreme data points. BL, between lineages. d, Proportions of human and animal-derived C. perfringens genomes in each lineage.
Genomic investigation reveals a hypovirulent lineage
Genome-based, strain-level virulence profiling was performed on all 80 infant-associated C. perfringens representative strains (derived from original 272, including reference strains; Fig. 2a). Toxinome analysis indicated that lineage III encoded more toxin genes (n = 9) than other lineages, whereas lineage V had a reduced presence of the pore-forming toxin gene pfoA (Fig. 2b) and mostly consisted of neonatal isolates (Fig. 2c). Overall, C. perfringens strains from lineages III, VI and VII carried significantly more virulence genes versus strains from lineages I and V (Fig. 2d). Significantly fewer toxin genes including pfoA and cpb2 were encoded within lineage V versus other lineages (Fig. 2e), with colonization-associated genes sialidases (nan), hyaluronidase (nag) and adhesin (cna) overrepresented in lineages III, VI and VII when compared with lineages I and V (Extended Data Fig. 1a).
a, A mid-point-rooted, maximum-likelihood phylogenetic tree of 80 infant-associated C. perfringens representative strains, derived from 272 C. perfringens isolate genomes using an ANI cutoff of 99.9% (strain-level) for dereplication of genomes. This tree was inferred from 175,479 core SNPs (in 1,995 single-copy core genes) aligned with clinical metadata and virulence profiles of each strain including toxin genes, colonization factors and antimicrobial resistance genes, in addition to virulence gene counts and genome sizes, as indicated. Branches are colour coded according to the lineages assigned in Fig. 1. AMR, antimicrobial resistance. b, Toxinome profiles of each lineage. c, Sample source proportions in each lineage. d, Virulence count (per strain) comparison across lineages. Significance was pairwise compared. The box in the boxplot represents 50% of the central data, in between the lower and upper quartiles, with a central line representing the median, whereas the whiskers show the most extreme data points. There was no difference between lineages I and V, whereas lineages III, VI and VII (no difference among these three lineages) are significantly different from lineages I and V. ****P < 0.0001. Statistical analysis was performed using Kruskal–Wallis test and post-hoc Dunn’s test. e, Proportion of each toxin gene across all lineages. The significance was pairwise compared within individual toxin genes. The significant difference was indicated in cpb2 genes across lineages V, VI and VII, and pairwise comparison indicated that lineage V has reduced proportionally in cpb2 genes compared with lineages VI and VII. ***P = 0.003, ****P < 0.0001, whereas proportion of pfoA gene is significantly reduced in lineage V versus all other lineages: *P < 0.05. f, Full toxin gene profiles of five CPA-NEC C. perfringens strains. g, Toxin gene proportion in isolates within global lineage V (which includes MAGs from human microbiome project) and infant-associated lineage V isolates. NS, non-significant (Note: lineages II, IV and VIII were not included in statistical investigations due to these lineages only encompassing singletons.) In e and g, statistical analysis was performed using Fisher’s test (two sided).
CPA-NEC strains (n = 5) nested within lineages I, IV, VI and VII, respectively (Fig. 2f), and encoded similar toxinome profiles. No CPA-NEC strains nested within lineage V, which predominantly contained strains with fewer virulence potentials (including apparent lack of pfoA), indicative of a potentially hypovirulent or commensal-like lineage (Fig. 2g).
SNP analysis indicates potential strain dissemination
We investigated circulation of C. perfringens strains based on 210 longitudinal isolates (≥2 sampling time points) obtained from 31 infants from 3 hospitals (Fig. 3a,b). Strain-level SNP distance was determined to be ~0.2 SNPs, whereas between-strain SNP distance was estimated at ~544 SNPs (Fig. 3c).
a, Number of C. perfringens isolates sequenced from five hospitals. b, Number of infants with two or more longitudinal sampling time points. c, Pairwise SNP distance in longitudinally sampled isolates (n = 209): strains (ANI > 99.9%) versus between strains (ANI < 99.9%). M, mean (SNPs). The statistical analysis was performed using Wilcoxon’s test (two sided). The box in the boxplot represents 50% of the central data, in between the lower and upper quartiles, with a central line representing the median, whereas the whiskers show the most extreme data points. d, Phylogenetics of 272 infant-associated C. perfringens isolates with branches colour coded according to the lineage assigned. The key strains, defined as isolates having highly similar ANI > 99.9%, as indicated on the tree, are further described in e–g. e, Tracking multi-host C. perfringens strain VIf (31 isolates) which involves six individuals in two sister hospitals, A and B, based on clinical metadata. Twin pairs are colour coded. A CPA-NEC infant (N20) was highlighted in red. f, Putative intrahospital circulation of strain Ib. Strain Ib was not detected in a 60-d window in the infant cohort. A CPA-NEC infant (N3) was highlighted in red. g, Comparison of SNP distance trees between isolates from twin pairs, suggesting dissemination of C. perfringens.
Sublineage VIf represented a clonal cluster consisting of 31 identical isolates at core-SNP level (0 SNPs) obtained from 6 infants residing in sister hospitals A and B (Fig. 3d,e). Clinical metadata linkage indicated that the index case of strain VIf was in infant N18 in hospital A. Later, this strain was detected in several individuals in hospital A (N17, N18 and N20), one of whom (N20) was diagnosed with CPA-NEC. Infant N20, who was also diagnosed with a condition predisposing to GI mucosal hypoxaemia, subsequently succumbed to NEC 10 d after strain VIf was first detected. N17 and N18 (index case for strain VIf) had an episode of possible NEC which led to administration of antibiotics (broad-range β-lactam + vancomycin) for 2 d, with strain VIf first detected ~8 d after antibiotic administration finished in N18. Strain VIf was also detected in individuals residing in hospital B (across the same 100-d window). This suggests potential spread of the C. perfringens strain between hospitals, whether transferred on/in patients, their parents or equipment or healthcare staff. We also observed that strain Ib in the index case N3 (who succumbed on day 27; Fig. 3f) was again detected in N10 (after a 60-d gap), suggesting possible horizontal transmission and persistence of C. perfringens strains, potentially in the NICU environment. Additional transmissions events were also implicated in four twin pairs carrying identical strains (Fig. 3g).
Virulence plasmid carriage among infant-associated isolates
We next investigated plasmid acquisition of two key C. perfringens (virulence) conjugative plasmid families, pCW3 (ref. 27) and pCP13 (ref. 28) (Fig. 4a). Plasmid pCW3 (with tcp system) encoded toxin genes such as cpb2 and cpe, iota binary toxin genes iap and ibp and tetracycline resistance determinants tetA(P) and tetB(P) (Fig. 4b). The adhesin gene cnaC was also detected in most pCW3 plasmids, potentially enhancing colonization capacity. NEC-associated isolates from sublineage VIf primarily carried pCW3 plasmids in either ~50-kb or ~70-kb sizes, encoding tetA(P) and tetB(P), cnaC and toxin gene cpb2, with the 50-kb plasmids missing ~20 genes due to transposases encoded by CPR_0630 on both flanks. Plasmid VIf, encoded by strain VIf, was not detected in strains from other lineages apart from a lineage V strain IQ074 (obtained from infant N10; Fig. 4c), indicating potential plasmid transmissible events between C. perfringens strains, because the NICU stay of N10 overlapped with infants carrying VIf strains (Fig. 4d).
a, Carriage/distribution of pCW3 and pCP13 conjugative virulence plasmids among infant-associated C. perfringens strains indicated in the infant C. perfringens phylogeny. b, Plasmid-coding sequences mapped to reference plasmid sequence (pLH112) to present a comparative overview of pCW3 plasmids computationally extracted from infant-associated C. perfringens isolates (n = 85). Data/sequences were aligned with a newly constructed plasmid phylogeny. c, SNP distance tree of plasmids from strain VIf (from seven individuals). d, Probable plasmid circulation map of plasmid VIf (identical plasmids encoded in strains VIf). e, Plasmid-coding sequences mapped to reference plasmid sequence (pLH112) to present a comparative overview of pCP13 plasmids computationally extracted from infant-associated C. perfringens isolates (n = 111). Data/sequences were aligned with a newly constructed plasmid phylogeny. f, SNP distance tree of identical plasmids encoded in nine isolates obtained from four individuals. g, Isolates’ positions from the four individuals indicated in f within the infant-associated C. perfringens phylogenetic tree. These four individuals harboured C. perfringens isolates (n = 9; from different hospitals) that encoded identical pCP13 conjugative virulence plasmids nested within distinct sublineages of lineage V. h, Proportions of conjugative virulence plasmids encoded by C. perfringens strains across lineages. Proportion of strain-level virulence plasmid carriage was significantly lower in lineage V compared with lineage VII (****P < 0.0001). Statistical analysis was performed using Fisher’s test (two sided). i, Proportions of infant-associated C. perfringens strains carrying one or more conjugative virulence plasmid(s), pCW3 and/or pCP13: a comparison across lineage V (n = 36), CPA-NEC strains (n = 5) and non-CPA-NEC strains (n = 73). Statistical analysis was performed using Fisher’s test (two sided). The proportional difference was not found to be significant.
The pCP13 plasmids (~45–60 kb) in our dataset harboured cpb2, becAB and edpA toxin genes29. Adhesin cnaB was detected in all extracted pCP13 plasmids. Importantly, sublineage IIIb strains carried pCP13 plasmids (54,591 bp) which encoded becAB binary toxin genes, previously associated with human gastroenteritis, with further analysis confirming identical plasmids from three individual infants (Fig. 4e and Extended Data Fig. 1b).
Another pCP13 plasmid (~62 kb) was detected in nine C. perfringens identical isolates (Fig. 4f), obtained from four individuals, that nested in three distinct sublineages in lineage V, highlighting the common transferable nature of conjugative plasmids among C. perfringens strains (Fig. 4g). Both plasmids pCW13 and pCP13 were less prevalent in C. perfringens strains in lineage V (19%); in contrast plasmid carriage in other lineages ranged from 50% to 86% (Fig. 4h). Plasmid carriage in CPA-NEC strains was 80% versus non-CPA-NEC strains at 44% (Fig. 4i).
Toxicity assays links toxin gene pfoA to epithelial damage
From the genomic data, it was apparent that C. perfringens strains from lineage V encoded significantly fewer virulence factors, most clearly evidenced by the absence of the pfoA toxin gene. We selected 30 infant-associated C. perfringens strains from all lineages, including pfoA+ and pfoA− strains for in vitro tests (Fig. 5a).
a, Schematic of strain selection (n = 31). b, Viability of C. perfringens strains (n = 31) exposed to ambient air over 14 d (data stratified by lineage). A lineage II strain was undetectable after 10 d. Data are the mean. c, Sporulation capacity of C. perfringens strains (n = 28) assessed using ethanol pre-treatment (4 h) to eliminate vegetative cells before culturing, with n = 3 biologically independent samples for each strain. The statistical significance was pairwise compared across the groups. LD, limit of detection. d, Toxicity of sterile-filtered supernatants (n = 31) determined by 2-h co-culturing with colonic cell line Caco-2 (n = 6 in two independent experiments). Cell death comparison was between lineages. The significance was compared with the untreated group. e, Cell toxicity of C. perfringens strains compared between pfoA+ (n = 15) and pfoA− groups (n = 15). The pfoA+ strains encode pfoA toxin gene; the pfoA− strains do not encode pfoA. The significance was compared with the untreated group, and pairwise between pfoA+ and pfoA− groups as indicated. f, Cell death percentage of individual strains aligned with toxin gene profiles. The presence of both toxin genes pfoA (rs = 0.823) and cpb2 (rs = 0.453) was positively correlated with the increase in Caco-2 cell death (P < 0.05). A correlation analysis was performed using the point-biserial-correlation test (Spearman’s rank correlation). The significance was compared with the untreated group: ***P < 0.001, **P < 0.01, *P < 0.05. g, Cell death comparison between pfoA+ strains (n = 7) and pfoA+cpb2+ strains (n = 9) showing potential synergistic effects of cpb2 toxin gene alongside pfoA. h, Cell death comparison between CPA-NEC pfoA+ strains (n = 4) and non-CPA-NEC pfoA+ strains (n = 26). i, Images of C. perfringens colonies (10 μl of confluent cultures per spot) on CBA after 24 h of anaerobic incubation. All representative pfoA+ strains (n = 6) exhibit complete haemolysis on CBA whereas pfoA− strains (n = 6) display partial haemolysis. In c–h, the box in the boxplot represents 50% of the central data, in between the lower and upper quartiles, with a central line representing the median, whereas the whiskers show the most extreme data points. All statistical analyses were performed using the Kruskal–Wallis test and post-hoc Dunn’s test of multiple comparisons, with P values adjusted using the Benjamini–Hochberg method, unless otherwise stated. Significance was tested against the untreated group, otherwise significance brackets were used in pairwise comparison.
Survivability/transmissibility assays indicated that lineage VII strains appeared to be most oxygen tolerant (Fig. 5b), surviving significantly better than strains from lineage V from 48 h to 192 h (Extended Data Fig. 2a), although C. perfringens strains from lineage VI produced significantly more spores than strains nested in lineage V (Fig. 5c).
C. perfringens-associated cellular toxicity assays showed supernatants of strains in lineages I, III, VI and VII inducing significant cell death in epithelial monolayers, whereas lineage V strains did not (Fig. 5d). Grouping by presence of the pfoA pore-forming toxin gene indicated a clear cytoxicity signal, whereas cell death induced by pfoA− strains was no different statistically from untreated controls (Fig. 5e). Cellular toxicity phenotypes correlated with the presence of pfoA, and to a lesser extent toxin gene cpb2 (Fig. 5f), with a potential synergistic effect observed in strains encoding both toxins (Fig. 5g). It is interesting that pfoA+ and CPA-NEC strains also exhibited significantly higher cell toxicity versus non-CPA-NEC strains (Fig. 5h). All pfoA+ strains also exhibited complete haemolysis on blood agar, whereas pfoA− strains displayed partial haemolysis (Fig. 5i). One pfoA− strain (LH019) induced higher cell death (Fig. 5f), probably due to the presence of alveolysin (alv) which shares ~86% homology (nucleotide identity) with pfoA30.
C. perfringens strains induce murine colonic histopathology
We developed an oral-challenge murine model to explore pathology related to pfoA using six C. perfringens (pfoA+/pfoA−) strains (Fig. 6a). Colonization rates indicated that most strains were completely cleared by day 4 (Fig. 6b and Extended Data Fig. 2b), although microbiota genus-level diversity profiles between pfoA+ and pfoA− groups did not differ statistically (Fig. 6c and Extended Data Fig. 2c,d). Differences in daily weight changes between groups were not observed (Fig. 6d); however, overall weight change indicated that the pfoA+ groups had significantly less weight gain versus controls (Fig. 6e).
This experiment was based on eight groups of mice (n = 5 for each group) including control (untreated) and antibiotic groups, and six experimental groups (three groups of pfoA+ strains and three groups of pfoA− strains). a, Schematic of in vivo infection model experimental design. PBS, phosphate-buffered saline. b, Intestinal colonization of C. perfringens monitored over a 7-d post-challenge pfoA+ (n = 15) versus pfoA− (n = 15). c, Inverse Simpson’s index: gut microbiome diversity of mice on days 0, 2 and 6, respectively, showing no significant genus-level differences when comparing groups pfoA+ (n = 15) and pfoA− (n = 15) to antibiotics (n = 5). The statistical analysis was performed with ANOVA and Tukey’s test (post-hoc pairwise). NS, non-significant (P > 0.05). d, Daily weight changes (%) of mice over experimental period across groups. Data are mean ± s.e.m. *P = 0.034. e, Total average weight changes from day 0 to day 6. The pfoA+ group (n = 15) has a significantly lower mean weight change compared with the control group (P < 0.05). The statistical analysis was performed with ANOVA and Tukey’s test (post-hoc pairwise). f, Histopathological scores for intestinal inflammation of murine distal colons (day 7 post-challenge), comparison between groups. Scoring was based on two H&E-stained images per distal colon. Increments in score value denote increasing severity. g, Representative H&E-stained murine distal colonic sections of control, antibiotics and experimental groups (pfoA+ and pfoA−), respectively, showing epithelial changes. The green arrow shows erosion and crypt hyperplasia, the black arrow loss of goblet cells and the red arrow immune cell infiltration. Scale bars, 50 µm. h, Distal colonic crypt length comparison. Measurement was performed based on 20 crypts per representative image. Data are mean ± s.e.m. Normal range for distal colonic crypt length is 133–175 µm. i, Colon weight (whole-colon) comparison across all groups. Data are mean ± s.e.m. In b, c, e and f, the box in the boxplot (where applicable) represents 50% of the central data, in between the lower and upper quartiles, with a central line representing the median, whereas the whiskers show the most extreme data points. All statistical analyses were performed using the Kruskal–Wallis test and a post-hoc Dunn’s test of multiple comparisons and P values adjusted using the Benjamini–Hochberg method, unless otherwise stated. The significance was tested against the control group; otherwise significance brackets were used in pairwise comparison.
Both C. perfringens-infected groups (pfoA+ and pfoA−) exhibited significantly higher histopathological scores for intestinal inflammation compared with control groups, with colonic sections from pfoA+ groups displaying more severe damage/pathology versus pfoA− groups (Fig. 6f). Distal colonic sections of both infected groups demonstrated similar pathological features, including marked goblet cell loss and inflammatory cell infiltrate (Fig. 6g and Extended Data Fig. 3), with pfoA+ groups having significantly longer crypt lengths versus pfoA− and control groups (Fig. 6h), although not when comparing colon weights (Fig. 6i). Immune profiling of colonic tissue showed significantly increased levels of tumour necrosis factor (TNF)-α, KC/GRO and interleukin (IL)-1β in the pfoA+ group versus controls (Extended Data Fig. 4) and, although not significant, elevated levels of IL-22 and IL-6 were observed.
Discussion
Genomic analysis of preterm infant-associated C. perfringens strains suggests distinct lineages that comprise potentially hypovirulent strains versus typical virulent strains linked with possible NICU circulation and carriage of virulence-carrying mobile elements, including CPA-NEC strains. Certain genotype-linked phenotypical traits, for example, pfoA+-associated cytotoxicity, appear to correlate with increased intestinal cell pathology, which may be associated with infection/disease outcomes.
Currently, public genome databases are biased towards diseased-associated isolates, with only a limited number of C. perfringens genomes obtained from healthy individuals (~5%)31,32. C. perfringens is also considered as a normal gut microbiota member33,34 and we observed a potentially hypovirulent or ‘commensal-like’, human-derived monophyletic lineage (V). Strains encoded significantly fewer virulence factors and plasmids and were thus distinct from more typically virulent C. perfringens, including CPA-NEC strains. Indeed, potentially hypovirulent lineage V strains may not encode the necessary virulence traits required to cause overt disease such as severe intestinal colitis or CPA-NEC.
A genomic signature apparently enriched in typical, virulent, infant-associated C. perfringens isolates (including CPA-NEC strains) was the presence of the gene pfoA that transcribes the pore-forming haemolytic toxin PFO. We consistently observed a direct correlation between the presence of pfoA and significantly enhanced cell toxicity in intestinal cell lines and complete haemolysis, when compared with pfoA− strains regardless of lineage (Fig. 5e,i). PFO (θ-toxin), a typical cholesterol-dependent cytolysin, is known as a pore-forming toxin35, which has been heavily associated with the pathogenesis of myonecrosis (gas gangrene), haemorrhagic enteritis in calves and more recently septicaemia (intravascular haemolysis) in humans, including neonates36,37,38,39,40. PFO can act synergistically with α-toxin in the pathology of gas gangrene, which is a disease that shares a high degree of symptom similarity with CPA-NEC in preterm infants, and has also been called ‘intestinal gas gangrene’19,41,42. The haemolysis rates of C. perfringens septicaemia clinical isolates were shown to strongly correlate with PFO expression alone, regardless of α-toxin production (whereas a pfoA-deficient strain did not induce haemolysis), with proinflammatory cytokine production (TNF-α, interleukin IL-5 and IL-6) significantly stronger than α-toxin40. Nevertheless, the mechanistic role of PFO in disease development is not completely defined, although previous studies have indicated additional modulation of host responses36,43,44,45. Indeed, our in vivo studies suggest a high degree of colonic immune cell infiltration and a higher proinflammatory cytokine milieu (and associated pathology) during pfoA+ C. perfringens infection. Although previous studies have indicated that the pfoA toxin gene is universally encoded in C. perfringens, aside from gastroenteritis-associated toxinotype F strains (nested in lineage VIII in Fig. 1a), we did not observe this in lineage V strains8,9,46,47. This prior assumption may be due to skewing in public databases for disease-associated isolates, so our data highlight the important role of this toxin in disease involvement, including CPA-NEC.
A concern with typical virulent pfoA+ C. perfringens strains is nosocomial transmission in at-risk populations, which is a huge issue in related Clostridioides difficile48,49. We observed a circulating strain VIf, which was detected in six individuals during their NICU stay, and seemingly circulated between two sister hospitals that regularly exchange patients. Transmission of C. perfringens has also been demonstrated in similar confined settings such as within elderly care homes (more vulnerable communities), where non-foodborne outbreaks of toxigenic C. perfringens take place frequently8,50. Notably, the index case of strain VIf was detected in an individual who had been treated with broad-spectrum antibiotics, suggesting that creation of a more ‘favourable’ niche may have allowed C. perfringens to flourish in the microbiota-depleted gut. This strain (IQ129) also nested within lineage VI linking to a stronger spore-forming capacity (Fig. 5c), a trait that may provide protection against antibiotics, as demonstrated by its detection 8 d after termination of antibiotic treatment. Although six infants harboured strain VIf, only N20 developed CPA-NEC and ultimately succumbed to the infection. This preterm infant had been previously diagnosed with congenital heart disease which may have caused reduced blood flow and oxygen supply to the gut, potentially altering the gut microbiota and barrier permeability that may favour the growth of certain virulent C. perfringens strains51,52,53. Further studies exploring wider microbiome dynamics longitudinally in preterm infant cohorts, complemented by in vitro studies (for example, model colon systems), are required to tease apart potential mechanisms.
The widespread dissemination of certain strains may be partly attributed to several traits. Typical pfoA+ C. perfringens strains (including those associated with CPA-NEC) appeared to be more oxygen tolerant and have enhanced spore-forming/germination ability, which may facilitate spread between infants and NICUs. This may include resilience against standard disinfectant practices, thus enabling persistence in even adverse ‘sterile’ hospital environments, and subsequent germination once spores enter the infant gut54,55. These traits pose a significant challenge for infection control measures, especially those wards where extremely low-birth-weight, preterm infants reside who have a heightened risk of developing CPA-NEC56. The presence of circulating strains also indicates that routine genomic surveillance of C. perfringens may be helpful in NICUs to monitor and prevent circulation of virulent strains.
Certain plasmids (for example, pCW3 and pCP13) have long been recognized as key mobile genetic elements for transfer of virulence genes in C. perfringens27,28. We observed nine isolates from three distinctive sublineages in lineage V carrying identical pCP13 conjugative plasmids, suggesting frequent conjugative plasmid transfer28. Alongside virulence factors, certain colonization factors are encoded on these plasmids such as adhesin genes cnaB and cnaC27, with collagen adhesin critical in pathogenesis by conferring additional cell-binding ability in host strains57. Notably, adhesin gene cnaC, also required for conjugative transfer, has previously been correlated with virulent poultry-NE C. perfringens strains, whereas a pCP13 plasmid that encodes cnaB and becAB was recently linked to acute gastroenteritis outbreaks6,27,30,58. Carriage of these plasmids was found to be more common in all lineages except hypovirulent lineage V, suggesting that plasmid-encoded virulence traits represent key genomic signatures linked to pathophysiology. This potential plasmid circulation links with recent findings that phylogenetically distinct type-F C. perfringens isolates were found to harbour identical CPE-encoding plasmids in a single gastroenteritis outbreak8. However, as our results are based on short-read sequence data (albeit at very high sequencing coverage), further studies using long-read-based sequencing with plasmid reconstruction/characterization are required to probe these findings in detail.
We identified pfoA as the major toxin gene positively associated with intestinal cell injury, whereas pfoA+ CPA-NEC strains also correlated with significantly higher cell death compared with pfoA+ non-CPA-NEC strains. Notably, overt infection and subsequent production of virulence factors in many cases will be prevented by gut microbiota colonization resistance mechanisms. However, the preterm infant gut microbiota is known to be significantly disrupted compared with full-term infants (in which NEC is extremely rare), which may allow overgrowth of C. perfringens1. The use of probiotics has been proposed as a prophylactic strategy to beneficially modulate the preterm gut microbiota and reduce incidence of NEC, and previous studies have indicated that supplementation with the early life microbiota genus Bifidobacterium spp. correlated with a reduction in Clostridium spp. and associated NEC59,60.
We added an antibiotic regimen (five-antibiotic cocktail + clindamycin) to deplete the murine microbiota, and more closely mimic the clinical scenario in preterm neonates, who are routinely exposed to empirical antibiotic therapy59. Distinctive disease symptoms were detected: pfoA+ strains induced moderate colonic inflammation versus milder colonic inflammation observed in the pfoA− groups, suggesting that PFO may act as a major virulence factor linked to intestinal pathology. The pathological changes observed in distal colonic sections also correlated with diarrhoea, which is similar to Citrobacter rodentium infection, a model for human enteropathogenic Escherichia coli61,62. It also appeared that immune-mediated pathology plays a role during pfoA+ C. perfringens infection, as evidenced by mucosal erosions, goblet cell reduction, immune cell infiltration and increased levels of the proinflammatory cytokine TNF-α, when compared with pfoA−-infected animals. This implicates the importance of the potential immune-driven aspects of C. perfringens intestinal infection, which warrants further investigation. Although pfoA+ CPA-NEC strains were used in this infection model, we did not observe severe CPA-NEC symptoms including complete intestinal necrosis and abdominal distension. This is probably due to differences between mice and humans and factors such as diet. The standard chow diet may not favour this protein-hungry pathogen, the use of adolescent mice (rather than neonatal mice) may lead to more efficient C. perfringens clearance through immune-mediated pathways, and murine rather than preterm gut microbiota may have impacted infection kinetics. Further optimization studies, including the use of ‘humanized’ mouse models, may allow development of a more clinically relevant CPA-NEC model63.
There are certain limitations associated with the present study. First, only five CPA-NEC strains from four CPA-NEC patients from two sister hospitals were included. Given the very rare occurrence of CPA-NEC (subset of NEC cases), we had access to a limited number of patients (including issues/restrictions related to the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic), and we were unable to reconstruct any additional C. perfringens metagenome-assembled genomes (MAGs) from previously published preterm infant microbiome studies, which was largely due to lack of sequencing depth64. Within our dataset/cohort we had ~6% NEC incidence. A rate of between 5% and 15% NEC is often reported, but its incidence varies widely between NICUs, with NEC cases often reported in ‘outbreaks’ that may link to emergence of a specific virulent strain in the hospital environment, a hypothesis supported by our SNP analysis60,65. This is particularly important, because previous terminology for the disease ‘intestinal gas gangrene’ signals the more severe and fulminant condition of CPA-NEC, compared with classic NEC, often leading to surgery within 24 h and associated higher mortality rates (78%)19,66. Larger (and longer) surveillance studies that incorporate numerous NICUs across a wider geographical region would be required to capture a larger number of samples from CPA-NEC-diagnosed infants. We also fully recognize NEC as a multifactorial intestinal disease that has been associated with several bacteria including C. perfringens and Klebsiella and Enterococcus spp.21,22,67. Indeed, two preterm neonates had previously been diagnosed with Klebsiella-associated NEC in the previous cohort study22 and, although they each harboured C. perfringens strains, these isolates exhibited a ‘commensal’ genomic signature (within hypovirulent lineage V and did not encode pfoA). We used Caco-2 cells to examine whether the secreted toxins of C. perfringens caused epithelial damage. We appreciate the usual caveats of using an immortal cell line and over-extrapolation of results; however, our further strain testing in vivo does strengthen our findings in terms of cell death and links to pathology. Future work is required to pinpoint the mechanistic role of the virulence factors identified, such as pfoA, potentially in a more clinically relevant, preterm infant enteroid model68 and/or using neonatal mice, although ethical and regulatory considerations limit these types of studies.
Overall, we have identified a potentially hypovirulent lineage of C. perfringens, characterized by lack of the pfoA toxin gene, colonization factors and survival-related capacities, in contrast to the typical pfoA+ virulent lineages of C. perfringens (including CPA-NEC strains), which linked to in vitro and in vivo pathological traits. Potential dissemination of virulent C. perfringens strains was observed between and within preterm infants and hospitals (including a case of CPA-NEC), highlighting the potential value of routine surveillance and enhanced infection control measures.
Methods
Clinical samples and bacterial isolation work
We conducted a retrospective genomic analysis on C. perfringens isolates obtained from faecal samples of 70 neonatal patients admitted to NICUs at hospitals A, B, C, D and E, respectively, in the UK between February 2011 and March 2016 (Supplementary Table 2). All cohort data and sequenced microbiota profiles can be accessed in the original clinical study publications; most samples were sequenced via 16S ribosomal RNA or metagenomic sequencing as per the original studies/publications in which the colonization dynamics had already been reported and discussed22,59,69. Dates of hospital admission and transfers were extracted electronically. Faecal samples collected were stored at −80 °C in a freezer before the experimental process to isolate C. perfringens using the ethanol-shock method (50% ethanol in Robertson’s cooked meat medium), followed by plating on fastidious anaerobic agar supplemented with defibrinated sheep blood and 0.1% sodium taurocholate; alternatively, faecal samples were plated directly on tryptose–sulfite–cycloserine egg yolk agar (TSC-EYA) before 37 °C anaerobic incubation for 18–24 h (ref. 70). Multiple or single distinct colonies on the plates were purified and maintained as pure isolates in autoclaved brain–heart infusion (BHI) broth with 30% glycerol for cryopreservation at −80 °C. Subjects without C. perfringens were defined as C. perfringens negative by culturing. We predefined CPA-NEC in the present study as definite NEC diagnosis (Bell stage II/III) with overabundant C. perfringens before NEC onset (Bell’s staging classification system: Bell stage II/III is considered to be definite NEC whereas Bell stage I covers non-specific signs) as described in Sim et al.22. This was a 2-year microbiota study that involved a cohort of >350 infants, with 4 CPA-NEC cases who showed statistically enriched C. perfringens before NEC onset in longitudinal samples (by comparing with samples from aged-matched non-NEC preterm infants), although CPA-NEC strains were the isolates collected at the final time points before disease onset.
Genomic DNA extraction and WGS
Pure isolates were cultured anaerobically overnight at 37 °C in BHI broth (~10–15 h) for genomic DNA extraction using either the phenol–chloroform extraction method or FastDNA SPIN Kit for soil according to the manufacturer’s instructions (hospital E samples only; MP Biomedicals). For the phenol–chloroform extraction method71, briefly, overnight 10 ml of pure cultures in BHI was harvested, followed by resuspension of bacterial pellets in 2 ml of 25% sucrose in 10 mM Tris and 1 mM EDTA, pH 8.0. Cells were then enzymatically lysed using 50 μl of lysozyme, 100 mg ml−1 (Roche). Next, 100 μl of Proteinase K, 20 mg ml−1 (Roche), 30 μl of RNase A, 10 mg ml−1 (Roche), 400 μl of 0.5 M EDTA, pH 8.0 and 250 μl of 10% Sarkosyl NL30 (Thermo Fisher Scientific) were added accordingly into the lysed suspension. This was then followed by 1 h of ice incubation and a 50 °C water bath overnight. On the second day, three rounds of phenol–chloroform–isoamyl alcohol (Merck) extraction were performed using 15 ml of gel-lock tubes (QIAGEN). Chloroform–isoamyl alcohol (Merck) extraction was then carried out followed by ethanol precipitation and 70% ethanol wash (once). DNA pellets (usually visible) were finally resuspended in 200–300 μl of 10 mM Tris, pH 8.0 and stored at −20 °C until further analysis. WGS of each isolate sample was performed on Illumina HiSeq 2500 to generate 101 125-bp paired-end reads72. Isolate samples from hospital E were sequenced on Illumina NextSeq 500 to generate 150-bp paired-end reads.
De novo genome assemblies and metagenome-assembled genomes
Raw sequence reads (FASTQ) generated from sequencers were quality filtered (-q 20) with fastp v.0.20.0 before de novo genome assembly using SPAdes v.3.14.1 (ref. 73) at default parameters. Contigs of <500 bp were discarded in each isolate genome assembly before subsequent analyses. Genome assembly statistics were generated via sequence-stats v.1.0 (ref. 74) (Supplementary Table 3). All infant-associated genome assemblies were subjected to contamination check via CheckM v.1.1.3 (ref. 75) and contaminated (contamination >5%) and/or incomplete (completeness <90%) genome assemblies were excluded from further analysis (n = 2). A total of 117 human-gut-associated, metagenome-assembled genomes and 17 isolate genomes of C. perfringens was retrieved from the Unified Human Gastrointestinal Genome v.2.0 (UHGG)76 collection (genome completeness >80% and contamination <5%).
Phylogenetic analyses
A total of 673 C. perfringens genome assemblies was investigated, including 171 public isolate genomes retrieved from GenBank, 272 infant-associated draft genomes newly generated in the present study, 96 food-poisoning-linked isolate genomes (from the UK) published previously and 117 high-quality metagenome-assembled genomes, alongside 17 isolate genomes from UHGG (human-gut-associated C. perfringens genomes)8,76. All genomes investigated in the present study passed contamination checks via CheckM v.1.1.3 (ref. 75) (completeness >80% and contamination <5%) and taxonomically assigned as C. perfringens by gtdb-tk v.1.5.1 (ref. 77) (average nucleotide identity (ANI) > 95% against type strain).
These high-quality C. perfringens genome assemblies (n = 673) were subsequently dereplicated at strain level (ANI > 99.9%) using dRep v.3.2.2 (ref. 78) before reconstruction of a core-gene alignment. A core-gene alignment (comprising 858 single-copy core genes) of 310 C. perfringens genomes based on Prokka v.1.14.6 (ref. 79) annotation of coding sequences (general feature format) was next generated using Panaroo v.1.2.8 (ref. 80) with options --merge_paralogs, --clean-mode strict and --aligner mafft81. In a similar manner, infant-associated, core-gene alignment (1,995 single-copy core genes) was generated using dereplicated strain-level genomes (n = 80) via Panaroo v.1.2.8. SNPs were next extracted from core-gene alignments using snp-sites v.2.3.3 (ref. 82) and further dereplicated at core-gene level via in-house script, resulting in 302 representative genomes in the global phylogenetic tree (Fig. 1) and 80 representative genomes in the infant-associated phylogenetic tree (Fig. 2; with two reference genomes, type strains ATCC13124 and NCTC8679). Maximum-likelihood trees were constructed based on the extracted and dereplicated core-gene SNP alignment using IQ-TREE v.2.0.5 (ref. 83) with ultra-fast bootstrap replicates -B 1000 and automatic evolutionary model selection option -m TEST (best fit model for both trees was determined as GTR + F + ASC + G4).
Phylogenetic lineages were assigned using R library RhierBAPS v.1.1.3 (ref. 84). Tree topology was checked and validated using mash-distance trees generated by Mashtree v.1.2.0 (ref. 85). Tree annotation was performed via iTOL v.6.5.8 (ref. 86).
Toxinotype assignment and virulence profiling
Toxinotypes A–G were assigned to each C. perfringens sample via database TOXIper v.1.1 (ref. 87). Virulence-related genes and colonization factor sequence search were performed via ABRicate v.1.0.1 with options --minid=90 and --mincov=80 based on in-house sequence databases (Supplementary Table 5). ResFinder v.4.0 database was used via ABRicate for profiling antimicrobial resistance genes88.
SNP analysis
SNPs were extracted from core-gene alignment with snp-sites v.2.3.3 (ref. 82) and snp-dists v.0.7 (ref. 89) was utilized for computing pairwise SNP distances. Transmission analysis was limited to hospital-associated strains (from 70 individuals) and within closest genetic distance, identified by phylogenetic topologies and pairwise SNP distances within 8,506 core-gene SNPs in 858 single-copy core-gene alignment. Probable transmission dynamics were predicted based on clinical metadata of the patients and the time of the individuals’ stay within the hospitals. SNP distance was further explored to determine the exact genetic distance to infer transmission.
Plasmid investigation
Plasmid analysis was limited to infant-associated C. perfringens genomes. Conjugative plasmids, both pCW3 and pCP13 families (known to encode multiple virulence factors), were extracted based on the identification of the tcp (pCW3 plasmids27; n = 12) and pcp (pCP13 (ref. 28) plasmids; n = 20) genes in the conjugative systems via ABRicate v.1.0.1 (ref. 90). Briefly, tcp and pcp loci and plasmid replication protein gene rep were comprehensively searched on 274 genomes; only single contigs within genome assemblies comprising more than five conjugative genes (rep is compulsory) were extracted and assumed as functional conjugative plasmids for further analyses. Predicted genes were mapped to reference plasmids LH112 (the largest plasmid size in both families) to allow genome architecture comparison of plasmid contents across both pCW3 and pCP13 plasmid families. Virulence factors were identified via ABRicate v.1.0.1 with in-house databases as described in the previous section. Plasmid sequences were aligned with MAFFT v.7.305b, SNP distance was compared via snp-dists v.0.7 (ref. 81). Easyfig v.2.2.2 (ref. 91) was utilized for visualization of plasmid sequence comparison.
Maintenance of cell line Caco-2
Frozen stocks were resuscitated/resuspended in pre-warmed (37 °C) Dulbecco’s Modified Eagle Medium (DMEM; Thermo Fisher Scientific), supplemented with 20% fetal bovine serum (FBS; Thermo Fisher Scientific). Cells were immediately transferred to T75 sterile culture flasks (Corning) and incubated at 37 °C in a 5% CO2 incubator for 24 h. Spent medium was removed after 48–72 h, replaced with 20–25 ml of fresh warm medium as described and incubated until confluency was reached at ~5–8 d. Caco-2 cell line passages 53–54 were used for the present study, cells were split at ratios 1:5–10 for each passage. Cells were counted on a haemocytometer (Neubauer Chamber).
In vitro phenotypic characterization
Cell toxicity assay
The Caco-2 cell line was maintained in liquid nitrogen at Quadram Institute Bioscience (American Type Culture Collection (ATCC), catalogue no. HTB-37). Cells (passages 53–54) were seeded at 20,000 cells per well in tissue-treated 96-well plates (Corning Costar). After 6–7 d of 5% CO2 incubation at 37 °C, a confluent monolayer was formed (manually confirmed using microscopy) and subjected to further experiments. Briefly, phenol-red DMEM was removed, cells were washed twice gently and replaced with phenol-red-free DMEM supplemented with 1% FBS. Diluted sterile-filtered (0.22-μm filter) bacterial supernatants (selected C. perfringens strains were cultured from 1:100 confluent overnight cultures for 10 h before collection of supernatants; all strains reached a density of ~109 colony-forming units (c.f.u.) ml−1 at the end of 10 h) were added neat into each well at 12.5% total volume followed by a 2-h 5% CO2 incubation at 37 °C. Cell death measurement (fluorescence signal of lactate dehydrogenase released from cells with a damaged cell membrane) was performed and estimated using CytoTox-ONE Homogeneous Membrane Integrity Assay according to the manufacturer’s instructions (Promega) using a black 96-well plate to reduce background readouts (Corning).
Haemolysis assay
Columbian blood agar (CBA; 5% sheep blood; Oxoid) was used to identify the haemolytic trait of C. perfringens isolates linked to pore-forming toxins produced. Briefly, C. perfringens was streaked on the blood agar and incubated anaerobically overnight (<20 h). Isolated colonies with clear transparent halos formed in and around the colonies indicate complete haemolysis. Assay outcomes were observed manually.
Oxygen tolerance assay
Pure cultures of 31 C. perfringens strains (including 30 infant-associated isolates and type strain ATCC13124) were grown anaerobically to confluency for 24 h in BHI. Subsequently, cultures were spotted in a dilution series on to BHI agar supplemented with 0.1% sodium taurocholate (potent germinant for spores). Spotted Petri dishes were dried and incubated under ambient (aerobic) conditions at room temperature (22 °C) for specified time periods before being returned to an anaerobic chamber for determination of the colony-forming unit count (maximum 14 d). All colony-forming units were counted after 15–20 h of anaerobic incubation at 37 °C. Cultures (spotted plates) that were not exposed to ambient conditions acted as controls for each strain. A viable percentage was shown after comparing with control cultures.
Sporulation assay
Cultures of 31 C. perfringens strains (including 30 infant-associated isolates and type strain ATCC13124) were induced to sporulate using the modified Duncan-strong medium (proteose peptone 16 g l−1, yeast extract 4 g l−1, cysteine hydrochloride 0.5 g l−1, disodium phosphate 10 g l−1 and raffinose 4 g l−1, with a final pH of 7.8) for 24-h anaerobic incubation using 1:100 overnight confluent cultures92. Cultures were pre-treated with sterile-filtered 70% ethanol for 4 h to eliminate vegetative cells and spot plated on BHI agar supplemented with 0.1% taurocholate (potent germinant) for spores. Agar plates were incubated anaerobically for 24 h before counting the colony-forming units.
In vivo studies
Ethics and licence
All animal experiments and related protocols described were performed under the Animals (Scientific Procedures) Act 1986 (ASPA) under project licence no. PP8873233 and approved by the Home Office and University of East Anglia (UEA) FMH Research Ethics Committee. Animals are monitored and assessed frequently during studies for physical condition and behaviour. Mice determined to have suffered from distress would be euthanized via the ASPA schedule 1 protocol (CO2 and cervical dislocation). Trained and qualified animal technicians carried out animal husbandry at UEA Disease Modelling Unit (DMU).
Animals, housing and animal study design
Specific pathogen-free C57BL/6 female mice (n = 40; 3-week-old Mus musculus), obtained from Charles River (France) were used in the animal study. Animals were transported to and housed in the UEA DMU acclimatization barn under specific pathogen-free conditions for 2 weeks before moving to the infection suite. During the infection study, mice were housed with autoclaved bedding (and cage), food (stock pellets) and water with a 12 h light cycle (12 h:12 h light:darkness). Cages were changed in a laminar flow cabinet. The rooms were maintained at 21 ± 2 °C and 55 ± 10% humidity. Air change rates are 12–15 per h.
Mice were caged into eight groups (five mice in each group), which include control group, ABX group (only treated with antibiotics) and six experimental groups (pfoA+ versus pfoA−). Group pfoA+ (n = 3) only was challenged with C. perfringens strains IQ146, IQ129 and LH115, which encode a pfoA toxin gene, whereas group pfoA− (n = 3) was given only C. perfringens strains that do not encode the pfoA toxin gene, IQ147, IQ133 and LH043, respectively.
Experimental groups (n = 6) and the ABX group were treated with a five-antibiotic cocktail comprising kanamycin (0.4 mg ml−1), gentamicin (0.035 mg ml−1), colistin (850 U ml−1), metronidazole (0.215 mg ml−1) and vancomycin (0.045 mg ml−1) in autoclaved drinking water for 3 d administered freely (Fig. 6a). Drinking water was then switched to sterile water and, 48 h later, mice in the experimental group and the ABX group were orally gavaged with 150 mg kg−1 of clindamycin. After 24 h, mice were gavaged with ~109 c.f.u. of C. perfringens in 100 μl and all mice were closely monitored for signs of disease symptoms, including significant weight loss (>20%). All animals were euthanized and the organs harvested at day 7 (Fig. 6a).
Bacterial strains and growth
Bacterial stocks were recovered on BHI agar for purity checks each time and subsequently cultured in BHI broth overnight to reach confluency (~109 c.f.u. ml−1, approximately 14–17 h). Bacterial pellets were washed and resuspended in sterile phosphate-buffered saline before feeding the mice via oral gavaging using a 20G plastic sterile feeding tube.
Faecal sample collection and colony-forming unit enumeration
Faecal samples were collected in sterilized tubes daily and stored at −80 °C until further analysis. Serially diluted faecal mixtures were plated on fresh TSC agar and the colony-forming units enumerated after <24 h anaerobic incubation. Pitch-black colonies were counted as C. perfringens colonies.
Tissue processing and H&E staining
Distal colons were fixed in 10% neutral buffered formalin (~4% formaldehyde; Merck) for <24 h and followed by 70% ethanol. Tissues were subsequently processed using an automated Leica Tissue Processor ASP-300-S and embedded in paraffin manually. Next, sectioning of tissues was performed using a Leica microtome (5-μm-thick sections) and left overnight for samples to air dry before staining and further analysis. Haematoxylin and eosin (H&E) staining was then performed using Leica Autostainer for structural imaging of intestinal samples.
Microscopy imaging
Bright-field microscopy was performed using Olympus BX60 with a microscope camera Jenoptik C10 and ProgRes CapturePro software v.2.10 (Supplementary Figs. 1–8).
Histopathological scoring
All distal colonic section images (two to three distal colonic images per animal) were examined and graded single-blinded by L.J.H. The histological severity of intestinal inflammation was graded using a histopathological scoring system (0–14; denoting increasing severity) based on three general pathological features: inflammatory cell infiltrate (score value: 0–4), epithelial changes that include epithelial hyperplasia and goblet cell loss (score value: 0–5) and mucosal architecture—villous blunting (score value: 0–5) as described by Erben et al.93. The overall additive score (0–14) was the sum of each component score value. An overall pathological score of 1–4 suggests minimal colitis, 5–8 mild colitis, 9–11 moderate colitis and >12 marked/severe colitis.
Morphometry
Crypt length of distal colons was examined digitally via ImageJ2 v.2.3.0 based on 20 crypts per representative image of distal colonic microscopy slides.
Cytokine analysis
Proximal murine colons were homogenized in lysis buffer (150 mmol l−1 of NaCl, 20 mmol l−1 of Tris, 1 mmol l−1 of EDTA, 1 mmol l−1 of EGTA and 1% Triton X-100 at pH 7.5) using a FastPrep-24 bead-beating grinder (MP Biomedicals) at speed 4.0 for 40 s followed by speed 6.0 for 40 s. Samples were then centrifuged for 12 min at 4 °C. Supernatants were transferred to clean tubes and further quantified (protein concentration) before cytokine analysis.
Protein concentration was quantified by BCA (bicinchoninic acid) assay (BioRad) and cytokines were analysed using a customized Mesoscale Discovery U-Plex assay (MSD) according to the manufacturer’s instructions. Cytokine concentrations were normalized to total protein input and samples were run in duplicates. The plate was read using an MSD QuickPlex SQ 120 imager and data analysed on the MSD Discovery Workbench software.
Gut microbiome analysis
Genomic DNA of murine faecal samples was extracted, sequenced and analysed following the protocol described previously6. Briefly, DNA was extracted from samples using FastDNA Spin Kit for Soil (MP Biomedicals) following the manufacturer’s instruction, while extending the bead-beating (in FastPrep-24 bead-beating grinder) step to 80 s at speed 6.0 (ref. 59). Next, genomic DNA extracts were subjected to the 16S rRNA gene V1 + V2 region library preparation before sequencing on an Illumina NovaSeq sequencing platform at 2 × 250 bp (paired-end sequencing)59. Sequencing raw reads (FASTQ) were first merged using PEAR v.0.9.6 (ref. 94), then underwent quality and chimera filtering via QIIME v.1.9.1 (ref. 95), subsequently with operational taxonomic unit (OTU) assignment using a SILVA_132 database96. BIOM output of OTU tables was read using MEGAN6 (ref. 97).
Data visualization
Phylogenetic trees (mid-point rooted and unrooted) were graphed using iTOL v.6.5.8 (ref. 86). Various statistical graphs including pie charts, line charts, dot plots, bar plots and box plots were drawn in R v.4.1.2 (ref. 98), using R libraries tidyverse v.1.3.1 (ref. 99), ggplot2 v.3.3.5 (ref. 100) and ggpubr v.0.4.0 (ref. 101). R library vegan v.2.6.2 (ref. 102) was used to construct non-metric multidimensional scaling (NMDS) plot for microbiome data and estimation of the inverse Simpson’s index.
Sample size, randomization and blinding
No statistical methods were used to predetermine sample size for experiments. Sample size for the in vivo study was selected following the 3Rs principles—reduction, replacement and refinement—ensuring sufficient sample size for meaningful statistical analysis.
Randomization was not attempted during the in vivo study to avoid cross-contamination because treatments were performed according to the individual cages; however, animals were initially allocated to each cage randomly. For in vitro experiments, randomization is not relevant due to the way bacterial isolates were selected with an aim of formally characterizing pfoA+ isolates (n = 15) and pfoA− isolates (n = 15).
Single blinding was used in the in vivo colon tissue histopathological scoring analysis to prevent bias. For in vivo sample collection and analysis, the investigators were blinded to group allocation. For other in vitro experiments, data collection and analysis were not performed blind to the conditions of the experiments.
Data exclusion
Two C. perfringens isolate genomes (sequenced in the present study) were excluded from genome analysis due to detected sequence contamination by CheckM v.1.1.3 (>10%). No animals or data points were excluded from analyses.
Statistics and reproducibility
Statistical tests were performed via R base packages stats v.4.1.2 (ref. 98), including Fisher’s exact test (two sided), Kruskal–Wallis test, Wilcoxon’s test, analysis of variance (ANOVA), post-hoc Tukey’s test and point-biserial correlation test. The Shapiro–Wilk normality test was used to formally test for data normality where appropriate. R library rstatix v.0.7.0 (ref. 103) was used for post-hoc Dunn’s test of multiple comparisons (two sided), whereas rcompanion v.2.4.16 (ref. 104) was utilized to conduct pairwise post-hoc tests (Fisher’s exact test).
For both oxygen tolerance and sporulation assays, experiments were performed on three biologically independent replicates per sample (bacterial strain; n = 3). In cell toxicity assay, cell death measurements were performed in two independent experiments, each in three biological replicates per sample (sterile supernatants of bacterial strains; n = 6). The animal study was conducted once with five C57BL/6 mice each group or cage (n = 40). For murine distal colonic samples, two to three subsections of each processed distal colon (cross-sections) were photographed using a digital bright-field microscope and used for further phenotypic examinations.
Ethics statement
Faecal collection from Norfolk and Norwich University Hospital (NNUH) and Rosie Hospital (BAMBI study) was approved by the Faculty of Medical and Health Sciences Ethics Committee at the UEA and followed protocols laid out by the UEA Biorepository (licence no. 11208). Faecal collection from Imperial Healthcare NICUs (NeoM and NeoM2 studies) was approved by West London Research Ethics Committee (REC) under the REC approval reference no. 10/H0711/39. In all cases, doctors and nurses recruited infants after parents had given written consent. Ethical approval for the SERVIS study was approved by the North East and N Tyneside committee 10/H0908/39, and signed parental consent was obtained from every parent. We have anonymized identifiers of patients, hospitals and associated clinical data.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Genome assemblies of 171 C. perfringens isolates were retrieved from the National Center for Biotechnology Information’s (NCBI) GenBank (downloaded on 2 April 2020), while 96 C. perfringens food-poisoning-associated, isolated genomes were downloaded (May 2022) from the European Nucleotide Archive under accession no. PRJEB25764 (Supplementary Table 3). Another 117 C. perfringens metagenome-assembled genomes and 17 C. perfringens isolated genomes were retrieved from the UHGG v.2.0 collection (May 2022; Supplementary Table 3). Sequencing raw reads and draft genome assemblies for 272 C. perfringens isolates generated in the present study are publicly available in the NCBI Sequence Read Archive (SRA) and GenBank respectively, under accession no. PRJNA755973 (Supplementary Table 3). The 16S rRNA gene amplicon sequence reads (in vivo microbiome study) are publicly available in SRA under accession no. PRJNA755974 (Supplementary Table 4). Accessible links to sequence databases including ResFinder v.4.0, TOXIper v.1.1, tcp loci, pcp loci and C. perfringens-associated virulence genes (Supplementary Table 5) used in the present study are available in Supplementary Table 6. Anonymized clinical metadata have been made available in Supplementary Table 2. Computationally extracted plasmid sequences, histology images and data used for the figures (Source data files) are listed in the inventory (‘Inventory_of_supplementary_info.xlsx’) which is openly shared via the GitHub repository: https://github.com/raymondkiu/Infant-Clostridium-perfringens-Paper (ref. 105). Source data are provided with this paper.
Code availability
R scripts associated with data visualization and statistical analysis are available and shared via GitHub (https://github.com/raymondkiu/Infant-Clostridium-perfringens-Paper)105.
References
Kiu, R. & Hall, L. J. An update on the human and animal enteric pathogen Clostridium perfringens. Emerg. Microbes Infect. 7, 141 (2018).
Li, J., Paredes-Sabja, D., Sarker, M. R. & McClane, B. A. Clostridium perfringens sporulation and sporulation-associated toxin production. Microbiol. Spectr. https://doi.org/10.1128/microbiolspec.TBS-0022-2015 (2016).
Rood, J. I. et al. Expansion of the Clostridium perfringens toxin-based typing scheme. Anaerobe 53, 5–10 (2018).
Ronco, T. et al. Genome analysis of Clostridium perfringens isolates from healthy and necrotic enteritis infected chickens and turkeys. BMC Res. Notes 10, 270 (2017).
Lacey, J. A. et al. Whole genome analysis reveals the diversity and evolutionary relationships between necrotic enteritis-causing strains of Clostridium perfringens. BMC Genom. 19, 379 (2018).
Kiu, R. et al. Genomic analysis on broiler-associated Clostridium perfringens strains and exploratory caecal microbiome investigation reveals key factors linked to poultry necrotic enteritis. Anim. Microbiome 1, 12 (2019).
Mahamat Abdelrahim, A. et al. Large-scale genomic analyses and toxinotyping of Clostridium perfringens implicated in foodborne outbreaks in France. Front. Microbiol. https://doi.org/10.3389/fmicb.2019.00777 (2019).
Kiu, R. et al. Phylogenomic analysis of gastroenteritis-associated Clostridium perfringens in England and Wales over a 7-year period indicates distribution of clonal toxigenic strains in multiple outbreaks and extensive involvement of enterotoxin-encoding (CPE) plasmids. Micro. Genom. https://doi.org/10.1099/mgen.0.000297 (2019).
Camargo, A. et al. Intra-species diversity of Clostridium perfringens: a diverse genetic repertoire reveals its pathogenic potential. Front. Microbiol. 13, 952081 (2022).
Neu, J. & Walker, A. W. Necrotizing enterocolitis. N. Engl. J. Med. 364, 255–264 (2011).
Lim, J. C., Golden, J. M. & Ford, H. R. Pathogenesis of neonatal necrotizing enterocolitis. Pediatr. Surg. Int. 31, 509–518 (2015).
Lin, P. W. & Stoll, B. J. Necrotising enterocolitis. Lancet 368, 1271–1283 (2006).
Rees, C. M., Eaton, S. & Pierro, A. National prospective surveillance study of necrotizing enterocolitis in neonatal intensive care units. J. Pediatr. Surg. 45, 1391–1397 (2010).
Shah, T. A. et al. Hospital and neurodevelopmental outcomes of extremely low-birth-weight infants with necrotizing enterocolitis and spontaneous intestinal perforation. J. Perinatol. 32, 552–558 (2012).
Bazacliu, C. & Neu, J. Necrotizing enterocolitis: long term complications. Curr. Pediatr. Rev. 15, 115–124 (2019).
Coggins, S. A., Wynn, J. L. & Weitkamp, J. H. Infectious causes of necrotizing enterocolitis. Clin. Perinatol. 42, 133–154 (2015).
Carlisle, E. M. & Morowitz, M. J. The intestinal microbiome and necrotizing enterocolitis. Curr. Opin. Pediatr. 25, 382–387 (2013).
Morowitz, M. J., Poroyko, V., Caplan, M., Alverdy, J. & Liu, D. C. Redefining the role of intestinal microbes in the pathogenesis of necrotizing enterocolitis. Pediatrics 125, 777–785 (2010).
Hanke, C. A. et al. Clostridium perfringens intestinal gas gangrene in a preterm newborn. Eur. J. Pediatr. Surg. 19, 257–259 (2009).
Claud, E. C. Neonatal necrotizing enterocolitis: inflammation and intestinal immaturity. Antiinflamm. Antiallergy Agents Med. Chem. 8, 248–259 (2009).
Heida, F. H. et al. A necrotizing enterocolitis-associated gut microbiota is present in the meconium: results of a prospective study. Clin. Infect. Dis. 62, 863–870 (2016).
Sim, K. et al. Dysbiosis anticipating necrotizing enterocolitis in very premature infants. Clin. Infect. Dis. 60, 389–397 (2015).
Butel, M. J. et al. Clostridial pathogenicity in experimental necrotising enterocolitis in gnotobiotic quails and protective role of bifidobacteria. J. Med. Microbiol. 47, 391–399 (1998).
Waligora-Dupriet, A. J., Dugay, A., Auzeil, N., Huerre, M. & Butel, M. J. Evidence of clostidial implication in necrotizing enterocolitis through bacterial fermentation in a gnotobiotic quail model. Pediatr. Res. 58, 629–635 (2005).
Kosloske, A. M., Ball, W. S. Jr, Umland, E. & Skipper, B. Clostridial necrotizing enterocolitis. J. Pediatr. Surg. 20, 155–159 (1985).
Kosloske, A. M., Ulrich, J. A. & Hoffman, H. Fulminant necrotising enterocolitis associated with clostridia. Lancet 312, 1014–1016 (1978).
Revitt-Mills, S. A., Watts, T. D., Lyras, D., Adams, V. & Rood, J. I. The ever-expanding tcp conjugation locus of pCW3 from Clostridium perfringens. Plasmid 113, 102516 (2020).
Watts, T. D. et al. pCP13, a representative of a new family of conjugative toxin plasmids in Clostridium perfringens. Plasmid 102, 37–45 (2019).
Lacey, J. A., Johanesen, P. A., Lyras, D. & Moore, R. J. In silico identification of novel toxin homologs and associated mobile genetic elements in Clostridium perfringens. Pathogens 8, 16 (2019).
Kiu, R. et al. Genomic analysis of Clostridium perfringens BEC/CPILE-positive, toxinotype D and E strains isolated from healthy children. Toxins 11, 543 (2019).
Feng, Y. et al. Phylogenetic and genomic analysis reveals high genomic openness and genetic diversity of Clostridium perfringens. Microb. Genom. https://doi.org/10.1099/mgen.0.000441 (2020).
Kiu, R., Caim, S., Alexander, S., Pachori, P. & Hall, L. J. Probing genomic aspects of the multi-host pathogen Clostridium perfringens reveals significant pangenome diversity, and a diverse array of virulence factors. Front. Microbiol. 8, 2485 (2017).
Lakshminarayanan, B. et al. Prevalence and characterization of Clostridium perfringens from the faecal microbiota of elderly Irish subjects. J. Med. Microbiol. 62, 457–466 (2013).
Hamza, D., Dorgham, S. M., Elhariri, M., Elhelw, R. & Ismael, E. New insight of apparently healthy animals as a potential reservoir for Clostridium perfringens: a public health implication. J. Vet. Res. 62, 457–462 (2018).
Savinov, S. N. & Heuck, A. P. Interaction of cholesterol with perfringolysin O: what have we learned from functional analysis? Toxins 9, 381 (2017).
Stevens, D. L. & Bryant, A. E. The role of clostridial toxins in the pathogenesis of gas gangrene. Clin. Infect. Dis. 35, S93–S100 (2002).
Verherstraeten, S. et al. The synergistics necrohemorrhagic action of Clostridium perfringens perfringolysin and alpha toxin in the bovine intestine and against bovine endothelial cells. Vet. Res. 44, 45 (2013).
Zhang, J. et al. Verbascoside protects mice from clostridial gas gangrene by inhibiting the activity of alpha toxin and perfringolysin O. Front. Microbiol. 11, 1504 (2020).
Li, Z., Yan, C., Gong, X. & Wang, J. Severe intravascular hemolysis from Clostridium perfringens septicemia in a neonate with necrotizing enterocolitis in China: a case report. Infect. Drug Resist. 15, 1461–1465 (2022).
Suzaki, A. et al. Pathogenic characterization of Clostridium perfringens strains isolated from patients with massive intravascular hemolysis. Front. Microbiol. 12, 713509 (2021).
Awad, M. M., Ellemor, D. M., Boyd, R. L., Emmins, J. J. & Rood, J. I. Synergistic effects of alpha-toxin and perfringolysin O in Clostridium perfringens-mediated gas gangrene. Infect. Immun. 69, 7904–7910 (2001).
Dittmar, E. et al. Necrotizing enterocolitis of the neonate with Clostridium perfringens: diagnosis, clinical course, and role of alpha toxin. Eur. J. Pediatr. 167, 891–895 (2008).
Park, J. M., Ng, V. H., Maeda, S., Rest, R. F. & Karin, M. Anthrolysin O and other gram-positive cytolysins are toll-like receptor 4 agonists. J. Exp. Med. 200, 1647–1655 (2004).
Yamamura, K. et al. Inflammasome activation induced by perfringolysin O of Clostridium perfringens and Its involvement in the progression of gas gangrene. Front. Microbiol. 10, 2406 (2019).
Yuan, J., Najafov, A. & Py, B. F. Roles of caspases in necrotic cell death. Cell 167, 1693–1704 (2016).
Rood, J. I. & Cole, S. T. Molecular genetics and pathogenesis of Clostridium perfringens. Microbiol Rev. 55, 621–648 (1991).
Verherstraeten, S. et al. Perfringolysin O: the underrated Clostridium perfringens toxin? Toxins 7, 1702–1721 (2015).
Kumar, N. et al. Genome-based infection tracking reveals dynamics of Clostridium difficile transmission and disease recurrence. Clin. Infect. Dis. 62, 746–752 (2015).
Garcia-Fernandez, S. et al. Whole-genome sequencing reveals nosocomial Clostridioides difficile transmission and a previously unsuspected epidemic scenario. Sci. Rep. 9, 6959 (2019).
Moffatt, C. R., Howard, P. J. & Burns, T. A mild outbreak of gastroenteritis in long-term care facility residents due to Clostridium perfringens, Australia 2009. Foodborne Pathog. Dis. 8, 791–796 (2011).
Ellis, C. L., Rutledge, J. C. & Underwood, M. A. Intestinal microbiota and blue baby syndrome: probiotic therapy for term neonates with cyanotic congenital heart disease. Gut Microbes 1, 359–366 (2010).
Malagon, I. et al. Gut permeability in paediatric cardiac surgery. Br. J. Anaesth. 94, 181–185 (2005).
McElhinney, D. B. et al. Necrotizing enterocolitis in neonates with congenital heart disease: risk factors and outcomes. Pediatrics 106, 1080–1087 (2000).
Browne, H. P. et al. Culturing of ‘unculturable’ human microbiota reveals novel taxa and extensive sporulation. Nature 533, 543–546 (2016).
Rolfe, R. D., Hentges, D. J., Campbell, B. J. & Barrett, J. T. Factors related to the oxygen tolerance of anaerobic bacteria. Appl. Environ. Microbiol. 36, 306–313 (1978).
Lemyre, B. et al. A decrease in the number of cases of necrotizing enterocolitis associated with the enhancement of infection prevention and control measures during a Staphylococcus aureus outbreak in a neonatal intensive care unit. Infect. Control Hosp. Epidemiol. 33, 29–33 (2012).
Dufrene, Y. F. & Viljoen, A. Binding strength of gram-positive bacterial adhesins. Front. Microbiol. 11, 1457 (2020).
Yonogi, S. et al. BEC, a novel enterotoxin of Clostridium perfringens found in human clinical isolates from acute gastroenteritis outbreaks. Infect. Immun. 82, 2390–2399 (2014).
Alcon-Giner, C. et al. Microbiota supplementation with Bifidobacterium and Lactobacillus modifies the preterm infant gut microbiota and metabolome: an observational study. Cell Rep. Med. 1, 100077 (2020).
Robertson, C. et al. Incidence of necrotising enterocolitis before and after introducing routine prophylactic Lactobacillus and Bifidobacterium probiotics. Arch. Dis. Child Fetal Neonatal Ed. 0, F1–F7 (2019).
Hall, L. J. et al. Natural killer cells protect against mucosal and systemic infection with the enteric pathogen Citrobacter rodentium. Infect. Immun. 81, 460–469 (2013).
Bouladoux, N., Harrison, O. J. & Belkaid, Y. The mouse model of infection with Citrobacter rodentium. Curr. Protoc. Immunol. 119, 19.15.11–19.15.25 (2017).
Lundberg, R. Humanizing the gut microbiota of mice: opportunities and challenges. Lab. Anim. 53, 244–251 (2019).
Tarracchini, C. et al. Unraveling the microbiome of necrotizing enterocolitis: insights in novel microbial and metabolomic biomarkers. Microbiol. Spectr. 9, e0117621 (2021).
Alsaied, A., Islam, N. & Thalib, L. Global incidence of necrotizing enterocolitis: a systematic review and meta-analysis. BMC Pediatr. 20, 344 (2020).
Haigh, E. Gas gangrene of the colon in a newborn infant; report of a case successfully treated by total colectomy. Br. J. Surg. 43, 659–661 (1956).
Raskind, C. H., Dembry, L. M. & Gallagher, P. G. Vancomycin-resistant enterococcal bacteremia and necrotizing enterocolitis in a preterm neonate. Pediatr. Infect. Dis. J. 24, 943–944 (2005).
Stewart, C. J., Estes, M. K. & Ramani, S. Establishing human intestinal enteroid/organoid lines from preterm infant and adult tissue. Methods Mol. Biol. 2121, 185–198 (2020).
Stewart, C. J. et al. Temporal development of the gut microbiome in early childhood from the TEDDY study. Nature 562, 583–588 (2018).
Kotsanas, D. et al. Novel use of tryptose sulfite cycloserine egg yolk agar for isolation of Clostridium perfringens during an outbreak of necrotizing enterocolitis in a neonatal unit. J. Clin. Microbiol. 48, 4263–4265 (2010).
Kiu, R. et al. Preterm Infant-Associated Clostridium tertium, Clostridium cadaveris, and Clostridium paraputrificum Strains: genomic and Evolutionary Insights. Genome Biol. Evol. 9, 2707–2714 (2017).
Collins, J. et al. Dietary trehalose enhances virulence of epidemic Clostridium difficile. Nature 553, 291–294 (2018).
Bankevich, A. et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012).
Kiu, R. sequence-stats: generate sequence statistics from FASTA and FASTQ files. GitHub https://github.com/raymondkiu/sequence-stats (2020).
Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P. & Tyson, G. W. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043–1055 (2015).
Almeida, A. et al. A unified catalog of 204,938 reference genomes from the human gut microbiome. Nat. Biotechnol. 39, 105–114 (2021).
Chaumeil, P. A., Mussig, A. J., Hugenholtz, P. & Parks, D. H. GTDB-Tk: a toolkit to classify genomes with the Genome Taxonomy Database. Bioinformatics 36, 1925–1927 (2020).
Olm, M. R., Brown, C. T., Brooks, B. & Banfield, J. F. dRep: a tool for fast and accurate genomic comparisons that enables improved genome recovery from metagenomes through de-replication. ISME J. 11, 2864–2868 (2017).
Seemann, T. Prokka: rapid prokaryotic genome annotation. Bioinformatics 30, 2068–2069 (2014).
Tonkin-Hill, G. et al. Producing polished prokaryotic pangenomes with the Panaroo pipeline. Genome Biol. 21, 180 (2020).
Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013).
Page, A. J. et al. SNP-sites: rapid efficient extraction of SNPs from multi-FASTA alignments. Micro. Genom. 2, e000056 (2016).
Minh, B. Q. et al. IQ-TREE 2: new models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).
Tonkin-Hill, G., John, A. L., Stephen, D. B., Simon, D. W. F. & Jukka, C. RhierBAPS: an R implementation of the population clustering algorithm hierBAPS. Wellcome Open Res. 3, 93 (2018).
Katz, L. S. et al. Mashtree: a rapid comparison of whole genome sequence files. J. Open Source Softw. 4, 1762 (2019).
Letunic, I. & Bork, P. Interactive Tree Of Life (iTOL) v4: recent updates and new developments. Nucleic Acids Res. 47, W256–W259 (2019).
Kiu, R. TOXIper: rapid Clostridium perfringens toxinotype assignment using genome assemblies. GitHub https://github.com/raymondkiu/TOXIper (2020).
Bortolaia, V. et al. ResFinder 4.0 for predictions of phenotypes from genotypes. J. Antimicrob. Chemother. 75, 3491–3500 (2020).
Seemann, T., Klotzl, F. & Page, A. J. snp-dists: pairwise SNP distance matrix from a FASTA sequence alignment. GitHub https://github.com/tseemann/snp-dists (2018).
Seemann, T. ABRicate: mass screening of contigs for antimicrobial and virulence genes. GitHub https://github.com/tseemann/abricate (2018).
Sullivan, M. J., Petty, N. K. & Beatson, S. A. Easyfig: a genome comparison visualizer. Bioinformatics 27, 1009–1010 (2011).
Duncan, C. L. & Strong, D. H. Improved medium for sporulation of Clostridium perfringens. Appl. Microbiol. 16, 82–89 (1968).
Erben, U. et al. A guide to histomorphological evaluation of intestinal inflammation in mouse models. Int J. Clin. Exp. Pathol. 7, 4557–4576 (2014).
Zhang, J., Kobert, K., Flouri, T. & Stamatakis, A. PEAR: a fast and accurate Illumina paired-end reAd mergeR. Bioinformatics 30, 614–620 (2014).
Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336 (2010).
Quast, C. et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 41, D590–D596 (2013).
Huson, D. H. et al. MEGAN community edition—interactive exploration and analysis of large-scale microbiome sequencing data. PLoS Comput. Biol. 12, e1004957 (2016).
R Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2021).
Wickham, H. et al. Welcome to the tidyverse. J. Open Source Softw. 4, 1686 (2019).
Wickham, H. ggplot2: Elegant Graphics for Data Analysis (Springer-Verlag, 2016).
Kassambara, A. ggpubr: ‘ggplot2’ based publication ready plots. R package version 0.4.0. https://CRAN.R-project.org/package=ggpubr (2020).
Dixon, P. VEGAN, a package of R functions for community ecology. J. Veg. Sci. 14, 927–930 (2003).
Kassambara, A. rstatix: pipe-friendly framework for basic statistical tests. R package version 0.7.0. https://CRAN.R-project.org/package=rstatix (2021).
Mangiafico, S. S. rcompanion: functions to support extension education program evaluation. R package version 2.4.16. https://CRAN.R-project.org/package=rcompanion (2022).
Kiu, R. et al. Particular genomic and virulence traits associated with preterm-derived toxigenic Clostridium perfringens strains. Zenodo https://doi.org/10.5281/zenodo.7756267 (2023).
Acknowledgements
This research was supported in part by the NBI Computing infrastructure for Science (CiS) group through the provision of a high-performance computing cluster. L.J.H. is supported by: Wellcome Trust Investigator Awards (nos. 100974/C/13/Z and 220876/Z/20/Z); the Biotechnology and Biological Sciences Research Council (BBSRC), Institute Strategic Programme Gut Microbes and Health BB/R012490/1, and its constituent projects BBS/E/F/000PR10353 and BBS/E/F/000PR10356. D.P. was funded by the Wellcome Trust. C.J.S. is supported by the Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (no. 221745/Z/20/Z). We thank the sequencing team at both the Wellcome Trust Sanger Institute and the Quadram Institute Bioscience (QIB) for genome sequencing. We would like to express our appreciation to our colleague S. Carding, and former colleagues S. Caim, L. Harnisch and B. Kirkup at QIB for their discussion on the data and assistance in various training sessions associated with this work.
Author information
Authors and Affiliations
Contributions
R.K., A.G.S, K.S, G. Douce, J.S.K. and L.J.H. conceived the study. R.K., S.A.D., S.D.R. and L.J.H. provided the methodology. R.K. provided the software. K.S., A.G.S. and L.J.H. validated the study. R.K. and L.J.H. did the formal analysis. R.K., K. S., A.G.S., A.A.G., C.A.P., S.A.D., W.J.F., E.C, J.M., C.A., S.P., Z.S. and L.J.H. carried out the investigations. R.K., A.G.S., E.C., D.P., G.B., J.E.B., G.R.Y., C.J.S., G. Dougan, P.C. and J.S.K. provided the resources. R.K., A.G.S, K.S. and C.A. curated the data. R.K. and L.J.H. wrote the original draft of the MS. R.K., A.G.S., K.S., Z.S., J.E.B., C.J.S., J.S.K. and L.J.H. reviewed and edited the MS. R.K. carried out the visualization. J.S.K. and L.J.H. supervised the study. R.K. administered the project. J.S.K and L.J.H. acquired funds.
Corresponding author
Ethics declarations
Competing interests
C.J.S. declares performing consultancy for Astarte Medical, receiving lecture honoraria from Nestlé Nutrition Institute and supervising a BBSRC collaborative training partnership PhD student for which Nestlé are involved (no salary or other personal payment is provided by Nestlé). The remaining authors declare no competing interests.
Peer review
Peer review information
Nature Microbiology thanks Horst Posthaus and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Genomic analysis of C. perfringens strains.
a, Colonisation factor comparison across all isolates by lineage. Colonisation factors are significantly more in lineages III, VI and VII than lineages I and V, while no significant difference between lineages III, VI and VII, similarly also lineages I and V. The box in boxplot represents 50% of the central data, in between lower and upper quartile, with a central line representing the median while whiskers show the most extreme data points. Statistical analyses were performed with Kruskal-Wallis test and post-hoc Dunn’s test of multiple comparison with P values adjusted by Benjamini-Hochberg method. b, Comparative genomic architecture analysis of identical plasmids from sub-lineage IIIb isolates.
Extended Data Fig. 2 Strain-level in vitro and in vivo assessment of C. perfringens virulence potentials.
a, Oxygen tolerance test. Comparison of viability of C. perfringens strains (n = 31) from lineages I, III, V, VI and VII exposed to ambient air (aerobic) over 14 days in 3 biologically independent samples for each strain. Cell viability in percentage was shown from 48 h to 192 h between lineage VII and lineage V with statistical difference. The box in boxplot represents 50% of the central data, in between lower and upper quartile, with a central line representing the median while whiskers show the most extreme data points (also in b). b, Intestinal colonisation of C. perfringens monitored over 7 days post-challenge. LD, limit of detection. c, NMDS plot of murine genus-level gut microbiome profiles on day 0, day 2 and day 6. d, Comparison of genus-level murine gut microbiome on day 0, day 2 and day 6. A total of 4 groups were compared, Control (n = 5), ABX (n = 5), pfoA+ (n = 15) and pfoA- (n = 15). DPC: days post-challenge. (All statistical analyses were performed with Kruskal-Wallis test and post-hoc Dunn’s test of multiple comparisons with P values adjusted by Benjamini-Hochberg method. The box in boxplots represents 50% of the central data, in between first and third quartile, with an inner line representing the median).
Extended Data Fig. 3 Histopathological analysis on murine distal colonic sections.
a-d, Histopathological scores for intestinal inflammation of murine distal colons comparison between groups in three main general criteria (features), with indicative severity scale (minimal, mild, moderate and marked). There were 8 groups of mice (n = 5 each group) including control groups (Control and ABX) and 6 experimental groups (3 groups of pfoA+ and 3 groups of pfoA-). The box in boxplot represents 50% of the central data, in between lower and upper quartile, with a central line representing the median while whiskers show the most extreme data points. All statistical analyses were performed with Kruskal-Wallis test and post-hoc Dunn’s test of multiple comparisons with P values adjusted by Benjamini-Hochberg method, unless otherwise stated. Significance was tested against untreated group, otherwise significance brackets will be used in pairwise comparison. a, Inflammatory infiltrate: immune cell infiltration (score value: 0–4). b, Epithelial changes: epithelial hyperplasia and goblet cell loss (score value: 0–5). c, Mucosal architecture: villous blunting (score value: 0–5). d, Total score (additive score value: 0–14). Increments in score value denotes increasing severity. Scoring was based on two H&E-stained images per distal colon.
Extended Data Fig. 4 Evaluation of pro-inflammatory cytokine profiles in proximal murine colons.
a-j, Concentration (pg/mg) of cytokines IFN-γ (a), IL-10 (b), IL-12p70 (c), IL-17 (d), IL-1β (e), IL-22 (f), IL-4 (g), IL-6 (h), KC/GRO (i) and TNF-α (j) respectively, in murine proximal colonic samples, stratified into groups: Control (n = 5), ABX (n = 5), pfoA+ (n = 15), pfoA- (n = 15). All individual samples were measured in technical duplicates (n = 2). The box in boxplot represents 50% of the central data, in between lower and upper quartile, with a central line representing the median while whiskers show the most extreme data points. All statistical analyses were performed with Kruskal-Wallis test and post-hoc Dunn’s test of multiple comparisons with P values adjusted by Benjamini-Hochberg method, unless otherwise stated. Significance was tested against untreated group, otherwise significance brackets will be used in pairwise comparison.
Supplementary information
Supplementary Information
Supplementary Figs. 1–8.
Supplementary Tables
Supplementary Table 1 Clinical statistics: a summary of sample screening, C. perfringens isolates and incidence (colonization data) in this multi-center collaborative study. Supplementary Table 2 Clinical metadata. Supplementary Table 3 Genome assembly statistics. Supplementary Table 4 Metadata for in vivo metagenomics study. Supplementary Table 5 GenBank accessions for virulence genes. Supplementary Table 6 Links to databases used in the present study.
Source data
Source Data Fig. 1
Statistical source data.
Source Data Fig. 2
Statistical source data.
Source Data Fig. 3
Statistical source data.
Source Data Fig. 4
Statistical source data.
Source Data Fig. 5
Statistical source data.
Source Data Fig. 6
Statistical source data.
Source Data Extended Data Fig. 1
Statistical source data.
Source Data Extended Data Fig. 2
Statistical source data.
Source Data Extended Data Fig. 3
Statistical source data.
Source Data Extended Data Fig. 4
Statistical source data.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Kiu, R., Shaw, A.G., Sim, K. et al. Particular genomic and virulence traits associated with preterm infant-derived toxigenic Clostridium perfringens strains. Nat Microbiol 8, 1160–1175 (2023). https://doi.org/10.1038/s41564-023-01385-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41564-023-01385-z
This article is cited by
-
Simulated digestions of free oligosaccharides and mucin-type O-glycans reveal a potential role for Clostridium perfringens
Scientific Reports (2024)
-
Detection of Clostridium perfringens in donor milk at a human breast milk bank: a case report
BMC Infectious Diseases (2023)
-
Superior performance of biofilm versus planktonic Limosilactobacillus reuteri in protection of the intestines and brain in a piglet model of necrotizing enterocolitis
Scientific Reports (2023)
