Clostridium celatum WC0700, Clostridium tertium WC0709, and Paraclostridium bifermentans WC0705 were isolated from fecal cultures enriched on mucin, in order to identify mucin-degrading bacteria (Raimondi et al., 2021b). The strains were cultured in mucin medium (MM) (Raimondi et al., 2021b) in anaerobic conditions into butyl-rubber stoppered tubes to determine the extent of grow in terms of OD₆₀₀ units.

The ability of the three strains to grow fermenting the single monomeric units that compose mucin oligosaccharides was assessed on basal M17 broth (BD Difco, Sparks, USA) supplemented of D-Glucose, D-Galactose (Sigma, Darmstadt, Germany), L-Fucose, Neu5Ac, GlcNAc, or GalNAc (Carbosynth, Staad, Switzerland), at the final concentration of 20 mM. M17 medium without any carbon source was used as negative control. The pH of the media was adjusted to 6.9–7.2 with 1 M HCl, then 12 mL of medium were dispensed into butyl-rubber stoppered tubes and sterilized. To obtain inocula where the carbon source was depleted, C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 were grown for two consecutive steps on M17 medium with lactose (2.9 mM) at 37°C for 48 h. These cultures were used to inoculate (10% v/v) tubes containing M17 supplemented with the diverse carbohydrates and the negative controls. For each medium, three subcultures were carried out, each one in triplicate, incubating the cultures at 37°C for 48 h. At the end of each step, OD₆₀₀ was measured. In order to assess whether the fermentation pathways produced hydrogen, qualitative analysis of the headspace was carried out with a μGC 3000A (Agilent Technologies, Milano, Italy) under the following conditions: injector temperature 90°C; column temperature 60°C; sampling time 20 s; injection time 50 ms; column pressure 25 psi; run time 45 s and nitrogen as carrier gas.

Organic acids of the culture supernatants were quantified by HPLC with refractive index detector (1,200 System, Agilent Technologies, Waldbronn, Germany) and Aminex HPX-87 H ion exclusion column. Isocratic elution was carried out at 60°C with 0.8 mL min⁻¹ of 5 mM H₂SO₄ (Amaretti et al., 2013).

C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 were tested for the fermentation of 49 carbohydrates and other carbohydrate derived molecules using API 50 CH test strips (bioMerieux, Marcy, l’Etoile, France). Bacterial biomass from the surface of M17-glucose agar plates (BD Difco, Sparks, USA) was harvested and resuspended in API 50 CHL medium at 2 McFarland units. The suspension was used to inoculate the strips, according to manufacturer instructions. Growth occurred in the anaerobic chamber for 48 h at 37°C.

The ability of C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 to form biofilm and to adhere to a mucin covered surface was tested. Purified mucin type III (Sigma-Aldrich, Darmstadt, Germany) was dissolved in phosphate-buffered saline (PBS) pH 7.4 to a final concentration of 1 mg/mL. Each well of 96-well polystyrene microtiter plates was loaded with 200 μL of the mucin suspension; the plates were maintained overnight at 4°C. Unbound mucin was removed washing wells with PBS, according to Sadiq et al. (2021). Cultures grown anaerobically in M17-glucose for 48 h were diluted (10% v/v) in fresh M17-glucose medium and seeded in mucin-coated and uncoated wells. Plates were incubated in anaerobiosis for 48 h at 37°C and biofilm formation was assayed by crystal violet (CV) staining according to Raimondi et al. (2019). Briefly, unattached cells were discarded, and each well was washed three times with PBS. Biofilm was stained with CV solution (0.1%) for 15 min, then excess of staining was removed and wells were washed three times with PBS. De-staining solution (80% v/v ethanol, and 20% v/v acetone) was added to release the stain, and biofilm was quantified by measuring OD₅₇₀. Specific biofilm formation (SBF) index was calculated as the ratio between CV absorbance at 570 nm and culture’s turbidity at 620 nm, setting a threshold of 1. The strains Escherichia coli 03.73 and Klebsiella pneumoniae 11.71 (Raimondi et al., 2019; Amaretti et al., 2020) were used as positive and negative controls, respectively.

Tolerance to oxygen and to high temperatures of C. celatum WC0700, C. tertium WC0709, P. bifermentans WC0705 were tested on 48 h cultures grown in M17-glucose broth at 37°C. Portions of the same culture were: (I) exposed to air in a 10x volume baffled flask maintained at 37°C for 1 h in an orbital shaker (180 rpm); (II) heated at 80°C for 30 min; (III) exposed to air, then heated at 80°C for 30 min (I + II). After each treatment, serial dilutions in M17-glucose were made in anaerobic tubes, that were incubated at 37°C for 48 h. The Most Probable Number (MPN) method was used to estimate microbial population size, using untreated cultures as controls.

The susceptibility to antibiotics of C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 was assayed with broth microdilution method according to the International Standard Organization (ISO 20776-1:2019), Cultures grown in M17-glucose medium for 48 h at 37°C were diluted in fresh M17-glucose broth to obtain a final concentration of 5 × 10⁵ cfu/mL. Ampicillin, gentamicin, and chloramphenicol were tested at doubling dilutions, from 0.06 to 64 mg/L; penicillin G and tetracycline from 0.015 to 16 mg/L. The antibiotic susceptibility was determined after 48 h of incubation at 37°C in anaerobiosis. MICs were defined according to the European Committee on Antimicrobial Susceptibility Testing breakpoints (EUCAST)¹ for Gram-positive anaerobes, except for tetracycline and penicillin G, for which breakpoints were defined according to Clinical and Laboratory Standards Institute (CLSI)². No defined breakpoints were available for gentamicin in the two databases. Indicatively, the gentamicin breakpoint of 16 was chosen, according to Wei et al. (2020), referring to the CLSI-M100-S23:2019 database.

The data reported are the means of at least three independent experiments, each carried out in triplicate. The statistical significance was analyzed with t-test (p < 0.05). Statistical analysis was performed using GraphPad Prism 7 software (GraphPad, San Diego, CA, United States). Statistical differences were assessed by one-way ANOVA and Bonferroni multiple-comparison post hoc tests. Differences were considered significant at p < 0.05.

Bacterial cells of C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 grown in MM at 37°C for 48 h under strictly anaerobic conditions was collected by centrifugation for 10 min at 12,000 g. The genomic DNA was extracted with DNeasy Blood & Tissue kit (Qiagen GmbH, Düsseldorf, Germany). Before DNA purification, the pre-treatment for Gram-positive bacteria was performed following manufacturer’s specific, with some modifications: longer incubation times (2 h at 37°C and 1 h at 56°C) and twofold the volume of enzymatic lysis buffer, proteinase K and Buffer AL. The quality of the DNA was checked with a Nanodrop spectrophotometer, and the concentration was quantified with a Qubit 3.0 fluorimeter (Thermo Fisher Scientific, Waltham, MA, USA). The samples were sequenced with Illumina NovaSeq 6000 by Eurofins Genomics (Ebersberg Germany). For each sample, 150-bp paired-end reads were obtained. Assembled genomes of C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 were submitted to NCBI with the accession number, PRJNA781812, PRJNA737738, PRJNA781829, respectively. Strains WC0700 and WC0705 were sequenced for this work, while WC0709 was published previously (Musmeci et al., 2021).

The raw reads were checked for quality with FastQC v0.11.8 (Andrews, 2010). To trim Illumina adapters and remove reads with a low quality score, Cutadapt v1.16 was used with the following parameters: overlap = 15, minimum length = 30 and quality-cutoff = 20 (Martin, 2011). The trimmed reads were assembled with SPAdes v3.13.0 (−-careful --cov-cutoff auto -k auto) (Bankevich et al., 2012). The quality of the assemblies was evaluated using QUAST 5.0.2 (Gurevich et al., 2013). Trimming, quality checking, and assembly were performed on Galaxy platform³ (Afgan et al., 2018). The taxonomy was assessed with TypeMat of Microbial Genomes Atlas (MIGA; Rodriguez-R et al., 2020). The completeness of the assemblies was verified with CheckM 1.0.8 (Parks et al., 2015).

The annotation was carried out with the tool Prokka v1.14.5 (Seemann, 2014) and the hypothetical proteins were further annotated with eggnog-mapper v2 (Huerta-Cepas et al., 2019; Cantalapiedra et al., 2021) and InterPro Scan 5.60–92.0 (Jones et al., 2014; Blum et al., 2021). KEGG tools, BlastKOALA and MAPPER (Kanehisa et al., 2016; Kanehisa and Sato, 2020), and the tool gapseq (Zimmermann et al., 2021) were used to predict functional properties of the genomes. The dbCAN2 metaserver was used for the annotation of Carbohydrate Active Enzyme (CAZymes) (Zhang et al., 2018), to identify the genes encoding glycoside hydrolases (GH), carbohydrate esterases (CE), glycosyl transferases (GT), carbohydrate-binding modules (CBM), and polysaccharide lyases (PL). Only the CAZymes annotated by two out of the three tools (HMMER, Hotpep, and Diamond) used by dbCAN2 were considered. The putative glycoside hydrolases involved in the mucin degradation (GH2, GH20, GH29, GH33, GH35, GH42, GH84, GH85, GH89, GH95, GH98, GH101, and GH129) were searched. Enzymes involved in mucin monomers (D-Galactose, L-Fucose, Neu5Ac, GlcNAc, and GalNAc) utilization were also searched through BLAST using the sequences identified by Ravcheev and Thiele (2017) as references.

In order to investigate the presence of an adaptive immunity systems, CRISPRs and Cas genes were searched using CRISPRCasFinder (Couvin et al., 2018), with default settings and subtype clustering of Cas genes. The presence of prophage sequences was analyzed with PHASTER (Arndt et al., 2016). To assess the potential of these strains to share genetic resistance determinants, the identification of antibiotic resistance genes was carried out using the web tool RGI (Resistance Gene Identifier) of CARD (Comprehensive Antibiotic Resistance Database), processing the contigs file for “Perfect, Strict and Loose hits” (Alcock et al., 2020). Insertion sequences were identified with ISFinder (Siguier et al., 2006). BAGEL 4 server was used to search bacteriocins genes (Van Heel et al., 2018).

The presence of putative virulence factors was investigated with Virulence factor database (VFDB; Liu et al., 2022) to assess the pathogenic potential of these strains. This tool allowed also to compare the pathogenicity of the three strains against C. perfringens and Clostridioides difficile. Pathogenicity and virulence factors were also investigated with PathogenFinder v1.1 (Cosentino et al., 2013).

The available genomes of C. celatum, C. tertium, and P. bifermentans strains were downloaded from GenBank (Supplementary Table S1) on 16 January 2023. Metagenome-assembled genomes (MAG) and genomes with a contamination >5% after quality check with CheckM were excluded. As a whole, comparative genomics investigated 4 genomes of C. celatum, 23 of P. bifermentans, and 12 of C. tertium. Average Nucleotide Identity (ANI) and digital-DNA/DNA hybridization (dDDH) were calculated using the ANI Matrix web tool (Rodriguez-R and Konstantinidis, 2016) and Genome-to-Genome Distance Calculator GGDC 2.1,⁴ with the thresholds for species demarcation of 95% for ANI and 70% for dDDH (Richter and Rossello, 2009). Panaroo was utilized to calculate the pangenome and to define core and accessory genes, utilizing Prokka annotation files (Tonkin-Hill et al., 2020). Default settings were applied to Panaroo’s runs (--clean-mode strict, --core_threshold 0.98, --len_dif_percent 0.98). Genes in the pangenome were categorized as core if present in all the strains, soft core if present in 95–99% of strains, shell genes in 95–15% of strains, cloud genes in less than 15% of strains.

The taxonomy of the strains WC0700, WC0709, and WC0705, which were initially identified as Clostridium disporicum, Clostridium tertium, and Paraclostridium benzoelyticum based on partial 16S rRNA gene sequencing (Raimondi et al., 2021b), was re-evaluated using TypeMat. The proper taxonomy of the isolates was Clostridium celatum WC0700, C. tertium WC0709, and Paraclostridium bifermentans WC0705. To confirm the taxonomic assignment, digital DNA–DNA hybridization (dDDH) and average nucleotide identity (ANI) values were calculated comparing each isolate with the relative reference strain available (Table 1). All the isolates were over the thresholds for species demarcation (70 and 95% for dDDH and ANI, respectively; Richter and Rossello, 2009).

In the 3 genomes, whose main features are presented in Supplementary Table S2, a high number of proteins were annotated as functionally uncharacterized hypothetical proteins using Prokka (41.9–44.8%). A deeper annotation carried out with eggnog-mapper and InterProScan allowed to reduce the number of hypothetical proteins to 12.7–15.7%.

None of the three strains possessed CRISPR-Cas systems (Stern et al., 2012). On the other hand, a number of genes encoding restriction systems were identified in the genomes of C. celatum WC0700 (22), C. tertium WC0709 (8), and P. benzoelyticum WC0705 (12). Only in the genome of C. celatum WC0700 a prophage sequence was identified. Seven ISs were identified in C. celatum WC0700, belonging to the families IS1595, IS6, and IS200/IS605. They presented similarities with others from Campylobacter jejuni and C. disporicum (IS1595), C. tetani (IS6), and C. perfringens (IS200/IS605). P. bifermentans WC0705 encompassed two ISs sequences of the families IS607 (C. botulinum) and IS1182 (Bacillus cytotoxicus). ISs of the families IS3 and IS200/IS605 were found in C. tertium WC0709 and presented similarities with the counterparts of C. beijerinckii and Staphylococcus epidermidis, respectively.

The pangenome of the three species was also analyzed, retrieving high quality genomes from GenBank (Supplementary Table S2; Supplementary Figure S1A). Given the low number of available genomes, for C. celatum it was not possible to calculate the γ value to establish if it was a closed or open pangenome (Supplementary Figure S1B). According to Heap’s law, the pangenomes of P. bifermentans and C. tertium were considered open, with a γ value of 0.44 and 0.53, respectively, (Tettelin et al., 2008).

The metabolic capabilities of C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 were assessed using API 50 CH (Table 5). C. tertium WC0709 exhibited the widest range of substrates. P. bifermentans WC0705 was the sole strain growing on gluconate, and 2-ketogluconate, while only C. celatum WC0700 fermented D-trehalose and L-fucose.

To evaluate the ability of C. celatum WC0700, C. tertium WC0709, and P. bifermentans WC0705 to ferment the monomers of mucin O-glycan, the strains were cultured in M17 containing the single monosaccharides L-fucose, Neu5Ac, Gal, GalNAc, and GlcNAc (Figure 3). C. celatum WC0700 grew on L-fucose, Gal, and GlcNAc. C. tertium WC0709 grew on Neu5Ac, Gal, GlcNAc and GalNAc. C. celatum WC0700 was not able to growth on GalNAc, whereas P. bifermentans WC0705 grew only on GalNAc and Gal.

To characterize mucin utilization, the strains were cultured in MM, where mucin represented the sole source of both C and N. The maximum biomass yield was generally reached after 24 h, except for C. tertium WC0709, which grew for 72 h (Figure 4). Mucin fermentation generated formic, acetic, propionic, butyric, isovaleric, and lactic acids, and ethanol. The abundance of each fermentation product varied among the strains, with acetic acid being generally the main primary metabolite generated. C. celatum WC0700 mainly produced acetic, formic, and propionic acids during growth, that reached 0.5, 0.2, and 0.1 g/L after 48 h, respectively. Growth was accompanied by the generation of isovaleric and lactic acids, both peaking at 24 h (0.2 and 0.06 g/L respectively), and by minor amounts of the other metabolites (always <0.1 g/L).

The main products of P. bifermentans WC0705 were acetic, propionic, formic, and isovaleric acid, which reached 0.25, 0.14, 0.07, and 0.05 g/L, respectively, after 48 h. In the cultures of C. tertium WC0709, acetic acid was the most abundant fermentation product, with 0.27 g/L after 72 h. Propionic and isovaleric acids were also abundantly produced in the first 24 h (0.15 and 0.16 g/L, respectively) but slightly decreased toward the end of the fermentation. In all the processes, the analysis of the headspace revealed the accumulation of CO₂ and H₂ as volatile fermentation products (data not shown).

The isovaleric produced in the first 24 h of fermentation decreased in all the strains, likely entering the leucine degradation pathway with energy consumption.

Genetic determinants for antibiotic resistances were searched in the genome of C. celatum WC0700, P. bifermentans WC0705, and C. tertium WC0709. The three genomes presented several genes encoding components of eukaryotic type ABC transporters, putatively involved in multidrug efflux system (EfrA/B, MdlA, BplA) and possibly participating in antibiotic resistance mechanisms (Supplementary Spreadsheet 1). The CARD analysis revealed the presence of tetracycline resistance in P. bifermentans WC0705, while KEGG detected β-lactam resistance in C. tertium WC0709. To confirm genomic findings, MICs estimation was performed applying the microdilution method. C. tertium WC0709 was shown to be resistant to ampicillin (Table 6) and presented intermediate resistance to penicillin. All the strains were susceptible to chloramphenicol. C. tertium WC0709 and P. bifermentans WC0705 were resistant to tetracycline. C. celatum WC0700 was susceptible to all the antimicrobials tested. Interestingly, all the strains exhibited growth in presence of high concentration of gentamicin, but breakpoints were not available for defining susceptibility or resistance to this antibiotic.

The three strains were tested for biofilm formation in presence and absence of mucin (Figure 5). The SBF of C. tertium WC0709 was under the threshold of 1, suggesting the inability to form biofilm in both conditions. Compared to the positive control E. coli 03.73, slight biofilm was produced by C. celatum WC0700 and P. bifermentans WC0705, with scores of 2.0 and 2.8, respectively. No statistically difference (paired samples t-test, p < 0.05) was observed in presence of mucin coating.

Survival under oxygen and high temperature exposure was assessed. Thermic and oxygen treatments, separately or combined, caused the decrease of one Log₁₀ of the MPN, suggesting that the majority of the cells on the 24 h cultures were resistance spores (Supplementary Table S4).