Abstract
Clostridium is a vital gut anaerobe in giant pandas (GPs), aiding bamboo digestion and gut homeostasis. The present study optimizes anaerobic culturing to isolate Clostridium species from GPs, evaluating their ecological roles in bamboo digestion while assessing associated pathogenic and antibiotic resistance threats. The results show that the enriching samples in liquid media facilitated the isolation of Clostridium species. A total of 14 species are obtained, with C. perfringens, C. sardiniense, and C. baratii being most prevalent. 86.30% of strains exhibit lignocellulose-degrading activity, with all C. butyricum strains displaying activity for β-glucosidase, xylanase, and manganese peroxidase. Genomic analysis identifies carbohydrate-active enzymes and metabolic pathways involved in lignocellulose degradation, short-chain fatty acid production, and essential amino acid biosynthesis. C. butyricum possesses the most hemicellulose- and cellulose-degrading genes. We also identify 19 antibiotic resistance genes (ARGs), predominantly glycopeptide-resistant van genes, and 23 virulence factors (VFs) encoded by 408 virulence genes (VGs). Notably, C. perfringens harbors the most ARGs and VFs, some of which are flanked by mobile genetic elements, suggesting risks of horizontal gene transfer. Overall, this study describes the dual role of Clostridium in GPs, contributing to dietary adaptation while also posing potential hazards due to pathogenic traits and antimicrobial resistance.
Similar content being viewed by others
Introduction
The giant panda (GP) (Ailuropoda melanoleuca) gastrointestinal tract harbors a complex and diverse gut microbiota, consisting of bacteria, fungi, protozoa, and viruses, that regulates many essential physiological processes and plays a vital role in the health of GPs1. Among these microbial constituents, bacterial communities have received special attention due to their predominance and functional significance. In recent years, next-generation sequencing (NGS) technology, including 16S rRNA, metagenomic, and meta-transcriptomes, has enhanced our understanding of the compositional and functional dynamics of the gut microbiota of GPs2,3,4. For example, it is widely recognized that three phyla—Firmicutes, Bacteroidetes, and Proteobacteria—dominate the gut microbiota of captive and wild GPs3,5. Additionally, research on the co-evolution between GPs and their gut symbionts demonstrated that the gut microbiome in captive GPs possessed more genes related to multi-drug resistance, whereas the gut microbiome in wild GPs possessed more genes for host genetically specific structures and functions6.
Numerous metabolic processes in healthy colonocytes involve high-oxygen consumption, which contributes to maintaining an anoxic or anaerobic gut environment, thereby providing a favorable habitat for an abundance of anaerobic bacteria7,8. Culture-independent sequencing approaches have identified several anaerobic genera in the gut of GPs, such as Clostridium, Bifidobacterium, Terrisporobacter, and Veillonella, with Clostridium being the predominant genera3. Members of the genus Clostridium are Gram-positive, endospore-forming, obligate anaerobes that have been reported to have both positive and negative effects on gut homeostasis and body health via directly or indirectly interacting with the other resident microbial populations, as well as providing fundamental and specific functions, especially in metabolic and immune processes9,10. Moreover, Clostridium has been shown to exert metabolic effects that positively influence gut inflammation, cancer therapy, and energy metabolism11. Clostridium colonizes the GP gut during breastfeeding early in infancy and becomes a member of the gut microbiome following weaning onto a bamboo diet5. Clostridium is understood to aid the adaptation of GPs to a high-fiber diet, as it is an effective lignocellulose degrader12,13. However, further studies are needed to determine the prevalence of lignocellulose-degrading Clostridium in GPs and if different Clostridium species have varying lignocellulose-degrading capabilities.
The genus Clostridium also includes opportunistic pathogens that are capable of causing gut diseases, largely dependent on the various types of virulence factors (VFs) produced14,15. These pathogens can induce histotoxic infections when toxins enter the bloodstream and damage internal organs16. Antibiotic treatments are frequently used to treat these infections, which may result in the storage and transmission of antibiotic resistance genes (ARGs) in the gut microbiota, raising concerns about increasing resistance among Clostridium species17,18. Current understanding of ARGs and VFs in giant panda gut microbiota primarily stems from metagenomic studies, which revealed that the genus Clostridium contributes to the ARG pool, with particular species emerging as dominant reservoirs6,19. However, existing studies have primarily focused on comparative analyses of ARG and VF profiles across geographical populations and management conditions (captive, reintroduced, and wild), rather than strain-level characterization20,21. Consequently, cultivation-based studies systematically evaluating ARG and VF distribution among different Clostridium species remain limited. This gap underscores the need to understand VFs and ARGs distribution within the Clostridium genus to develop more effective therapeutic strategies for relevant diseases.
In recent years, the generation of genomic binning has made it possible to reconstruct microbial genomes from shotgun metagenomic sequencing data, which provides insights into the microbial communities and their potential functions at the species level22,23. A large number of metagenome-assembled genomes (MAGs) have been recovered from the gut of GPs. These MAGs were comprised of several bacterial strains within the Clostridium species, including C. perfringens, C. ventriculi, C. paraputrificum, and C. neonatale4. However, metagenomic assembly approaches only catch a limited portion of bacterial populations, leaving minority species unidentified. Thus, the majority of MAGs remain unclassified at the species level24. Given that a substantial proportion of gut bacteria are considered ‘unculturable’25, there is a need to optimize culturing methods in order to expand our understanding of the Clostridium species.
Microbial cultivation remains indispensable for isolating specific bacterial strains and providing a continuous supply of cells for phenotypic and genetic analyses. This approach enables the determination of bacterial features, such as growth characteristics, physiology, and metabolism, as well as the interactions between these features26. With cultivation, the functional attributes of bacteria inferred from genomic data could be validated in experiments, improving reproducibility and statistical confidence27. Compared with aerobic bacteria, obligate anaerobic bacteria from the gut of GPs are more difficult to culture. Fecal samples must be collected and processed in an oxygen-free environment immediately after defecation to minimize exposure to air. Additionally, the diversity of cultivable strains is strongly influenced by media selection and formulation. Although previous researches have established optimized culture strategies for isolating gut anaerobes from humans, monkeys, and mice28,29,30, including the development of appropriate selective media, physicochemical parameters, and growth factors, these approaches remain largely untested for giant pandas (GPs). The specialized bamboo diet of GPs creates a unique intestinal microenvironment with distinct nutritional requirements that differ from those of humans or model animals. This emphasizes the importance of developing culture media that better mimic the gut environment of GPs.
Although targeted isolation methods have successfully recovered specific Clostridium species from human and animal guts31,32, including the isolation of C. butyricum from GPs11, few studies have systematically evaluated the cultivable diversity and functional characteristics of Clostridium in GPs.
In this study, we optimized the anaerobic culture conditions for Clostridium to assess the diversity, composition and genomic feature of culturable Clostridium in GPs of different age groups (sub-adult, adult, and geriatric). The genotype and phenotype of the Clostridium strains, as well as their lignocellulose-degrading capabilities, were assessed. We further investigated the prevalence of VFs and ARGs to provide a better understanding of the functional and pathogenic potential of Clostridium species in the GP gut microbiome.
Results
Cultured Clostridium from the gut of GPs
Duplicate Clostridium species from the same samples isolated via the same cultural strategy (e.g., isolated from GAM without pre-incubation) were removed. In total, 98 Clostridium isolates were used to assess the various cultural strategies (Supplementary Table 1). Based on the alignment of 16S rRNA gene sequences against the EzBioCloud and NCBI BLAST databases (with a >97.0% similarity threshold for species assignment), the isolates were classified into 14 Clostridium species, including C. perfringens (n = 25), C. sardiniense (n = 19), C. baratii (n = 11), C. nigeriense (n = 9), C. paraputrificum (n = 7), C. butyricum (n = 6), C. neonatale (n = 5), C. sporogenes (n = 5), C. saudiense (n = 3), C. ihumii (n = 2), C. moniliforme (n = 2), C. tertium (n = 2), C. carnis (n = 1), and C. sartagoforme (n = 1). The most prevalent species detected among GPs were C. perfringens (61.29%, 19/31), C. sardiniense (45.16%, 14/31), and C. baratii (29.03%, 9/31).
We further compared the distribution of Clostridium species across different age groups (Fig. 1). In general, adult GPs (n = 11) exhibited a greater diversity of Clostridium species as compared to sub-adult (n = 9) and geriatric (n = 6) GPs. Three species, C. perfringens, C. sardiniense, and C. sporogenes, were common to all age groups. In contrast, C. carnis and C. moniliforme were only isolated from adult GPs, and C. sartagoforme and C. saudiense were only isolated from geriatric GPs.
SGP: sub-adult giant panda, AGP: adult giant panda, GGP: geriatric giant panda.
Evaluation of the culture conditions for Clostridium isolation
Here, we found that 72 isolates, representing 14 species, were obtained with pre-incubation treatment, whereas 27 isolates, representing 7 species, were isolated without pre-incubation treatment (Fig. 2). Considering the effect of culture agars, most Clostridium species were produced from TM (n = 10), followed by RCM (n = 9) and GAM (n = 8). Fewer Clostridium species were isolated from CM (n = 6) and YCFA (n = 3). The results further illustrated that C. sartagoforme and C. carnis were exclusively recovered from GAM and TM agars, respectively, and C. saudiense and C. tertium were isolated only from RCM agar.
TM: thiolglycollate medium, GAM: Gifu anaerobic medium, RCM: reinforced clostridial medium, YCFA: YCFA medium, CM: Columbia medium. *Without pre-incubation treatment; **with pre-incubation treatment.
Without pre-incubation treatment, RCM (n = 5) and GAM (n = 3) agar obtained the most Clostridium species, whereas fewer Clostridium species were identified from YCFA (n = 1) and TM (n = 3) as compared to RCM or GAM agar. However, no Clostridium was detected on CM agar. Under pre-incubation treatment, we found that more Clostridium species grew on RCM agar (n = 9) and TM agar (n = 9), followed by GAM agar (n = 7), CM agar (n = 6), and YCFA agar (n = 3). Additionally, all Clostridium species identified without pre-incubation treatment were also found after pre-incubation treatment, and 7 species, including C. ihumii, C. butyricum, C. carnis, C. moniliforme, C. neonatale, C. sartagoforme, and C. tertium were only identified after pre-incubation enrichment conditions. Of note, the combination of RCM, TM, and GAM with pre-incubation isolated all 14 Clostridium species. Thus, the combination of pre-incubation strategies and certain culture media was effective in increasing both the diversity and richness of culturable Clostridium.
Analysis of lignocellulose degradation
After dereplicating the initial collection of 98 Clostridium strains (i.e., removing same-species isolates from the same GP), 73 strains were retained for further testing (SGP, n = 17; AGP, n = 46; GGP, n = 10). Notably, all of the Clostridium strains were identified via the primary screening as having the potential ability to degrade cellulose (79.45%, n = 58), hemicellulose (54.79%, n = 40), and lignin (57.53%, n = 42), as evidenced by exhibiting corresponding β-glucosidase, xylanase, and MnP activities (Fig. 3a). In the SGP and AGP groups, there were more Clostridium strains with cellulose-degrading ability than those with hemicellulose- and lignin-degrading abilities. Also, Clostridium strains with lignin-degrading ability were less prevalent than those with hemicellulose- and cellulose-degrading abilities in the GGP group. Most Clostridium (35.62%, n = 26) possessed all three types of lignocellulose enzymatic capabilities, followed by 34.25% (n = 25) showing two types, and 16.44% (n = 12) showing one type of activity. Additionally, 13.70% (n = 10) of Clostridium lacked the capacity to degrade lignocellulose (Fig. 3b). In the SGP group, all strains exhibited at least two enzymatic activities, with some strains demonstrating all three enzymatic activities. In the AGP and GGP groups, 84.44% and 80.00% of Clostridium strains, respectively, possessed at least one enzymatic activity, with a higher proportion of strains exhibiting two enzymatic activities (Fig. 3b). The mean value of β-glucosidase activity in the AGP group was higher than that in the SGP and GGP groups, whereas the mean value of xylanase and Mnp activities was lower than that observed in the SGP and GGP groups. However, three enzymatic activities did not show significant differences among the various groups (p > 0.05) (Fig. 3c).
a Prevalence of cellulose-, hemicellulose-, and lignin-degrading abilities in Clostridium isolated from giant pandas of different ages via primary screening. b Proportion of Clostridium strains by enzyme type (β-glucosidase, xylanase, and MnP activities) in different age groups. c Determination of xylanase, β-glucosidase, and manganese activities (U/mL) in the different age groups of giant pandas. Data are presented as mean ± SD. d Determination of xylanase, β-glucosidase, and manganese activities (U/mL) from different Clostridium species. Data are presented as mean ± SD. SGP: sub-adult giant panda, AGP: adult giant panda, GGP: geriatric giant panda.
The enzymatic activities for lignocellulose degradation varied among Clostridium species (Fig. 3d). All species, except C. sartagoforme, exhibited β-glucosidase activity, with particularly high enzymatic activity observed in certain strains of C. nigeriense (n = 30.37 U/mL), C. paraputrificum (n = 20.19 U/mL), and C. sardiniense (n = 20.62 U/mL). A higher level of xylanase activity was observed in certain strains of C. butyricum (n = 30.49 U/mL), C. perfringens (n = 28.62 U/mL), and C. baratii (n = 28.55 U/mL), whereas no detectable xylanase activity was observed in C. carnis and C. moniliforme. MnP activity assessment identified C. perfringens strains as the most efficient producers, whereas C. sartagoforme, C. sardiniens, and C. moniliforme strains displayed no detectable MnP activity. Statistical analysis of species with ≥5 measurable replicates revealed significantly higher MnP activity in C. perfringens compared to C. baratii (p < 0.05). Nine different Clostridium species with three lignocellulose enzymatic activities were observed. Interestingly, all C. butyricum strains tested showed activity for all three lignocellulose-degrading enzymes, underscoring its exceptional lignocellulolytic potential. In contrast, C. sartagoforme, C. sardiniens, and C. moniliforme lacked detectable activity for one or more enzymes, indicating their limited lignocellulose degradation capability.
Whole genome sequencing of lignocellulose-degrading Clostridium
General genomic features of strains
Genome sequencing was performed on 26 Clostridium strains exhibiting all three lignocellulose-degrading enzyme activities. All genomic data have been deposited in BioProject database at NCBI under accession number PRJNA1258569. The assembly statistics and genome characteristics are presented in Supplementary Table 2. There were 169–2691 contigs (734 in average) yielded from Clostridium strains, with the total length ranging from 2,802,519 to 5,458,347 bp (3,884,414 bp in average). The N50 of the assembled genomes ranged from 100,005 to 2,363,697 bp (852,980 bp in average). The GC content ranged from 26.95% to 31.44% (29.05% in average), and the number of coding sequences (CDS) ranged from 2632–4562 (3396 in average) in the Clostridium strains.
Functional analyses of lignocellulose-degrading Clostridium
We annotated the genes of the Clostridium strains using the CAZy and KEGG databases to better understand their functions. There were 4294 genes identified to encode 211 CAZyme families, in which most were related to glycoside hydrolases (GHs) (n = 114), followed by carbohydrate-binding protein module (CBM) (n = 41), glycosyl transferases (GTs) (n = 28), polysaccharide lyases (PLs) (n = 9), carbohydrate esterases (CEs) (n = 14), and auxiliary activities (AAs) (n = 5). We further analyzed 41 CAZyme families that exhibited potential activity in degrading lignocellulose, including 30 GHs, 7 CEs, and 4 AAs (Fig. 4). Among these, we found that 28 of the 41 CAZyme families were involved in hemicellulose degradation, harboring enzymes, such as acetylxylan esterase (EC 3.1.1.72), xylan β-1,4-xylosidase (EC 3.2.1.37), acetylesterase (EC 3.1.1.6), endo-β-1,4-xylanase (EC 3.2.1.8), α-L-arabinofuranosidase (EC 3.2.1.55), and feruloyl esterase (EC 3.1.1.73). The CE4 (acetylxylan esterase), CE1 (feruloyl esterase, acetylxylan esterase), and CE3 (acetylxylan esterase) families were present in all Clostridium strains. Enzymes, such as endo-β-1,4-glucanase (EC 3.2.1.4), β-glucosidase (EC 3.2.1.21), exo-β-1,4-glucanase (EC 3.2.1.74), exo-β-1,3-glucanase (EC 3.2.1.58), cellobiohydrolase (EC 3.2.1.91), and cellodextrinase (EC 3.2.1.74)—all related to cellulose breakdown—were found in 15 CAZyme families, with GH1 (exo-β-1,4-glucanase, cellodextrinase, β-glucosidase), and GH4 (β-glucosidase) exhibiting the highest gene counts. Additionally, we identified 6 CAZyme families involved in both cellulose and hemicellulose degradation. Furthermore, 4 AA families (AA1/2/4/6) with [copper-containing] dihydrogeodin oxidase (EC 1.10.3.-), laccase (EC 1.10.3.2), manganese peroxidase (EC 1.11.1.13), p-benzoquinone reductase (NADPH) (EC 1.6.5.6), and vanillyl-alcohol oxidase (EC 1.1.3.38), which all contribute to lignin degradation, were observed. The number of CAZyme families varied greatly among Clostridium strains, ranging from C. ihumii A467 (n = 7) to C. neonatale B160 (n = 27). Within the same species, several similarities were found. For instance, several CAZyme families, including GH43_4/43_10/10/115/30_2/67, which are all involved in the breakdown of hemicellulose, were specific to C. neonatale. Additionally, C. butyricum was shown to have more CAZyme families involved in cellulose breakdown, with GH5_44 and GH16_21 being unique to this species.
Distribution of CAZymes associated with lignocellulose degradation in different species.
Of all Clostridium strains, a total of 245 KEGG pathways were obtained and assigned into 6 categories at KEGG level 1. The majority of these pathways were associated with metabolism (n = 97–112), followed by human diseases (n = 35–41), and organismal systems (n = 21–29). Fewer pathways were linked to genetic information processing (n = 14), environmental information processing (n = 12–15), and cellular processes (n = 11–17). This distribution underscores the central role Clostridium plays in metabolic functions within the gut of GPs. Subsequently, we focused on the top 35 enriched metabolic pathways, which were selected based on their highest average gene counts across all tested genomes. As shown in Fig. 5a, the number of functional genes in each metabolic pathway was consistent in strains within the same species. Notably, C. butyricum and C. neonatale harbored a greater abundance of functional genes, whereas C. perfringens and C. baratii displayed comparatively fewer. These pathways were classified into 7 categories at KEGG level 2, with the majority of genes enriched in carbohydrate metabolism. The key enzymes involved in the pathway of hemicellulose (xylan, mannose, and galactose) digestion were identified, including xylan 1,4-beta-xylosidase [EC 3.2.1.37], mannose PTS system EIIA component [EC:2.7.1.191], mannose-6-phosphate isomerase [EC:5.3.1.8], aldose 1-epimerase [EC:5.1.3.3], galactokinase [EC:2.7.1.6], UDPglucose--hexose-1-phosphate uridylyltransferase [EC:2.7.7.12], and phosphoglucomutase [EC:5.4.2.2] (Fig. 4b). Similarly, cellulose-digesting enzymes, such as endoglucanase [EC:3.2.1.4], beta-glucosidase [EC:3.2.1.21], cellobiose PTS system EIIA component [EC:2.7.1.205], 6-phospho-beta-glucosidase [EC:3.2.1.86], and glucokinase [EC:2.7.1.2], were detected (Fig. 5b). The distribution of these enzymes varied among different species, with C. butyricum exhibiting more diverse enzymes associated with hemicellulose and cellulose degradation (Fig. 4c). Furthermore, the metabolic products, glucose and fructose, could be fermented to produce short-chain fatty acids (SCFAs)—in particular acetate, propionate, and butyrate—as minor end-products. Within the “amino acid metabolism”, enrichment was observed in arginine biosynthesis and lysine biosynthesis pathways. Clostridium species mainly facilitated arginine synthesis via KEGG module M00028 (glutamate => ornithine) and KEGG module M00844 (ornithine => arginine) (Fig. S1), whereas lysine synthesis was mediated via KEGG module M00016 (succinyl-DAP pathway, aspartate => lysine) and M00527 (DAP aminotransferase pathway, aspartate => lysine) (Fig. S1). Additionally, genes linked to the “metabolism of cofactors and vitamins” were also identified in pathways, such as porphyrin metabolism and folate biosynthesis.
a Distribution of metabolic pathways enriched in various Clostridium species. The gene counts were normalized using z-scores (scale from −2 to +2). b Key enzymes in the metabolic networks involved in hemicellulose and cellulose degradation. c Distribution of key enzymes involved in hemicellulose and cellulose degradation in various Clostridium species. White grids represent species without the enzyme; gray-blue grids represent species with the enzyme.
Antibiotic resistance genes in Clostridium
In general, 197 genes belonging to 19 types of ARGs were annotated in the genomes of Clostridium strains isolated from the gut of GPs (Fig. 6). Most types of ARGs were related to glycopeptide resistance (vanG/H/R/T/W/XY/Y), followed by tetracycline resistance [tet(Q), tetA(P), and tetB(P)], multidrug resistance (cplR, sdrM, and ermQ), and disinfectant resistance (qacG and qacJ). Additionally, the genes gyrB, blaR1, aph (3’), and mprF, which may confer resistance to fluoroquinolone, beta-lactam, aminoglycoside, and peptides, were also observed. Antibiotic target alteration (52.63%, 10/19) was the major resistance mechanism of these ARGs, followed by antibiotic efflux (21.05%, 4/19), antibiotic target protection (15.79%, 3/19), and antibiotic inactivation (10.53%, 2/19). All strains, except C. butyricum A511, possessed the genes vanW and vanT; the genes vanY (73.08% 19/26), cplR (61.54%, 16/26), and gyrB (53.85%, 14/26) were also prevalent. The distribution and prevalence of ARGs varied by Clostridium species. For example, more types of ARGs (n = 7–9) were enriched in C. perfringens, whereas C. butyricum A511 had only 2 ARGs, tetA(P) and tetB(P). The genes vanH and mprF were unique in C. perfringens strains, and vanG, vanR, and aph(3’) were only detected in C. ihumii A467.
Distribution of antibiotic resistance genes found in Clostridium strains.
Furthermore, based on the KEGG analysis of pathways related to antibiotic resistance, we found that most genes were involved in β-lactam resistance (ko01501), and vancomycin resistance (ko01502) was the most abundant among all strains. Clostridium strains that resist β-lactam could result from the regulation of the blaZ gene, which encodes a class A β-lactamase enzyme to hydrolyze the β-lactam ring, or the alteration of the β-lactam targets of penicillin-binding proteins (PBPs), including PBP1a/2, PBP2, and ftsI. Vancomycin resistance pathway is regulated by vanR and vanS (VanS-VanR two-component system), and several KOs, such as K01929 (murF), K01000 (mraY), K02563 (murG), K01921 (ddl), and K01775 (alr), are involved to prevent vancomycin binding by replacing the D-alanyl-D-alanine (D-Ala-D-Ala) target to either D-alanyl-D-lactate (D-Ala-D-Lac) or D-alanyl-D-serine (D-Ala-D-Ser), which results in different levels of resistance to vancomycin.
Prevalence of virulence factors in Clostridium strains isolated from GPs
All strains comprised 23 VFs expressed by 408 VGs, which could be categorized as follows: adherence (n = 7), exoenzyme (n = 4), immune modulation (n = 4), exotoxin (n = 4), motility (n = 2), regulation (n = 1), and stress survival (n = 1) (Fig. 7). Of these, the adherence VFs were more prevalent among Clostridium strains, whereas the VFs, EF-Tu and GroEL, were shared by all strains. The Streptococcal plasmin receptor/GAPDH and FbpA/Fbp68 were found in 92.31% (24/26) and 53.85% (14/26) of strains, respectively. The majority of VFs related to exoenzyme, exotoxin, and regulation were predominantly identified in C. perfringens. Of particular interest were the VFs that were uniquely found in all strains of C. perfringens, including alpha-clostripain, kappa-toxin, mu-toxin, alpha-toxin (CpPLC), and the VirR/VirS two-component system. Additionally, most VGs were found in type IV pili (n = 83), mu-toxin (n = 62), and EF-Tu (n = 46), in which type IV pili covered most types (n = 16), including pilA/M/N/T. Additionally, the most frequently detected VGs in strains were tufA (EF-Tu, 46/408) and groEL (GroEL, 26/408).
Distribution of virulence factors found in Clostridium strains.
Given that the VGs of toxins, which are largely distributed in C. perfringens, are responsible for disease pathogenesis, we further compared the prevalence of toxin genes in C. perfringens strains from herbivores (sheep and bovine) and humans as previously studied (Supplementary Table 3)32. Our results indicated that the majority of toxin genes were found in C. perfringens strains isolated from sheep, followed by bovine, human, and GPs. All 13 toxin genes observed in GPs were also found either in herbivores or humans (Fig. 8a). The detection rates varied among different toxin genes. All strains from GPs, herbivores, and humans had nanH, plc, and cloSI, whereas only 3.64% of C. perfringens strains from sheep had the cpe gene. The detection rate of cpb2 in C. perfringens strains from GPs was significantly higher compared to sheep and humans, and the detection rate of nagL in C. perfringens strains from GPs was significantly higher than in sheep. The detection rate of pfoA in C. perfringens strains from GPs was significantly lower than in sheep (Fig. 8b).
a Detection rate of toxin genes from different hosts. b Significantly different detection rates of toxin genes between giant pandas and humans and herbivores were observed.
Potential mobile genetic elements in genomes
A total of 260 transposases were annotated among genomes (Supplementary Table 4), in which 99 transposases were classified in 16 different IS families, with the majority being found in IS3 (n = 19), IS200/IS605 (n = 18), and IS1595 (n = 15) transposase families. The maximum number of transposases was found in two C. neonatale strains (B155, n = 26; B160, n = 24), whereas the lowest number was found in the two C. baratii strains (n = 2). The number of transposases in different strains of C. perfringens (n = 3–20) and C. butyricum (n = 5–22) varied significantly. Some IS families were found to be shared among genomes. For example, IS3 and IS200/IS605 were prevalent in 42.31% (11/26) and 38.46% (10/26) of strains, respectively. Additionally, ISL3, IS91, IS701, and IS1470 transposase families were uniquely identified in different strains. Furthermore, we examined the genetic background of transposase in contigs of strains to investigate the potential risk of ARGs and VGs dissemination (Fig. 9). We found that the peptide gene, mprF, had a similar genetic background in three C. perfringens (122, 134, C90) and was distributed upstream of transposase insQ, and C. perfringens A216 had the vanY genes downstream of insQ. The distribution of vanY in four C. perfringens strains (122, 134, C90, and C216) was similar, as it sequentially associated with the IS200/IS605 family element transposase accessory protein tnpB.
The genetic environment of antibiotic resistance and virulence genes associated with transposase.
Discussion
Clostridium is a heterogeneous genus of obligate anaerobes belonging to the phylum Firmicutes, which comprises over 300 species (data collected in EzBioCloud database, https://www.ezbiocloud.net/). Different Clostridium patterns have been cultivated from a variety of habitats, such as human and animal intestine, soil, and mud pits28,29,33,34. To improve our understanding of difficult-to-culture Clostridium species and their functions in the guts of GPs while optimizing experimental efficiency, we systematically evaluated culture conditions through parallel comparison of pre-incubation versus non-pre-incubation approaches, and comparative assessment of isolation media types. Our findings demonstrate the effect of a pre-incubation strategy on enhancing bacterial diversity, which aligns with a previous study that highlighted the benefit of culture enrichment in increasing bacterial species35. While Lagier et al. recommended that fecal extraction and sheep blood are crucial growth enhancers in blood culture bottles for human gut anaerobes36, we found pre-incubation in GAM liquid media or blood culture bottles containing sheep blood failed to enhance species diversity compared to RCM and TM liquid media without sheep blood. This underscores the variability in optimal cultivation conditions across different samples, which requires considerable work to screen for appropriate conditions37. Through comparative analysis of five media types commonly used in anaerobic culturing24,31, we identified suboptimal media that could be excluded from further isolation workflows. Ultimately, this study indicated that a combination of pre-incubation and RCM, TM, and GAM agar yielded the highest number of species, providing a valuable reference for standardizing Clostridium isolation protocols in GP gut microbiota research.
Compared with culture-independent studies regarding the Clostridium composition in the gut of GPs4,6, the culture-dependent approach used in this study allowed for outcomes of more Clostridium species, such as C. sardiniense, C. carnis, C. ihumii, and C. moniliforme, thereby expanding the catalog of gut Clostridium genomes in GPs. Under normal circumstances, Clostridium species in the gastrointestinal tract play complex physiological functions and establish a symbiotic relationship with the host14. Based on this, a wide range of mechanistic studies have been conducted in order to improve the development of microbial therapies that maintain gut homeostasis10. As one of the most promising probiotics, evidence on the roles that C. butyricum plays in treating intestinal conditions, such as intestinal injury, irritable bowel syndrome, inflammatory bowel disease, and colorectal cancer, has been summarized38. C. butyricum mediates tolerogenic APC signaling to increase the Treg response, which in turn suppresses proinflammatory effector T cell responses when pathogenic or inflammatory circumstances are present in the intestinal epithelium38. Previous studies have suggested that the C. butyricum isolated from GP feces alleviates colitis in mice by increasing the expression of intestinal barrier proteins and the abundance of probiotic species and by inhibiting immune responses11. Furthermore, the main metabolite of the intestinal epithelium, butyrate, not only reduces epithelial damage in the colonic tissue of diarrhea mice by increasing mucin production via enhancing the expression of the MUC gene39, but also potentially by increasing the body weight of GPs by upregulating the hepatic circadian gene to promote lipid biosynthesis40. In this study, we isolated C. butyricum from the gut of SGPs and AGPs, but not from GGPs. This might suggest a lower prevalence of C. butyricum with age. With advancing age, GPs experience gastrointestinal function decline and tooth wear, which impair crude fiber (CF) digestion through reduced chewing efficiency and intestinal processing. Given the well-documented probiotic properties of C. butyricum—particularly its immunomodulatory effects, butyrate production, and demonstrated lignocellulolytic activity—dietary supplementation in GGPs may ameliorate age-related gut dysfunction while enhancing bamboo digestion and nutrient utilization capacity. We therefore propose that C. butyricum supplementation represents a promising nutritional strategy to maintain both intestinal homeostasis and digestive efficiency in GGPs. Additionally, C. sporogenes, which was observed across all age groups, has demonstrated probiotic potential as its bioactive metabolite indole-3-propionic acid (IPA) helps skeletal muscle development and protects against chronic inflammation41. Taken together, it is valuable to further investigate the mechanisms of Clostridium on molecular responses and nutrient utilization in order to better understand how Clostridium species benefit GPs.
Although all Clostridium strains we obtained are a normal component of the gut microbiome from healthy GPs, several species isolated, including C. perfringens, C. paraputrificum, and C. tertium, have been reported as potentially hazardous agents to humans and/or animals in certain cases42,43,44. As one of the most important opportunistic pathogens, C. perfringens can cause several gastrointestinal diseases, such as necrotic enteritis, enterotoxemia, and enterocolitis, primarily due to its ability to produce various toxins45,46. The short generation time of C. perfringens, doubling within 10 min under optimal conditions47, likely contributes to its high isolation rate in GPs and may exacerbate disease progression. These findings highlight the dual nature of Clostridium species in the gut of GPs, encompassing both beneficial effects and potential health risks. Further research is warranted to elucidate the balance between these opposing roles and their implications for GP health management.
Symbiosis with a lignocellulose-degrading microbiome appears to compensate for an inability to generate lignocellulolytic enzymes in GPs5. Compared with fungi, bacteria exhibit several advantages in lignocellulose degradation, including higher growth rate, broad substrate specificity, and easy expression of multi-enzyme complexes48. These attributes position bacteria as promising candidates for the efficient degradation and utilization of lignocellulosic biomass in various biotechnological applications. Under anaerobic conditions, Clostridium proliferates on the surface of organic matter, releasing complex enzymes to degrade lignocellulose and further transform it into SCFAs, such as acetate, butyrate, and propionate, that the host needs49. Previous high-throughput sequencing studies have revealed Clostridium as a main resource for cellulose degradation among gut bacteria in GPs12,50. Likewise, our results regarding in vitro lignocellulose enzyme assays in combination with genome analysis of lignocellulose-degrading enzymes strongly suggested that the Clostridium in the gut of GPs secrete multiple enzymes to support lignocellulose metabolism. The types of CAZyme families associated with hemicellulose degradation were more than cellulose and lignin degradation, which may be because the complexity of the composition of hemicellulose requires the action of a wide set of enzymes51. Key enzymes involved in cellulose degradation, such as β-glucosidase (EC 3.2.1.21), endo-β-1,4-glucanase (EC 3.2.1.4), and cellobiohydrolase (EC 3.2.1.91), are distributed in various GH families (e.g., GH1, GH4, GH8, GH39, GH30_1, GH51_1, GH16_21). These enzymes act synergistically at various stages of cellulose hydrolysis, ultimately converting cellulose into glucose, which is readily absorbed by the host52. Lignin is a heterogeneous aromatic heteropolymer that binds cellulose and hemicellulose and is quite recalcitrant to degradation53. Our primary screening revealed a reduced incidence of lignin-degrading Clostridium in GGPs as compared to SGPs and AGPs. A prior study discovered that GGPs who consumed more bamboo shoots and less bamboo had a lower intake and digestion of crude fiber54. Hence, we theorized that the reduced lignin-degrading ability of Clostridium in GGPs could be due to their lower bamboo fiber intake. In return, a higher prevalence of lignin-degrading Clostridium in SGPs and AGPs may reflect an adaptive evolution to a higher fiber diet. As one of the best characterized lignin-modifying enzymes, the activity of MnP was detected in all strains with potential lignin-degrading ability, which has the capacity of converting lignin phenolic compounds to phenoxy radicals. In addition to lignin degradation, the ability of MnP to detoxify multiple mycotoxins, including aflatoxin, deoxynivalenol, and patulin55, may aid GPs in the reduction of potential mycotoxins in food. However, MnP was represented with low prevalence in strains via CAZyme annotation, which could be due to the low copy number of these enzymes in the Clostridium genomes and/or the low number of representative sequences available in the CAZymes database56.
The ability of Clostridium species to ferment a wide range of nutrients as chemoorganotrophic bacteria is widely recognized, and the majority of the metabolites they generate have multiple beneficial impacts on gut health57. In this study, we found a wide range of gut Clostridium in GPs with lignocellulose-degrading ability not only harbored a variety of genes related to SCFAs, but also contributed to the synthesis of essential amino acids via arginine and lysine biosynthesis pathways, thus serving as an energy source for a series of bodily functions. Arginine plays a central metabolic role as the biosynthetic precursor for several physiologically critical molecules, including nitric oxide, urea, ornithine, and citrulline. Nitric oxide particularly serves as a key regulator of both immune function and metabolic homeostasis, modulating glucose, fatty acid, and amino acid metabolism in mammalian systems58,59. Furthermore, lysine undergoes catabolism in mammalian systems to provide energy, while also serving as an essential precursor for numerous bioactive compounds, such as carnitine, glutamate, and collagen. Through these derivatives, lysine plays an indispensable role in cellular metabolism, energy homeostasis, and neural function60. These functional attributes of Clostridium may represent an evolutionary response to the bamboo diet of GPs, underscoring the role of the gut microbiota in strengthening the energy supply and maintaining metabolic homeostasis in GPs.
Since antibiotics are the most widely used and effective treatment for gastrointestinal disorders in GPs, it is inevitable that they may have side effects and exacerbate antibiotic resistance. Previous studies using metagenomic sequencing have gained insight into the overall distribution of ARGs within the gut microbiome of GPs, demonstrating that the ARGs mediated by the mechanism of antibiotic efflux were the most expressed19,61. The type of bacteria associated with the abundance and diversity of ARGs was also investigated previously3. For example, E. coli was recognized as the primary reservoir of ARGs, whereas specific Clostridium species were the main source of ARGs in wild GPs6,19. Additionally, using an aerobic culture-dependent approach, the antibiotic phenotype and genotype were monitored from the prevalent bacteria in GPs, such as E. coli, Enterococcus spp., Enterobacter spp., and Klebsiella pneumoniae, providing more precise guidance for appropriately prescribing antibiotics for diseases caused by specific bacterial infections62. Herein, we broaden the understanding of ARG distribution in Clostridium, a representative anaerobe in the gut microbiome of GPs. Our results demonstrated that Clostridium strains exhibit ARGs resistant to different classes of antibiotics, with the van genes, which act as a primary source of glycopeptide resistance determinants in environmental bacteria and pathogens63, were most prevalent. Similarly, Mustafa et al.64 also found a positive correlation between Clostridium and various van genes through the metagenome sequencing of gut bacteria in GPs. Glycopeptide antibiotics are a last resort in treating multidrug-resistant, gram-positive bacterial infections. These antibiotics work via binding to the D-Ala-D-Ala terminus of peptidoglycan, thus preventing the synthesis of the bacterial cell wall65. However, as the first therapeutically approved glycopeptide, long-term use of vancomycin triggered widespread bacterial resistance by the mechanism of antibiotic target alteration66. Notably, vancomycin is seldom used in clinical treatments for captive GPs, suggesting that the prevalence of genes in Clostridium is unlikely to be driven by direct antibiotic selection pressure. This conclusion is further supported by the previous identification of Clostridium as the primary reservoir of van genes in the gut microbiome of wild GPs67, where antibiotic exposure is minimal. To further investigate the potential for horizontal transfer of these ARGs, we performed an analysis of mobile genetic elements (MGEs). Although the use of draft genomes assembled from short-read sequencing data can introduce a bias whereby MGEs are placed at contig termini, manual curation confirmed that all major MGEs associated with ARGs were located internally within large contigs. This important finding guarantees the integrity of their sequences and flanking genetic contexts. That said, it cannot be ruled out that ARGs on MGEs might be missed due to the presence of only draft genomes. Future studies with complete genome sequences will provide a more complete picture of ARGs on MGEs in the giant panda gut microbiome. Our genomic analysis indeed revealed that these van genes are physically linked to MGEs, indicating a high potential for horizontal gene transfer within species. Beyond this, recent studies suggest that resistance may also stem from complex environmental and dietary interactions68. Specifically, dietary bamboo has been identified as a reservoir of ARGs that harbors van genes, with MGE-mediated mechanisms facilitating their incorporation into the GPs’ gut resistome69. Meanwhile, the shared presence of van genes and mobile genetic elements (MGEs) across bacterial communities in GP feces, dietary bamboo, and surrounding soil suggests their involvement in cross-species gene transfer70. Furthermore, metals are known to exert selection pressure promoting ARG proliferation in bacterial communities71. In GPs, identical metals detected in bamboo were consistently identified in gastrointestinal and fecal samples, along with significant associations observed among gastrointestinal metals, microbiome, metal resistance genes (MRGs), ARGs and MGEs61. Additionally, other studies have shown drug-resistant Bacillus carries a variety of copper resistance genes, van genes and MEGs in the plasmid that enables both vertical and horizontal gene transfer72. In summary, the available findings suggest van gene prevalence likely results from synergistic effects of dietary exposure to both resistance genes and metal selective pressures, combined with horizontal transfer from environmental and commensal bacterial reservoirs.
Bacteria use a variety of VFs to increase their ability to evade host defenses and transmit disease73. Our research demonstrated that most genes encoded VFs for adhesion, which may be the primary factor threatening the health of GPs. As the most prevalent VF, EF-Tu attaches to the host extracellular matrix components, such as plasminogen, fibronectin, mucins, and factor H, which can serve as effective adhesion targets for pathogens. EF-Tu facilitates invasion and colonization, aids immune system evasion, and increases virulence74,75. GroEL, a member of the heat-shock protein (HSP) family, is widely found in both prokaryotes and eukaryotes and primarily serves to increase bacterial heat tolerance. GroEL is also thought to be a critical molecule in infectious and inflammatory diseases induced by pathogens76, and it acts as an adhesin for several pathogens, including C. difficile and Chlamydia pneumoniae, to aggravate specific diseases77,78. Moreover, Cronobacter sakazakii GroEL is characterized as an inflammation stimulator in the gut, which causes host cell necrosis by activating the NF-κB-signaling pathway to release more proinflammatory cytokines (TNF-α, IL-6, and IL-8), and assisting pathogens in crossing the intestinal barrier79.
Notably, our findings revealed that C. perfringens is the main carrier of VFs. A number of type IV pili it encodes mediate adhesion to host cells80, which help other VFs disrupt epithelial barrier function and induce histotoxic infections or tissue necrosis17,81. The different extracellular toxins and enzymes that C. perfringens produces are largely responsible for its pathogenicity. Previous studies have identified seven different types of C. perfringens toxins (A–G) based on the combination of six major toxins (α-toxin, β-toxin, ι-toxin, ε-toxin, C. perfringens enterotoxin, and necrotic enteritis B-like toxin)32,45,82. We found that all eight of the C. perfringens isolates listed here displayed comparable toxin production patterns, and they are all categorized as type A, which solely generates α-toxin. The α-toxin is a zinc-containing phospholipase C enzyme with the ability to hydrolyze cell membranes and impair the innate immune response15,83. The other non-typing toxins or extracellular degradative enzymes (e.g., beta-2 toxin, theta toxin, Mu toxin, and sialidase) enriched in C. perfringens are supposed to play an important role in colonization and immunomodulation, cell injury, and, as a result, are now coming under intensive validation study83,84.
Though the commensal gut bacteria Clostridium in GPs exhibited carried VGs, it should be noted that the determinant of pathogenesis is not only dependent on the virulence factors encoded by the pathogen, but is also dependent upon several environmental stressors and accessory genes85. This could explain the presence of opportunistic pathogens, such as C. perfringens, in the guts of GPs or other healthy humans and animals47,86. Our findings offer some insight into the prevalence and horizontal transfer of ARGs and VGs in lignocellulose-degrading Clostridium, which could be useful in anticipating and averting the development of virulence and antibiotic resistance. The acquisition of ARGs and VGs may facilitate bacterial evolution and environmental adaptation, potentially enhancing the ecological fitness of these microorganisms. Furthermore, the presence of ARGs and VGs in Clostridium species may promote successful gut colonization and increase resistance to environmental stressors, thereby supporting their functional role in lignocellulose degradation within the gastrointestinal ecosystem. These findings contribute to our understanding of the complex interplay between genetic determinants and functional adaptation in the gut microbiota, particularly in relation to lignocellulose degradation processes.
Methods
Collection and processing of fecal samples
Fecal samples were collected from 31 GPs living in the China Conservation and Research Center for the Giant Panda (CCRCGP), including 7 sub-adult GPs (SGPs) (2–3 years old), 17 adult GPs (AGPs) (6–18 years old), and 7 geriatric GPs (GGPs) (21–29 years old) (Supplementary Table 5). All healthy individuals were fed a diet with a consistent composition and did not receive any drugs for at least 30 days prior to sampling. To preserve anaerobic conditions, freshly voided fecal samples were immediately transferred to sterile anaerobic gas-generating bags (Hope Bio-Technology Co., Ltd) and maintained at 4 °C during transport. All samples were processed within 2 h of collection in an anaerobic chamber (BactronEZ-2, SHELLAB, USA) under an atmosphere of 90% N2, 5% H2, and 5% CO2 to maintain anaerobic conditions.
Culture strategy for Clostridium
Four different liquid media conditions and five different solid media conditions were used in this study, selected based on their established efficacy for gut anaerobe cultivation24,31,87, manufacturer recommendations, and successful Clostridium isolation in our preliminary tests. Detailed information regarding the culture media and additives is presented in Supplementary Table 6. Fecal suspensions were prepared by homogenizing 10 g of the most unexposed inner part of each fecal sample in 50 ml of phosphate-buffered saline (PBS) supplemented with 0.05% cysteine. Then, 5 ml of each fecal suspension was inoculated into 45 ml of various liquid culture enrichment media, including Gifu anaerobic medium (GAM), thiolglycollate medium (TM), reinforced clostridial medium (RCM), and anaerobic blood culture medium. Each culture was incubated at 37 °C in an anaerobic chamber. After 0 days (without pre-incubation) and 5 days (with pre-incubation) of anaerobic culture, 1 ml of liquid from the GAM, TM, or anaerobic blood culture was diluted, plated onto solid GAM, TM and YCFA plates, respectively, and then incubated at 37 °C under anaerobic conditions for 48–72 h. Also, 1 ml of liquid from the RCM culture was diluted and plated onto RCM and Columbia medium (CM) at 37 °C under anaerobic conditions for 48–72 h. Colonies with different appearances (e.g., color, size, and shape) were picked from each plate and purified in GAM agar at 37 °C under anaerobic conditions for 48 h. All purified isolates were amplified using universal 16S rRNA primers 27F (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492R (5′-TACGGYTACCTTGTTACGACTT-3′) and sequenced by Sangon Bioengineering Co., Ltd (Shanghai, China). Each strain was amplified using PCR three separate times. Then, PCR products were sequenced and aligned, confirming identical species identification with high reproducibility. Nucleotide sequences were analyzed using EzBioCloud (https://www.ezbiocloud.net) and NCBI. If multiple strains isolated from the same sample under the same culture conditions were identified as the same species, then one was randomly selected for further evaluation. Confirmed strains were stored in GAM broth containing 30% glycerol at −80 °C.
Primary screening of lignocellulose-degrading Clostridium
A dye decolorization assay was used to assess the isolated Clostridium for hemicellulose-, cellulose-, and lignin-degrading activity as previously described88,89. Three types of selective media containing xylan, carboxymethyl cellulose (CMC), or sodium lignin sulfonate as the sole carbon source were used to select isolates with potential hemicellulose-, cellulose-, and lignin-degrading activity, respectively. Detailed information on the media used is listed in Supplementary Table 6. Briefly, duplicate Clostridium species from the same GP were removed from analysis. The purified strains were evenly spread across GAM and incubated at 37 °C under anaerobic conditions for 48 h. Next, cylindrical agar blocks (5 mm in diameter) were excised from agar with confluent growth and placed on selective media. The agar plate containing the agar block was incubated at 37 °C under anaerobic conditions for 48 h. The Clostridium species with hemicellulose-, cellulose-, and/or lignin-degrading ability were indicated by decolorization zones. The isolates that exhibited decolorization zones in triplicate were selected to be assayed by a lignocellulose enzyme assay.
Lignocellulose enzyme assay
To quantitatively verify the bamboo lignocellulose degradation activity, the Clostridium species with potential hemicellulose-, cellulose-, and/or lignin-degrading activity were identified by testing their β-glucosidase, xylanase, and manganese peroxidase (MnP) activity, respectively. Briefly, bamboo stem and leaf (Pleioblastus amarus) were dried at 40 °C and then ground and passed through a 50-mesh sieve. To minimize the effect of the bamboo microbiome on the enzyme assay, bamboo powder was irradiated using a Coalt-60 gamma irradiator with doses of 8.00 kGy (Zhongjin Irradiation Co., Ltd.). No surviving colonies were detected in the bamboo powder after radiation. Subsequently, single colonies of each strain were inoculated into GAM broth and incubated at 37 °C under anaerobic conditions for 24 h. Then 1 ml of the bacterial suspension (OD600 = 0.6) was added to bamboo fermentation broth (Supplementary Table 6) containing 5% (w/v) bamboo powder as the sole carbon source. After 24 h of anaerobic incubation, the bamboo fermentation solution was centrifuged (8000 × g, 10 min, 4 °C) and the supernatant was collected for enzymatic analyses. Enzyme activity was quantified using commercial assay kits (Solarbio, China) following the manufacturer’s protocols, with absorbance measurements performed on a Varioskan LUX multimode microplate reader (Thermo Scientific, USA); β-glucosidase activity (BC 2560) was measured via the p-nitrophenyl-β-D-glucopyranoside hydrolysis method, with the enzymatic release of p-nitrophenol quantified at 400 nm; neutral xylanase activity (BC 2595) was assessed using the 3,5-dinitrosalicylic acid (DNS) method based on reducing sugar release from xylan substrate and quantified at 540 nm; and MnP activity (BC 1625) was determined by monitoring the enzymatic oxidation of guaiacol to tetraguaiacol at 465 nm in the presence of Mn2+. Each strain was analyzed using enzyme activity assays three separate times, with the final enzymatic activity value representing the mean of the three measurements.
Whole-genome sequencing of lignocellulose-degrading Clostridium
The genomic characteristics of those Clostridium strains that possessed all three tested enzymatic activities were subjected to further analysis. Total genomic DNA of Clostridium was extracted using the DNeasy Blood and Tissue Kit (Qiagen) according to the manufacturer’s instructions. The sequencing library was generated using NEB Next® UltraTM DNA Library Prep Kit for Illumina (NEB, USA), and the library quality was verified by Qubit®3.0 Flurometer (Invitrogen, USA) and NGS3K/Caliper. After sequencing on an Illumina PE150 platform (150-bp paired-end reads) in Novogene, the raw reads were trimmed to remove low-quality reads via Trimmomatic, and the de novo assembly was performed using SPAdes90. Genome assemblies were assessed using QUAST for metrics, such as total contig length, GC content, N50, and N9091. The genomes were identified using GTDB-Tk v2.4.1 for precise taxonomic classification92. The draft genomes were annotated using Bakta v1.11.093. Gene functions of the genomes were annotated with the Kyoto Encyclopedia of Genes and Genomes (KEGG) database using eggNOG-mapper v2.1.1294. Carbohydrate-active enzymes (CAZyme) were annotated using HMMER v3.4 (http://www.hmmer.org/) based on dbCAN395. ARGs were searched via RGI v6.0.3 based on the comprehensive antibiotic resistance database (CARD)96, and the VFs and virulence genes (VGs) were predicted using the VFanalyzer (https://www.mgc.ac.cn/cgi-bin/VFs/v5/main.cgi), VFs of pathogenic bacteria (VFDB)97.
Statistics and reproducibility
The IBM SPSS Statistics software (v29.0.1.1) was used for data analysis. GraphPad Prism software (v10.1.1) was used to generate graphs. Significance analysis of lignocellulose enzyme assay was conducted across age groups and species after verifying data assumptions. Appropriate statistical tests (t-test/ANOVA or Mann-Whitney/Kruskal-Wallis) were employed for groups containing ≥5 replicates. p < 0.05 indicated statistically significant differences.
For strain identification, each strain was amplified using PCR three separate times and only those showing consistent results across all three replicates were selected for further study. In the lignocellulose enzyme assay, each strain was tested in three replicates, and the final enzyme activity values represent the mean of these measurements for each strain.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Conclusion
In summary, this study systematically evaluated different culture conditions for Clostridium isolation, establishing optimized cultivation strategies that enable researchers to obtain a greater diversity of Clostridium isolates from the gut of GPs with less workload. Importantly, we elucidated the mechanism underlying the role of Clostridium in the digestion, absorption, and metabolism of nutrients from bamboo. Lignocellulose degradation capabilities were universally observed across Clostridium species, with substantial interspecies variability. This result provides a foundation for the screening and application of beneficial Clostridium isolates. However, widespread presence and dissemination of ARGs and VFs, especially in C. perfringens, necessitate the implementation of systematic monitoring to assess antibiotic resistance patterns and evaluate potential pathogenic risks. Our data holds significance for understanding the complex interplay between Clostridium and GP health. Furthermore, the availability of isolates generated in this study enables further exploration into their potential applications in regulating the gut microbiota, alleviating gut disorders, and promoting host health.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The 16S rRNA sequencing data have been deposited in BioProject database at NCBI under accession number PRJNA1320993. The whole-genome sequencing data have been deposited in BioProject database at NCBI under accession number PRJNA1258569. All data generated or analyzed during this study are included in this published article and its supplementary files.
References
Yang, S. et al. Metagenomic analysis of bacteria, fungi, bacteriophages, and helminths in the gut of giant pandas. Front. Microbiol. 9, 1717–1732 (2018).
Liu, F. Y. et al. Age-related alterations in metabolome and microbiome provide insights in dietary transition in giant pandas. mSystems 8, e0025223 (2023).
Yang, S. et al. Reference gene catalog and metagenome-assembled genomes from the gut microbiome reveal the microbial composition, antibiotic resistome, and adaptability of a lignocellulose diet in the giant panda. Environ. Res. 245, 118090 (2023).
Deng, F. et al. The unique gut microbiome of giant pandas involved in protein metabolism contributes to the host’s dietary adaption to bamboo. Microbiome 11, 180–193 (2023).
Jin, L. et al. Diet, habitat environment and lifestyle conversion affect the gut microbiomes of giant pandas. Sci. Total Environ. 770, 145316 (2021).
Huang, G. et al. PandaGUT provides new insights into bacterial diversity, function, and resistome landscapes with implications for conservation. Microbiome 11, 221–235 (2023).
Litvak, Y., Byndloss, M. X. & Baumler, A. J. Colonocyte metabolism shapes the gut microbiota. Science 362, 1017 (2018).
Lee, J. Y., Tsolis, R. M. & Baumler, A. J. The microbiome and gut homeostasis. Science 377, eabp9960 (2022).
Disha, B., Flores, C., Christine, W. M. & Anna, M. S. Diversity and prevalence of Clostridium innocuum in the human gut microbiota. mSphere 8, 00569–00522 (2023).
Lopetuso, L. R., Scaldaferri, F., Petito, V. & Gasbarrini, A. Commensal Clostridia: leading players in the maintenance of gut homeostasis. Gut Pathog. 5, 23–30 (2013).
Yu, S. R. et al. Clostridium butyricum isolated from giant panda can attenuate dextran sodium sulfate-induced colitis in mice. Front. Microbiol. 15, 1361945 (2024).
Zhu, L., Wu, Q., Dai, J., Zhang, S. & Wei, F. Evidence of cellulose metabolism by the giant panda gut microbiome. PNAS 108, 17714–17719 (2011).
Prawitwong, P. et al. Direct glucose production from lignocellulose using Clostridium thermocellum cultures supplemented with a thermostable β-glucosidase. Biotech. Biofuels 6, 184–194 (2013).
Silva-Andrade, C., Martin, A. J. & Garrido, D. Comparative genomics of Clostridium baratii reveals strain-level diversity in toxin abundance. Microorganisms 10, 213–229 (2022).
Hassani, S., Pakbin, B., Bruck, W. M., Mahmoudi, R. & Mousavi, S. Prevalence, antibiotic resistance, toxin-typing and genotyping of Clostridium perfringens in raw beefmeats obtained from Qazvin City, Iran. Antibiotics (Basel) 11, 340–352 (2022).
Li, J. et al. Toxin plasmids of Clostridium perfringens. Microbiol. Mol. Biol. Rev. 77, 208–233 (2013).
Feng, Y. et al. Phylogenetic and genomic analysis reveals high genomic openness and genetic diversity of Clostridium perfringens. Microb. Genom. 6, mgen000441 (2020).
Cobo, F. Antimicrobial susceptibility and clinical findings of anaerobic bacteria. Antibiotics (Basel) 11, 351–353 (2022).
Deng, F. et al. A comprehensive analysis of antibiotic resistance genes in the giant panda gut. Imeta 3, e171 (2024).
Feng, W. et al. Metagenome analysis reveals changes in gut microbial antibiotic resistance genes and virulence factors in reintroduced giant pandas. Microorganisms 13, 1616–1618 (2025).
Zhao, M. Y., Li, Y. X., Wei, W., Zhang, Z. J. & Zhou, H. The distribution variation of pathogens and virulence factors in different geographical populations of giant pandas. Front. Microbiol. 14, 1264786 (2023).
Blanco-Miguez, A. et al. Extending and improving metagenomic taxonomic profiling with uncharacterized species using MetaPhlAn 4. Nat. Biotechnol. 41, 1633–1644 (2023).
Benoit, G. et al. High-quality metagenome assembly from long accurate reads with metaMDBG. Nat. Biotechnol. 42, 1378–1383 (2024).
Alou, M. T. et al. State of the art in the culture of the human microbiota new interests and strategies. Clin. Microbiol. Rev. 34, e00129–00119 (2021).
Browne, H. P. et al. Culturing of ‘unculturable’ human microbiota reveals novel taxa and extensive sporulation. Nature 533, 543–546 (2016).
Lichtenegger, A. S., Posadas-Cantera, S., Badr, M. T. & Hacker, G. Comparison of the diversity of anaerobic-cultured gut bacterial communities on different culture media using 16S rDNA sequencing. J. Microbiol. Met. 224, 106988 (2024).
Lewis, W. H., Tahon, G., Geesink, P., Sousa, D. Z. & Ettema, T. J. G. Innovations to culturing the uncultured microbial majority. Nat. Rev. Microbiol. 19, 225–240 (2021).
Li, D. et al. The monkey microbial biobank brings previously uncultivated bioresources for nonhuman primate and human gut microbiomes. mLife 1, 210–217 (2022).
Zou, Y. et al. 1520 reference genomes from cultivated human gut bacteria enable functional microbiome analyses. Nat. Biotechnol. 37, 179–185 (2019).
Liu, C. et al. The mouse gut microbial biobank expands the coverage of cultured bacteria. Nat. Commun. 11, 79–90 (2020).
Gong, R., Ye, X., Wang, S. & Ren, Z. Isolation, identification, and biological characteristics of Clostridium sartagoforme from rabbit. PLoS One 16, e0259715 (2021).
Wu, K. et al. Genomic adaptation of Clostridium perfringens to human intestine. iMetaOmics 1, e38 (2024).
Chai, L. J. et al. Profiling the Clostridia with butyrate-producing potential in the mud of Chinese liquor fermentation cellar. Int. J. Food Microbiol. 297, 41–50 (2019).
Obiekezie, S. O., Ekeleme, I. K., Makut, M. D. & Owuna, G. Isolation, identification and production of biobutanol by different Clostridium species isolated from soil using waste paper and sugar cane molasses. S. Asian J. Res. Microbiol. 2, 1–9 (2018).
Chang, Y. et al. Optimization of culturomics strategy in human fecal samples. Front. Microbiol. 10, 2891–2991 (2019).
Lagier, J. C. et al. The rebirth of culture in microbiology through the example of culturomics to study human gut microbiota. Clin. Microbiol. Rev. 28, 237–264 (2015).
Hou, F. et al. Application of LpxC enzyme inhibitor to inhibit some fast-growing bacteria in human gut bacterial culturomics. BMC Microbiol. 19, 308–315 (2019).
Stoeva, M. K. et al. Butyrate-producing human gut symbiont, Clostridium butyricum, and its role in health and disease. Gut Microbes 13, 1–28 (2021).
Hagihara, M. et al. Clostridium butyricum modulates the microbiome to protect intestinal barrier function in mice with antibiotic-induced dysbiosis. iScience 23, 100772 (2020).
Huang, G. et al. Seasonal shift of the gut microbiome synchronizes host peripheral circadian rhythm for physiological adaptation to a low-fat diet in the giant panda. Cell Rep. 38, 110203 (2022).
Du, L., Qi, R., Wang, J., Liu, Z. & Wu, Z. Indole-3-Propionic acid, a functional metabolite of Clostridium sporogenes, promotes muscle tissue development and reduces muscle cell inflammation. Int. J. Mol. Sci. 22, 12345–12360 (2021).
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).
Munoz, M. et al. Comparative genomics identifies potential virulence factors in Clostridium tertium and C. paraputrificum. Virulence 10, 657–676 (2019).
Grenda, T. et al. Clostridium perfringens—opportunistic foodborne pathogen, its diversity and epidemiological significance. Pathogens 12, 768–779 (2023).
Duc, H. M. et al. Prevalence and antibiotic resistance profile of Clostridium perfringens isolated from pork and chicken meat in Vietnam. Pathogens 13, 400–409 (2024).
AlJindan, R. et al. Genomic Insights into virulence factors and multi-drug resistance in Clostridium perfringens IRMC2505A. Toxins (Basel) 15, 359–371 (2023).
Mehdizadeh Gohari, I. et al. Pathogenicity and virulence of Clostridium perfringens. Virulence 12, 723–753 (2021).
Sumranwanich, T. et al. Evaluating lignin degradation under limited oxygen conditions by bacterial isolates from forest soil. Sci. Rep. 14, 13350 (2024).
Chukwuma, O. B., Rafatullah, M., Tajarudin, H. A. & Ismail, N. A review on bacterial contribution to lignocellulose breakdown into useful bio-products. Int. J. Environ. Res. Pub. He. 18, 6001–6027 (2021).
Jin, L. et al. Bamboo nutrients and microbiome affect gut microbiome of giant panda. Symbiosis 80, 293–304 (2020).
Lopez-Mondejar, R., Zuhlke, D., Becher, D., Riedel, K. & Baldrian, P. Cellulose and hemicellulose decomposition by forest soil bacteria proceeds by the action of structurally variable enzymatic systems. Sci. Rep. 6, 25279 (2016).
Li, H. et al. Accelerated degradation of cellulose in silkworm excrement by the interaction of housefly larvae and cellulose-degrading bacteria. J. Environ. Manag. 323, 116295 (2022).
Yang, C. X., Wang, T., Gao, L. N., Yin, H. J. & Lu, X. Isolation, identification and characterization of lignin-degrading bacteria from Qinling, China. J. Appl. Microbiol. 123, 1447–1460 (2017).
Wang, C. et al. Nutrient utilization and gut microbiota composition in giant pandas of different age groups. Animals (Basel) 14, 2324–2338 (2024).
Wang, S. et al. Detoxification of the mycotoxin citrinin by a manganese peroxidase from Moniliophthora roreri. Toxins 14, 801–811 (2022).
Diaz-Garcia, L., Bugg, T. D. H. & Jimenez, D. J. Exploring the lignin catabolism potential of soil-derived lignocellulolytic microbial consortia by a gene-centric metagenomic approach. Microb. Ecol. 80, 885–896 (2020).
Guo, P., Zhang, K., Ma, X. & He, P. Clostridium species as probiotics: potentials and challenges. J. Anim. Sci. Biotechnol. 11, 24–33 (2020).
Jobgen, W. S., Fried, S. K., Fu, W. J., Meininger, C. J. & Wu, G. Regulatory role for the arginine-nitric oxide pathway in metabolism of energy substrates. J. Nutr. Biochem. 17, 571–588 (2006).
Stuehr, D. J. Enzymes of the L-arginine to nitric oxide pathway. J. Nutr. 134, 2748S–2751 (2004).
Wu, G. et al. Arginine metabolism and nutrition in growth, health and disease. Amino Acids 37, 153–168 (2009).
Jin, L. et al. Gastrointestinal microbiome, resistance genes, and risk assessment of heavy metals in wild giant pandas. Sci. Total Environ. 899, 165671 (2023).
Wang, X. et al. Antimicrobial resistance of Escherichia coli, Enterobacter spp., Klebsiella pneumoniae and Enterococcus spp. isolated from the feces of giant panda. BMC Microbiol 22, 102–112 (2022).
Yushchuk, O., Binda, E. & Marinelli, F. Glycopeptide antibiotic resistance genes: distribution and function in the producer Actinomycetes. Front. Microbiol. 11, 1173 (2020).
Mustafa, G. R. et al. Metagenomic analysis revealed a wide distribution of antibiotic resistance genes and biosynthesis of antibiotics in the gut of giant pandas. BMC Microbiol. 21, 15 (2021).
Meziane-Cherif, D. et al. Structural and functional adaptation of vancomycin resistance VanT serine racemases. mBio 6, e00806 (2015).
Yim, G., Thaker, M. N., Koteva, K. & Wright, G. Glycopeptide antibiotic biosynthesis. J. Antibiotics (Tokyo) 67, 31–41 (2014).
Hu, T. et al. Geographic pattern of antibiotic resistance genes in the metagenomes of the giant panda. Microb. Biotechnol. 14, 186–197 (2021).
Yan, Z. et al. The impact of bamboo consumption on the spread of antibiotic resistance genes in giant pandas. Vet. Sci. 10, 630–643 (2023).
Yan, Z. et al. Dietary microbiota-mediated shifts in gut microbial ecology and pathogen interactions in giant pandas (Ailuropoda melanoleuca). Commun. Biol. 8, 864–877 (2025).
Fu, Y. et al. Co-occurrence patterns of gut microbiome, antibiotic resistome and the perturbation of dietary uptake in captive giant pandas. J. Hazard. Mater. 471, 134252 (2024).
Deng, W. W. et al. Heavy metals, antibiotics and nutrients affect the bacterial community and resistance genes in chicken manure composting and fertilized soil. J. Environ. Manag. 257, 109980 (2020).
Yan, M. et al. Emerging antibiotic and heavy metal resistance in spore-forming bacteria from pig manure, manure slurry and fertilized soil. J. Environ. Manag. 371, 123270 (2024).
Dong, W. et al. An expanded database and analytical toolkit for identifying bacterial virulence factors and their associations with chronic diseases. Nat. Commun. 15, 8084–8099 (2024).
Yang, Q. et al. Evaluation of immunogenicity and protective efficacy of the elongation factor Tu against Streptococcus agalactiae in tilapia. Aquaculture 492, 184–189 (2018).
Harvey, K. L., Jarocki, V. M., Charles, I. G. & Djordjevic, S. P. The diverse functional roles of elongation factor Tu (EF-Tu) in microbial pathogenesis. Front. Microbiol. 10, 2351–2369 (2019).
Muscariello, L., De Siena, B. & Marasco, R. Lactobacillus cell surface proteins involved in interaction with mucus and extracellular matrix components. Curr. Microbiol. 77, 3831–3841 (2020).
Wuppermann, F. N., Molleken, K., Julien, M., Jantos, C. A. & Hegemann, J. H. Chlamydia pneumoniae GroEL1 protein is cell surface associated and required for infection of HEp-2 cells. J. Bacteriol. 190, 3757–3767 (2008).
Hennequin, C. et al. GroEL (Hsp60) of Clostridium difficile is involved in cell adherence. Microbiology (Reading) 147, 87–96 (2001).
Zhu, D. et al. Characterization of molecular chaperone GroEL as a potential virulence factor in Cronobacter sakazakii. Foods 12, 3404–3419 (2023).
Melville, S. & Craig, L. Type IV pili in Gram-positive bacteria. Microbiol. Mol. Biol. Rev. 77, 323–341 (2013).
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, 2485–2502 (2017).
Rood, J. I. et al. Expansion of the Clostridium perfringens toxin-based typing scheme. Anaerobe 53, 5–10 (2018).
Camargo, A. et al. Intra-species diversity of Clostridium perfringens: A diverse genetic repertoire reveals its pathogenic potential. Front. Microbiol. 13, 952081 (2022).
Kiu, R. et al. Particular genomic and virulence traits associated with preterm infant-derived toxigenic Clostridium perfringens strains. Nat. Microbiol. 8, 1160–1175 (2023).
Hafner, L. et al. Differential stress responsiveness determines intraspecies virulence heterogeneity and host adaptation in Listeria monocytogenes. Nat. Microbiol. 9, 3345–3361 (2024).
Forti, K. et al. Molecular characterization of Clostridium perfringens strains isolated in Italy. Toxins (Basel) 12, 650–669 (2020).
Chang, Y. X., Hou, F. Y., Bi, Y. J. & Yang, R. F. Optimization of culturomics methods. Microbiome Protoc. ebook Bio-101, e2003639 (2021).
Sharker, B. et al. Characterization of lignin and hemicellulose degrading bacteria isolated from cow rumen and forest soil: unveiling a novel enzymatic model for rice straw deconstruction. Sci. Total Environ. 904, 166704 (2023).
Haque, M. A. et al. Rapid deconstruction of cotton, coir, areca, and banana fibers recalcitrant structure using a bacterial consortium with enhanced saccharification. Waste Biomass-. Valor. 12, 4001–4018 (2020).
Bankevich, A. et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012).
Gurevich, A., Saveliev, V., Vyahhi, N. & Tesler, G. QUAST: quality assessment tool for genome assemblies. Bioinformatics 29, 1072–1075 (2013).
Chaumeil, P. A., Mussig, A. J., Hugenholtz, P. & Parks, D. H. GTDB-Tk v2: memory friendly classification with the genome taxonomy database. Bioinformatics 38, 5315–5316 (2022).
Schwengers, O. et al. Bakta: rapid and standardized annotation of bacterial genomes via alignment-free sequence identification. Microb. Genom. 7, 000685 (2021).
Cantalapiedra, C. P., Hernandez-Plaza, A., Letunic, I., Bork, P. & Huerta-Cepas, J. eggNOG-mapper v2: functional annotation, orthology assignments, and domain prediction at the metagenomic scale. Mol. Biol. Evol. 38, 5825–5829 (2021).
Zheng, J. et al. dbCAN3: automated carbohydrate-active enzyme and substrate annotation. Nucleic Acids Res. 51, W115–W121 (2023).
Jia, B. et al. CARD 2017: expansion and model-centric curation of the comprehensive antibiotic resistance database. Nucleic Acids Res. 45, D566–D573 (2017).
Liu, B., Zheng, D., Jin, Q., Chen, L. & Yang, J. VFDB 2019: a comparative pathogenomic platform with an interactive web interface. Nucleic Acids Res. 47, D687–D692 (2019).
Acknowledgements
This work was supported by the International Cooperation Funding Project for Giant Pandas (The Giant Panda Microbiome Research and Biobank Establishment).
Author information
Authors and Affiliations
Contributions
L. Zou, K. Zhao, and C. Li contributed to the design of this study and revised the manuscript. Y. Huang and D. Li contributed to supervising this study. W. Deng and C. Liu performed the experiment and collected data. W. Deng, and S. Yang performed data analysis and drafted the manuscript. T. Li and D. Wu contributed to sample collection. Y. He and R. Li conducted scientific direction and provided valuable comments. All authors reviewed and agreed to the final version of the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Communications Biology thanks Francesco Candeliere and Lei Wang for their contribution to the peer review of this work. Primary Handling Editors: Sabina La Rosa and Tobias Goris. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, 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 licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence 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 licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Deng, W., Li, C., Huang, Y. et al. Lignocellulose degradation capabilities and distribution of antibiotic resistance genes and virulence factors in Clostridium from the gut of giant pandas. Commun Biol 8, 1602 (2025). https://doi.org/10.1038/s42003-025-08943-7
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s42003-025-08943-7
