
==== Front
Microbiome
Microbiome
Microbiome
2049-2618
BioMed Central London

39300575
1863
10.1186/s40168-024-01863-4
Research
Trait-based study predicts glycerol/diol dehydratases as a key function of the gut microbiota of hindgut-fermenting carnivores
Li Qing 14
Ruscheweyh Hans-Joachim 2
Østergaard Lærke Hartmann 1
Libertella Micael 1
Simonsen Kim Skalborg 3
Sunagawa Shinichi 2
Scoma Alberto 1
Schwab Clarissa schwab@bce.au.dk

1
1 https://ror.org/01aj84f44 grid.7048.b 0000 0001 1956 2722 Department of Biological and Chemical Engineering, Aarhus University, Gustav Wieds Vej 10, 8000 Arhus, Denmark
2 https://ror.org/05a28rw58 grid.5801.c 0000 0001 2156 2780 Department of Biology, Institute of Microbiology and Swiss Institute of Bioinformatics, ETH Zürich, Vladimir-Prelog-Weg 4, 8093 Zurich, Switzerland
3 Givskud Zoo - Zootopia, Løveparkvej 3, 7323 Givskud, Denmark
4 https://ror.org/04qtj9h94 grid.5170.3 0000 0001 2181 8870 Present address: National Food Institute, Technical University of Denmark, Kgs. Lyngby, Denmark
19 9 2024
19 9 2024
2024
12 17824 2 2024
25 6 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by/4.0/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. 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/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Background

Microbial pdu and cob-cbi-hem gene clusters encode the key enzyme glycerol/diol dehydratase (PduCDE), which mediates the transformation of dietary nutrients glycerol and 1,2-propanediol (1,2-PD) to a variety of metabolites, and enzymes for cobalamin synthesis, a co-factor and shared good of microbial communities. It was the aim of this study to relate pdu as a multipurpose functional trait to environmental conditions and microbial community composition. We collected fecal samples from wild animal species living in captivity with different gut physiology and diet (n = 55, in total 104 samples), determined occurrence and diversity of pdu and cob-cbi-hem using a novel approach combining metagenomics with quantification of metabolic and genetic biomarkers, and conducted in vitro fermentations to test for trait-based activity.

Results

Fecal levels of the glycerol transformation product 1,3-propanediol (1,3-PD) were higher in hindgut than foregut fermenters. Gene-based analyses indicated that pduC harboring taxa are common feature of captive wild animal fecal microbiota that occur more frequently and at higher abundance in hindgut fermenters. Phylogenetic analysis of genomes reconstructed from metagenomic sequences identified captive wild animal fecal microbiota as taxonomically rich with a total of 4150 species and > 1800 novel species but pointed at only 56 species that at least partially harbored pdu and cbi-cob-hem. While taxonomic diversity was highest in fecal samples of foregut-fermenting herbivores, higher pduC abundance and higher diversity of pdu/cbi-cob-hem related to higher potential for glycerol and 1,2-PD utilization of the less diverse microbiota of hindgut-fermenting carnivores in vitro.

Conclusion

Our approach combining metabolite and gene biomarker analysis with metagenomics and phenotypic characterization identified Pdu as a common function of fecal microbiota of captive wild animals shared by few taxa and stratified the potential of fecal microbiota for glycerol/1,2-PD utilization and cobalamin synthesis depending on diet and physiology of the host. This trait-based study suggests that the ability to utilize glycerol/1,2-PD is a key function of hindgut-fermenting carnivores, which does not relate to overall community diversity but links to the potential for cobalamin formation.

Video Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s40168-024-01863-4.

Keywords

Diet
Gut physiology
Glycerol diol dehydratase
Metagenomic sequencing
Biomarker
http://dx.doi.org/10.13039/501100003043 EMBO 9152 Li Qing Aarhus Universitet Forsknings FondenAUFF-F-2020-7 Schwab Clarissa issue-copyright-statement© BioMed Central Ltd., part of Springer Nature 2024
==== Body
pmcIntroduction

While ancestral mammals were carnivores, the transition from carnivores to herbivores was a significant milestone in mammalian history [1]. This evolutionary shift facilitated the proliferation of herbivorous species, which now constitute approximately 80% of mammals. Adaption to a herbivorous diet related to a longer gut retention time through an enlargement of the foregut or hindgut [1] enabling the intestinal microbial communities to access, degrade, and ferment plant-based feedstock. Today, most omnivores and carnivores are hindgut fermenters while herbivores ferment in the foregut or hindgut [2]. Fecal microbial communities differ between foregut- and hindgut-fermenting herbivores, and hindgut-fermenting omnivores and carnivores [2], and diet explained the α- and β-diversity variation to a larger extent than geography, habitat, genome-based host phylogeny, and technical factors [3]. Concurrently, microbial genes and metabolites recovered from mammalian feces were significantly clustered by diet and gut physiology [3, 4] highlighting that environmental host parameters link to intestinal microbial activity and functionality. Fecal microbiota is frequently used as a proxy of intestinal microbial communities [2, 3, 5].

The operons pdu and cob-cbi-hem (Fig. 1A) were previously identified as host-related traits of the gut symbiont Limosilactobacillus reuteri indicating a possible link between microbial function and intestinal microbial lifestyle [6]. The key enzyme of pdu, glycerol/propanediol dehydratase (PduCDE) encoded by pduCDE genes, catalyzes the transformation of 1,2-propanediol (1,2-PD) to propanal, which can be further metabolized to propanol and to the short-chain fatty acid (SCFA) propionate mediated by enzymes of the Pdu pathway (Pdu, Fig. 1B). Propionate provides energy to the host, modulates the immune system, and promotes intestinal homeostasis [7]. As a second substrate, PduCDE uses glycerol to form 3-hydroxypropanal with the end products 1,3-propanediol (1,3-PD) and 3-hydroxypropionate [8]. 3-Hydroxypropanal can spontaneously degrade to the double unsaturated reactive aldehyde acrolein; both compounds are part of the antimicrobial reuterin system [9]. Part of the Pdu metabolism occurs in bacterial proteinaceous microcompartments (BMC) formed by PduA, PduB, and PduJ [10]. Such BMC enhance enzyme efficiency and protect bacteria from toxic intermediates like propanal [11]. Both substrates of PduCDE are readily available in gut ecosystems as degradation products of triglycerides (glycerol) and as an intermediate of the microbial metabolism of the deoxyhexoses fucose and rhamnose (1,2-PD) [12]. Fucose and rhamnose are components of dietary plant and host-derived glycans, e.g., pectin and mucus [13, 14].Fig. 1 Pdu and cobalamin biosynthesis pathways. A Gene structure of pdu/cob-cbi-hem operons in L. reuteri with highlighted key genes [8]. Genes were not drawn according to scale. B Pdu encoded by the pdu operon mediates the degradation of 1,2-propanediol and glycerol to form propanol and propionate, and 1,3-propanediol and 3-hydroxypropionate as final metabolites, respectively. C Cobalamin synthesis pathways are encoded by genes on cob-cbi-hem operons

Cobalamin, a cobamamide, is a cofactor for PduCDE and other enzymes involved in the metabolism of nucleic acids, amino acids, and fatty acids [15]. Microbes can synthesize cobalamin via anaerobic, aerobic, C4/C5, and salvage pathways (Fig. 1C). In L. reuteri and other bacterial species, cobalamin forming enzymes are encoded by cob-cbi-hem, which is adjacent to the pdu operon [16, 17] (Fig. 1A). It has been predicted that > 80% of the members of gut and environmental microbial communities depend on external supply of cobamides, while only 30–40% have the capacity for synthesis [15].

Due to its intrinsic relationship with gut symbiont lifestyle, dietary nutrients, and microbial interactions, pdu and cob-cbi-hem operons can be considered as highly relevant functions to study the distribution and role of specific traits of intestinal microbial communities. Using a novel multi-pronged approach, the aim of this study was to elucidate the occurrence and function of pdu and cob-cbi-hem operons of different gut microbial ecosystems. We used an experimental set-up (Fig. 2) based on n = 104 fecal microbial communities collected from 55 species of captive Mammalia and Aves covering different diet schemes (herbivorous, carnivorous, omnivorous) and gut physiology (foregut and hindgut fermenters) in combination with quantitative PCR and metabolite analysis to identify fecal biomarkers (Table S1), with metagenomics to determine taxonomic composition and to predict functional potential encoded by pdu and cbi-cob-hem, and with strict anaerobic fermentation to determine Pdu activity in vitro (Fig. 2).Fig. 2 Study design. This study utilized fecal microbiota collected from 55 animal species housed at Givskud and Copenhagen Zoo (Denmark) to conduct metabolic and genetic marker analysis, metagenomic sequencing, and in vitro cultivations. This figure was created using BioRender

Results and discussion

Occurrence of the glycerol transformation metabolite 1,3-PD links to diet and to fecal microbial communities of hindgut fermenters

Pdu driven metabolism leads to the formation of the SCFA propionate and of the glycerol-specific metabolite 1,3-PD. To identify the potential of captive wild animal gut microbiota for Pdu activity, fecal metabolite profiles of 104 fecal samples were assessed for 1,3-PD and primary SCFAs (e.g., acetate, propionate, butyrate) using HPLC-RI (Table 1, Table S1). Table 1 SCFA and 1,3-propanediol levels and occurrence in feces. Fecal levels of major SCFAs (acetate, propionate, and butyrate) and 1,3-propanediol (1,3-PD) were determined with HPLC-RI

Factor	Fecal SCFA and 1,3-PD (median; 25 and 75% quartile)	
Acetate	Propionate	Butyrate	Total SCFA	1,3-PD	
Levels (µmol g−1)	Proportion (%)	Levels (µmol g−1)	Proportion (%)	Levels (µmol g−1)	Proportion (%)	Levels (µmol g−1)	Levels (µmol g−1)	
Gut physiology	Foregut	25.4; 16.8; 34.4	85.2; 82.6; 87.1	0; 0; 0.7	0; 0; 1.9	3.9; 3.4; 4.8	13.1; 10.6; 15.1	30.0; 20.4; 39.2	0; 0; 5.0	
Hindgut	37.4**; 19.8; 61.0	84.7; 77.4, 98.1	0; 0; 2.6	0; 0; 6.0	4.5; 0; 6.4	11.2; 0; 16.1	46.7**; 23.4; 73.7	4.6**; 0; 6.3	
Diet	Carnivores	31.3ab; 12.3; 62.0	86.2ab; 67.5; 100	0; 0; 4.7	0; 0; 5.3	4.3; 0; 7.5	12.1ab; 0; 18.0	33.6ab; 17.2; 73.9	4.8ab; 0; 6.2	
Omnivores	55.7a; 30.3; 95.4	89.2a; 82.0; 100.0	0; 0; 1.3	0; 0; 4.3	3.7; 0; 9.6	6.1bc; 0; 11.8	71.9a; 32.6; 104.2	5.3a; 0; 7.4	
Herbivores	26.0bc; 18.2; 37.0	84.4bc; 79.9; 87.0	0; 0; 1.7	0; 0; 4.1	4.1; 3.5; 4.2	13.6a; 10.0; 16.3	31.0bc; 23.0; 44.7	0; 0bc; 5.4	
* indicates significant differences in abundance and occurrence of (p < 0.01) between hindgut-fermenting animals and foregut-fermenting animals. Amount and percentage of SCFAs and 1,3-PD from feces of carnivores, omnivores, and herbivores do not share a common lowercase superscript differ significantly (p < 0.05)

Total SCFA contents ranged from 4.9 to 188.8 µmol∙g−1, with acetate contributing the highest proportion (range 50.9 to 100%), followed by butyrate (0 to 40.4%) and propionate (0 to 31.5%) (Table 1). The concentration of acetate was higher in fecal samples of hindgut than foregut fermenters (37.4 vs. 25.4 µmol∙g−1, p < 0.01), while concentration and proportion of acetate was higher in samples of omnivores rather than herbivores (55.7 vs. 26.0 µmol∙g−1, p < 0.001, 89.3 vs. 84.0%, p < 0.05). The proportion of butyrate was higher in herbivores than in omnivores (13.7 vs. 6.1%, p < 0.001) (Table 1), and it was suggested before that diet is a major determinant in defining composition of the fecal butyrate producing community of animals [18].

In foregut fermenters, microbial degradation and fermentation almost exclusively occur in the upper gastrointestinal tract, while in hindgut fermenters microbes are most abundant and active in the large intestine, with overall shorter retention times [19]. This systematic difference might at least partly explain the higher acetate levels determined in fecal samples of hindgut- than of foregut-fermenting animals. Concentrations or proportions of propionate were not affected by diet or physiology, possibly because propionate cannot only be formed from 1,2-PD but also from lactate and succinate [20].

The glycerol metabolite 1,3-PD was detected in 56% of the samples at levels ranging from 4.5 to 14.6 µmol∙g−1 (Table 1). 1,3-PD occurred more frequently (68.0 vs. 33.3%, p < 0.05) and at higher median levels (5.3 vs. 0.0 µmol∙g−1, p < 0.01) in fecal samples of omnivores than herbivores. 1,3-PD was detected in 60.0% of the samples collected from carnivores (median level 4.8 µmol∙g−1). The fecal concentration of 1,3-PD was higher in hindgut than foregut fermenters (median level 4.6 vs. 0.0 µmol∙g−1, p < 0.01). Glycerol is the only (known) carbon source that acts as substrate for the microbial biosynthesis of 1,3-PD. A higher content of triglycerides in the meat-containing diet of omnivores and carnivores as compared to herbivores, together with differences in lipid digestion efficiency [21], can lead to higher glycerol levels in the hindgut and thus to the microbial formation of 1,3-PD. Considering 1,3-PD as a diet- and substrate-dependent marker of pdu metabolism, hindgut-fermenting carnivores and omnivores showed higher glycerol transformation than foregut-fermenting herbivores.

Occurrence and abundance of pduC contributing taxa was higher in hindgut than foregut fermenters

In parallel to fecal metabolite analysis, qPCR was performed on fecal microbial communities targeting pduC as a genetic marker of PduCDE function (Table S2). We quantified taxa known for the presence of pdu that were abundant in fecal microbiota of humans, which can be considered a hindgut-fermenting omnivore. The tested taxa included L. reuteri, Anaerobutyricum hallii, Blautia obeum, Veillonella dispar, Flavonifractor plautii, and Ruminococcus gnavus [22, 23]. We also included Clostridium perfringens, which was identified in carnivores/wild captive animals before and used glycerol via PduCDE [18, 24, 25] (Suppl. Methods).

The majority of samples (85.4%) harbored at least one taxon carrying pduC (median 2) with an abundance ≥ 4.7 log pduC g−1 (Table S1), indicating the common occurrence of pduC harboring taxa in fecal microbiota of captive wild animals. The total abundance of tested pduC taxa was higher in hindgut fermenting or carnivore microbiota than foregut fermenting (median level 5.5 vs. 4.9 log pduC∙g−1, p < 0.001) or omnivore and herbivore microbiota (median level 8.5 vs. 5.2 vs. 5.2 log pduC∙g−1, p < 0.05) (Table 2). The significantly (p < 0.001) most recurrent pduC-harboring taxon was L. reuteri (66.3% of samples) (Table 2, Table S1), in agreement with recent studies linking the presence of pdu/cob-cbi-hem to omnivorous (human, primates, pigs, and chicken) and herbivorous hosts [6, 26]. Yet, L. reuteri generally had low abundance (median level 5.4 log pduC g−1). Table 2 Overview of pduC abundance and occurrence from selected bacteria of fecal microbiota by qPCR. The cell copies major taxa known to harbour pduC

Taxon	Factor	Cell counts (median; 25 and 75% quartile; log pduC∙g−1)/occurrence (%)	
Anaerobutyricum hallii	Limosilactobacillus reuteri	Ruminococcus gnavus	Veillonella dispar	Blautia obeum	Clostridium perfringens	Sum	
Median	Occurrence	Median	Occurrence	Median	Occurrence	Median	Occurrence	Median	Occurrence	Median	Occurrence	Median	Occurrence	
Gut physiology	Foregut	0; 0, 0	19.0	4.8; 0, 4.9	54.8	0**; 0, 0	11.9*	0; 0, 0	14.3	0; 0, 0	4.8	0; 0, 0	0	4.9; 4.7, 5.2	76.2	
Hindgut	0**; 0, 6.5	45.9*	4.8; 0; 4.9	75.4	0; 0, 0	0	0; 0, 0	23.0	4.8***; 0, 4.9	50.8***	0**; 0, 0	14.8*	5.5***; 5.1, 7.8	90.3	
Diet	Carnivores	0; 0, 0	20.0	4.8; 4.7, 4.9	80.0	0; 0, 0	0	0; 0, 0	20.0	4.9a; 1.2, 4.9	70.0a	8.5a; 0, 9.1	60.0a	8.5a; 5.9, 9.1	90.0	
Omnivores	0; 0, 5.0	33.3	4.8; 0, 4.9	70.8	0; 0, 0	0	0; 0, 4.9	29.2	0b; 0, 4.9	41.2ab	0b; 0, 0	4.2b	5.2b; 4.9, 6.5	88.0	
Herbivores	0; 0, 5.0	37.7	4.8; 0, 4.9	63.8	0; 0, 0	7.2	0; 0, 0	15.9	0b; 0, 0	23.2b	0b; 0, 0	2.9b	5.2b; 4.8, 6.2	82.6	
Total occurrence	34.6%	63.3%		4.8%		19.2%		31.7%		8.7%		85.4%		
*, **, and *** represent higher abundance and occurrence of p < 0.05, p < 0.01, and p < 0.001, respectively, between hindgut-fermenting animals and foregut-fermenting animals. Abundance and occurrence of pduC from selected bacteria identified in fecal microbiota of carnivores, omnivores, and herbivores do not share a common lowercase superscript differ significantly (p < 0.05). Analysis was based on qPCR results from 103 fecal samples without one of Barbary macaque (sample nr. 9)

The omnipresent A. hallii species contributing PduCDE to human fecal microbiota [22] was detected in 34.6% (median level 5.4 log pduC g−1) of the samples followed by B. obeum, V. dispar, C. perfringens, and R. gnavus (Table 2). The occurrence of A. hallii and B. obeum was higher in the fecal microbiota of hindgut- than foregut-fermenting animals (A. hallii, median level 45.9 vs. 19%, p < 0.05; B. obeum, median level 50.8% vs. 4.8%, p < 0.001). Concurrently, the fecal microbiota of hindgut-fermenting animals harbored higher levels of pduC assigned to A. hallii (p < 0.01), C. perfringens (p < 0.05), and B. obeum (p < 0.001). B. obeum and C. perfringens were most often detected in feces of carnivores (Table 2) and abundance of C. perfringens and B. obeum pduC was higher in carnivores than in herbivores and omnivores (p < 0.001 and p < 0.05) (Table 2). Interestingly, pduC of R. gnavus was exclusively present in fecal microbiota of foregut-fermenting animals. While most of the tested species showed a preference for glycerol or 1,2-PD in vitro, there was no or marginal growth of A. hallii, B. obeum, L. reuteri, and R. gnavus when either substrate was provided as sole carbon source [23]. Occurrence and abundance can depend on the availability of alternative carbon sources, for example, A. hallii can cross-feed on the fermentation intermediates lactate and acetate [27], while C. perfringens is proteolytic, profiting from protein-rich substrate supply.

Together, our observations made after fecal metabolite analysis and qPCR suggest that both 1,3-PD and pduC can be considered as indicators that relate to pdu. To compare both markers, we analyzed the recurrence of 1,3-PD and pduC in samples collected from the same species with multiple sample points (n = 2–3), which were not necessarily obtained from the same animal. 1,3-PD could be repeatedly detected in 18.2% of the species while pduC was present more frequently (60.0%, p < 0.001), suggesting pduC as a more consistent marker of pdu than 1,3-PD (p < 0.001). Indeed, 38.5% of the samples harbored pduC and 1,3-PD and 46.2% only pduC, while only 1,3-PD was recovered from 5.8% of the samples. Neither pduC nor 1,3-PD was detected in 9.6% of the samples. The higher occurrence of pduC than of 1,3-PD might relate to the diet-dependent formation of 1,3-PD and also points out that the potential of a microbial community to confer a specific function is not necessarily indicative that the corresponding pathway is active.

The potential for PduCDE activity is shared by few taxa

To investigate the occurrence and diversity of pdu harboring microbes using an untargeted approach, we generated metagenomic sequences of 25 fecal samples representing different diet types (17 herbivores, three carnivores, five omnivores) and gut physiology (11 foregut and 14 hindgut fermenters). Quality controlled reads from shotgun sequencing were assembled into scaffolds, which were subsequently used for metagenomic assembled genome (MAG) construction.

In total, 5040 MAGs were constructed, and 4958 MAGs were annotated as belonging to 25 phyla of the domain Bacteria (Fig. 3A). MAGs of Bacteria were grouped into 4150 species-level clusters and from every cluster, the representative MAG (rMAG) with the highest Qscore was selected for phylogenetic analysis. Among all rMAGs, 61.4 and 20.0% were assigned to Bacillota and Bacteroidota, respectively (Fig. 3A). The rMAGs assigned to Verrucomicrobiota, Pseudomonadota, and Actinomycetota contributed 5.1, 4.5, and 3.2%, respectively. A total of 1848 rMAGs (44.5%) were suggested as novel species. These rMAGs belonged to 18 phyla, which were mainly Bacillota (n = 1094), Bacteroidota (n = 362), and Verrucomicrobiota (n = 114) similar to a previous study that discovered > 50% of novel species from the fecal microbiota of wild and captive Mammalia, Aves, Reptilia, and Osteichthyes, suggesting that wild animals living in captivity are a promising source of novel microbes [5].Fig. 3 Occurrence of pdu/cob-cbi-hem related genes in rMAGs that harbored pduCDE. A Phylogenetic analysis of species-level representative MAGs (rMAGs). Genome sequences of rMAGs were used in phylogenetic analysis conducted by Phylophlan 3.0. The heatmaps were added via gheatmap. Shown are annotated phyla of rMAGs, presence of pduCDE in rMAGs, and host diet and gut physiology. Phyla that consisted of less than 100 MAGs were labeled as “Others” in the heatmap. The maximum ANI values were used for novelty categorization as described [5]. Pdu coding genes that are present in MAGs were annotated as a layer of heatmap on the phylogenetic tree. MAGs harboring at least one of pduCDE were labeled with “Positive”; otherwise, they were labeled with “Negative.” B Presence of individual genes at pdu/cob-cbi-hem operons in pdu harboring rMAGs. Genome sequences of rMAGs with > one of pduCDE were extracted, and the presence or absence of individual genes at pdu/cob-cbi-hem loci is shown. The annotation on the left side is taxon|sample number_rMAG id|completeness of rMAG|diet of host|gut physiology of host

Genes retrieved from MAGs and assemblies were clustered, quantified, and annotated to construct a gene catalog from captive wild animal fecal microbiota. To obtain taxa with the potential to metabolize glycerol and 1,2-PD, we identified rMAGs that harbored at least one gene of pduCDE and extracted pdu and cob-cbi-hem related genes (Table S3) of those rMAGs from the gene catalog. In total, 56 rMAGs (1.3% of all rMAGs) possessed at least one subunit encoding gene of pduCDE (Fig. 3B). The recovery in approx. 1% of rMAGs suggested pduCDE as a microbial function shared by few taxa in agreement with results of qPCR and as reported in humans [22, 23]. The 56 rMAGs were derived from 21/25 samples (84%), which was similar to the pduC occurrence identified with qPCR (85%). The majority (n = 54) of rMAGs harboring pduC, pduD, and/or pduE were assigned to Bacillota, one to Actinomycetota (Propionibacteriaceae family) and one to Pseudomonadota (Rhodobacteraceae family). The predominant family harboring pduC was Lachnospiraceae including the genera Blautia, Ruminococcus, Anaerobutyricum, Acetatifactor, and Faecalimonas. Twelve of 56 rMAGs were annotated to species level, including R. gnavus and C. perfringens (Fig. 3B), validating the results obtained by qPCR. No rMAG from the species L. reuteri, a frequently occurring contributor of pduC, was reconstructed due to its low abundance as shown by qPCR.

Pdu function generally co-occurs with cobalamin synthesis and the potential for BMC formation

In the extensively researched L. reuteri, most of the strains that possessed pdu also harbored cob-cbi-hem [6] and only few genomes with only pdu and lacking cob-cbi-hem were detected [26]. A similar co-occurrence has been reported for other Lactobacillaceae, including Loigolactobacillus coryniformis [28] and Furfurilactobacillus rossiae [29], and several pathogens [30] while gene synteny was not as preserved for A. hallii [31].

We tested for the co-occurrence of pdu and cob-cbi-hem operons (genes are listed in Table S3) in the 56 rMAGs that harbored pduCDE (Fig. 3B) and found that 19 rMAGs possessed ≥ 10 genes encoding cobalamin production enzymes (cob-cbi-hem) including C. perfringens, R. gnavus, and F. plautii. Besides producing the co-factor for PduCDE, cobalamin synthesis is considered as an example of the “Black Queen Hypothesis” (BQH), which suggests a selective advantage to microorganisms that lose energetically costly functions. Only a minority population of “helper microbes” may maintain the ability to provide cobalamin as an indispensable public good to the community [32]. In agreement, only 179 of 4150 rMAGs had the potential for cobalamin synthesis based on the presence of 10 or more genes on cob-cbi-hem loci. These 179 rMAGs included the 19 rMAGs that also harbored pdu suggesting that taxa, which possessed pdu and cob-cbi-hem, contribute an important community function.

PduCDE driven substrate transformation occurs in a BMC [10]. Using pduA and pduB as marker genes for the potential to initiate BMC formation, the majority of rMAGs (83.9%) that were positive for pduCDE harbored genes encoding PduA or PduB, highlighting the co-occurrence of PduCDE and BMC formation in species present in fecal microbiota of captive wild animals (Fig. 3B). From a biotechnological perspective, BMC offer solutions for metabolic and biomedical engineering that can increase enzyme stability and catalysis rates [33, 34]; captive animal microbiota can be considered as a novel source of BMC.

The potential for glycerol/1,2-PD transformation and final metabolite production was higher in carnivore than herbivore fecal microbiota

To gain understanding on how diet and gut physiology impact abundance of genes related to the pdu (including genes homologous to pduABN) and cob-cbi-hem operons in metagenomes of fecal microbiota, we extracted relevant genes (Table S3) from the gene catalog and constructed host-dependent pdu/cob-cbi-hem profile. Based on functional profiles, pduCDE and most other genes of pdu were present in samples of all 25 animal species as suggested by qPCR analysis indicating the common potential of captive animal microbiota for PduCDE activity (Fig. 4A). In addition, the majority of cbi, cob, and hem genes of the anaerobic (except cbiA), aerobic (except cob), C3, and C4 pathways of cobalamin biosynthesis, respectively, were detected in fecal microbiota of the 25 animal species (Fig. 4A). Using DESeq2 [35], we tested for differences in abundance of genes of pdu/cob-cbi-hem profiles of fecal microbial communities. The abundance of pduCDE was significantly higher in carnivore than herbivore microbiota, indicating a higher potential for PduCDE activities (Fig. 4B). Higher abundance of pduLQ in omnivore than herbivore fecal microbiota (Fig. 4C) and in microbiota of hindgut compared to foregut fermenters (Fig. 4D) suggested a higher potential for the formation of final metabolites of the Pdu pathway (e.g., propanol and propionate, or 1,3-PD and hydroxypropionate, Fig. 1). Except for hemD, two cob genes, and cysG, most of genes related to cobalamin biosynthesis did not differ in abundance in fecal microbiota of hosts with different diet and gut physiology (Fig. 4B, C, D).Fig. 4 Abundance of genes of the pdu/cob-cbi-hem operons in the gene catalog of fecal microbiota of captive wild animals. A Count matrix of genes in pdu/cob-cbi-hem operons. Counts of individual genes of the pdu/cob-cbi-hem operons were transformed by variance-stabilizing transformations through Deseq2. B/C/D Differential analysis of gene abundance of pdu/cob-cbi-hem operons in fecal microbiota of animals with different diet or gut physiology. The abundance of 42 genes from the pdu/cob-cbi-hem operons was extracted from the gene catalog and was compared: herbivore vs. carnivore (B), herbivore vs. omnivore (C), and hindgut vs. foregut (D)

pdu/cob-cbi-hem related genes were more diverse in carnivore than herbivore fecal microbiota

To investigate whether composition and diversity of the microbial community related to the presence of pdu, we generated taxonomic profiles from metagenomic sequences with mOTUs [36] and normalized, analyzed, and visualized phylum abundance with Phyloseq [37]. In parallel, we extracted gene abundance related to the pdu (including genes homologous to pduABN) and cob-cbi-hem operons (Table S3) from the gene catalog to construct a pdu/cob-cbi-hem functional profile. We calculated and compared diversity indices from mOTUs and a pdu/cob-cbi-hem functional profile with the R package “vegan.” Due to the small number of samples, the diversity indices of PDU functional profile were combined for comparison of carnivores/omnivores to herbivores.

In total, 7798 mOTUs were identified and assigned to 26 phyla, comparable to 27 phyla of rMAGs. In accordance with rMAGs phylogeny, the fecal microbiota of foregut- and hindgut-fermenting herbivores and omnivores was dominated by Bacteroidota and Bacillota in most samples (Fig. 5A, B). Alpha- and β-diversity of fecal microbiota were analyzed based on mOTU abundance. The α-diversity of fecal microbiota of herbivores was more rich, even, and diverse than those of carnivores/omnivores, based on Chao1, Simpson’s evenness, and Shannon indices, respectively (non-parametric Wilcoxon test, p < 0.05, Fig. S1). Both Chao1 and Shannon indices were higher in fecal microbiota of foregut than hindgut fermenters (p < 0.01, Fig. S1) and β-diversity was significantly influenced by diet and gut physiology (permutational multivariate analysis of variance based on Bray–Curtis index of the relative abundances of mOTUs, p < 0.05, Fig. S2) in agreement with other studies [2, 5, 24]. Lower compositional diversity of fecal microbiota of hindgut-fermenting carnivores compared to herbivores may be due to a faster gut passage/lower retention time of the dietary material [1] reducing the possibility for effective microbial degradation and fermentation.Fig. 5 Fecal microbiota composition and diversity. Taxonomic profiles of fecal microbiota from 25 animal species were constructed with mOTUs. Samples were grouped by diet (A) and fermentation organs (B). Candidatus Melainabacteria and Candidatus Saccharibacteria are provisional names for characterized but uncultured organisms. The ratio of richness (C), evenness (D), and alpha diversity (E) of pdu/cob-cbi-hem functional profile to mOTU taxonomic profile were calculated

While α-diversity of pdu/cob-cbi-hem functional profiles of fecal microbiota was not different across diets and gut physiology (Fig. S3), β-diversity based on Bray–Curtis index was significantly related to diet and gut physiology (p < 0.01, Fig. S4). In previous studies, α-diversity of human fecal microbiota (= hindgut-fermenting omnivores) was positively correlated with the diversity of selected glycosyl hydrolase, and compositional diversity related strongly to functional diversity in thermal spring sediments [38, 39]. For fecal microbiota of wild animals living in captivity, compositional α-diversity was not reflected in pdu/cob-cbi-hem diversity with a significantly higher ratio of pdu/cob-cbi-hem to compositional diversity of omnivores/carnivores to herbivores, or hindgut to foregut fermenters (Fig. 5C, D, E). These observations suggest the community diversity is not predictive of the richness of pdu/cbi-cob-hem, which point at high relevance of functions encoded by pdu/cbi-cob-hem especially in hindgut-fermenting omnivores and carnivores.

PDU driven glycerol and 1,2-PD metabolism in vitro is predicted by genetic and metabolic biomarkers

As our observational data suggested that the majority of fecal microbial communities was capable of PduCDE activity, albeit with higher potential if collected from hindgut-fermenting carnivores, we compared the impact of 1,2-PD and glycerol addition on substrate metabolism and overall fermentation activity during in vitro fermentation. Fecal samples (n = 23) from nine hindgut-fermenting carnivores, omnivores, and herbivores (each n = 3) and three foregut-fermenting herbivores that were included in metagenomic sequencing were selected for in vitro fermentation (Table S1). Samples were anaerobically incubated for 24 h using Macfarlane as control medium (MF-CON), which contains complex plant polysaccharides, protein and peptide sources and mucin (Suppl. Methods). Macfarlane medium was supplemented with glycerol (MF-GLYC) and 1,2-PD (MF-12PD, both 60–70 mM), which were compared to MF-CON. We collected samples at 0, 6, and 24 h during the fermentation for analysis of substrate utilization and metabolite formation.

Microbiota of hindgut fermenters produced higher level of total SCFA in MF-CON than foregut-fermenting microbiota (p < 0.05, Table S4), while microbiota of carnivores produced higher levels of total SCFAs than herbivore (p < 0.05) and/or omnivore (p < 0.05) microbiota at all conditions. In agreement with previous studies, supplementation with 1,2-PD increased the proportion of propionate and decreased the proportion of butyrate produced by carnivore, herbivore, and hindgut-fermenter microbiota (p < 0.05, Table S4), while glycerol supplementation reduced the proportion of propionate (p < 0.05, Table S4) [22, 40].

With MF-GLYC and MF-12PD, the microbiota of carnivores consumed ~90% of the provided glycerol and 1,2-PD (Fig. 6A, C), respectively, which was consistent with higher abundance of pduC/pduCDE based on qPCR and DeSeq2 analysis. Carnivore fecal microbiota consumed more glycerol or 1,2-PD at 6 and/or 24 h (p < 0.01) and produced more 1,3-PD and propanol (Fig. 6A, B, C, D) than omnivores and herbivores. Hindgut-fermenting microbiota had higher glycerol/1,2-PD consumption and 1,3-PD/propanol formation than foregut fermenters at 6 or 24 h (p < 0.01) (Fig. 6F, G, H, I). With MF-12PD, carnivores and hindgut fermenters produced more propionate with 1,2-PD than herbivores or foregut fermenters (p < 0.01) (Fig. 6E, J), corresponding to higher abundance of pduLQW as suggested by DeSeq2 analysis.Fig. 6 Metabolite production from glycerol and 1,2-PD from fermented fecal samples. Substrate utilization and metabolite formation in samples collected from carnivores, omnivores and herbivores (A-E) and hindgut and foregut fermenters (F-J) were determined with HPLC-RI. The consumption of glycerol (A/F) and 1,2-PD (C/H) and during fermentation with fecal microbiota of animals following different diets is shown, as well as the corresponding metabolites, 1,3-PD (B/G), propanol (D/I), and propionate (E/J). The fermentations were performed with Macfarlane medium (MF-CON) or Macfarlane medium with glycerol (A, B, F, H) or 1,2-PD (C-E, H-J) anaerobically at 37 °C for 24 h. Two samples from each species (exception of the Chilean flamingo) were used, and each sample was fermented in triplicates. The figures were plotted using SigmaPlot, box plots show median, 25 and 75% percentiles, and dots indicate individual values

Factor analysis of mixed data (FAMD) was conducted to investigate the association between quantitative data including content of total and individual SCFAs, substrate consumption, metabolite production, absolute pduC abundance, and α-diversity indices, with qualitative data, including diet and gut physiology. The first two principal components explained 54.9% of the variation (Fig. 7). Combining qualitative and quantitative variables, α-diversity indices were positively related to fecal microbiota derived from herbivores and foregut fermenters, and negatively to carnivores, hindgut fermenters, and pduC abundance (Fig. 7). Carnivore microbiota was positively related to substrate utilization and metabolite concentrations except for butyrate production from MF-12PD. The consumption of glycerol and 1,2-PD positively related to the production of 1,3-PD, propionate, and propanol. These results re-emphasize that pduC abundance and Pdu related activity rather are a function of hindgut-fermenting carnivores than of foregut fermenter/herbivores.Fig. 7 Factor analysis of mixed data including quantitative and qualitative variables. Factor analysis of mixed data (FAMD) was used to analyze association between all variables; both qualitative (diets and gut physiologies; A) and quantitative variables (α-diversity indices, substrate consumption, metabolite production, and individual SCFA concentrations; B) are shown

Conclusion

The trait-based approach used in this study established the relationship between diversity and selective functioning of the microbial community from animal fecal samples and allowed the prediction of a common function in animal fecal microbiota based on genetic and metabolic biomarkers. Our results suggest that gut physiology and animal diet influence gut microbial composition and the potential of gut microbiota to utilize glycerol/1,2-PD and to produce cobalamin. We conclude that abundance of pduC/pduCDE can be used as biomarkers to predict potential glycerol or 1,2-PD utilization via Pdu activity from a complex microbial community. In addition, this study identifies the gut microbiota of wild animals as a largely untapped resource for the discovery of novel microbes and functions that might be of relevance in biotechnological approaches.

Material and methods

Sample collection

Fresh fecal samples were collected using a noninvasive method by zookeepers in 2020 and 2021. Fecal matter was collected using a sterile tool, transferred into sterile 50-ml containers, and immediately frozen at − 20 °C until further processing. No contact with animals occurred during sample collection. In total, 104 fresh fecal samples from 55 animal species were obtained: ten samples from hindgut-fermenting carnivores, 25 samples from hindgut-fermenting omnivores, 42 samples from foregut-fermenting herbivores, and 27 samples from hindgut-fermenting herbivores (Table S1). The first batch of samples included 49 samples from Givskud Zoo and 7 from Copenhagen Zoo, both in Denmark. At the second collection, 48 fecal samples were obtained from Givskud from the same animal species as the first batch with the exception of the Chilean flamingo.

DNA isolation from fecal samples

DNA from 0.2 to 0.3 g frozen fecal samples was isolated using the FastDNA Spin Kit for Soil (MP Biomedicals) following the instructions with exceptions. Briefly, samples were lysed twice at 6.0 m∙s−1 for 40 s using Lysing Matrix E tubes and a FastPrep-24 instrument (MP Biomedicals). DNA was eluted with nuclease-free water. The quality of DNA was evaluated by agarose gel electrophoresis to test DNA degradation, and the concentration of DNA was measured with Qubit 2.0 by Novogene Sequencing Europe.

Quantitative PCR

Quantitative PCR (qPCR) was conducted to quantify selected bacterial groups harboring pduC that were previously shown to be prevalent in human fecal microbiota and artificial intestinal microbiota that was derived from animals, including A. hallii, L. reuteri, B. obeum, R. gnavus, F. plautii, and V. dispar [22]. In addition, primers were generated targeting pduC of C. perfringens (Table S2, Suppl. Methods).

Standards were prepared, and qPCR were run as previously described [41]. Briefly, a tenfold dilution series of each linearized plasmid containing the target gene was included in the run to determine the linear range and the limits of detection (Table S2). Each run contained negative controls without template DNA. PCR protocols were run for 40 cycles as outlined in Suppl. Methods followed by melting curve analysis. For results below the detection limit of the standards, ½ log gene counts at the detection limit were used for statistical analysis.

Metagenomic sequencing and data processing

DNA samples from 25 captive animal species were selected based on DNA quality, 1,3-PD concentration of samples, and host properties, and sent for shotgun sequencing using an Illumina HiSeq (Novogene Sequencing Europe). These samples encompassed different gut physiology (11 from foregut and 14 from hindgut) and diets (17 from herbivores and three/five from carnivores/omnivores) (Table S1).

Metagenomic sequencing datasets were processed as previously described [42]. Briefly, BBMap (v.38.71) was used to quality control sequencing reads from all samples by removing adapters from the reads, removing reads that mapped to quality control sequences (PhiX genome), and discarding low-quality reads (trimq = 14, maq = 20, maxns = 1, and minlength = 45). Quality-controlled reads were merged using bbmerge.sh with a minimum overlap of 16 bases, resulting in merged, unmerged paired, and single reads. The reads from metagenomic samples were assembled into scaffolded contigs (hereafter scaffolds) using the SPAdes assembler (v3.15.2) [43] in metagenomic mode. Scaffolds with a length of ≥ 500 bp were used for gene calling using prodigal (v2.6.3, -c -q -m -p meta) [44].

Scaffolds were length-filtered (≥ 1000 bp) and quality-controlled reads from each metagenomic sample were mapped against the scaffolds of each sample. Mapping was performed using BWA (v0.7.17-r1188; -a) [45]. Alignments were filtered to be at least 45 bp in length, with an identity of ≥ 97% and a coverage of ≥ 80% of the read sequence. The resulting BAM files were processed using the jgi_summarize_bam_contig_depths script of MetaBAT2 (v2.12.1) to compute within- and between-sample coverages for each scaffold [46]. The scaffolds were binned by running MetaBAT2 on all samples individually (–minContig 2000 and –maxEdges 500). Metagenomic bins were annotated with Anvio (v7.1.0) [47] and quality-controlled using the CheckM (v1.0.13) [48] lineage workflow (completeness ≥ 50% and contamination < 10%) to generate 4958 prokaryotic metagenomic assembled genomes (MAGs). Completeness was predicted using Prokka (v1.14.6) [49], and MAGs were taxonomically annotated with GTDBtk (v1.7) [50]. A representative set of MAGs (rMAGs, n = 4150) was generated by clustering all MAGs using the dRep (v3.2.2, S_ani = 0.95) [51] dereplicate workflow.

The phylogenetic relationship of 4150 rMAGs was inferred with a maximum likelihood alignment-based approach with PhyloPhlAn3 [52]. Visualization and annotation of the tree was performed with GGTREE [53].

Construction of a captive animal fecal microbial gene catalog

Genes from the complete set of MAGs (n = 7,985,556), genes from the scaffolds (n = 58,355,880), and manually selected pduCDE genes (n = 166) were clustered at 95% identity using CD-HIT (v4.8.1) with the parameters -c 0.95 -M 0 -G 0 -aS 0.9 -g 1 -r 0 -d 0 -b 1000 while keeping the longest sequence as representative. Representative gene sequences (n = 46,443,565) were aligned against the KEGG database (release April 2021) using DIAMOND (v2.0.15) [54] and filtered to have a minimum query and subject coverage of 70% and requiring a bitScore of at least 50% of the maximum expected bitScore (i.e., score when the reference sequence is aligned against itself) to construct a gene catalog.

Quality-controlled metagenomic sequencing reads were aligned against the gene catalog and read abundances were normalized to a gene length of 1000 bp. Gene abundances were divided by the median of the abundances of ten universal single copy marker genes (COG0012, COG0016, COG0018, COG0172, COG0215, COG0495, COG0525, COG0533, COG0541, COG0552) to derive per-cell abundances and multiplied by 1000 [55, 56].

Taxonomic profiling and microbiota diversity analysis

Version 3.0.1 of the mOTUs database was augmented with the 4958 prokaryotic MAGs using the mOTUs-extender tool (https://github.com/motu-tool/mOTUs-extender). Metagenomic sequencing samples were then taxonomically profiled using mOTUs [36] with the extended database as reference.

Phylum abundances were analyzed and plotted with Phyloseq [37] after rarefication with the median sum of all mOTUs. Species richness (Chao1), evenness (Simpson’s evenness), and diversity (Shannon’s) indices were calculated after rarefaction of mOTUs abundance with minimum sum of all mOTUs via the vegan package as measures of α-diversity [57]. Beta-diversity was determined with non-metric multidimensional scaling (NMDS) based on dissimilarity matrix of Bray–Curtis index calculated from the relative abundances of mOTUs using the vegan package [57].

Identification of pdu/cob-cbi-hem related genes and analysis of functional profiles

For rMAGs with at least one of three pduCDE genes, the presence of pdu/cob-cbi-hem related genes (Table S3) were tested to explore the co-occurrence of pdu with cob-cbi-hem genes. To determine the distribution of operons among different animals, the abundance of pdu/cob-cbi-hem related genes (Table S3) in metagenomic samples was extracted from the gene catalog to construct a pdu/cob-cbi-hem functional profile.

The pdu genes encoding PduGH and PduABB’JKNTU to construct the shell of BMC in Salmonella enterica [58] have not been assigned to KO numbers and were therefore not identified in the gene catalog that was annotated based on the KEGG database. We extracted genes encoding proteins that were homologous to PduABN from our gene catalog and added them into pdu/cob-cbi-hem functional profile (Table S3). In detail, we added EutM and EutK, which were identified a homologous proteins to PduA in Salmonella typhimurium and to CcmK in Synechocystis sp. strain PCC6803 [59, 60]. EutL and EutS were homologs to PduB [60], and EutN and CcmL were homologs to PduN [60].

Gene counts the pdu/cob-cbi-hem functional profile were rarefied with the minimum sum of all genes within animal samples after rounded into integers for α-diversity indices calculation. Alpha-diversity indices were calculated with rarefied gene counts of pdu/cob-cbi-hem functional profiles. The relative abundance of rarefied gene counts was used for β-diversity ordination with NMDS based on Bray–Curtis index. Rounded gene counts of the pdu/cob-cbi-hem functional profile were used for functional differential analysis based on diets or gut physiology by Deseq2 (v1.36.0), and the results were processed by LFC shrinkage with “ashr” [61]. The functional profile processed via DESeq2 was corrected with variance stabilizing transformation to remove the dependence of the variance on the mean for count matrix plotting [35]. The significance level was 0.05, and the log2-fold change cutoff was 1.

Anaerobic batch fermentations to determine the PDU-dependent metabolic activity of selected fecal microbiota

To assess the potential PDU activity of selected fecal microbiota, anaerobic fermentation of fecal samples was performed under standardized conditions using anaerobically prepared modified Macfarlane medium [62] containing complex peptides and carbohydrates as nutrients (Suppl. Methods) that mimicked substrate availability in intestinal environments and was suitable for the diverse samples included in this study. Macfarlane medium does not contain free glycerol or fucose [63] to disrupt fermentation. Fecal samples (n = 23) from 12 animal species (three hindgut-fermenting carnivores, omnivores, and herbivores and three foregut-fermenting herbivores, Table S1) were selected for anaerobic fermentation. Macfarlane media was adapted to conditions for batch fermentations by increasing buffer capacity [64] and was supplemented with approx. 60 mM glycerol and 70 mM 1,2-PD (both Merck). Fecal slurries were prepared by resuspending fecal material in anaerobic peptone water to prepare a 10% (w/v) fecal suspension in an anaerobic chamber (Baker Ruskinn). The slurry supernatant was inoculated into fresh Macfarlane media to obtain a 1% inoculation. Samples were incubated at 37 °C for 24 h and collected at 0, 6, and 24 h for substrate and metabolite analysis. Two independently collected fecal samples were used from each species except for the Chilean flamingo. Two or three independent biological replicates were run for every fecal sample.

Substrate and metabolite analysis using high-performance liquid chromatography with a refractive index detector (HPLC-RI)

A 1260 Infinity II LC with RI (Agilent) was used to determine substrates and metabolites in fecal samples and during anaerobic fermentation of fecal samples, including glycerol, 1,2-PD, 1,3-PD, propanol, propionate, acetate, and butyrate as described previously [41]. Metabolites were extracted from 200 to 300 mg of animal feces with 5 mM H2SO4; supernatants from batch fermentations were analyzed directly using a Hi-Plex-H column (300 × 7.7 mm) that was connected to a guard column (7.7 × 50 mm, 8 µm, all Agilent). The mobile phase was 5 mM H2SO4 with a flow rate of 0.6 ml∙min−1 at 40 °C. External standards were used for quantification. The minimum detection limit was 0.01 mM.

Statistical analysis

Every dataset was tested for normality using the Shapiro–Wilk test. Alpha-diversity indices were compared using nonparametric Wilcoxon tests to identify differences between diets and gut physiologies. Permutational multivariate analysis of variance was used to test whether the difference in composition of the microbial communities was explained by the grouping of these samples in different diets or gut physiologies. Carnivore and herbivore samples were combined because of the limited number of samples for metagenomic sequencing. The differences of pduC abundance in the fecal microbiota of animals with different diets and gut physiologies were compared by the nonparametric Wilcoxon test. The occurrence of pduC dependent on diet and gut physiology and the proportion of pduC from a given species within different diets or gut physiologies were compared with the z test.

The significance of the differences in substrate consumption, metabolite production, and SCFAs at 6 and 24 h was assessed with Kruskal–Wallis test followed by Mann–Whitney pairwise test for microbiota derived from carnivores, omnivores, and herbivores, and with Mann–Whitney test to compare between microbiota of hindgut and foregut fermenters. Factor analysis of mixed data (FAMD) was used to analyze association between all variables, both quantitative and qualitative variables via FactoMineR [65] and results were visualized via factoextra [66].

Supplementary Information

Additional file 1: Supplement methods.

Acknowledgements

We would like to express our gratitude for the review of this article provided by Prof. Angela Schwarm from Norwegian University of Life Sciences. Additionally, we acknowledge laboratory support from Flavie Rouss.

Authors’ contributions

CS and QL designed and coordinated the study and supervised the research work. KS collected fecal samples at Givskud Zoo-Zootopia. AS provided frozen fecal samples collected at Copenhagen Zoo. QL, LØ, and ML performed experiments and collected data. HR performed metagenomic sequence processing and supported further data analysis by QL. QL and CS organized the data, conducted statistical analysis, and were responsible for data visualization. CS, QL, and HR wrote the manuscript with support of SS and AS. All authors made substantial contributions and approved the final version of the manuscript.

Funding

This work was supported by the Aarhus Universitet Forsknings Fonden Start-up Grant (grant AUFF-F-2020–7) to CS. QL acknowledges funding from EMBO (Scientific Exchange Grant: nr. 9152), which supported the stay at ETH Zürich.

Availability of data and materials

Metagenomic datasets are available at ENA, accession number: PRJEB62607, link for reviewers: https://www.ebi.ac.uk/ena/browser/view/PRJEB62607. All other data is presented in the manuscript or in supplementary materials.

Declarations

Ethics approval and consent to participate.

Fresh fecal samples were collected using a non-invasive method by zookeepers under the guidance of zoo veterinarians in 2020 and 2021 during regular maintenance of the housing facilities.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Stevens CE, Hume ID, editors. Comparative physiology of the vertebrate digestive system. 2nd ed. Cambridge, UK: Cambridge University Press; 2004.
2. Ley RE Hamady M Lozupone C Evolution of mammals and their gut microbes Sci 2008 320 1647 1651 10.1126/science.1155725
Ley RE, Hamady M, Lozupone C, et al. Evolution of mammals and their gut microbes. Sci. 2008;320:1647–51.
3. Youngblut ND Reischer GH Walters W Host diet and evolutionary history explain different aspects of gut microbiome diversity among vertebrate clades Nat Commun 2019 10 2200 10.1038/s41467-019-10191-3 31097702
Youngblut ND, Reischer GH, Walters W, et al. Host diet and evolutionary history explain different aspects of gut microbiome diversity among vertebrate clades. Nat Commun. 2019;10:2200.31097702
4. Gregor R Probst M Eyal S Mammalian gut metabolomes mirror microbiome composition and host phylogeny ISME J 2021 16 1262 1274 10.1038/s41396-021-01152-0 34903850
Gregor R, Probst M, Eyal S, et al. Mammalian gut metabolomes mirror microbiome composition and host phylogeny. ISME J. 2021;16:1262–74.34903850
5. Levin D Raab N Pinto Y Diversity and functional landscapes in the microbiota of animals in the wild Sci 2021 372 eabb5352 10.1126/science.abb5352
Levin D, Raab N, Pinto Y, et al. Diversity and functional landscapes in the microbiota of animals in the wild. Sci. 2021;372:eabb5352.
6. Li F Li X Cheng CC A phylogenomic analysis of Limosilactobacillus reuteri reveals ancient and stable evolutionary relationships with rodents and birds and zoonotic transmission to humans BMC Biol 2023 21 53 10.1186/s12915-023-01541-1 36907868
Li F, Li X, Cheng CC, et al. A phylogenomic analysis of Limosilactobacillus reuteri reveals ancient and stable evolutionary relationships with rodents and birds and zoonotic transmission to humans. BMC Biol. 2023;21:53.36907868
7. van der Hee B Wells JM Microbial regulation of host physiology by short-chain fatty acids Trends Microbiol 2021 29 700 712 10.1016/j.tim.2021.02.001 33674141
van der Hee B, Wells JM. Microbial regulation of host physiology by short-chain fatty acids. Trends Microbiol. 2021;29:700–12.33674141
8. Engels C Ruscheweyh HJ Beerenwinkel N The common gut microbe Eubacterium hallii also contributes to intestinal propionate formation Front Microbiol 2016 7 713 10.3389/fmicb.2016.00713 27242734
Engels C, Ruscheweyh HJ, Beerenwinkel N, et al. The common gut microbe Eubacterium hallii also contributes to intestinal propionate formation. Front Microbiol. 2016;7:713.27242734
9. Engels C Schwab C Zhang J Acrolein contributes strongly to antimicrobial and heterocyclic amine transformation activities of reuterin Sci Rep 2016 6 36246 10.1038/srep36246 27819285
Engels C, Schwab C, Zhang J, et al. Acrolein contributes strongly to antimicrobial and heterocyclic amine transformation activities of reuterin. Sci Rep. 2016;6:36246.27819285
10. Yang M Wenner N Dykes GF Biogenesis of a bacterial metabolosome for propanediol utilization Nat Commun 2022 13 2920 10.1038/s41467-022-30608-w 35614058
Yang M, Wenner N, Dykes GF, et al. Biogenesis of a bacterial metabolosome for propanediol utilization. Nat Commun. 2022;13:2920.35614058
11. Kerfeld CA Aussignargues C Zarzycki J Bacterial microcompartments Nat Rev Microbiol 2018 16 277 290 10.1038/nrmicro.2018.10 29503457
Kerfeld CA, Aussignargues C, Zarzycki J, et al. Bacterial microcompartments. Nat Rev Microbiol. 2018;16:277–90.29503457
12. Louis P Flint H Formation of propionate and butyrate by the human colonic microbiota Environ Microbiol 2017 19 29 41 10.1111/1462-2920.13589 27928878
Louis P, Flint H. Formation of propionate and butyrate by the human colonic microbiota. Environ Microbiol. 2017;19:29–41.27928878
13. Koropatkin NM Cameron EA Martens EC How glycan metabolism shapes the human gut microbiota Nat Rev Microbiol 2014 10 323 335 10.1038/nrmicro2746
Koropatkin NM, Cameron EA, Martens EC. How glycan metabolism shapes the human gut microbiota. Nat Rev Microbiol. 2014;10:323–35.
14. Burton RA Gidley MJ Fincher GB Heterogeneity in the chemistry, structure and function of plant cell walls Nat Chem Biol 2010 6 724 732 10.1038/nchembio.439 20852610
Burton RA, Gidley MJ, Fincher GB. Heterogeneity in the chemistry, structure and function of plant cell walls. Nat Chem Biol. 2010;6:724–32.20852610
15. Sokolovskaya OM Shelton AN Taga ME Sharing vitamins: cobamides unveil microbial interactions Sci 2020 369 48
Sokolovskaya OM, Shelton AN, Taga ME. Sharing vitamins: cobamides unveil microbial interactions. Sci. 2020;369:48.
16. Morita H Hidehiro TOH Fukuda S Comparative genome analysis of Lactobacillus renteri and Lactobacillus fermentum reveal a genomic island for reuterin and cobalamin production DNA Res 2008 15 151 161 10.1093/dnares/dsn009 18487258
Morita H, Hidehiro TOH, Fukuda S, et al. Comparative genome analysis of Lactobacillus renteri and Lactobacillus fermentum reveal a genomic island for reuterin and cobalamin production. DNA Res. 2008;15:151–61.18487258
17. Sriramulu DD Liang M Hernandez-Romero D Lactobacillus reuteri DSM 20016 produces cobalamin-dependent diol dehydratase in metabolosomes and metabolizes 1,2-propanediol by disproportionation J Bacteriol 2008 190 4559 4567 10.1128/JB.01535-07 18469107
Sriramulu DD, Liang M, Hernandez-Romero D, et al. Lactobacillus reuteri DSM 20016 produces cobalamin-dependent diol dehydratase in metabolosomes and metabolizes 1,2-propanediol by disproportionation. J Bacteriol. 2008;190:4559–67.18469107
18. Vital M Gao J Rizzo M Diet is a major factor governing the fecal butyrate-producing community structure across Mammalia Aves and Reptilia ISME J 2015 9 832 843 10.1038/ismej.2014.179 25343515
Vital M, Gao J, Rizzo M, et al. Diet is a major factor governing the fecal butyrate-producing community structure across Mammalia. Aves and Reptilia ISME J. 2015;9:832–43.25343515
19. Cummings JH Pomare EW Branch HWJ Short chain fatty acids in human large intestine, portal, hepatic and venous blood Gut 1987 28 1221 1227 10.1136/gut.28.10.1221 3678950
Cummings JH, Pomare EW, Branch HWJ, et al. Short chain fatty acids in human large intestine, portal, hepatic and venous blood. Gut. 1987;28:1221–7.3678950
20. Reichardt N Duncan SH Young P Phylogenetic distribution of three pathways for propionate production within the human gut microbiota ISME J 2014 8 1323 1335 10.1038/ismej.2014.14 24553467
Reichardt N, Duncan SH, Young P, et al. Phylogenetic distribution of three pathways for propionate production within the human gut microbiota. ISME J. 2014;8:1323–35.24553467
21. Hecker N Sharma V Hiller M Convergent gene losses illuminate metabolic and physiological changes in herbivores and carnivores Proc Natl Acad Sci U S A 2019 116 3036 3041 10.1073/pnas.1818504116 30718421
Hecker N, Sharma V, Hiller M. Convergent gene losses illuminate metabolic and physiological changes in herbivores and carnivores. Proc Natl Acad Sci U S A. 2019;116:3036–41.30718421
22. Ramirez Garcia A Zhang J Greppi A Impact of manipulation of glycerol/diol dehydratase activity on intestinal microbiota ecology and metabolism Environ Microbiol 2021 23 1765 1779 10.1111/1462-2920.15431 33587772
Ramirez Garcia A, Zhang J, Greppi A, et al. Impact of manipulation of glycerol/diol dehydratase activity on intestinal microbiota ecology and metabolism. Environ Microbiol. 2021;23:1765–79.33587772
23. Zhang J Lacroix C Wortmann E Gut microbial beta-glucuronidase and glycerol/diol dehydratase activity contribute to dietary heterocyclic amine biotransformation BMC Microbiol 2019 19 99 10.1186/s12866-019-1483-x 31096909
Zhang J, Lacroix C, Wortmann E, et al. Gut microbial beta-glucuronidase and glycerol/diol dehydratase activity contribute to dietary heterocyclic amine biotransformation. BMC Microbiol. 2019;19:99.31096909
24. Zoelzer F Burger AL Dierkes PW Unraveling differences in fecal microbiota stability in mammals: from high variable carnivores and consistently stable herbivores Anim Microbiome 2021 3 77 10.1186/s42523-021-00141-0 34736528
Zoelzer F, Burger AL, Dierkes PW. Unraveling differences in fecal microbiota stability in mammals: from high variable carnivores and consistently stable herbivores. Anim Microbiome. 2021;3:77.34736528
25. Guo Y Dai L Xin B 1,3-Propanediol production by a newly isolated strain Clostridium perfringens GYL Bioresource Technol 2017 233 404 412
Guo Y, Dai L, Xin B, et al. 1,3-Propanediol production by a newly isolated strain. Clostridium perfringens GYL Bioresource Technol. 2017;233:404–12.
26. Greppi A Asare PT Schwab C Isolation and comparative genomic analysis of reuterin-producing Lactobacillus reuteri from the chicken gastrointestinal tract Front Microbiol 2020 11 1166 10.3389/fmicb.2020.01166 32670217
Greppi A, Asare PT, Schwab C, et al. Isolation and comparative genomic analysis of reuterin-producing Lactobacillus reuteri from the chicken gastrointestinal tract. Front Microbiol. 2020;11:1166.32670217
27. Duncan SH Louis P Flint HJ Lactate-utilizing bacteria, isolated from human feces, that produce butyrate as a major fermentation product Appl Environ Microbiol 2004 70 5810 5817 10.1128/AEM.70.10.5810-5817.2004 15466518
Duncan SH, Louis P, Flint HJ. Lactate-utilizing bacteria, isolated from human feces, that produce butyrate as a major fermentation product. Appl Environ Microbiol. 2004;70:5810–7.15466518
28. Torres AC Vannini V Bonacina J Cobalamin production by Lactobacillus coryniformis: biochemical identification of the synthetized corrinoid and genomic analysis of the biosynthetic cluster BMC Microbiol 2016 16 240 10.1186/s12866-016-0854-9 27737643
Torres AC, Vannini V, Bonacina J, et al. Cobalamin production by Lactobacillus coryniformis: biochemical identification of the synthetized corrinoid and genomic analysis of the biosynthetic cluster. BMC Microbiol. 2016;16:240.27737643
29. De Angelis M Bottacini F Fosso B Lactobacillus rossiae, a vitamin B12 producer, represents a metabolically versatile species within the genus Lactobacillus PLoS ONE 2014 9 e107232 10.1371/journal.pone.0107232 25264826
De Angelis M, Bottacini F, Fosso B, et al. Lactobacillus rossiae, a vitamin B12 producer, represents a metabolically versatile species within the genus Lactobacillus. PLoS ONE. 2014;9:e107232.25264826
30. Staib L Fuchs TM From food to cell: nutrient exploitation strategies of enteropathogens Microbiol 2014 160 1020 1039 10.1099/mic.0.078105-0
Staib L, Fuchs TM. From food to cell: nutrient exploitation strategies of enteropathogens. Microbiol. 2014;160:1020–39.
31. Fekry MI Engels C Zhang J The strict anaerobic gut microbe Eubacterium hallii transforms the carcinogenic dietary heterocyclic amine 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) Environ Microbiol Rep 2016 8 201 209 10.1111/1758-2229.12369 26711372
Fekry MI, Engels C, Zhang J, et al. The strict anaerobic gut microbe Eubacterium hallii transforms the carcinogenic dietary heterocyclic amine 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP). Environ Microbiol Rep. 2016;8:201–9.26711372
32. Morris JJ Lenski RE Zinser ER The black queen hypothesis: evolution of dependencies through adaptive gene loss mBio 2012 3 e0003612 10.1128/mBio.00036-12
Morris JJ, Lenski RE, Zinser ER. The black queen hypothesis: evolution of dependencies through adaptive gene loss. mBio. 2012;3:e0003612.
33. Lee MJ Palmer DJ Warren MJ Biotechnological advances in bacterial microcompartment technology Trends Biotechnol 2019 37 325 336 10.1016/j.tibtech.2018.08.006 30236905
Lee MJ, Palmer DJ, Warren MJ. Biotechnological advances in bacterial microcompartment technology. Trends Biotechnol. 2019;37:325–36.30236905
34. Kirst H Kerfeld CA Bacterial microcompartments: catalysis-enhancing metabolic modules for next generation metabolic and biomedical engineering BMC Biol 2019 17 79 10.1186/s12915-019-0691-z 31601225
Kirst H, Kerfeld CA. Bacterial microcompartments: catalysis-enhancing metabolic modules for next generation metabolic and biomedical engineering. BMC Biol. 2019;17:79.31601225
35. Love MI Huber W Anders S Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 Genome Biol 2014 15 550 10.1186/s13059-014-0550-8 25516281
Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550.25516281
36. Ruscheweyh HJ Milanese A Paoli L Cultivation-independent genomes greatly expand taxonomic-profiling capabilities of mOTUs across various environments Microbiome 2022 10 212 10.1186/s40168-022-01410-z 36464731
Ruscheweyh HJ, Milanese A, Paoli L, et al. Cultivation-independent genomes greatly expand taxonomic-profiling capabilities of mOTUs across various environments. Microbiome. 2022;10:212.36464731
37. McMurdie PJ Holmes S Phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data PLoS ONE 2013 8 e61217 10.1371/journal.pone.0061217 23630581
McMurdie PJ, Holmes S. Phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE. 2013;8:e61217.23630581
38. Flores R Shi J Gail MH Association of fecal microbial diversity and taxonomy with selected enzymatic functions PLoS ONE 2012 7 e39745 10.1371/journal.pone.0039745 22761886
Flores R, Shi J, Gail MH, et al. Association of fecal microbial diversity and taxonomy with selected enzymatic functions. PLoS ONE. 2012;7:e39745.22761886
39. Ruhl IA Sheremet A Smirnova A Microbial functional diversity correlates with species diversity along a temperature gradient mSystems 2022 7 e0099121 10.1128/msystems.00991-21 35166562
Ruhl IA, Sheremet A, Smirnova A, et al. Microbial functional diversity correlates with species diversity along a temperature gradient. mSystems. 2022;7:e0099121.35166562
40. Ramirez Garcia A Greppi A Constancias F Anaerobutyricum hallii promotes the functional depletion of a food carcinogen in diverse healthy fecal microbiota Front Microbiomes 2023 2 1194516 10.3389/frmbi.2023.1194516
Ramirez Garcia A, Greppi A, Constancias F, et al. Anaerobutyricum hallii promotes the functional depletion of a food carcinogen in diverse healthy fecal microbiota. Front Microbiomes. 2023;2:1194516.
41. Yazdi FG Barner L Li Q Long-term daily high-protein, drained yoghurt consumption alters abundance of selected functional groups of the human gut microbiota and fecal short-chain fatty acid profiles in a cohort of overweight and obese women J Funct Foods 2022 93 105089 10.1016/j.jff.2022.105089
Yazdi FG, Barner L, Li Q, et al. Long-term daily high-protein, drained yoghurt consumption alters abundance of selected functional groups of the human gut microbiota and fecal short-chain fatty acid profiles in a cohort of overweight and obese women. J Funct Foods. 2022;93:105089.
42. Paoli L Ruscheweyh HJ Forneris CC Biosynthetic potential of the global ocean microbiome Nature 2022 607 111 118 10.1038/s41586-022-04862-3 35732736
Paoli L, Ruscheweyh HJ, Forneris CC, et al. Biosynthetic potential of the global ocean microbiome. Nature. 2022;607:111–8.35732736
43. Nurk S Meleshko D Korobeynikov A Pevzner PA MetaSPAdes: a new versatile metagenomic assembler Genome Res 2017 27 824 834 10.1101/gr.213959.116 28298430
Nurk S, Meleshko D, Korobeynikov A, Pevzner PA. MetaSPAdes: a new versatile metagenomic assembler. Genome Res. 2017;27:824–34.28298430
44. Hyatt D Chen G-L Locascio PF Prodigal: prokaryotic gene recognition and translation initiation site identification BMC Bioinformatics 2010 11 119 10.1186/1471-2105-11-119 20211023
Hyatt D, Chen G-L, Locascio PF, et al. Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinformatics. 2010;11:119.20211023
45. Li H Durbin R Fast and accurate short read alignment with Burrows-Wheeler transform Bioinformatics 2009 25 1754 1760 10.1093/bioinformatics/btp324 19451168
Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25:1754–60.19451168
46. Kang DD Li F Kirton E MetaBAT 2: an adaptive binning algorithm for robust and efficient genome reconstruction from metagenome assemblies PeerJ 2019 7 e7359 10.7717/peerj.7359 31388474
Kang DD, Li F, Kirton E, et al. MetaBAT 2: an adaptive binning algorithm for robust and efficient genome reconstruction from metagenome assemblies. PeerJ. 2019;7:e7359.31388474
47. Eren AM Esen OC Quince C Anvi’o: an advanced analysis and visualization platform for ’omics data PeerJ 2015 2015 e1319 10.7717/peerj.1319
Eren AM, Esen OC, Quince C, et al. Anvi’o: an advanced analysis and visualization platform for ’omics data. PeerJ. 2015;2015:e1319.
48. Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes Genome Res 2015 25 1043 1055 10.1101/gr.186072.114 25977477
Parks DH, Imelfort M, Skennerton CT, Hugenholtz P, Tyson GW. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 2015;25:1043–55.25977477
49. Seemann T Prokka: rapid prokaryotic genome annotation Bioinformatics 2014 30 2068 2069 10.1093/bioinformatics/btu153 24642063
Seemann T. Prokka: rapid prokaryotic genome annotation. Bioinformatics. 2014;30:2068–9.24642063
50. Chaumeil PA Mussig AJ Hugenholtz P Parks DH GTDB-Tk: a toolkit to classify genomes with the genome taxonomy database Bioinformatics 2020 36 1925 1927 10.1093/bioinformatics/btz848
Chaumeil PA, Mussig AJ, Hugenholtz P, Parks DH. GTDB-Tk: a toolkit to classify genomes with the genome taxonomy database. Bioinformatics. 2020;36:1925–7.
51. Olm MR Brown CT Brooks B Banfield JF DRep: a tool for fast and accurate genomic comparisons that enables improved genome recovery from metagenomes through de-replication ISME J 2017 11 2864 2868 10.1038/ismej.2017.126 28742071
Olm MR, Brown CT, Brooks B, Banfield JF. DRep: a tool for fast and accurate genomic comparisons that enables improved genome recovery from metagenomes through de-replication. ISME J. 2017;11:2864–8.28742071
52. Asnicar F Thomas AM Beghini F Precise phylogenetic analysis of microbial isolates and genomes from metagenomes using PhyloPhlAn 3.0 Nat Commun. 2020 11 2500 10.1038/s41467-020-16366-7 32427907
Asnicar F, Thomas AM, Beghini F, et al. Precise phylogenetic analysis of microbial isolates and genomes from metagenomes using PhyloPhlAn 3.0. Nat Commun. 2020;11:2500.32427907
53. Yu G Smith DK Zhu H GGTREE: an R package for visualization and annotation of phylogenetic trees with their covariates and other associated data Methods Ecol Evol 2017 8 28 36 10.1111/2041-210X.12628
Yu G, Smith DK, Zhu H, et al. GGTREE: an R package for visualization and annotation of phylogenetic trees with their covariates and other associated data. Methods Ecol Evol. 2017;8:28–36.
54. Buchfink B Reuter K Drost HG Sensitive protein alignments at tree-of-life scale using DIAMOND Nat Methods 2021 18 366 368 10.1038/s41592-021-01101-x 33828273
Buchfink B, Reuter K, Drost HG. Sensitive protein alignments at tree-of-life scale using DIAMOND. Nat Methods. 2021;18:366–8.33828273
55. Ruscheweyh HJ Milanese A Paoli L Cultivation-independent genomes greatly expand taxonomic-profiling capabilities of mOTUs across various environments Microbiome 2022 10 212 10.1186/s40168-022-01410-z 36464731
Ruscheweyh HJ, Milanese A, Paoli L, et al. Cultivation-independent genomes greatly expand taxonomic-profiling capabilities of mOTUs across various environments. Microbiome. 2022;10:212.36464731
56. Salazar G Paoli L Alberti A Gene expression changes and community turnover differentially shape the global ocean metatranscriptome Cell 2019 179 P1068 P1083 10.1016/j.cell.2019.10.014
Salazar G, Paoli L, Alberti A, et al. Gene expression changes and community turnover differentially shape the global ocean metatranscriptome. Cell. 2019;179:P1068–83.
57. Oksanen J, Simpson G, Blanchet F, et al. Vegan: community ecology package. 2022. Available from: https://CRAN.R-project.org/package=vegan.
58. Cheng S Sinha S Fan C Genetic analysis of the protein shell of the microcompartments involved in coenzyme B12-dependent 1,2-propanediol degradation by Salmonella J Bacteriol 2011 193 1385 1392 10.1128/JB.01473-10 21239588
Cheng S, Sinha S, Fan C, et al. Genetic analysis of the protein shell of the microcompartments involved in coenzyme B12-dependent 1,2-propanediol degradation by Salmonella. J Bacteriol. 2011;193:1385–92.21239588
59. Stojiljkovic I Bäumler AJ Heffron F Ethanolamine utilization in Salmonella typhimurium: nucleotide sequence, protein expression, and mutational analysis of the cchA cchB eutE eutJ eutG eutH gene cluster J Bacteriol 1995 177 1357 1366 10.1128/jb.177.5.1357-1366.1995 7868611
Stojiljkovic I, Bäumler AJ, Heffron F. Ethanolamine utilization in Salmonella typhimurium: nucleotide sequence, protein expression, and mutational analysis of the cchA cchB eutE eutJ eutG eutH gene cluster. J Bacteriol. 1995;177:1357–66.7868611
60. Kofoid E Rappleye C Stojiljkovic I Roth J The 17-gene ethanolamine (eut) operon of Salmonella typhimurium encodes five homologues of carboxysome shell proteins J Bacteriol 1999 181 5317 5329 10.1128/JB.181.17.5317-5329.1999 10464203
Kofoid E, Rappleye C, Stojiljkovic I, Roth J. The 17-gene ethanolamine (eut) operon of Salmonella typhimurium encodes five homologues of carboxysome shell proteins. J Bacteriol. 1999;181:5317–29.10464203
61. Stephens M False discovery rates: a new deal Biostatistics 2017 18 275 294 27756721
Stephens M. False discovery rates: a new deal. Biostatistics. 2017;18:275–94.27756721
62. Macfarlane GT Macfarlane S Gibson GR Validation of a three-stage compound continuous culture system for investigating the effect of retention time on the ecology and metabolism of bacteria in the human colon Microb Ecol 1998 35 180 187 10.1007/s002489900072 9541554
Macfarlane GT, Macfarlane S, Gibson GR. Validation of a three-stage compound continuous culture system for investigating the effect of retention time on the ecology and metabolism of bacteria in the human colon. Microb Ecol. 1998;35:180–7.9541554
63. Høgsgaard K Vidal NP Marietou A Fucose modifies short chain fatty acid and H2S formation through alterations of microbial cross-feeding activities FEMS Microbiol Ecol. 2023 99 fiad107 10.1093/femsec/fiad107 37777844
Høgsgaard K, Vidal NP, Marietou A, et al. Fucose modifies short chain fatty acid and H2S formation through alterations of microbial cross-feeding activities. FEMS Microbiol Ecol. 2023;99:fiad107.37777844
64. Bircher L Schwab C Geirnaert A Lacroix C Cryopreservation of artificial gut microbiota produced with in vitro fermentation technology Microb Biotechnol. 2018 11 163 75 10.1111/1751-7915.12844 28980453
Bircher L, Schwab C, Geirnaert A, Lacroix C. Cryopreservation of artificial gut microbiota produced with in vitro fermentation technology. Microb Biotechnol. 2018;11:163–75.28980453
65. Lê S Josse J Husson F FactoMineR: an R package for multivariate analysis J Stat Softw 2008 25 1 18 10.18637/jss.v025.i01
Lê S, Josse J, Husson F. FactoMineR: an R package for multivariate analysis. J Stat Softw. 2008;25:1–18.
66. Kassambara A, Mundt F. Factoextra: extract and visualize the results of multivariate data analyses. 2020. Available from: https://CRAN.R-project.org/package=factoextra.
