Symbiotic bacteria mediate volatile chemical signal synthesis in a large solitary mammal species

Composition of chemical constituents and bacterial communities in AGS and feces indicates separate, unique odor profiles

The gas chromatography–mass spectrometry analyses revealed that AGS volatiles of wild and captive pandas were comprised of a multicomponent blend of 30–50 chemical compounds, including fatty acids, aldehydes, ketones, aliphatic ethers, amides, aromatics, alcohols, steroids and squalene (Fig. 2a and Supplementary Table S2). These compounds are typical components of chemosignals across species due to their volatility, detectability and other characteristics facilitating chemoreception [3, 26, 32]. By contrast, feces contained mostly fatty acid ethyl ester, and a small number and quantity of fatty acids, amides, steroids and indole (Fig. 2b and Supplementary Table S3). Our results show that the relative abundance of steroids, aldehydes and fatty acids were remarkably higher in AGS than in feces (Fig. 3a), and the number of chemical components of aldehydes, fatty acids, and ketones in AGS was also significantly higher than found in feces (Fig. 3b). These results indicate that the chemical constituents of AGS are much better suited for chemosignaling than those from feces.

The composition of bacterial communities in AGS and feces was markedly different at the phylum (Fig. 3c) and genus levels (Fig. 3d), based on taxonomic classifications of predicted gene sequences. Principal Co-ordinates Analysis (PCoA) (Fig. 3e) and hierarchical clustering analyses (Fig. 3f) revealed cluster patterns based on provenance (captive/wild) and sample type (AGS/fecal). Notably, the microbiota composition of AGS from different individuals or living environments was more similar than were AGS and fecal samples from the same individuals. Actinobacteria (X² = 26.33, P < 0.0001) and Bacteroidetes (X² = 25.88, P < 0.0001) were the dominant phyla in AGS microbial communities, but scarcely appeared in feces (Fig. 3c). By contrast, Clostridium (X² = 18.61, P < 0.001) and Escherichia (X² = 14.26, P < 0.01) were prevalent in feces, but rare in AGS microbiomes (Supplementary Fig. S2a). Unlike the gut microbiomes, Corynebacterium (X² = 27.40, P < 0.0001), Pseudomonas (X² = 26.12, P < 0.0001), Porphyromonas (X² = 27.81, P < 0.0001) and Psychrobacter (X² = 20.80, P < 0.0001) dominated AGS microbiomes (Fig. 3d), and these, along with Peptoniphilus (X² = 26.89, P < 0.0001), Anaerococcus (X² = 23.99, P < 0.0001), and Trueperella (X² = 25.72, P < 0.0001) occurred in significantly higher proportions in AGS than fecal microbiomes (Supplementary Fig. S2a). Pseudomonas (W = 66, P < 0.05) abundance was higher in the AGS of wild than captive pandas, while Psychrobacter (W = 64.5, P < 0.05) was lower in wild pandas (Supplementary Fig. S2a). These findings provide clear evidence that the microbial communities of feces and the anogenital glands of pandas differ in important ways that could produce divergent metabolites with plausible communicatory significance. Actually, AGS microbes are not a simple extension of gut microbes, and the anogenital gland harbors a unique community of microbes likely adapted to the environmental conditions prevailing in the gland.

Comparative analysis of microbial KEGG pathways in AGS and feces

The Kyoto Encyclopedia of Genes and Genomes (KEGG) database provides an extensive non-redundant catalogue of microbiome genomics and metabolic pathways, providing opportunities to identify bacterial functions for the fermentation of the volatile chemical signals in AGS. To clarify microbial functions, we annotated the function of protein coding genes identified in whole metagenome data according to the KEGG orthology. The comparative analysis of microbial metabolic profiles showed a significant increase in functional genes for lipid metabolism in AGS compared to feces (Supplementary Fig. S3).

Further LEfSe analysis identified six fatty acid metabolic pathways with significantly disparate representation between the AGS and feces (Fig. 3g). Three lipid metabolism pathways, including fatty acid biosynthesis (ko00061) (X² = 9.32, P < 0.001), synthesis and degradation of ketone bodies (ko00072) (X² = 7.17, P < 0.05) and steroid biosynthesis (ko00100) (X² = 13.34, P < 0.05) were significantly higher in AGS groups than feces. By contrast, glycerophospholipid metabolism (ko00564) (X² = 13.87, P < 0.01), primary bile acid biosynthesis (ko00120) (X² = 16.91, P < 0.01) and secondary bile acid biosynthesis (ko00121) (X² = 24.91, P < 0.0001) were higher in the feces. Twelve pathways involved in lipid metabolism were significantly different (Supplementary Fig. S2b), with fatty acid biosynthesis (ko00061), biosynthesis of unsaturated fatty acids (ko01040) (X² = 8.94, P < 0.05), synthesis and degradation of ketone bodies (ko00072) and ether lipid metabolism (ko00565) (X² = 13.15, P < 0.001) higher in AGS than feces (Supplementary Fig. S2b).

The genes involved in lipid metabolism in AGS were primarily from Actinobacteria (42.67%), Proteobacteria (35.72%) and Firmicutes (14.48%) (Fig. 3h). Seven families of Actinobacteria (Nocardiopsaceae, Mycobacteriaceae, Microbacteriaceae, Nocardiaceae, Micrococcaceae, Corynebacteriaceae and Intrasporangiaceae in Actinobacteria) and three families of Proteobacteria (Bradyrhizobiaceae, Xanthomonadaceae and Desulfobacteraceae) were the top 10 family involved in lipid metabolism in AGS (Fig. 3h). Species and functional contribution analysis indicated that Clostridium and Escherichi were primary contributors supporting lipid metabolism in feces (Fig. 3i). It is interesting to note that Corynebacterium was the genus with greatest contributions to lipid metabolism in AGS for both wild and captive individuals. The next largest genus contributing to lipid metabolism was Pseudomonas in wild pandas and Psychrobacter in captive pandas (Fig. 3i). In pathway level 3, Corynebacterium, Pseudomonas and Psychrobacter also played a significant role in lipid metabolism, such as fatty acid biosynthesis (ko00061) and degradation (ko00071), glycerophospholipid metabolism (ko00564), glycerolipid metabolism (ko00561), biosynthesis of unsaturated fatty acids (ko01040), synthesis and degradation of ketone bodies (ko00072) found in metabolic pathways in AGS, but seldom appeared in feces (Supplementary Fig. S4).

Taken together, these results highlight important differences in genes supporting metabolic pathways for lipid metabolism, though there are taxonomic differences in the microbes functioning in feces versus AGS pathways. Differences appear related to function, with AGS microbes yielding important odorants, such as ketones and steroids, while fecal microbes are involved in digestive function (e.g., bile synthesis pathways).

Metabolic pathways associated with the production of volatile chemical odorants

The triacylglycerols (TAGs) comprise much of the secretions on the skin surface (Fig. 1). Triacylglycerol lipase (EC: 3.1.1.3) acts as a key enzyme to break down the ester bond which links fatty acid moieties with the glycerol backbone. In glycerolipid metabolism (ko00561), triacylglycerol lipase (EC: 3.1.1.3) cooperates with other enzymes to decompose TAGs into glycerol and fatty acids, which then enter the fatty acid degradation pathway (Supplementary Fig. S5). Interestingly, the abundance of triacylglycerol lipase (EC: 3.1.1.3) was significantly higher in AGS than feces (X² = 19.87, P < 0.001) (Supplementary Fig. S6). A total of 22 species of bacteria in AGS microbiomes contained triacylglycerol lipase (EC: 3.1.1.3) belonging to Actinobacteria (9), Proteobacteria (10) and Firmicutes (5). Further, most of these bacteria also contain other enzymes involved in lipid metabolism (Supplementary Fig. S7).

The synthesis of chemical signals secreted by the AGS may involve multiple lipid metabolism pathways (Fig. 4). In the fatty acid degradation pathway (ko00071), long-chain acyl-CoA synthetase (EC: 6.2.1.3) (X² = 20.50, P < 0.001), triacylglycerol lipase (EC: 3.1.1.3) translates the hexadecanoic acid to hexadecanoyl-CoA, and other enzymes contribute to degrade the hexadecanoyl-CoA to acetyl-CoA step by step (Fig. 4 and Supplementary Fig. S8). The relative abundance of five enzymes involved in fatty acid degradation in the AGS microbiome was significantly higher than in feces, including unspecific monooxygenase (EC: 1.14.14.1) (X² = 27.43, P < 0.0001), alkane 1-monooxygenase (EC: 1.14.15.3) (X² = 26.98, P < 0.0001), glutaryl-CoA dehydrogenase (EC: 1.3.8.6) (X² = 25.23, P < 0.0001) and carnitine O-palmitoyltransferase (EC: 2.3.1.21) (X² = 11.17, P < 0.05) (Supplementary Fig. S9). The function of aldehyde dehydrogenase (NAD + ) (EC: 1.2.1.3) (X² = 10.44, P < 0.05) is to translate fatty acids to aldehydes, while alcohol dehydrogenase 1/7 (EC: 1.1.1.1) (X² = 27.64, P < 0.0001) continues to convert them into alcohols (Fig. 4 and Supplementary Fig. S10).

Acetyl-CoA acts as the initial substrate for synthesis of medium- and long-chain fatty acids in the fatty acid biosynthesis pathway (ko00061) (Fig. 4 and Supplementary Fig. S10). Acetyl-CoA carboxylase (EC: 6.4.1.2) (X² = 6.41, P = 0.052), [acyl-carrier-protein] S-malonyl transferase (EC: 2.3.1.39) (X² = 7.04, P = 0.070), 3-oxoacyl-[acyl-carrier-protein] synthase II (EC: 2.3.1.179) (X² = 6.80, P = 0.078), 3-oxoacyl-[acyl-carrier protein] reductase (EC: 1.1.1.100) (X² = 9.39, P < 0.05), fatty acid synthase, animal type (EC: 2.3.1.85) (X² = 18.51, P < 0.001) and long-chain acyl-CoA synthetase (EC: 6.2.1.3) (X² = 20.50, P < 0.001) were highly enriched in this metabolic pathway and function to assist with synthesis of medium- and long-chain fatty acids (Fig. 4 and Supplementary Fig. S11).

As one of the primary chemical signals in AGS, ketones may be synthesized by the synthesis and degradation pathway of ketone bodies (ko00072) in bacteria (Fig. 4 and Supplementary Fig. S12). Five of the six enzymes involved in synthesis and degradation of ketone bodies significant higher than in feces, which contained 3-hydroxybutyrate dehydrogenase (EC: 1.1.1.30) (X² = 21.57, P < 0.0001), 3-oxoacid CoA-transferase (EC: 2.8.3.5) (X² = 21.82, P < 0.0001), hydroxymethylglutaryl-CoA lyase (EC: 4.1.3.4) (X² = 23.19, P < 0.0001), acetoacetate decarboxylase (EC: 4.1.1.4) (X² = 14.82, P < 0.01) and hydroxymethylglutaryl-CoA synthase (EC: 2.3.3.10) (X² = 16.37, P < 0.001) (Fig. 4 and Supplementary Fig. S13).

Thus, the panda’s anogenital gland is heavily populated with microbes rich in enzymes involved in biosynthesis of important chemosignals, especially those involving fatty acid degradation and the production of ketones and aldehydes. By comparison, these enzymes and metabolic pathways are relatively rare or absent in the microbiota of feces.

Source: Ecology - nature.com