Convergent evolution of metabolic functions: Evidence from the gut microbiomes of humans and dogs

Dogs have undergone convergent evolution in their genome with humans in long-term co-living. Studies have shown that the gut microbiomes of dogs exhibit greater similarity to those of humans compared to mice and pigs. In our study, we first compared the gut microbiomes of macaques, three dog groups, and two wolf groups to those in humans. Results showed that gut microbiomes in dogs have undergone convergent evolution with those in humans. Functional enrichment analysis revealed that dogs exhibit the strongest carbohydrate metabolism ability compared to other hosts including starch and sucrose metabolism. Working dogs and pet dogs showed distinct functional specialization in metabolism and human disease. We suggest that dogs could serve as a potential model for research related to human gut microbial responses to human health, disease, and environmental changes. Dogs and humans show convergent evolution in their genomes as a result of long-term co-living [1-3]. Gut microbiomes, crucial for digestion and metabolism, play a vital role in environmental adaptation and influence health in both animals and humans [4-8]. Studies have shown that the gut microbiomes of dogs exhibit greater similarity to those of humans compared to mice and pigs [9]. We hypothesize that dogs could serve as a potential model for research related to human gut microbial responses to human health, disease, and environmental changes. However, as the critical models for human biomedical research, nonhuman primates (NHPs) share close evolutionary relationships with humans, as well as physiological, neuroanatomical, and developmental similarities. Here, we firstly compared the gut microbiomes of macaques (NHPs), three dog groups, and two wolf groups to those in humans. Results showed that gut microbiomes in dogs have undergone convergent evolution with those in humans. Functional enrichment analysis revealed that dogs exhibit the strongest carbohydrate metabolism ability compared to other hosts, including starch and sucrose metabolism, which are the critical functions in dietary adaptation during dogs' domestication. Working dogs and pet dogs showed distinct functional specialization in metabolism and human disease. We suggest that dogs could serve as a potential model for research related to human gut microbial responses to human health, disease, and environmental changes. Our study included metagenomic seequences of fecal samples from six groups: pet dogs (n = 43), working dogs (n = 34), Chinese indigenous dogs (n = 39), wild wolves (n = 17), zoo wolves (n = 15), and macaques (n = 18) (Table S1). After a battery of bioinformatic processing for sequence data (Supplementary Methods), we got 21,289 assembled genomes of gut microorganisms. Then, we obtained a total of 6667 metagenome-assembled genomes (MAGs) after quality filtering (Table S2). MAGs were dereplicated at a 95% average nucleotide identity (ANI) threshold for species-level genomes and a 99% ANI threshold for strain-level genomes. In total, we obtained 1837 species-level genome bins (SGBs) and 2892 genome bins at the strain level, respectively. Among all groups, indigenous dogs have the most SGBs, followed by macaque, and zoo wolves exhibited the lowest detection rate of SGBs. We compared SGBs from six groups to 4744 well-organized human metagenomes [10]. Comparisons were conducted at both species and strain levels. Results showed that macaques have the highest overlap ratio (60.16%) with human metagenomes at the species level, followed by pet dogs (59.17%) and indigenous dogs (55.16%). Wild wolves have the lowest overlap ratio (13.64%) (Table S2). Zoo wolves showed a higher overlap ratio than wild wolves (p = 2.21E-8), close to working dogs (p = 1) but lower than pet dogs (p = 2.16E-4) and macaques (p = 3.18E-5) (all p values in Table S2). The high overlap ratio of gut microbiomes between macaques and humans is expected because of their closer genetic distance compared to canids lupus (Figure S1). While it is surprising that dogs have a very similar overlapping ratio with humans at the species level as macaque, especially pet dogs and indigenous dogs. At the strain level, we found that indigenous dogs have the highest overlap ratio with human metagenomes, at 13.08%, followed by pet dogs (12.12%) and working dogs (9.75%). Despite being the closest organism to humans in this study, macaques had a significantly lower overlap (5.32%) than dogs, even compared to working dogs (p = 7.8E-3), which means dogs have more similar microbiome strains to humans than macaques. Wild wolves still have the lowest overlap ratio with humans compared to other groups. The observed microbiome similarities between humans and dogs represent a unique evolutionary pattern between these two species, and suggesting convergent adaptation of gut microbial genomes. Notably, the striking divergence in microbial communities between dogs and their wild progenitors underscored the critical role of gut microbiomes in canine domestication and subsequent coevolution with humans. We identified the taxonomic information of the overlapped metagenomes at the strain level and summarized the families across six groups (Figure S2). Results indicate that dogs covered a broader range of families compared to other groups. Notably, dogs have MAGs overlapped with humans from Actinobacteriota, which are absent in macaques, contain the Bifidobacteriaceae, known as probiotics, and have a positive correlation with human health. We also compared the presence of Lachnospiraceae between macaques and dogs, a crucial bacterial family in human guts [11, 12], and found that pet dogs have a significantly higher level than macaques (p = 0.014). We would infer that dogs might be more similar to humans in gut microbiomes compared to macaques, which are inconsistent with the phylogeny distance among hosts, while consistent with studies that the environment makes more contributions to the shaping of gut microbiomes [13, 14]. To analyze the gut microbiomes' structure and composition of seven groups (including humans), we built a data set of 2686 SGBs from 245 samples. The estimator of species richness and Shannon index was calculated to assess the alpha diversity (Figure S3), and Bray–Curtis distance was calculated for the beta diversity (Figure S4). Results showed that macaques have the highest Species richness and Shannon diversity among groups, and humans exhibited the highest variation in samples. The principal coordinates analysis (PCoA) plot (Figure 1A) showed that gut microbial communities were distinguished by groups. Specifically, the microbial communities of two wolf groups are distinctly separated, just as the comparisons of overlapped SGBs. Three dog groups shared some similar microbial communities but exhibited unique patterns. The results also indicate that indigenous dogs appear to be closer to humans than macaques. Interestingly, zoo wolves seem to have a greater distance from wild wolves compared to working dogs. Together with the comparisons of overlapped SGBs (with humans) in the last section, we prompted interesting questions regarding the mechanisms underlying gut microbiome adaptation in response to changes in environmental conditions. It is plausible that zoo wolves, upon transitioning into captive environments, underwent both active and passive alterations in their gut microbiomes to adapt to novel dietary habits and ecological niches, which also could induce the decrease in the number of microorganisms (Table S2). However, further research is needed to elucidate the specific factors driving these observed patterns and their implications for host physiology and health. The UpSetR intersection diagram (Figure 1B) of SGBs unique to and shared among groups showed that macaques have the most unique SGBs (585 SGBs), followed by humans (432 SGBs) and other groups. Humans share the most SGBs with macaques (327 SGBs), followed by indigenous dogs (103 SGBs). We constructed a microbial dendrogram of bacterial communities within each group (Figure 1C, left). The dendrogram recealed that the microbial communities from humans are positioned in a clade closer to that of macaques, reflecting the pattern shown in the phylogenetic tree of humans and macaques. In contrast, the microbial communities of dogs and wolves were on a clade even they showed considerable divergence. These results are consistent with the phylogenetic relationships among the hosts, and it is reasonable that humans are much closer to macaques in the gut microbiome community than dogs since they are much closer in phylogeny (Figure S1), and the host genetics has strong effects on the gut microbiome [15, 16]. However, our analysis of overlapped SGBs revealed significant similarities in the gut metagenomes of humans and dogs. We expected to obtain related explanations from the functional annotation data. Thus, we conducted a matrix representing the abundance of KO numbers across various samples to build a dendrogram for functional enrichment information (Figure 1C, right). The resulting tree structure showed that humans and dogs are grouped in the same clade. The paired comparisons were conducted on the metabolism pathways among four host species. And results (Figure 2A) showed that dogs possess the strongest ability in most carbohydrate metabolism, particularly in starch and sucrose metabolism, which likely evolved to better adapt to human food and living conditions. Humans and macaques demonstrated weaker abilities in starch and sucrose metabolism, and wolves were the weakest among the four hosts. These results provide evidence for the convergent evolution of the functions of gut microbiomes in humans and dogs. Suggesting that dogs' gut microbes may have evolved stronger starch digestion capabilities than humans to compensate for their biological limitations, enabling them to adapt to new diets and survival environments. This is consistent with the convergent evolution between humans and dogs, as dogs have undergone positive selection for genes associated with starch digestion across their genome [1-3]. We also compared metabolism functions among three dog groups (Figure S5), and the results showed that pet dogs performed better in starch and sucrose metabolism, even though their foods are similar to those of working dogs. The comparisons of functional enrichment showed distinct functional specialization between working dogs and pet dogs (Figure 2B). Working dogs exhibited significant enrichment in microbial pathways involved in lipid metabolism and xenobiotic degradation. While pet dogs showed stronger microbial functions on the metabolism of amino acids, carbohydrates, and energy. Notably, the disease-associated pathways uncovered specific characteristics: Working dogs harbored signature pathways linked to human diabetic cardiomyopathy and Alzheimer's disease, while pet dogs demonstrated pathways associated with Helicobacter pylori infection. These results underscore the impact of environment and lifestyles. The enrichment of lipid metabolism and xenobiotic degradation pathways in working dogs aligns with their high-energy demands and environmental exposures, suggesting adaptive microbial provisioning of metabolic fuels and detoxification capacity. Notably, the functional similarities between working dog and human metabolic/neurodegenerative disease pathways may reflect conserved host-microbe interactions under physiological stress, and the PPAR signaling pathway is important to diabetic cardiomyopathy [17]. Pet dogs' microbial function enrichment strikingly overlaps with Helicobacter pylori-associated pathways, raising important questions about whether modern pet-keeping practices inadvertently cultivate microbial ecosystems resembling human dysbiotic states. This pattern mirrors observations in urbanized human populations, where reduced microbial diversity correlates with increased inflammatory conditions [18]. According to our results, we boldly speculate that domestic dogs may serve as potential models for studying human gut microbial responses to human disease related to metabolism and environmental changes. Our study demonstrated significant convergence in gut microbiome functionality between dogs and humans—particularly in starch metabolism pathways—the potential influence of contemporary diet requires careful consideration. The sampled dog populations exhibited distinct dietary regimes: pet and working dogs primarily consumed commercial dog food, while indigenous dogs relied on human food scraps with variable starch content. These differences in diets could contribute to observed microbial profiles. Nevertheless, we acknowledge that controlled feeding experiments would be invaluable to disentangle transient dietary effects from evolved adaptations. Future studies tracking microbiome dynamics during dietary switches could further resolve these mechanisms. We utilized three sequencing platforms (HiSeq, NovaSeq, DNBseq) to generate sequence data, which could theoretically introduce technical variability in our research. To mitigate the impact of platforms, we used standard quality control, adapter removal, and low-quality reads filtering strategies to reduce the influence on the bioinformatic analysis process. Three platforms were based on the PCR sequence principle, we all sequenced in 150 bp pair-end in our study, and existing research has demonstrated that these sequencing platforms show comparable performance in terms of genomic coverage and strain-level detection sensitivity in metagenome analysis [19]. Thus, we think the effect of different platforms could not influence the results of our study. However, although the effect of different platforms is tolerable, bridge PCR techniques may lead to detection errors for extremely low-abundance species. Future research could combine long-read sequencing technology to address the common error and bias issues encountered by short-read platforms. In our study, we only chose macaques as a representative of NHPs, which can represent some NHP groups but cannot represent all NHPs, such as apes. We used the eggNOG database [20] for the function annotation in this study, which has a broad coverage of prokaryotic and eukaryotic organisms. However, the resolution remains limited and includes overrepresentation of model organisms. It should be combined with pathway-centric databases for comprehensive functional interpretation if needed. Guo-Dong Wang, Zhigang Zhang, and Zhenxin Fan: Conceptualization; methodology; project administration; funding acquisition; supervision; writing—review and editing. Xiaoyang Wang: Investigation; data curation; formal analysis; writing—original draft; writing—review and editing; visualization. Lei Zhu: Investigation; data curation. Yue Lan: investigation; data curation. Chen Tian: Investigation; data curation. Guimei Li: Resources. Tong Zhou: Software. Qingguo Huang: Investigation. Tifei Yuan: Writing—review and editing. All authors have read the final manuscript and approved it for publication. We are grateful to Teng Li and Xin Cheng for the advice of data analysis, Yajiang Wu, Zhaoli Ding, and Guigang Zhao for the resource support. We also thank the Genome Center of Biodiversity, Kunming Institute of Zoology, Chinese Academy of Sciences (Kunming 650201, China) for the resource support. This study was supported by grants from the STI2030-Major Projects (2021ZD0203900), the National Natural Science Foundation of China (32170436, 32360122, U2002206), Spring City Plan: the High-level Talent Promotion and Training Project of Kunming (2022SCP001), the Second Tibetan Plateau Scientific Expedition and Research (STEP) program (2021QZKK0103, 2019QZKK0503), the Biological Resources Program, Chinese Academy of Sciences (KFJ-BRP-004), and the Central Guidance on Local Science and Technology Development Fund of Yunnan Province (202407AA110009). Guo-Dong Wang was supported by Yunnan Fundamental Research Projects (202201AV070011). Zhenxin Fan was supported by the Sichuan Science and Technology Program (2023NSFSC1935). Zhigang Zhang was supported by the Yunnan Province Xing Dian Talent Support Program (Yunling Scholar) (Z.G.Z.). We apologize for not being able to cite additional work owing to space limitations. The authors declare no conflicts of interest. No animals or humans were involved in this study. The datasets generated and analyzed during the current study were available at GSA (Genome Sequence Archive) with accession Number CRA018764 (https://ngdc.cncb.ac.cn/gsa/browse/CRA018764). Codes and datasets are available at GitHub: https://github.com/WangXy1230/wangIMO-0186. Supplementary materials (methods, figures, tables, graphical abstract, slides, videos, Chinese translated version, and update materials) may be found in the online DOI or iMetaOmics http://www.imeta.science/imetaomics/. imo270059-sup-0001-IMO-2025-0186_Supp_V9.docx Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.