Gut bacterial regulation of primate adaptive thermogenesis at high altitude

Abstract

Background

Adaptive thermogenesis is a fundamental physiological process by which mammals maintain their body temperature through shivering and non-shivering thermogenesis to adapt to environmental changes. The activation of brown adipose tissue (BAT) is the core of non-shivering thermogenesis. However, the specific mechanisms by which gut-derived bacteria trigger BAT to activate adaptive thermogenesis remain poorly understood. Environmental variations across different altitudes create unique temperature gradients, providing an ideal condition for identifying bacteria associated with cold adaptation. In this study, we identified Pantoea ananatis, a highland-enriched bacterium, as a key species regulating adaptive thermogenesis and lipid metabolism in Macaca mulatta.

Results

Through multi-omics and gavage experiments, we discovered that the gut microbiota of high-altitude macaques can enhance the nutritional absorption capacity of the small intestine of mice, increase the concentration of propionic acid, activate the glycerolipid metabolism and strengthen lipid metabolism. Furthermore, we found that P. ananatis, as one of the main effect bacteria of the high-altitude gut microbiota, can activate BAT, reduce white adipose tissue (WAT) storage, and enhance triglyceride metabolism. Finally, we preliminarily verified that ferulic acid, as one of the potential effector metabolites of P. ananatis, also contributes to the reduction of WAT accumulation.

Conclusions

Our work uncovers P. ananatis as a high-altitude-adapted potential probiotic that activates BAT and promotes systemic fat reduction through a gut microbiota-driven mechanism. This breakthrough provides a safe, effective alternative to cold-induced thermogenesis, with profound implications for obesity intervention.

Download video file ^{(50MB, mp4)}

Video Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s40168-026-02425-6.

Background

Adaptive thermogenesis refers to the heat generated by the body through shivering and non-shivering thermogenesis in response to environmental changes [1–4]. This process is crucial not only for mammals' adaptation to their surroundings but also for counteracting the excessive caloric intake associated with obesity [5–7]. There is evidence across multiple mammals that brown adipose tissue (BAT) is an important organ for adaptive thermogenesis and usually activated by cold/non-cold stimuli to consume fat and produce heat [5, 8]. Consequently, BAT is considered not only valuable for research on adaptation but also a vital target for obesity treatment. The activation of adaptive thermogenesis by cold stimuli often imposes a burden on the cardiovascular and cerebrovascular systems [9]. However, the mechanisms underlying adaptive thermogenesis in response to non-cold stimuli remain poorly understood.

Increasingly, gut microbiota is recognized for its role in regulating energy metabolism in mammals' environmental adaptation [10, 11]. Recent studies indicate that cold exposure alters the gut microbiota's structure, which subsequently activates BAT and promotes adaptive thermogenesis by influencing bile acid metabolism [10, 12]. In activated BAT, dietary glucose and triglycerides serve as fuel to meet the heightened energy demands during thermogenesis [13, 14]. However, it remains unclear whether gut microbiota contributes to the non-cold activation of BAT, as short-term cold exposure does not provide sufficient insight to address this question. The answer to this question will help expand our understanding of the mechanisms of cold adaptation in mammals and provide more diverse ways for weight control and obesity treatment.

Here, we utilized rhesus macaque (Macaca mulatta) at varying altitudes to establish a temperature gradient and conducted fecal bacterial transplantation experiments to mitigate differences arising from dietary and environmental factors. This approach allowed us to examine the response of the gut microbiota to cold temperatures, specifically to investigate whether it exerts an activating effect on BAT and to explore its underlying mechanisms. We collected fecal samples from wild rhesus macaque populations living at altitudes from 50 to above 4,000 m with average annual temperatures of 22.5℃ and 5.8℃, respectively (Resource and Environmental Science Data Platform; Fig. 1). Consequently, this species represents an excellent model to investigate cold adaptation mechanisms of gut microbiota [15]. Our findings suggest that, at room temperature, the high-altitude gut microbiota significantly decreases white fat, as well as serum total bile acid and triglyceride levels. Metagenomic analysis identified the primary lipid metabolic pathways associated with high-altitude gut microbiota, highlighting Pantoea ananatis as a key bacterial species. Additionally, we established a high-fat diet model using isogenic male mice, which demonstrated that P. ananatis could activate the UCP1 protein, thereby enhancing BAT thermogenesis, reducing white adipose tissue (WAT) content, and preventing obesity. Finally, non-targeted lipid metabolism analysis, based on the P. ananatis gavage experiment, identified metabolites potentially involved in the activation of BAT by gut microbiota, and revealed that ferulic acid may have the potential to activate BAT.

Fig. 1.

Pipeline to explore the adaptation mechanism of gut microbiota to high altitude, by combining the M. mulatta integrated gene catalog, metagenome-assembled genomes (MAGs) and validation experiments. Metagenomic sequencing data from the samples spanning seven altitudes from 87 M. mulatta metagenome data were integrated and used to construct a gene catalog and MAGs, including 22,803,526 non-redundant genes and 1,157 MAGs. In addition, we transplanted the feces from the highest and lowest altitudes into mice, and sequenced and measured their metagenomes, metabolome and various physiological indicators to observe effects of high-altitude gut microbiota on metabolism. Based on comparative metagenomic analyses and results of fecal microbiota transplantation (FMT) experiments, Pantoea ananatis was selected as a key bacterium for high-altitude adaptation, and it was administrated into high-fat diet mice to verify its ability to affect lipid metabolism. The final experiment verified the fat-reducing effect of ferulic acid, a metabolite of the P. ananatis

Results

Generation and quality assessment of metagenome-assembled genomes (MAGs) in different-altitude M. mulatta populations

To comprehensively explore the characteristics and functions of high-altitude gut bacteria [16–19], we collected metagenomic data from the gut microbiota of seven wild populations of M. mulatta (Supplementary Table 1). A total of 1,157 metagenome-assembled genomes (MAGs) were obtained at both the strain and species levels, with a completeness of ≥ 80% and contamination of ≤ 10%. This set represents a non-redundant collection of gut microbiota from M. mulatta (Fig. 2A; Supplementary Table 2). Subsequently, 725 MAGs were identified as of high-quality, meeting the criteria of completeness ≥ 90% and contamination ≤ 5% (Fig. 2B). The size of the MAGs is primarily distributed around 2.3 Mb, with a median contig number of 115 and N50 ranging from 2.1 kb to 837 kb (Fig. 2C-E). However, 77.4% of the MAGs remain unclassified, as they did not match any reference genomes in the Genome Taxonomy Database (GTDB) (Fig. 2A; Supplementary Table 2). Additionally, we constructed a gene catalog for M. mulatta, comprising 22,803,526 non-redundant genes (Fig. 2F, G).

Fig. 2.

Taxonomic characteristics and functional annotation of MAGs in different M. mulatta populations. A Circular cladogram representation of the phylogenetic relationships and taxonomic classification of the 1,157 strain-level MAGs from the seven M. mulatta populations. B Quality assessment (i.e. completeness and contamination statistics) of the 1,157 non-redundant MAGs. Each point represents a single MAG. Orange dots indicate high-quality genomes with ≥ 90% completeness and ≤ 5% contamination. All other MAGs have > 80% completeness and ≤ 10% contamination. C, D and E show the distribution of genome sizes, the number of contigs per genome, and N50 values for the 1,157 MAGs, respectively. F Number (percentages) of shared bacterial taxa among different proportions of samples at the phylum (blue), genus (green), and species (red) level. The percentage of shared items and the proportion of shared samples are represented on the y- and x-axis, respectively. Also indicated are number and percentage of each item shared by 20, 50, 90, and 100% of the samples. G Numbers (percentages) of shared function items among different proportions of samples for NR gene (blue), CAZy family (green), and KEGG orthologues (red). Other legends as in (F). A Alternatively, each panel or group of panels can be described separately

Then, we identified the key bacteria and metabolic pathways involved in high-altitude adaptation. A total of 13 MAGs were significantly more abundant in higher altitude populations (Supplementary Fig. 1A-I; p < 0.05). These bacteria belong to the UBA932, Acutalibacteraceae, Lachnospiraceae, CAG-272, CAG826, and Oscillospiraceae families, as well as the Gammaproteobacteria class. Among these 13 bacteria, two are known species-level groups (SGBs), while 11 are unknown SGBs, specifically UBA6857 sp900555805 and P. ananatis (Supplementary Fig. 1A). However, most of the metabolic pathways (116 out of 128) were more enriched in high altitude and therefore need further screening (Supplementary Fig. 2; p < 0.05).

High-altitude gut microbiota altered lipid metabolism

To investigate the role of gut microbiota in mediating high-altitude adaptation in M. mulatta, we conducted fecal microbiota transplants (FMT) from M. mulatta populations living at higher and lower elevations, as well as from mice, into bacteria-restricted mice (Fig. 3A). Our analysis revealed no significant differences in body weight, daily food intake, or organ weight between high-FMT and low-FMT mice (Supplementary Fig. 3A-F). However, we observed significant differences in the percentage of body weight gain (Fig. 3C; p = 0.007; Supplementary Fig. 4), calculated as (final body weight—initial body weight)/initial body weight, while WAT weight significantly decreased (Fig. 3D; p = 0.02) and BAT weight significantly increased (Fig. 3E; p = 0.02). These results suggest that high-altitude gut microbiota may play a role in altering lipid metabolism.

Fig. 3.

High-altitude feces affect body weight and short-chain fatty acids (SCFAs) metabolism. A Scheme of the experimental design for fecal microbiota transplantation (FMT) of high-altitude feces affecting lipid metabolism in C57BL/6 J mice. Twenty-six male mice were randomly divided into three groups: high-altitude M. mulatta FMT (high-FMT, n = 9), low-altitude M. mulatta FMT (low-FMT, n = 9), and Self-FMT Control (n = 8), and pairwise comparisons between groups were performed using the Wilcoxon rank sum test. B-E Changes observed after high-altitude FMT (n > 8 per group); (B) body weight change over time, (C) increase of body weight, (D) weight of WAT, and (E) and BAT. F Transverse section of the small intestine, as well as (G) weight, (H) crypt depth, and (I) villus length of small intestines. J-N Changes in SCFAs after high-altitude and low-altitude FMT

Given the substantial literature linking high-altitude adaptation to short-chain fatty acids (SCFAs) levels16-18 and that SCFAs have been reported to affect physiological indicators of the small intestine,19 we first examined small intestinal sections and then tested the SCFAs levels of the colon contents. Initially, we observed no significant differences in the weight and length of small intestinal villi (Fig. 3F, G, I); however, the crypts of high-FMT mice were deeper than those of low-FMT mice (Fig. 3H). Furthermore, an analysis of the SCFAs-targeted metabolome of the colon contents revealed that the concentrations of propionic acid were higher in high-FMT mice compared to low-FMT mice (Fig. 3J-N). These findings suggest that altered gut microbiota can influence SCFA function and lead to changes in small intestinal morphology.

Next, we investigated the impact of high-altitude gut microbiota on triglyceride (TG) metabolism, given the differences in WAT and BAT weights. Serum analyses revealed that the levels of total bile acids (Fig. 4A; p = 0.007) and TG (Fig. 4G; p = 0.003) were significantly lower in high-FMT mice compared to low-FMT mice, however other indicators did not differ (Fig. 4B-F). Physiological data guided our investigation into the lipid metabolome of mice colon contents and liver. A comprehensive lipomics analysis revealed 3,805 lipid species across six classes (Supplementary Fig. 5A), including 41 lipid sub-classes in the colon contents (Supplementary Fig. 6). The use of orthogonal projections to latent structures—discriminant analysis (OPLS-DA) facilitated the assessment of lipid profile changes among high-FMT, low-FMT, and control groups (Supplementary Fig. 5B; p < 0.01). Notably, the top-ranked enriched sub-classes in the colon contents lipid composition included TG, ceramides (Cer), glucosylsphingosine (Her1Cer), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), and diglyceride (DG). High-FMT mice exhibited significant reductions in TG and Cer lipid species, while showing increased levels of DG, PA, PC, PE, PI, and Her1Cer lipid species (Fig. 4H; Supplementary Table 3A; |log2FC|> 1 and p < 0.05). Results of lipid-targeted metabolome analyses of the liver are generally consistent with this finding, that is, TG is enriched at low altitudes, while DG is enriched at high altitudes (Fig. 4I; Supplementary Table 3B). Combined with the results of the serum analysis, the lipidomic data further emphasized the influence of high-altitude gut microbiota on lipid metabolism, particularly regarding on TG levels.

Fig. 4.

High-altitude feces affect lipid metabolism and particularly reduces triglyceride (TG) levels. A-G Effects of high-altitude feces on serum indexes. (A) Total bile acids, (B) cholesterol, (C) LDL-C, (D) HDL-C, (E) glucose in serum, (F) insulin, and (G) TGs levels in three groups. Significantly different non-target lipid metabolites of (H) colon contents and (I) liver between high- and low-altitude FMT mice. High-altitude feces reduce significantly the level of TG in serum (p < 0.05) and colon content (|log2FC|> 1 and p < 0.05) in FMT mice. J Glycerolipid metabolism pathway, involved in TG, DG, PG, PE, PC metabolism, is significantly enriched in higher altitude M. mulatta groups and high-FMT group

Identifying key bacterium in response to lipid metabolism

Metagenomic data from the colon contents of FMT mice was compared to investigate the potential bacteria and metabolic pathways influencing lipid metabolism. Differential KEGG metabolic pathway analyses revealed that a majority of pathways (101 out of 123) are more abundant in high-FMT mice (p < 0.05). Moreover, 45 metabolic pathways showed significant abundance in both high-FMT mice and high-altitude M. mulatta populations (Supplementary Fig. 7A). Among them, the glycerolipid metabolism (map00561) and glycerolipid metabolism (map00100) pathways, involved in regulating lipid metabolism, are more abundant in both high-altitude M. mulatta and high-FMT mice (Supplementary Fig. 7B, C). Lipids such as TG, DG, PC, PE, PS, and PI are significantly enriched in high-FMT mice and involved in the glycerolipid metabolism pathway (Fig. 4J), as well as the genes involved in DG synthesis, and TG degradation (Fig. 4J; Supplementary Fig. 8). These results suggest that the gut microbiota may play a role in regulating lipid metabolism in mice through the glycerolipid metabolism pathway. This pathway primarily focuses on the conversion cycle of glycerolipids (GLs), specifically TG, DG, and FA, which consumed ATP and releases heat [20, 21]. This pathway is believed to be crucial in maintaining body temperature [22–25], so we propose that the enrichment of this pathway may be related to cold adaptation in high-altitude environments.

Thirty-eight MAGs are significantly abundant in the colon contents of high-FMT mice compared with low-FMT mice (Supplementary Fig. 9; p < 0.05). Among them, 28 MAGs are unknown SGBs and 10 are SGBs. We screened for key bacteria related to high-altitude adaptation and found seven bacterial families/classes (Acutalibacteraceae, CAG-272, CAG-826, Lachnospiracea, Oscillospiraceae, UBA932, and Gammaproteobacteria) that were significantly enriched in high-altitude M. mulatta (above 4000 m) and high-FMT mice (Supplementary Fig. 10,11). The abundance of eight MAGs belonging to these families/classes increases as the altitude rises in M. mulatta populations (Supplementary Fig. 1B-I) and 19 MAGs were significantly more abundant in high-FMT mice (Supplementary Fig. 12). MAGs belonging to Gammaproteobacteria, Acutalibacteraceae, Lachnospiraceae, and Oscillospiraceae are significantly correlated to lipids (DG, PC, PE, PI, and Hex1Cer) and were significantly abundant in high-FMT mice (Supplementary Fig. 13). Among these, MAGs belonging to c_Gammaproteobacteria (p(Bins-0630) = 0.009; p(Bins-0663) = 0.019) are most significantly correlated with altitude (Supplementary Fig. 1F; Supplementary Fig. 11E), and the genome of Bins-0663 (P. ananatis) contains a broad range of genes involved in the glycerolipid metabolism (Supplementary Fig. 14). These genes are widely involved in glycerolipid metabolism and are mainly related to the synthesis of lysophosphatidic acid (LPA) including the conversion of DG (Supplementary Fig. 15). These results suggest that the identified MAGs, particularly P. ananatis, may affect lipid metabolism.

Pantoea ananatis can effectively prevent obesity

To verify the role of P. ananatis in lipid metabolism, we administered P. ananatis to mice on a high-fat diet (HF-B) (Fig. 5A), and observed that P. ananatis leads to significantly reduced body weight (Fig. 5B, C; p < 0.05), BAT weight (Fig. 5D; p < 0.05), and WAT weight (Fig. 5E-G; p < 0.01). To investigate the effects of P. ananatis on lipid metabolism, we first assessed the levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bile acids (CHOL) and TG in the liver. Our findings suggest that P. ananatis significantly reduced liver TG levels without significant affecting other indicators (Fig. 5H-K). This suggests that P. ananatis has a notable lipid-lowering effect, although it does not lower CHOL levels. Analyses of BAT slices revealed that the volume of mice orally administered P. ananatis was more comparable to that of the normal diet group (ND-Con) than to the high-fat diet group (HF-Con) without administered (Fig. 5L). Furthermore, the serum levels of tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β), interleukin-6 (IL-6), and C-reactive protein (CRP) did not show significant changes following oral gavage with Pantoea ananatis (Supplementary Fig. 16), suggesting that this intervention did not induce a systemic inflammatory response. Additionally, western blot analysis demonstrated that UCP1 protein expression in the BAT of HF-B mice was significantly higher than that in the HF-con group (p < 0.05; Fig. 5M, N). To further investigate this, we performed immunofluorescence against UCP1 on BAT sections and found that the expression of UCP1 was more pronounced in BAT of HF-B mice (Supplementary Fig. 18; Supplementary Fig.  20A). These findings indicate that P. ananatis can promote UCP1 protein expression in BAT. Interestingly, the transcriptional levels of lipoprotein lipase (LPL)—which breaks down triglycerides (TG) and releases free fatty acids (FFA) to serve as fuel for thermogenesis—in brown adipose tissue (BAT) and white adipose tissue (WAT) did not exhibit significant changes following oral gavage with P. ananatis (Supplementary Fig. 17), indicating that this bacterium did not activate this thermogenic pathway.

Fig. 5.

Pantoea ananatis reduces TG levels and activates BAT. A Scheme of the experimental design for oral gavage of P. ananatis affecting lipid metabolism in C57BL/6 J mice. Twenty-four mice were randomly assigned to three groups: normal diet (ND-Con, n = 8), high-fat diet (HF-Con, n = 8), and high-fat diet + P. ananatis (HF-B, n = 8), and pairwise comparisons between groups were performed using the Wilcoxon rank sum test. B-G Changes observed after oral gavage of P. ananatis: (B) body weight change over time, (C) body weight, (D) BAT weight, (E) subcutaneous fat, (F) peritestis fat, and (G) abdominal fat. H–K Levels of alanine aminotransferase (ALT), (h), aspartate aminotransferase (AST), (i), total bile acids (CHOL), (j) and triglyceride (TG), (k) in the liver. I Paraffin sections of BAT tissue. M P. ananatis gavage increased significantly UCP1 expression in BAT of HFD mice. N Quantitative analysis of UCP1 protein expression in BAT after P. ananatis gavage

Pantoea ananatis produces different metabolites

To investigate the potential metabolites produced by P. ananatis that may activate BAT, we compared the colon contents non-target metabolome of PT-Con and HF-Con, as well as HF-B and HF-Con. A total of 132 metabolites were significantly enriched in the ND-Con compared with HF-Con (Supplementary Table 4; |fold2change|> 2; p < 0.005). Additionally, 72 metabolites were significantly enriched in the HF-B group relative to HF-Con (Supplementary Table 4; |fold2change|> 2; p < 0.005). Notably, there were 12 overlapping metabolites between the two groups, including 1-methylguanosine, 2'-deoxyadenosine, coumarin, ferulic acid, adenine, D-(+)-maltose, and isoferulic acid (Fig. 6A, B). Metabolomic data showed that after oral gavage of P. ananatis, the gut microbiota of HF-B mice tended to have metabolites similar to those of the gut microbiota of mice on a normal diet, because many of their substances were significantly enriched, such as ferulic acid. Among them, ferulic acid has been shown to play a role in weight control [26].

Fig. 6.

Ferulic acid may be an important product of BAT activation by Pantoea ananatis. A, B Comparative analysis of the non-targeted metabolome of colon contents. A Volcanic diagram of differential metabolites between normal diet (ND-Con) and high-fat diet (HF-Con). B Volcanic diagram of differential metabolites of P. ananatis gavage (HF-B) and high-fat diet (HF-Con). C Scheme of the experimental design for oral gavage of ferulic acid affecting lipid metabolism in C57BL/6 J mice. Thirty mice were randomly assigned to three groups: high-fat diet (HF-Con, n = 10), high-fat diet + 0.25% ferulic acid (HF-LFA, n = 10), and high-fat diet + 0.5% ferulic acid (HF-HFA, n = 10), and pairwise comparisons between groups were performed using the Wilcoxon rank sum test. D-I Changes observed after oral gavage of ferulic acid: (D) body weight change over time, (E) body weight (F), BAT weight, (G) subcutaneous fat (H), peritestis fat, and (I) abdominal fat. J Paraffin sections of BAT tissue. K Ferulic acid gavage increased significantly UCP1 expression in BAT of HFD mice. L Quantitative analysis of UCP1 protein expression in BAT after ferulic acid treatment

We conducted feeding experiments on mice using 0.25% (HF-LFA) and 0.5% (HF-HFA) ferulic acid under high-fat conditions, establishing a high-fat diet control group without additives (HFD) (Fig. 6C). Ferulic acid was observed to increase body weight (Fig. 6D and E; p = 0.01), while the weight of BAT (Fig. 6F; p < 0.05) and WAT (Fig. 6G–I; p < 0.01) was significantly reduced. Analysis of BAT sections and western blot results revealed that HF-HFA group exhibited a tendency toward reduced BAT adipocyte volume (Fig. 6J), with significantly higher UCP1 protein expression in BAT compared to HFD group (p < 0.05; Fig. 6K, L). These observations indicate that ferulic acid facilitates the stimulation of UCP1 protein expression in BAT [27]. To further investigate this, we performed immunofluorescence against UCP1 on BAT sections and found that the expression of UCP1 was more pronounced in BAT of mice fed ferulic acid compared to those not receiving it (Supplementary Fig. 19, Supplementary Fig. 20 B). Additionally, we observed that UCP1 expression was more active at the edge of BAT compared to the center.

P. ananatis and ferulic acid promote thermogenesis for body temperature maintenance in cold environments

To investigate whether P. ananatis and FA stimulate thermogenesis under cold stress, mice were administered short-term oral gavage while maintained on a normal diet (ND) (Fig. 7A). On the fifth day of gavage, the core body temperatures of mice in the ND + P. ananatis and ND + FA groups were slightly higher than those in the ND + PBS control group (Supplementary Fig. 21A, B). Following acute cold exposure at 4 °C, the core body temperatures of the ND + P. ananatis and ND + FA groups were significantly elevated compared to the ND + PBS group at both 2 h and 4 h time points (Fig. 7B-D). Furthermore, RT-qPCR analysis of key thermogenic genes in brown adipose tissue (BAT), including Ucp1, Pgc1a, Prdm16, and Cidea, revealed an upward trend in the expression levels of the ND + P. ananatis and ND + FA groups. However, these differences did not reach statistical significance, potentially due to the relatively short duration of the gavage period (Supplementary Fig. 21C-F). Taken together, these results demonstrate that both P. ananatis and ferulic acid enhance thermogenesis and contribute to the maintenance of body temperature under cold environments.

Fig. 7.

Pantoea ananatis and ferulic acid can effectively activate BAT to maintain body temperature and consume fat. a Schematic diagram of the oral gavage and cold exposure experiment in C57BL/6 J mice with P. ananatis and ferulic acid. Thirty mice were randomly assigned to three groups: normal diet + PBS (n = 10), ND + P. ananatis (n = 10), and ND + FA (n = 10), and pairwise comparisons between groups were performed using the Wilcoxon rank sum test. (b) and (c) Difference plot of core body temperature in mice at 2 h and 4 h after 44 °C cold exposure. d Temperature differences of three groups of mice under the view of an infrared imager after 4 h of cold exposure

Discussion

Our understanding of the detailed functions of gut bacterium in activating adaptive thermogenesis remains limited. To fill this gap, we conducted a multifaceted study that integrates binning, FMT, single-bacteria gavage experiments and multi-omics analysis. This comprehensive approach enabled us to identify and validate key bacterium associated with high-altitude adaptation and to elucidate their functional mechanisms. In this study, we assembled 1,157 bacterial genomes and constructed a gene catalogue comprising 22,803,526 non-redundant genes from 82 samples across 7 rhesus populations at varying altitudes. This extensive dataset allowed us to identify the key bacterium and metabolic pathways involved in high-altitude adaptation. Our findings indicate that high-altitude fecal microbiota significantly affects lipid metabolism and may mediate adaptive thermogenesis through the glycerolipid metabolism pathway. Additionally, we identified P. ananatis as the key bacterium, that can effectively activate BAT and promote the conversion of WAT by stimulating UCP1. This activation subsequently enhances thermogenesis and facilitates the maintenance of core body temperature in mice during cold exposure. This study uncovers a novel gut microbiota-mediated BAT activation mechanism in wild rhesus monkeys, thereby expanding our understanding of the mechanisms underlying cold adaptation and providing new insights for obesity treatment.

The plateau environment above 4000 m presents significant challenges for animal survival. Factors such as persistently low temperatures (−5.1 °C), reduced oxygen availability, and limited food resources complicate energy metabolism in these animals. Previous studies on gut microbiota across various species have demonstrated that altitude can alter the composition of gut microbiota [26, 28], primarily enhancing the production capacity of SCFAs, particularly propionic acid and butyric acid [29–32]. SCFAs not only facilitate plateau adaptation from an energy metabolism perspective, but they have also been implicated in blood pressure regulation [33–37]. Our results indicate that gut microbiota extends beyond SCFA metabolism, utilizing an alternative glycerolipid metabolism pathway, alongside reduced triglyceride levels, total bile acid concentrations, and WAT storage. This pathway is closely associated with ATP consumption and thermogenesis [20, 21], which are critical for maintaining body temperature in cold environments [22–25]. This metabolic shift may represent a key adaptive mechanism to high-altitude cold stress.

The role of gut microbiota regulated by cold exposure in thermogenesis is increasingly recognized [10]. Among these, the gut microbiota has received particular attention for its role in enhancing UCP1-dependent thermogenesis, as it simultaneously exhibits both cold resistance and anti-obesity capabilities [11, 28–31]. For example, Claire Chevalier et al. reported that cold exposure induces WAT browning [10], while other studies have shown that bile acid and SCFA-related gene expression increases in the myocardium of piglets under cold conditions [38]. Additionally, in Buchner’s field, disrupted microbiota and norepinephrine signaling synchronize via cyclic AMP to regulate energy expenditure and thermogenesis [39]. However, it remains unknown which bacteria play key roles in this regulatory process. This uncertainty undoubtedly hinders our understanding and application of the thermogenic mechanisms regulated by the gut microbiota. Short-term (5-day) oral gavage of P. ananatis and its key metabolite, ferulic acid, effectively facilitated core body temperature maintenance in mice under cold stress, while long-term (8-week) administration significantly reduced adipose accumulation. These findings elucidate a novel cold-adaptive bacterium and its functional effector in wild rhesus monkeys that trigger thermogenesis, highlighting their significant therapeutic potential for obesity intervention.

P. ananatis has often been reported as a plant pathogenic bacterium [33, 34], but it is also widely present in the gut microbiota of animals [35–37]. We have for the first time discovered the activating effect of this bacterium on BAT, and verified the similar effect of its related product, ferulic acid. It has been reported that certain bacteria can influence metabolism and promote weight control, such as Akkermansia muciniphila, which primarily enhances glucose uptake and metabolism by improving insulin sensitivity [40–43]. However, studies in both humans and mice have found that while A. muciniphila can moderately reduce lipid levels or waist circumference, the effects are not statistically significant [40, 41, 43]. In contrast, our data indicate that P. ananatis specifically activates the UCP1 protein, thereby initiating the thermogenic process in BAT5 and significantly reducing TG levels [44], particularly in relation to obesity, without affecting blood glucose levels. Therefore, further investigation into these processes is warranted, as P. ananatis may be a more promising candidate for use as a probiotic specifically aimed at fat reduction compared to A. muciniphila.

Combining multi-omics analysis with experimental validation, this study systematically revealed the cold adaptation mechanism of gut microbiota in high-altitude M. mulatta. This mechanism primarily operates through three pathways to help the organism adapt to the cold high-altitude environment: enhancing the nutrient absorption capacity of the small intestine, increasing the concentration of propionic acid, and activating thermogenic pathways while strengthening lipid metabolism. On this basis, Furthermore, we identified Pantoea ananatis as the key microorganism mediating high-altitude adaptation, and found that this strain can activate BAT, reduce WAT storage. Meanwhile, we preliminarily verified the regulatory function of ferulic acid as a key effector metabolite of P. ananatis. This study clarifies a novel adaptation strategy of wild animals in response to high-altitude environments, which provides a research basis with potential application value for fields such as obesity intervention.

Limitations of the study

Our results are based on binning analyses, which may not capture all bacteria relevant to high-altitude adaptation, possibly due to potential loss during FMT or because the mice were kept in a non-high-altitude environment. Therefore, our findings represent only a preliminary attempt using a comprehensive bacterial mining approach. Additionally, during the analysis, we observed that genes associated with replication and repair are often enriched in high-altitude populations, a trend that has also been supported by our comparative analysis of gut microbiota. While this enrichment may be related to altitude adaptation, the specific causes and mechanisms remain unclear. Further investigation of this phenomenon could significantly enhance our understanding of human adaptation to high altitudes. Finally, while our experiments have initially linked P. ananatis (and its associated ferulic acid) to BAT activation and fat reduction, the exact relationship between these three requires further validation. Notably, ferulic acid induced less pronounced UCP1 protein expression than P. ananatis itself. We hypothesize that P. ananatis produces key effector metabolites beyond ferulic acid, highlighting the need for more extensive investigations into its fat-reducing mechanisms. In addition, although we measured a series of indicators including inflammatory factors, which demonstrated that P. ananatis did not induce an inflammatory response, further safety assessments are still required for validation. And the fat-reducing effect of P. ananatis also requires further experimental validation in obese mice. Furthermore, it is crucial to investigate whether key effectors such as ferulic acid can directly activate BAT and explore their underlying mechanisms of action.

Materials and methods

Animals

Two-month-old specific-pathogen-free (SPF) male C57BL/6 J mice (SPF Biotechnology Co., Ltd. and Beijing Vital River Laboratory Animal Technology Co., Ltd) were housed in the animal facility of the Institute of Zoology, CAS (CNASLA0014). Prior to any treatment, all animals were given a 7-day acclimation period. The mice were housed in a room with temperatures maintained at 23 ± 0.5℃ and a photoperiod of 14 h of light followed by 10 h of darkness (with lights were turned on at 06:00) and humidity (30%–70%). During the experiment, mice had ad libitum access to food and autoclaved water. Two diets were utilized in the experiment: a standard diet (containing 18% protein, 3% fat, 12% fiber, and 47% carbohydrate, Beijing KeAo Bioscience Co.) and a high fat diet (60% of calories derived from fat, Research Diets, New Brunswick, NJ; D12492). The animals were euthanized at the end of the experiment and blood and organs were collected for further testing and analysis. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Institute of Zoology, Chinese Academy of Sciences (IOZ, CAS, reference number IOZ-IACUC-2024–137).

Experimental design

Experiment 1: Fecal samples were collected from 82 individuals across seven populations of M. mulatta, spanning an altitude range from 50 to 4,317 m asl. Binning technology based on metagenomic shotgun sequencing was used to discover new bacterial species, to assemble the genome of major species, and we also generated a gene catalog of gut microbiota.

Experiment 2: To investigate the role of high-altitude gut microbiota and metabolites, we transplanted fecal microbiota from high- and low-altitude M. mulatta into mice. A total of twenty-six male mice were randomly assigned to three groups: recipient of high-altitude M. mulatta fecal bacteria group (high-FMT, n = 9), recipient of low-altitude M. mulatta fecal bacteria group (low-FMT, n = 9), and control group (Control, recipient of self-group fecal bacteria, n = 8). All mice were oral gavage with fresh composite antibiotics (200 µL/day) for 4 days, consisting of 100 µg/mL neomycin, 50 µg/mL streptomycin, and 100 U/mL penicillin (Sigma, Germany). During the experimental period, all mice were administered 200 µL of fecal suspension daily, with the study conducted over a four-week period. Throughout the experiment, all mice were maintained on a standard diet.

Experiment 3: To investigate the role of P. ananatis, which is significantly more abundant in high-altitude populations of M. mulatta, in lipid metabolism, we conducted a single-bacteria oral gavage experiment based on high-fat diet (HFD). P. ananatis was cultured aerobically at 30℃ on BL agarplates. The bacterial cells were washed with sterile PBS on the plate, then the bacterial solution was diluted to 10⁹ colony-forming units (CFU) [45]. Twenty-four male C57BL/6N mice were randomly divided into three groups: a normal diet control group (ND-Con, n = 8), high-fat diet control group (HF-Con, n = 8), high-fat diet and P. ananatis group (HF-B, n = 8, P. ananatis 10⁹ CFU). P. ananatis was prepared fresh daily and administered orally once per day. Body mass and food intake were measured weekly over the nine-week period of the study.

Experiment 4: To investigate the role of ferulic acid in lipid metabolism, which was significantly enriched in the intestinal tract of mice in the P. ananatis (HF-B) group compared to the normal diet group (ND-Con), we conducted a study focusing on lipid metabolism under a high-fat diet. The feeding experiment involved the administration of 0.25% and 0.5% ferulic acid in the high-fat feed [26], thereby creating low-dose and high-dose ferulic acid feeds, respectively. Thirty male C57BL/6N mice were randomly assigned to three groups: the high-fat diet control group (HF-Con, n = 10), the 0.25% ferulic acid high-fat diet group (HF-LFA, n = 10), and the 0.5% ferulic acid high-fat diet group (HF-HFA, n = 10). Throughout the six-week study period, the high-fat feed was stored at −20℃ and fed to the mice at room temperature daily, while monitoring food intake and mouse body weight.

Experiment 5: To evaluate whether oral administration of P. ananatis and ferulic acid (FA) effectively triggers brown adipose tissue (BAT) thermogenesis, we conducted gavage trials under a normal diet (ND). Thirty male C57BL/6N mice were randomly assigned to three experimental groups (n = 10 per group): the control group receiving PBS (ND-PBS), the P. ananatis group receiving 10⁹ colony-forming units (CFU) per day (ND-P. ananatis), and the FA group receiving 1.25 mg/day (ND-FA). Following a one-week stabilization period, mice were gavaged daily for five consecutive days. Core body temperatures were monitored at 0.5 h and 1 h post-gavage. Subsequently, mice were subjected to cold exposure at 4 °C, with core temperatures recorded at 2 h and 4 h post-exposure. Thermal differences across groups were captured via infrared thermography at the 4-h mark of cold exposure. Finally, BAT was harvested for quantitative PCR (qPCR) analysis to assess the mRNA expression levels of key thermogenic genes, including Ucp1, Pgc1α, Prdm16, and Cidea, to determine the regulatory patterns of upstream and downstream pathways involved in short-term BAT activation.

Collection of fecal samples from Macaca mulatta

A total of 82 fecal samples were collected from seven M. mulatta populations at varying altitudes: seven samples from Markam county in Xizang Province (4317 m asl), 12 samples from Yajiang county in Garze Prefecture, Sichuan Province (4120 m asl), 13 samples from Linzhi city in Xizang Province (3100 m asl), ten samples from Shennongjia National Park in Hubei Province (1600 m asl), nine samples from Jiyuan city in Henan Province (1058 m asl), ten samples from Beijing Zoo in Beijing City (575 m asl), and 21 samples from Dangan Island in Guangdong Province (50 m asl). Upon collection, the samples were promptly placed into sterile 15 mL centrifuge tubes and stored on dry ice. Subsequently, the samples were transported while maintained on dry ice for up to a week, and then stored at −80℃ at the IOZ for further analysis.

DNA extraction, sequencing, and data quality control

Microbial DNA was extracted from fecal samples using the QIAamp DNA Soil Mini Kit (Qiagen, Valencia, CA, USA) following the standard protocol. The quality and quantity of the extracted DNA were assessed using a Nanodrop spectrophotometer (model ND-1000, Nanodrop Technologies, Wilmington, DE, United States) and agarose gel electrophoresis. The DNA samples were stored at −20 °C until further use. For shotgun sequencing, an Illumina NovaSeq 6000 platform was employed, with a minimum of 10 Gb of sequencing data generated per sample with read length 150 bp.

Metagenome assembly

Raw sequence reads were quality trimmed using Trimmomatic version 0.36 [46]. Sequences with an average quality below 20 in a 4-base sliding window and reads shorter than 70 bp were removed. Subsequently, to eliminate contamination, the data were aligned to the genomes of M. mulatta (assembly GCF_003339765.1) using Bowtie2 (v2.3.5) [47]. This process ensured the retrieval of clean, non-contaminated sequencing data. Metagenome assembly was performed with MEGAHIT (v1.1.3) [48] using default parameters, with each sample’s sequence data assembled individually. Genes were then predicted within contigs longer than 300 bp using MetaGeneMark [49].

Genome reconstruction

Genome reconstruction of gut microbes using metagenomic sequencing data was conducted with the function modules of metaWRAP (v1.1.1) [50], a comprehensive pipeline that offers various modules for the analysis of metagenomic bins. The default minimum contig length for bins construction with metaBAT2 [51], Maxbin2 [52], and CONCOCT [53] was set at 1000 bp. Subsequently, metagenome-assembled genomes (MAGs) refinement was carried out using the bin_refinement module of metaWRAP to identify the best MAG based on the highest scoring function (S = Completion- 5 * Contamination value). A total of 31,096 MAGs were constructed.

Dereplication and species-level clustering of MAGs

The dereplication of all 31,096 MAGs was performed using dRep (v2.2.3) [54], with parameters set as: “-comp 80 -con 10 -str 100 -strW 0”. Out of these, 2,922 nonredundant MAGs met the quality evaluation (completeness ≥ 80% and contamination ≤ 10%) criteria of CheckM (v1.0.12) [55]. dRep can identify and select the best representative genome from highly similar genomes in a set. Additionally, MAGs were grouped into species-level genome bins (SGBs) based on a 95% average nucleotide identity (ANI) threshold using the ‘cluster’ tool in dRep (v2.2.3). SGBs with at least one reference genome from the Genome Taxonomy Database (GTDB, https://gtdb.ecogenomic.org/) were categorized as known SGBs, while those without references were classified as unknown SGBs. A total of 1,157 representative MAGs were identified, comprising 261 known SGBs and 896 unknown SGBs. The Minimum Information about a Metagenome-assembled Genome (MIMAG) standards established by the Genomic Standards Consortium [56], which mandate the presence of 23S, 16S, and 5S rRNA genes, at least 18 tRNAs genes and above 90% completeness and below 5% contamination in each MAGs. The identification of these genetic elements was performed with Prokka [57].

Phylogenetic analysis

Phylogenetic trees of the 1,157 representative MAGs were built by PhyloPhlAn (v3.0.51) [58] and visualized using iTOL (v5.6.2).

Estimation of the abundances of MAGs and taxa

Clean reads of each sample were aligned to representative MAGs using the metaWRAP QUANT_bin module to quantify the abundance of each MAG in each sample. Based on both the abundance and phylogenetic relationships of the MAGs, they were classified into various taxonomic levels including phylum, class, order, family, and genus. The abundances of the MAGs were then aggregated to generate an abundance table at each taxonomic level. It is important to note that bacterial abundance calculations in the metagenomes of colon contents from FMT-mice utilized a library of macaque MAGs and the same quantification method to compare bacterial populations before and after transplantation.

Generating a gene catalogue of Macaca mulatta gut microbiota

To produce a comprehensive library of functional genes from the gut microbiota of M. mulatta, we combined the gut microbiota genes from each individual. Initially, we employed MEGAHIT (v1.1.3) [48] to generate contigs from the clean data on an individual basis. Subsequently, genes were predicted within these contigs longer than 300 bp using MetaGeneMark [49]. Redundancy was then eliminated using CD-HIT [59] software with the following parameters: “-c 0.95 -aS 0.9 -g 1.0”. Ultimately, a gene catalog comprising 22,803,526 non-redundant genes was produced.

Estimation of the abundances of genes and functional terms

The clean reads of each sample were aligned to the gene catalog using Bowtie2 and the bam file was formatted by Samtools (v1.7) [60]. Bedtools (v2.26.0) [61] was used to count the mean depth of each gene from the bam file for each sample as gene abundance. Additionally, the gene catalog was annotated to Kyoto Encyclopedia of Genes and Genomes (KEGG, v50) [62] and carbohydrate-active enzyme (CAZy) database [63] using DIAMOND (v0.9.24) [64]. The abundance of KEGG Orthology (KO), KEGG pathway, and CAZyme were calculated by a perl script based on gene abundance. Gene and functional term abundance calculations in the metagenomes of colon contents from FMT-mice utilized a library of macaque gene catalog and the same quantification method was used to compare functional genes before and after transplantation.

Measurement of body mass, food intake and tissue weight of FTM-mice

During domestication, body mass and food intake were measured at 9:00 a.m. using an electronic balance. Food intake (g) was calculated by subtracting uneaten food weight from initial food weight. Following the transplantation experiment, the mice were euthanized using carbon dioxide and the necessary samples were collected, such as: carcass, brown fat, white fat, heart, liver, spleen, lung, kidney, blood, colon contents.

Determination of blood glucose, lipids, total bile acids and insulin in blood of FTM-mice

Serum glucose, serum total bile acids concentrations, total serum cholesterol (CHOL), LDL-C (low-density lipoprotein cholesterol) and HDL-C (high-density lipoprotein cholesterol), and triglycerides (TG) levels were determined using kits from Beijian XinChuangYuan, Beijing, China. Insulin levels were quantified using radioimmunoassay with a mouse RIA kit from China Institute of Atomic Energy, Beijing, China, within a detection range of 5–160 μIU/mL.

Hematoxylin and eosin staining of small intestine and BAT

The small intestines of FMT-mice were fixed in 4% buffered paraformaldehyde at room temperature and then embedded in paraffin. Subsequently, tissues were sectioned at a thickness of 5–6 µm and stained with hematoxylin and eosin (H&E). The villus length and crypt depth were analyzed using ImageJ software. The BAT of mice administered P. ananatis via oral gavage was paraffin-embedded, sectioned, and subjected to H&E staining and scanning to investigate the impact of P. ananatis on fat content in BAT.

Short-chain fatty acid (SCFAs) targeting metabolome analysis of FMT-mice colon contents

To compare the metabolic capacity of SCFAs in the gut microbiota of M. mulatta at high and low elevations, the colon contents from FMT-mice were homogenized in water with glass beads, centrifuged, and then subjected to GC–MS analysis. A Thermo Fisher Scientific gas chromatograph was used for the analysis, with specific parameters for injection, column temperature, and mass spectrometric detection. Calibration curves were employed to determine the concentration of target compounds in the samples. Metabolite concentrations below 0 were reported as not detected (ND). To evaluate the technical precision of each experiment, the relative standard deviation of peak areas was calculated for all compounds detected in the quality control (QC) sample (RSD = 100 * standard deviation/average of peak areas) with an ideal RSD of less than 15%.

Lipidomics analysis of liver and colon contents by LC–MS/MS

To compare the lipid metabolic capabilities of gut microbiota in macaques living at different altitudes, colon contents and liver were collected for lipidomics analyses. Lipid extraction and mass spectrometry were conducted by Applied Protein Technology. Individual samples from each group were pooled to create a QC sample for system stability and data reliability. LC–MS/MS analysis was performed using a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) coupled with a UHPLC Nexera LC-30A (SHIMADZU) with a 2.1 mm × 10 cm Waters, ACQUITY UPLC CSH C18 column with a particle size of 1.7 µm. The column temperature was maintained at 45 °C with a flow rate of 300 μL/min. The mobile phase for A consisted of acetonitrile aqueous solution (acetonitrile: water = 6:4, v/v) + 0.1% formic acid + 0.1 mM ammonium formate, while for B it was acetonitrile isopropanol solution (acetonitrile: isopropanol = 1:9, v/v) + 0.1% formic acid + 0.1 mM ammonium formate. Both positive and negative ion modes of electro spray ionization (ESI) were used for detection. Mass spectrometry analysis was carried out using the Q Exactive series mass spectrometer (Thermo Fisher Scientific) with specific ESI source conditions: heater temp: 300 °C; sheath gas flow rate: 45 arb; aux gas flow rate:15 arb; sweep gas flow rate: 1arb; spray voltage: 3000 V; capillary temp: 350 °C; s-lens RF level: 50%; MS1 scan ranges: 200–1800. Lipid identification, peak extraction, alignment, and quantification were done with LipidSearch software (v4.1, Thermo Fisher Scientific). Only variables with more than 50% nonzero measurement values in at least one group were considered in the extracted ion features.

Untargeted metabolomics of colon contents by LC–MS/MS

To identify differential metabolites associated with P. ananatis, untargeted metabolomics (LC–MS/MS) were conducted on the colon contents of mice following a single oral gavage of the bacteria. QC samples, composed of equal volumes from the experimental samples, were utilized to calibrate the chromatography-mass spectrometry system and to monitor instrument performance, thereby assessing system stability throughout the experiment. The LC–MS/MS analysis involved chromatographic separation on a 2.1 mm × 10 cm Hypersil Gold C18 column with a particle size of 1.9 µm (Thermo Fisher Scientific), coupled with a Q Exactive™ HF/Q Exactive™ HF-X mass spectrometer (Thermo Fisher Scientific) and a Vanquish UHPLC chromatograph (Thermo Fisher Scientific). The column temperature was maintained at 40 °C with a flow rate of 200 μL/min. The mobile phase consisted of A: water with 0.1% formic acid, and B: methanol. The mass spectrometer operated in both negative and positive scanning modes (100–1,500 m/z) with the following source parameters: spray voltage at 3,500 V; sheath gas flow rate at 35 psi; auxiliary gas flow rate at 10 L/min; capillary temperature at 320 °C; S-lens RF level at 60; and auxiliary gas heater temperature at 350 °C. Offline data were imported into CD3.3 database search software for processing and compared against mzCloud, mzVault, and Masslist databases, followed by standardization to obtain relative peak areas. Compounds with a coefficient of variation (CV) in relative peak areas exceeding 30% in QC samples were excluded, resulting in the final identification and relative quantification of metabolites.

Liver biochemical index detection

To observe the effects of oral gavage of P. ananatis on liver health and lipid content in HFD mice, we measured alanine aminotransferase (ALT), aspartate aminotransferase (AST), cholesterol (CHOL) and triglycerides (TG) using alanine substrate method, aspartate substrate method, CHOD-PAP method and GPO-PAP method, respectively.

Western plot of UCP1 in BAT of mice

To investigate whether P. ananatis can activate the energy metabolism function of BAT, we extracted UCP1 from the BAT [63] of mice subjected to oral gavage with P. ananatis. We then assessed the differences in UCP1 amount through Western blot (WB) experiments. Specifically, we used 15 µg of protein for the WB analysis. SDS-PAGE electrophoresis and membrane transfer were performed according to standard protocols. The UCP1 antibody (Abcam Shanghai Trading Co., Ltd.) was diluted to a ratio of 1:2000 and incubated at 4 °C overnight. Subsequently, a goat anti-rabbit HRP secondary antibody was applied at a dilution of 1:5000 and incubated at room temperature for 1 h. The chemiluminescent reaction was initiated using ECL luminescent solution (Abclonal, China), and the results were visualized using the Bio-Rad ChemiDoc XRS + system. For the quantification of Western Blot results, ImageJ software was used to analyze the integrated density (IntDen) of protein bands. After background subtraction, the net integrated density (Net IntDen) was obtained. The relative expression level of the target protein was calculated as follows: Relative expression level of target protein = Net IntDen of target protein/Net IntDen of alpha-tubulin.

Enzyme-linked immunosorbent assay (ELISA)

To evaluate whether oral gavage of Pantoea ananatis induced an inflammatory response, serum concentrations of tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β), interleukin-6 (IL-6), and C-reactive protein (CRP) were quantified. These markers were measured using the following enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturer’s instructions: Mouse TNF-α ELISA Kit (Cat No. SU-BN20852), Mouse IL-1β ELISA Kit (Cat No. SU-BN20174), Mouse IL-6 ELISA Kit (Cat No. SU-BN20188), and Mouse CRP ELISA Kit (Cat No. SU-BN24852) (all from Beijing Kewei Biological Technology Co., Ltd., Beijing, China). Additionally, to investigate potential alterations in alternative triglyceride hydrolysis pathways, the levels of lipoprotein lipase (LPL) in brown adipose tissue (BAT) and white adipose tissue (WAT) were determined. LPL concentrations were measured using a Mouse LPL ELISA Kit (Cat No. SU-BN20838, Beijing Kewei Biological Technology Co., Ltd.) following the specified protocol.

Real-time RT-PCR quantification assays

Total RNA (200 ng) extracted from tissues was reverse transcribed into cDNA using SuperScript III Reverse Transcriptase (Invitrogen). Quantitative PCR was performed on an Applied Biosystems 7500 Fast Real-Time PCR System. For thermogenic gene expression (Ucp1, Prdm16, Cidea, and Pgc1a), reactions were conducted using specific primers (Ucp1: Forward Primer (5'−3')- ACTGCCACACCTCCAGTCATT; Reverse Primer (5'−3')- CTTTGCCTCACTCAGGATTGG; Prdm16: Forward Primer (5'−3')- GAAGTCACAGGAGGACACGG; Reverse Primer (5'−3')- CTCGCTCCTCAACACACCTC; Cidea: Forward Primer (5'−3')- TGCTCTTCTGTATCGCCCAG; Reverse Primer (5'−3')- GCCGTGTTAAGGAATCTGCTG; Pgc1a: Forward Primer (5'−3')- ACAGCTTTCTGGGTGGATTG; Reverse Primer (5'−3')- TGAGAACCGCTAGCAAGTTT) with beta-Actin (Forward Primer (5'−3')- CACCAGTTCGCCATGGATGAC; Reverse Primer (5'−3')- GCCTCGTCACCCACATAGGAGT) as the internal control. Relative transcript levels were analyzed using the ΔΔC T method. Three mice were used per group and all samples were assayed in duplicate.

Immunofluorescence of UCP1 in BAT of mice

We also conducted immunofluorescence experiments on paraffin sections of BAT tissue of mice administered with ferulic acid to explore the expression of UCP1 protein. BAT sections were attached on slides. The slides were dewaxed and rehydrate by xylene, and 100, 95, and 70% ethanol series. Heat-based antigen retrieval was performed using 0.1 M citrate buffer. All slides were treated with 0.2% Triton X-100 for 45 min, washed three times with PBS and finally blocked with 5% bovine serum albumin for 60 min. All samples were incubated with anti-UCP1 antibodies (Abcam Shanghai Trading Co., Ltd.) with 1:200 dilution at 4˚C overnight, after washed three times with PBS, incubation with secondary antibodies for 1 h at 37˚C, followed by washing with PBS and DAPI staining. Images were taken using microscopes.

Quantification and statistical analysis

Pairwise comparisons were performed with the Wilcoxon rank sum test. Differences among M. mulatta populations and functional terms of the gut microbiome at different altitudes were assessed with the Generalized Linear Model (GLM) implemented in the “glm2” R package (v1.2.1) [65].

Authors’ contributions

L.Ming, W.X.C., and B.T.B. Methodology: W.X.C., Z.M.Y., L.Meng, X.X.M., S.Yue., B.T.B., Y.T., Z.J.P., Z.D., Y.Y.Q., Investigation: W.X.C., Z.M.Y., X.H.L., and X.Z.F. Visualization: W.X.C., Z.M.Y., L.Meng, S.Yue., Y.T., S.Ying., and Q.J.W. Funding acquisition: L. Ming, W.X.C., and B.T.B. Project administration: W.X.C., Z.M.Y., X.X.M., L. Meng, and L.Ming Supervision: L.Ming Writing – original draft: W.X.C., Z.M.Y., and L. Ming Writing – review & editing: W.X.C., Z.M.Y., L.Meng, S.Yue., B.T.B., C.R., and L.Ming.

Funding

The work was supported by the National Natural Science Foundation of China (32300348, 32470487) and State Key Laboratory of Animal Biodiversity Conservation and Integrated Pest Management (Grant No. SKLA2504).

Data availability

Metagenome data have been deposited at the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA022869) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa. Original metabolomic data have been deposited at Mendeley at [https://doi.org/10.17632/7b68x3pgzm.1]. Original western blot images have been deposited at Mendeley at [https://doi.org/10.17632/fpwk6y5df2.1] and are publicly available as of the date of publication. Slicing data reported in this paper will be shared by the lead contact upon request. Codes are available at https://sandbox.zenodo.org/records/275838.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.