The Eubacterium Rectale Derived Extracellular Vesicles Alleviate Cancer Cachexia Induced Lipolysis by Inhibiting Macrophage Polarization

ABSTRACT

Cancer cachexia (CC) is a complex pathological condition associated with cancer progression and poor prognosis, particularly in gastrointestinal cancers, and is closely linked to the gut microbiota. Lipolysis in CC may play a key role in driving cachexia progression. However, the role of the gut microbiota in CC‐induced lipolysis and the underlying mechanisms remains unclear. In this study, we found that significant gut microbiota dysbiosis in pancreatic cancer cachexia, characterized by diminished abundance of Eubacterium rectale (E.rectale), was negatively associated with lipolysis severity. Furthermore, we demonstrated for the first time that E. rectale derived extracellular vesicles (EVs) can ameliorate lipolysis across multiple CC models. Mechanistically, these effects are mediated by modulating pro‐inflammatory macrophages through inhibition of NF‐kB signalling, reducing macrophage polarization and inflammatory cytokine secretion. Our findings indicated E. rectale EVs could offer potential therapeutic benefits for CC and emphasize the importance of host‐microbiota interactions.

Trial Registration: This study was registered at ClinicalTrials.gov (NCT06378853).

1. Introduction

Cancer cachexia (CC) is a complex metabolic syndrome characterized by progressive weight loss, systemic inflammation, muscle and adipose tissue wasting, affecting 40%–80% of gastrointestinal cancer patients (Baracos et al. 2018; Argilés et al. 2023; Yoon and Grundmann 2023). Particularly in pancreatic cancer (PC), CC exhibits the highest prevalence (>70%) and severity, significantly impairing survival outcomes, treatment efficacy, and quality of life (Shi et al. 2022; Mizrahi et al. 2020). While previous research has predominantly focused on muscle wasting in CC, emerging evidence underscores the pivotal role of adipose tissue loss, which precedes muscle atrophy and exacerbates systemic metabolic dysfunction (Rohm et al. 2016; Rydén et al. 2008). Adipose tissue loss occurs earlier than skeletal muscle atrophy and contributes to further muscle wasting (Fouladiun et al. 2005). The underlying mechanisms of adipose tissue wasting involve complex interactions among immune cells, pro‐inflammatory cytokines such as IL‐6 and TNF‐α, and tumor‐derived factors, leading to excessive lipolysis and subsequent release of free fatty acids (FFA) and glycerol, which perpetuate systemic inflammation and energy imbalance (Rupert et al. 2021; Vaitkus and Celi 2017; Porporato 2016).

Recent advances have highlighted the gut microbiota as a critical regulator of host immunity and physiological homeostasis (Sepich‐Poore et al. 2021). There is growing evidence that gut microbiota dysbiosis is closely associated with cancer progression and the development of CC (Yang et al. 2023). PC patients frequently exhibit marked changes in the composition of gut microbiota, characterized by reduced microbial diversity, increased pathogenic bacteria, and decreased beneficial species (Thomas and Jobin 2020). This dysbiotic state correlates strongly with systemic inflammation, compromised gut barrier integrity, and metabolic dysregulation, further aggravating CC‐associated lipid metabolism disorders.

Although the role of the gut microbiota in CC has been increasingly acknowledged (Ni et al. 2021; Liu et al. 2023), its impact on lipolysis in CC remains poorly understood. The targeted modulation of the gut microbiota via probiotics has emerged as a promising therapeutic approach for cancer and its associated conditions (Badgeley et al. 2021). However, how probiotics attenuate excessive lipolysis in CC has yet to be fully explored. This study systematically investigated the link between gut microbiota dysbiosis and lipolysis in pancreatic CC, pioneering the use of probiotic interventions to mitigate CC associated lipolysis.

Bacterial extracellular vesicles (EVs), nanosized bilayer membrane structures actively released by bacteria, have emerged as key mediators of host‐microbiota interactions (Xie et al. 2022). These vesicles encapsulate diverse biological components from parent bacteria, ensuring stability and facilitating long‐distance communication (Sartorio et al. 2021; Wang et al. 2023). Notably, bacterial EVs can transfer to distant organs and target immune cells, particularly macrophages and neutrophils (Erttmann et al. 2022; Díaz‐Garrido et al. 2021). In CC progression, inflammatory macrophages accumulate around adipose tissue (Molfino et al. 2022); however, the potential regulatory role of gut microbiota‐derived EVs in this process remains unexplored. Therefore, gaining insight into these interactions is crucial for developing microbiota‐based therapeutic approaches to alleviate CC‐related metabolic disturbances and improve clinical outcomes.

Here, we identified significant gut microbiota dysbiosis in both pancreatic cancer (PC) patients with CC and mouse models, characterized by a reduced abundance of Eubacterium rectale. Furthermore, we demonstrated for the first time that E. rectale attenuates CC‐induced lipolysis through its EVs, which modulate macrophage function in adipose tissue via the NF‐kB signalling pathway. In summary, our findings revealed a novel mechanism of host‐microbiota interaction in CC‐induced lipolysis and highlight the therapeutic potential of probiotic interventions in CC.

2. Results

2.1. Lipolysis and gut Dysbiosis With a Significant Reduction of E. rectale in PC Patients

To investigate the alterations in nutritional status and adipose tissue in PC patients with CC, we enrolled pancreatic ductal adenocarcinoma (PDAC) patients, further dividing the PDAC cohort into PC (pancreatic cancer‐associated cachexia) and non‐PC (non‐pancreatic cancer‐associated cachexia) groups. The baseline characteristics and clinical parameters of the two groups are summarised in Table S1. Significant differences were observed in IL‐6 levels, FFA, and Patient‐Generated Subjective Global Assessment (PG‐SGA) scores between the two groups, indicating poor nutritional status, elevated systemic inflammation, and pronounced lipolysis in PC patients. Computed tomography (CT) and body composition analysis further revealed a marked reduction in fat distribution and more severe adipose tissue depletion in the PC group (Figure 1A–C). Survival analysis based on fat mass index (FMI) revealed that patients with lower FMI had significantly shorter overall survival (Figure 1D). Additionally, analysis of subcutaneous adipose tissue (SAT) obtained during surgery demonstrated the upregulation of lipolysis‐related genes in PC patients (Figure 1E). H&E staining showed a significant decrease in adipocyte area, and IHC staining indicated increased white adipose tissue (WAT) browning and macrophage infiltration in SAT from PC patients (Figure 1F). These findings collectively demonstrated that PC is characterized by severe adipose tissue loss, which is closely associated with poor clinical outcomes.

FIGURE 1.

Lipolysis and gut dysbiosis with a significant reduction of E. rectale in PC patients. (A) Representative CT scans showing SAT and VAT distribution. (B) Comparison of the areas of SAT and VAT across the three groups (Non‐PC, n = 32; PC, n = 35) according to CT imaging. (C) Comparison of fat mass index (FMI) among the three groups. (D) Survival analysis based on FMI levels. (E) qPCR analysis of lipolysis‐related gene expression in intraoperatively collected SAT samples (n = 12). (F) Histological analysis of SAT, including: H&E staining, IHC staining for UCP‐1 (browning marker) and CD68+ (macrophage marker). Scale bars, 100 µm. (G–J) bacterial 16S rRNA gene sequencing in Non‐PC and PC patients (Non‐PC, n = 20; PC, n = 20). (G) Alpha diversity analysis comparing the Chao1, Shannon index and ACE between Non‐PC (blue) and PC (yellow) groups, with each dot representing an individual sample. (H) Principal coordinate analysis (PCoA) of bacterial communities showed distinct differences apart from Non‐PC. (I) Stacked bar plot illustrates the relative abundance of the top ten species across PC and Non‐PC groups. (J) LDA scores were determined for species with varying abundance between the NC and PC groups. The length indicates the effect size for each species. Only LDA scores exceeding 1.0 are displayed. (K) Cladogram of varying species across Non‐PC and PC groups. (L) Spearman correlation analysis assessing associations between selected bacterial species and clinical parameters. (M) Relative abundance of Eubacterium rectale between the Non‐PC and PC groups. Man–Whitney U test was used to calculate p values between Non‐PC and PC in (B–C). Log‐rank test was used to calculate p values in (D); Student's t‐test was used to calculate p values in (E), (G) and (M). *p < 0.05, **p < 0.01, ***p < 0.001.

Given the critical role of gut microbiota in various cancers and diseases, we further explored its potential involvement in PC. Fecal samples were collected from patients to characterize their gut microbial communities (Figure 1G–K). We observed alterations in gut microbiota diversity and richness—evidenced by decreased Chao1 and ACE indices, as well as the Shannon index (Figure 1G)—and distinct clustering in the PCoA analysis(Figure 1H). Our findings also revealed that certain microbial communities were more abundance in PC patients compared to the control group. At the species level, PC patients exhibited reduced microbial diversity (Figure 1I). To identify candidate species contributing to PC, we performed linear discriminant analysis effect size (LEfSe), which revealed a notable decrease in the abundance of Eubacterium rectale compared to the control group (Figure 1J–K. Subsequently, we performed correlation analyses between the top 10 most significantly differentially abundant classified species and clinical indicators (as identified at the species level and by LEfSe). The abundance of E. rectale was negatively correlated with the severity of PC induced lipolysis and inflammatory cytokine levels, while positively correlated with FMI and nutritional indicators such as PG‐SGA (Figure 1L). The relative abundance of E. rectale showed a similar trend (Figure 1M). These findings indicated that the reduction of E. rectale may play a vital role in PC induced adipose tissue loss.

To validate these observations, we established an orthotopic PC mouse model using Pan02 cells. PC mice exhibited a significant cachectic phenotype (Figure S1A–B). The abundance of E. rectale in the PC group mice was also lower than that in the NC (negative control) group (Figure S1C). Changes in body weight (Figure S1D), body composition of fat mass percentage (Figure S1E), grip strength (Figure S1F), serum FFA levels (Figure S1G), muscle mass, and adipose tissue mass (Figure S1H) indicated significant weight loss, along with muscle and fat loss in the PC mice. Both inguinal white adipose tissue (iWAT) and epididymal fat white adipose tissue (eWAT) showed pronounced depletion (Figure S1I–J and M–N), accompanied by decreased triglyceride (TG) levels in adipose tissue (Figure S1K and O). Key lipolytic proteins, including hormone‐sensitive lipase (HSL) and adipose triglyceride lipase (ATGL), were significantly upregulated in the PC model (Figure S1L, P, Q and R). These findings suggested that E. rectale may be critical in regulating cachexia‐associated lipolysis.

2.2. Gut Microbiota Is Closely Associated With Lipolysis

To explore the effect of gut microbiota on adipose tissue metabolism, we established antibiotic (ABX)‐treated models in both NC and PC mice to evaluate phenotypic changes in adipose tissue following gut microbiota depletion (Figure 2A). Absolute quantitative results confirmed a significant reduction in gut microbiota abundance in both NC and PC mice after 7 days of ABX treatment (Figure 2B and Figure S3A). Following ABX intervention, both NC and PC mice exhibited a marked decrease in body weight (Figure 2C and Figure S3B) and body fat percentage (Figure 2D and Figure S3C), accompanied by a significant increase in serum FFA levels (Figure 2E and Figure S3D). Additionally, fat mass and TG content in adipose tissue were significantly reduced (Figure 2F–I, S2A–D and S3E–J), while the expression of lipolysis‐related proteins was markedly upregulated (Figure 2J–K, S2E–F and S3L–P). These findings further emphasise the pivotal role of gut microbiota in regulating adipose tissue wasting.

FIGURE 2.

Gut microbiota is closely associated with lipolysis. (A) Workflow of experiment: PC mice received an antibiotic cocktail as a pretreatment (n = 6). (B) Evaluation of gut microbiota depletion after ABX treatment. (C) Changes in body weight after ABX treatment. (D) Alterations in body composition of fat mass percentage. (E) Changes in FFA levels. (F–K) Lipolysis in eWAT after ABX treatment: representative images of eWAT after ABX treatment in PC mice (F), eWAT mass (G) and TG content (H), H&E staining (I), western blot (J) and quantification (K) of ATGL and phosphorylated HSL levels. Scale bars, 100 µm (L) Workflow of FMT: PC mice were pretreated with ABX and then received FMT three times a week. After 5‐week treatment, the mice were sacrificed and used for subsequent experiment (n = 6). (M) Changes in body weight after FMT. (N) Alterations in body composition of fat mass percentage. (O) Changes in FFA levels. (P–U) Lipolysis in eWAT after FMT: representative images of eWAT after FMT (P), eWAT mass (Q) and TG content (R), H&E staining (S), western blot (T) and quantification (U) of ATGL and phosphorylated HSL levels. Scale bars, 100 µm. Student's t test was used to calculate p values between PC, PC‐ABX in (B), (C), (D), (E), (G), (H) and between Non‐CA^FMT, CA^FMT in (M), (N), (O), (Q), (G), (R) and (U). *p < 0.05, **p < 0.01, ***p < 0.001.

In our previous results, PC patients exhibited pronounced lipolysis and significant gut microbiota dysbiosis (Figure 1). To investigate the connection between gut microbiota and lipolysis, and to assess whether restoring gut microbiota homeostasis could ameliorate this pathological process, we collected fecal samples from non‐cachexia and cachexia patients. These samples were used for fecal microbiota transplantation (FMT) in mice over a 5‐week period to evaluate whether microbiota could mitigate cachexia‐induced lipolysis (Figure 2L).

As anticipated, mice receiving FMT from Non‐CA donors showed significantly higher body weight (Figure 2M) and body fat percentage (Figure 2N) compared to those receiving FMT from CA patients. Similarly, serum FFA levels and changes in white adipose tissue (iWAT and eWAT) followed the similar trends. Specifically, cachexia mice receiving FMT from Non‐CA donors exhibited a significant reduction in FFA levels (Figure 2O) and improved fat mass (Figure 2P–S and Figure S2G–J) and TG content. The expression of lipolysis‐related proteins was also down‐regulated in this group (Figure 2T–U and Figure S2K–L).

2.3. E. rectale Alleviates PC Induced Lipolysis

The results in Figure 1 suggest that E. rectale may play a significant role in regulating lipolysis during PC. To verify this hypothesis, we conducted the single‐strain transplantation experiment to assess the effects of E. rectale on attenuating lipolysis in PC mouse model. We cultured E. rectale and Escherichia coli (E. coli) and compared their effects on PC induced lipolysis(Figure 3A).

FIGURE 3.

E. rectale alleviates lipolysis induced by PC. Workflow of E. rectale gavage protocol: PC mice were administered weekly gavage of E. rectale after a 5‐week treatment. E. rectale burden of mice in Control, E.rectale and E.coli groups (n = 6). (C) Changes in body weight after E. rectale transplatation. (D) Alterations in body composition of fat mass percentage. (E) Changes in FFA levels. (F–K) Lipolysis in eWAT after E. rectale transplatation: representative images of eWAT after E. rectale transplatation in PC mice (F), eWAT mass (G) and TG content (H), H&E staining (I), western blot (J) and quantification (K) of ATGL and phosphorylated HSL levels. Scale bars, 100 µm One‐way ANOVA was used to calculate p values in (B), (C), (D), (E), (G), (H) and (K). * p < 0.05, ** p < 0.01, *** p < 0.001.

The results of the single‐strain transplantation experiment indicated that after 5 weeks of transplantation, the abundance of E. rectale in the feces of the E. rectale group was significantly higher than that in the other two groups (Figure 3B). The experimental results demonstrated that mice treated with E. rectale showed significantly higher overall body weight (Figure 3C) and body fat percent (Figure 3D) compared to the control and E. coli‐treated groups. Meanwhile, the FFA levels in the E. rectale group were significantly reduced (Figure 3E). The mass and TG content of white adipose tissue (iWAT and eWAT) were also notably increased (Figure 3F–I and Figure S4A–D). The expression of lipolysis‐related proteins showed a similar trend(Figure 3J–K and Figure S4E–F). These results indicated that E. rectale can partially alleviate PC induced lipolysis, providing a potential microbial intervention strategy for the treatment of PC.

2.4. E. rectale Alleviates PC‐Induced Lipolysis via EVs

The secretion of EVs is one of the key pathways for microbiota‐host interactions. To investigate whether the protective effect of E. rectale on lipolysis during PC is dependent on EVs, we conducted an intervention experiment using E. rectale and GW4869‐pretreated E. rectale in mice (GW4869 is a neutral sphingomyelinase inhibitor known to suppress EV release (Liu et al. 2021; Hong et al. 2023). We next evaluated the impact of intragastric delivery of GW4869‐pretreated E.rectale on CC induced lipolysis in mice. The diagram of the experimental procedures was shown in Figure 4A. The effect of GW4869 on E. rectale viability was assessed using the colony counting assay. We found that the treatment with GW4869 did not significantly affect the viability or growth of E. rectale (Figure 4B and Figure S5A). In addition, bacterial viability was assessed using DMAO/PI dual staining. As shown in Figure S5B, GW4869 treatment did not affect the viability of E. rectale. Given that metabolic activity is also an important indicator of bacterial growth status, we further measured ATP levels to evaluate bacterial metabolic activity. The results in Figure S5C indicated that GW4869 pretreatment did not alter the metabolic activity of E. rectale. Furthermore, scanning electron microscopy (SEM) was performed to examine whether this treatment affects bacterial membrane integrity and structure. As shown in Figure S5D, the SEM image also demonstrated that GW4869 treatment did not compromise the membrane integrity of E. rectale. EV protein content serves as a proxy for EV abundance, and normalizing EV levels to total protein concentration is a straightforward and commonly applied approach for EV quantification (Vanaja et al. 2016; Choi et al. 2015). Figure 4C demonstrated that the treatment with GW4869 for 4 days markedly suppressed the EVs production, reflected by the lower total protein content of isolated EVs. Although GW4869 was removed and the treated E.rectale was cultured for another 4 days, EV protein levels in the GW4869‐pretreated group remained substantially lower than those in the vehicle‐treated group, which is shown in Figure 4C. To further validate this observation, we quantified EV particle numbers from GW4869‐pretreated or vehicle‐treated E. rectale through Nanoparticle Tracking Analysis (NTA). In line with the previous observations, EV particle numbers in E. rectale were substantially reduced following 4 days of GW4869 treatment, and the decrease persisted even after the inhibitor was removed and the E.rectale were cultured for an additional 4 days (Figure 4D). In addition, we compared the particle‐to‐protein ratios of EV preparations across the different experimental groups (Figure S5E). The results showed no significant differences in this ratio among the groups, indicating that the purity of EV preparations was comparable across conditions. The results also suggest that, under our experimental conditions, both EVs protein concentration and particle number can serve as consistent indicators of relative EV abundance. Together, these findings demonstrated that GW4869 exerts a persistent inhibitory effect on EV secretion.

FIGURE 4.

E. rectale alleviates CC‐induced lipolysis via extracellular vesicles. (A) Experiment design for testing the impact of GW4869 on the secretion of EVs by E. rectale. The control and PC mice were orally treated with vehicle, E.rectale or GW4869‐pretreated E.rectale 3 times a week for 4 weeks. (B) Growth of E. rectale treated with vehicle (DMSO) and GW4869 (n = 3). OD 600, optical density at 600 nm. (C–D) Total protein contents (C) and particle numbers (D) of EVs from E.rectale treated or pretreated with vehicle or GW4869 for 4 days (n = 3). (E–J) PC mice were administered PBS (vehicle), E. rectale, or GW4869‐pretreated E. rectale (n = 6). Body weight changes (E), alterations in body composition of fat mass percentage (F), FFA levels (G), eWAT mass (H) and TG content (I), H&E staining (J). Scale bars, 100 µm, (K and L) Transmission electron microscopy (TEM) (K) and nanoparticle tracking analysis (NTA) (L) of isolated E.re EVs from cultured E. rectale. Scale bars, 100 nm. (M) Representative images of different tissues after administration of Dil‐labeled EVs and quantification of the fluorescence intensity in the organs of mice (n = 3). (N) The presence of Dil red fluorescence in the epididymal fat tissue after intervention of Dil‐labeled EVs. (O–T) Effects of E. re EVs administration on lipolysis in PC mice. PC mice received E. re EVs, PBS (vehicle) and mock EVs via oral gavage. Body weight changes (O), alterations in body composition of fat mass percentage (P), FFA levels (Q), eWAT mass (R) and TG content (S), H&E staining (T). Scale bars, 100 µm. Two‐way ANOVA was used to calculate p values in (B), (C), (D) and (M). One‐way ANOVA was used to calculate p values in (E–I) and (O–S). * p < 0.05, ** p < 0.01, *** p < 0.001.

In the GW4869‐pretreated E. rectale intervened mice, the protective effect of E. rectale on lipolysis was weakened. Specifically, these mice exhibited weight loss (Figure 4E), reduced body fat percent (Figure 4F), elevated serum FFA levels (Figure 4G), and a significant decrease in the mass and TG content of both iWAT and eWAT (Figure 4H–J and Figure S5F–H). These results suggested that GW4869 pretreatment significantly suppressed the beneficial effects of E. rectale, indicating that EVs may play a crucial role in E. rectale‐mediated alleviation of CC induced lipolysis.

To further verify this, we isolated and characterized the EVs secreted by E. rectale. The E.rectale EVs displayed cup‐like morphologies with median diameters of 122.4 nm and the concentration was 3.4 × 10¹⁰ particles/mL, as shown by transmission electron microscopy (TEM) and NTA (Figure 4K and L). We then labelled the E.rectale derived EVs with the Dil dye and tracked them in vivo (Figure 4M). Notably, we observed that EVs accumulated in adipose tissue, as confirmed by the fluorescent signal in adipose tissue. Compared with the PBS control group, EV treatment led to a clear accumulation of fluorescence signals in adipose tissue (Figure 4N). These results indicated that E. rectale EVs may exert their effects by transferring to adipose tissue.

In order to confirm whether the accumulation of EVs in adipose tissue can improve CC induced lipolysis, we intervened in PC mouse model with E. rectale EVs, the control group (Vehicle) and mock EVs group (the culture medium is subjected to the same procedures as the samples). The experimental results showed that mice treated with EVs exhibited significantly increased body weight change (Figure 4O) and fat composition (Figure 4P), while their serum FFA levels markedly decreased (Figure 4Q). Additionally, the mass and TG content of both iWAT and eWAT significantly increased (Figure 4R–T and Figure S5I–K). The expression levels of lipolysis‐related proteins were significantly decreased as well (Figure S5L–M). However, treatment with mock EVs did not exert a significant ameliorative effect. In addition, the treatment with E.re EVs or mock EVs had no effect on tumor size in mice (Figure S5N–O). These findings demonstrate that E. rectale EVs effectively attenuate CC‐induced lipolysis.

Given that WAT browning contributes to lipid mobilization, energy expenditure, and systemic inflammation during CC, we further investigated whether E.re EVs influence this process. Analysis of key browning‐related proteins in adipose tissue (Figure S5P) revealed that E.re EVs treatment suppressed CC–induced WAT browning, futher suggests E. rectale EVs can effectively alleviate CC induced lipolysis.

Given that E.re EVs exhibited the ability to alleviate cachexia‐induced lipolysis, we next performed a preventive intervention study. EVs treatment was initiated prior to tumor cell inoculation, and body weight, fat loss, as well as lipolysis and inflammation related parameters were subsequently monitored to evaluate whether EVs could delay the onset of cachexia. As shown in Figure S6, preventive administration of E.re EVs markedly attenuated body weight loss (Figure S6A) and increased body fat percentage (Figure S6B) in cachectic mice. In addition, EV treatment reduced FFA levels (Figure S6C) and improved adipose tissue mass (Figure S6D–F) and TG content (Figure S6E). Consistently, the expression of lipolysis‐associated proteins was also decreased (Figure S6G). Furthermore, evaluation of systemic inflammatory markers revealed that preventive administration of E.re EVs significantly reduced serum levels of IL‐6 and TNF‐α (Figure S6H–I), indicating that E.re EVs alleviate cachexia‐associated systemic inflammation.

Considering that gut microbiota dysbiosis and impaired intestinal barrier function can trigger systemic inflammation, and previous studies have reported intestinal barrier dysfunction and microbiota dysbiosis in patients with CC (Costa et al. 2019), we further investigated whether E.re EVs could improve intestinal barrier integrity. Notably, we also observed significant microbial dysbiosis in patients with pancreatic CC in Figure 1. As shown in Figure S6J–K, mice with pancreatic CC exhibited elevated serum endotoxin levels and pronounced intestinal pathological damage, both of which were alleviated by EVs treatment. Furthermore, we examined the expression and localization of key intestinal barrier proteins. Immunofluorescence and protein expression analyses revealed that EVs treatment increased the expression of Occludin, MUC2, and ZO‐1, which are critical components of the intestinal barrier (Figure S6L–M). Collectively, these findings indicate that E.re EVs effectively ameliorate intestinal barrier dysfunction associated with CC.

Given the role of E. rectale EV in improving lipolysis during CC, we further characterized the protein components of E. rectale EVs using label‐free proteomics (Figure S6N). A total of 5946 peptides were identified, with 5343 being specific peptides, and 1044 proteins were identified, of which 958 were quantitatively analysed (Figure S6O). Gene Ontology (GO) annotation revealed that these proteins are involved in cellular processes, metabolic processes, catalytic activity, and binding functions (Figure S6P). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis suggested that these proteins could be involved in essential biological processes such as amino acid and carbohydrate metabolism (Figure S6Q). These results suggested that E. rectale EVs can selectively enrich various functional proteins and exert regulatory effects on other cells. Based on the literature review and previous studies (Mantovani et al. 2004), we found that these metabolic pathways and biological functions are closely related to macrophage polarization. Therefore, we hypothesised that E.rectale derived EVs may exert their effects by modulating macrophage polarization.

2.5. E. rectale EVs Alleviated PC‐Induced Lipolysis by Inhibiting M1 Macrophage Polarization

Next, we aimed to explore the mechanism through which E. rectale EVs exert their effects. Previous studies have shown substantial macrophage infiltration in the adipose tissues of PC patients and animal models, while macrophage depletion can significantly alleviate the lipolytic phenotype (Molfino et al. 2023, Lu et al. 2020). These findings suggested the pivotal role of macrophages in PC‐induced lipolysis. Consistent with these findings, our study (Figure 1F) also observed pronounced macrophage infiltration in the adipose tissue of PC patients. In order to investigate the contribution of macrophages to PC‐induced lipolysis, we performed macrophage depletion experiments in PC mouse model. Treatment with clodronate liposomes effectively reduced macrophage presence in adipose tissue (Figure S7A). Notably, after macrophage depletion, the mice exhibited increased adipose tissue mass (Figure S7B and C), along with elevated TG content (Figure S7D), and a significant reduction in IL‐6 and TNF‐α levels (Figure S7E). These findings confirmed the essential role of macrophages in PC‐induced lipolysis.

Since macrophages are key phagocytic cells responsible for identifying and internalising bacterial effectors, we hypothesised that E. rectale EVs might improve PC‐associated lipolysis by modulating macrophage function. To evaluate this hypothesis, we administered E. rectale EVs to cachexia models and observed a significant reduction in macrophage infiltration and levels of lipolytic cytokines in adipose tissue after EV treatment (Figure 5A and B). However, in macrophage‐depleted mice, E. rectale EVs failed to further ameliorate the lipolytic phenotype (Figure S7A–E), suggesting that the effects of E. rectale EVs are dependent on macrophages.

FIGURE 5.

E. rectale EVs alleviated CC‐induced lipolysis by inhibiting M1 macrophage polarization. (A–B) PC mice received E. rectale EVs, vehicle and mock EVs via oral gavage, and macrophage‐related parameters were assessed: IHC staining for the macrophage marker F4/80 in eWAT (A) and levels of lipolytic cytokines IL‐6 and TNF‐α (B, n = 6). Scale bars, 100 µm. (C) Uptake of E. rectale EVs by RAW264.7 cells. Confocal fluorescence images showing E. rectale EVs uptake by RAW264.7 cells, with or without inhibition of major endocytic pathways using Genistein (caveolae‐dependent inhibitor) and CPZ (clathrin‐mediated inhibitor). Scale bars, 20 µm. (D) Flow cytometry to quantitatively validate EV uptake by macrophages. (E) Immunofluorescent staining was performed to detect M1 macrophage marker CD86 in eWAT. Scale bars, 50 µm (F) Immunofluorescence assessment was performed to detect M1 marker CD86 and M2 marker CD206 in RAW264.7 cells. Scale bars, 50 µm (G) Flow Cytometry Assessment of M1 macrophage and M2 macrophage in RAW264.7 cells. (H) ELISA results of lipolytic cytokines IL‐6 and TNF‐α. (I) Representative Oil Red O and BODIPY staining images showing lipid accumulation and lipolysis in 3T3‐L1 adipocytes after co‐culture with RAW264.7 cells supernatant. Scale bars, 100 µm. One‐way ANOVA was used to calculate p values in (B) and (H). * p < 0.05, ** p < 0.01, *** p < 0.001.

To further verify the direct interaction between E. rectale EVs and macrophages, we cultured RAW264.7 cells in vitro and labelled E. rectale EVs with PKH26 to trace their internalisation. We found that macrophages could internalise E. rectale EVs (Figure 5C). To explore the endocytic mechanism involved, we treated the macrophages with inhibitors of the two endocytosis pathways, including genistein (which blocks caveolae‐mediated endocytosis) and chlorpromazine (CPZ, which inhibits clathrin‐mediated endocytosis). Both inhibitors significantly reduced the uptake of E. rectale EVs by macrophages (Figure 5C). In addition, flow cytometry was used to quantitatively assess the uptake of EVs by macrophages. As shown in Figure 5D, the results demonstrated that E.re EVs were readily internalised by macrophages. Notably, treatment with the endocytosis inhibitors Genistein and CPZ significantly reduced EVs uptake. These findings suggested that EVs internalisation occurs via caveolae or clathrin‐mediated endocytosis.

Macrophages primarily mediate pro‐inflammatory and anti‐inflammatory effects through M1 and M2 polarization. Previous studies have reported an accumulation of M1 macrophages in the eWAT of PC models (Han et al. 2022). In our study, ELISA results showed that the levels of IL‐6 and TNF‐α in PC mice were significantly elevated, and E. rectale EVs treatment markedly reduced these cytokine levels (Figure 5B). Given that IL‐6 and TNF‐α are key cytokines derived by M1 macrophages, we hypothesised that E. rectale EVs might exert their effects by suppressing M1 macrophage polarization. Immunofluorescence results revealed that the expression of M1 macrophage marker CD86 was significantly increased in the eWAT of PC mice, while E. rectale EVs treatment notably inhibited the expression of these markers (Figure 5E). And M2 macrophage markers CD206 were decreased in the eWAT of PC mice, while E. rectale EVs treatment improved the expression (Figure S7F). However, treatment with mock EVs did not exert a significant ameliorative effect. To further validate the impact of E. rectale EVs on macrophage polarization, we first treated RAW264.7 cells with LPS to derive polarization into M1 phenotype. LPS treatment resulted in a marked upregulation of CD86+, both of which were notably suppressed by E. rectale EV treatment (Figure 5F). It is also shown that LPS treatment reduced the expression of the M2 marker CD206, whereas EVs treatment increased the expression. Flow cytometry analysis further confirmed the effects of EVs on the expression of the M1 and M2 markers (Figure 5G). Next, we examined the expression of the lipolytic factors IL‐6 and TNF‐α, which also followed a similar trend. (Figure 5H). To further validate this finding, we also conducted experiments in THP‐1 cells. THP‐1 cells were first treated with PMA to induce differentiation into the M0 phenotype, followed by stimulation with LPS and IFN‐γ to promote M1 polarization, during which the effects of EVs were also evaluated. We found that stimulation with LPS and IFN‐γ significantly upregulated the expression of the M1 marker CD86, while markedly downregulating the M2 marker CD206. Intervention with E.re EVs could ameliorate this effect, whereas mock EVs had no significant impact on THP‐1 cells (Figure S7G–H). The expression of lipolytic factors showed a similar pattern (Figure S7I). In addition, to exclude the potential effects of non‐vesicular components, such as soluble proteins or cell wall fragments, we treated macrophages with the supernatant obtained during the EVs isolation process and examined whether it could influence macrophage polarization. The results (Figure S7J) showed that treatment with the supernatant did not affect macrophage polarization. To investigate the relationship between macrophage polarization and lipolysis, we induced differentiation of 3T3‐L1 adipocytes and co‐treated them with supernatant from RAW264.7 cells. The supernatant from M1 RAW264.7 cells significantly reduced lipid accumulation in 3T3‐L1 cells and promoted lipid droplet breakdown, whereas the supernatant from E. rectale EVs treated M1 RAW264.7 cells alleviated this phenomenon (Figure 5I). The expression levels of lipolysis‐related proteins also followed the similar trend (Figure S7K). To determine whether E. rectale EVs can directly affect adipocytes, we treated 3T3‐L1 cells with E. rectale EVs. The results showed that direct treatment of 3T3‐L1 cells with E. rectale EV failed to alleviate the lipolytic phenotype (Figure S7L–M), suggesting that E. rectale EV exert their anti‐lipolytic effects primarily through the modulation of macrophage function.

2.6. E. rectale EVs Regulate Transcriptional Remodelling to Modulate the Pro‐Inflammatory Macrophage Phenotype

To investigate the mechanisms by which E. rectale EVs modulate and attenuate the polarization of pro‐inflammatory M1 macrophages, we performed transcriptomic sequencing analysis on three groups of in vitro‐cultured macrophages. 2D Principal Component Analysis (PCA) revealed distinct separation in gene expression profiles among the three groups (Figure 6A), and hierarchical clustering heat maps further confirmed the transcriptional differences (Figure 6B).

FIGURE 6.

E. rectale EVs regulate transcriptional remodeling to modulate the pro‐inflammatory macrophage phenotype. (A) 2D principal component analysis (PCA) of gene expression among the Control group, LPS group, and LPS+EVs group (n = 3). (B) Heatmap of differentially expressed genes (DGEs) identified in RAW264.7 cells of three groups. (C–E) Transcriptomic changes in RAW264.7 cells upon LPS stimulation. Volcano plot illustrating upregulated and downregulated genes (C), bubble plot for GO analysis (D) and KEGG pathway analysis (E) of DEGs in LPS treated RAW264.7 cells. (F–H) Transcriptomic changes in RAW264.7 cells after EVs treatment. Volcano plot illustrating upregulated and downregulated genes (F), bubble plot for GO analysis (G) and KEGG pathway analysis (H) of DEGs after EV treatment in RAW264.7 cells. (I) Venn diagram displaying the overlap between the genes upregulated by LPS treatment and genes downregulated after EVs intervention. (J) GO and KEGG pathway analyses of the overlapping genes to identify key biological processes and pathways influenced by EVs treatment.

Comparative analysis between the control and LPS‐treated group identified 1,808 significantly upregulated genes (Figure 6C). GO analysis indicated significant enrichment of immune processes among the DEGs, including inflammatory responses, immune cell activation, and cytokine production (Figure 6D). KEGG pathway analysis indicated that LPS treatment significantly activated pro‐inflammatory pathways such as the TNF signalling pathway and the NOD‐like receptor signalling pathway, confirming that LPS effectively induces macrophage polarization toward the pro‐inflammatory phenotype (Figure 6E).

Subsequent transcriptomic analysis of EVs‐treated macrophages revealed 487 significantly downregulated genes (Figure 6F). GO analysis showed that these genes were primarily involved in macrophage responses to bacterial products, immune regulation, and inflammatory responses. Additionally, E. rectale EVs significantly affected intracellular membrane responses and vesicular transport‐related pathways in macrophages (Figure 6G). KEGG analysis further indicated that E. rectale EVs treatment markedly suppressed the TNF signalling pathway, signal transduction pathways, and cytokine‐receptor interaction pathways (Figure 6H).

To further elucidate the regulatory effects of EVs on macrophage polarization, we identified the intersection between genes upregulated in the LPS‐treated group and those downregulated in the EVs‐treated group. The Venn diagram yielded 162 overlapping genes (Figure 6I), which were primarily enriched in the NF‐kB signalling pathway and macrophage responses to bacterial‐derived products (Figure 6J). These results suggested that after being internalised by macrophages, E. rectale EVs may modulate NF‐kB signalling pathway to alter macrophage polarization, thereby exerting immunoregulatory effects.

2.7. E. re EVs Suppress Macrophage Polarization by Modulating the NF‐κB Signalling Pathway

The regulatory role of the NF‐kB signalling pathway in M1 macrophage polarization is well‐documented, with nuclear translocation of NF‐kB and its binding to DNA being critical steps in its activation (Peng et al. 2025). To investigate whether E.re EVs inhibit macrophage polarization by targeting the NF‐kB pathway, we assessed the expression of key genes within this pathway in macrophages. Our results showed that LPS treatment significantly upregulated the expression of NF‐kB‐related genes (nfkb1, nfkb2, nfkbia, nfkbiz, and rela), whereas E.re EV intervention markedly downregulated expression levels (Figure 7A). Additionally, the expression of key NF‐kB signalling proteins, phosphorylated p65 (p‐p65) and phosphorylated IkBα (p‐IkBα), was significantly decreased following EVs treatment, indicating that E.re EVs effectively suppress NF‐kB pathway activation (Figure 7B and C). Consistent with previous studies, our further experiments revealed that LPS treatment induced substantial nuclear translocation of p65 in macrophages, while E.re EVs intervention partially mitigated this effect (Figure 7D–F). Given that suppression of NF‐κB signalling and macrophage polarization toward an anti‐inflammatory phenotype can also arise as a consequence of efferocytosis, we further examined the expression of key genes and proteins involved in phosphatidylserine‐dependent efferocytosis pathways to distinguish E.re EV‐mediated signalling effects from those associated with efferocytosis‐related mechanisms. First, we examined whether E.re EVs treatment induces apoptosis in target cells, which could secondarily trigger efferocytosis. After EV treatment, TUNEL staining was performed to test whether E.re EVs intervention induces apoptosis in target cells. As shown in Figure S8A, there was no significant difference between the EV‐treated group and LPS group, indicating that E.re EVs treatment does not affect cell apoptosis under our experimental conditions. Next, we assessed key components of phosphatidylserine‐dependent efferocytosis pathways. Specifically, we evaluated the expression of representative efferocytosis‐related protiens and genes, including MerTK and TIM4 at the protein level, as well as tim4, mertk, s1pr1, and axl at the gene level. As shown in Figure S8B–C, we found EVs treatment did not significantly alter the expression of these efferocytosis‐associated markers, which means that E.re EVs suppress NF‐κB signalling and promote anti‐inflammatory macrophage polarization independently of efferocytosis. These findings imply mechanistic insights into how E.re EVs modulate macrophage function and attenuate cachexia‐associated lipolysis.

FIGURE 7.

E.re EVs suppress macrophage polarization by modulating the NF‐kB signalling pathway. (A) Relative mRNA expression levels of key NF‐kB genes in RAW264.7 cells after treatment with LPS and E. re EVs (n = 3). (B and C) Western blot images (B) and quantification (C) showing protein expression levels of p‐p65 and p‐IkBα in RAW264.7 cells. (D–E) Western blot images showing protein levels of p65 in both cytoplasm and nuclear fractions. GAPDH served as internal control in cytosolic fractions, while Lamin B1 served as internal control in nuclear fractions. Quantification of p65 was performed for cytoplasm and nucleus. (F) Immunofluorescence images showing that E. re EVs inhibits LPS induced nuclear translocation of p65 in RAW264.7 cells. Scale bars, 10 µm. One‐way ANOVA was used to calculate p values in (A), (C), and (E). * p < 0.05, ** p < 0.01, *** p < 0.001.

2.8. E. rectale EVs Alleviate Lipolysis Induced by Colorectal and Gastric CC

To determine whether E. rectale EVs exert a broad protective effect against CC‐induced lipolysis across different cancer models, we established mouse models of colorectal cancer (CRC) (Thibaut et al. 2025) and gastric cancer (GC) cachexia (Cui et al. 2019). Compared to the control group, tumor‐bearing mice in both models exhibited significant weight loss (Figure S9A and J) and a reduction in body fat composition (Figure S9B and K). Additionally, elevated FFA levels (Figure S9C and L), decreased WAT mass (Figure S9D and M), and reduced TG content (Figure S9E and N) collectively indicated the presence of CC induced lipolysis in both models. Consistent with observations in PC model, IHC and immunofluorescence images revealed extensive macrophage infiltration in eWAT, with a significant increase in CD86+ M1 macrophages in both models. Notably, this inflammatory response was significantly suppressed after the treatment of E. rectale EVs (Figure S9F and O). The expression of lipolysis‐related proteins also showed similar trends (Figure S9G–H and P–Q). However, treatment with E. rectale EV significantly ameliorated all these lipolysis effects. Further ELISA assay showed that levels of pro‐inflammatory cytokines were markedly elevated in CRC and GC cachexia models, but their levels significantly decreased following E. rectale EVs intervention (Figure S9I and R). These results indicated that E. rectale EVs not only mitigate lipolysis in PC model but also suppress lipolysis in other CC models. The underlying mechanism involves suppression of M1 macrophage polarization in adipose tissue and downregulation of IL‐6 and TNF‐α expression, ultimately alleviating CC induced lipolysis.

3. Materials and Methods

3.1. Study Population

Between 2022 and 2023, we enrolled patients diagnosed with PDAC from the general surgery department at Jinling Hospital. All eligible subjects completed and signed informed consent forms. To be included in the study, participants had to be at least 18 years old and able to provide informed consent. Those excluded from the study included individuals with chronic or acute conditions associated with malnutrition, such as chronic heart failure, liver cirrhosis, chronic kidney disease, or infections, as well as those with cognitive impairments, difficulty swallowing, or gastrointestinal obstructions. Subjects lacking clinical examination and CT scan data will also be excluded. Patients with recent use of antibiotics or glucocorticoids will also be excluded.

Fresh fecal samples and serum from both patients and controls were collected in the morning on the second day after admission to the hospital, while SAT was obtained during the operation. Blood samples were collected after an overnight fast, and then hematological and other biochemical indicators were analysed using a 7600 fully automated biochemical analyzer (Hitachi). The remaining samples were stored at ‐80°C. For fecal samples, the researchers will transfer the fecal samples to the laboratory within 4 h. Approximately 0.5 mL of fecal sample will be added to 4.5 mL of buffer, which is PBS containing 0.1% cysteine and 0.3% riboflavin, and then homogenized. The mixture will be left to stand for 3–5 min to precipitate insoluble particles. The supernatant will then be mixed with an equal volume of storage solution (the buffer mentioned above and containing 30% glycerol and 10% sucrose) for storage. The samples will be stored at ‐80°C until transplantation.

For CT images, the third lumbar vertebrae SAT and VAT (visceral adipose tissue) and their areas were segmented and calculated through HY‐QCT medical software version V2.5.0. and TomoVision SliceOmatic software. This study adhered to the ethical guidelines of the 1975 Declaration of Helsinki and received approval from the Ethics Committee of Nanjing Jinling Hospital (2023DZKY‐049‐03). It was also registered at ClinicalTrials.gov (NCT06378853), and written informed consent was obtained from the study participants. The baseline characteristics of the study population are presented in Table S1.

3.2. Diagnosis of Cachexia and Nutritional Status Assessment

Cancer patients were categorized as cachectic or non‐cachectic according to international criteria. (A) Weight loss of more than 5% within 6 months (in the absence of simple starvation); (B) BMI < 20 kg/m² and any degree of weight loss >2%; (C) Loss of skeletal muscle mass, with concomitant weight loss of more than 2%. (Fearon et al. 2011) The skeletal muscle area (cm²) and adipose tissue area (cm²) at the level of the third lumbar vertebra were quantified using CT and the FMI was quantified by InBody. The nutritional status of subjects was assessed by PG‐SGA.

3.3. Body Composition Assessment

Human body composition analysis was conducted on the same morning as the blood tests. We performed the analysis using the S10 bioelectrical impedance analyzer (Inbody). The body mass index (BMI) and FMI were then calculated using formulas. (Gagnon et al. 2024)

Mouse body composition was assessed using the MiniQMR animal body analyzer (Small Animal MRI System‐Permanent magnet MRI NM21‐060H‐I, NIUMAG) and Ultra Focus DXA (Faxitron, USA) to evaluate the body composition of the mice. All tests were performed in a blinded manner.

3.4. Mice

Male C57BL/6 and 615‐line mice aged 6–8 weeks were purchased from Gem Pharmatech (Nanjing, China) and Chinese Academy of Medical Sciences (Tianjin, China), respectively. Mice were maintained in temperature‐controlled chambers under specific pathogen‐free (SPF) conditions. The mice were given standard rodent chow diet (MDIWC‐001, Suzhou Modern Animal Feed Co., Ltd), water and fed ad libitum, and maintained on a 12‐h light/dark cycle. Before the experiment began, all mice underwent a one‐week adaptation period for feeding. To avoid the cage effect, mice were housed individually in separate cages, with drinking water and bedding changed regularly. All experiments and protocols were approved by the Institutional Animal Care and Use Committee of Nanjing University.

3.5. Cell Culture

Mouse cell lines Pan02, MC38, MFC,3T3‐L1, RAW264.7 and THP‐1 were purchased from the Shanghai Zhong Qiao Xin Zhou Biotechnology Co., Ltd., American Type Culture Collection (ATCC) and Pricella Life Science & Technology Co., Ltd. All cell lines were authenticated by short tandem repeat analysis and ensured to be free of mycoplasma infection.

The Pan02 and MFC cells were cultured in high‐glucose DMEM supplemented with 10 % fetal bovine serum (A5670801, FBS, Gibco) and 1 % penicillin—streptomycin (PS, ST488S, Beyotime). The MC38 cells were cultured in RPMI 1640 medium supplemented with 10 % FBS and 1% PS. Once the cells reached 80–90 % confluence, they were collected for the cachexia model establishment.

3T3‐L1 cells were grown in DMEM supplemented with FBS and 1 % PS solution. The point at which the cells achieved confluence was defined as day ‐2. On day 0, two days post‐confluence, the 3T3‐L1 cells were prompted to differentiate by being exposed to DMEM enriched with 10% FBS, alongside 1 mg/ml of insulin (I6634, Sigma‐Aldrich), 0.5 mM of 1‐methyl‐3‐isobutylxanthine (IBMX) (I5879, Sigma‐Aldrich), and 0.25 mM of dexamethasone (D4902, Sigma‐Aldrich) until day 2. Following this initial period, the cells were cultured in DMEM with 10% FBS and 1 mg/ml of insulin for another two days. Afterward, 3T3‐L1 cells were continuously maintained in the DMEM containing 10% FBS, with the medium being refreshed every other day.

RAW264.7 macrophage cells utilised DMEM with 10% FBS and 1% PS. Upon reaching 70% confluence, 100 ng/ml LPS was added to the culture medium in order to induce pro‐inflammatory phenotype. Then collect the supernatant to co‐culture with 3T3‐L1 cells.

The human monocytic cell line THP‐1 utilised 1640 with 10% FBS and 1% PS. THP‐1 was induced to adhere by employing a final concentration of 100 ng/ml phorbol 12‐myristate 13‐acetate (PMA). After 24 h of induction, the cells adhered to become primary macrophages (M0). 20 ng/ml IFN‐γ and 100 ng/ml LPS were added to the culture medium. After 24 h, the cells were induced to a pro‐inflammatory phenotype.

3.6. Cancer Cachectic Mouse Model

For the CC model, three mouse models of PC, GC and colon cancer were established. C57BL/6 mice were used for pancreatic and colon cancer models, and 615‐line mice were used for GC models.

All mice were anesthetised with 2% isoflurane. A transverse incision approximately 5 mm long was made in the right upper abdomen of the mouse, we carefully exposed the pancreas. For PC model, using a micro‐syringe, 50 µL of single‐cell suspension containing 1 × 10⁶ Pan02 tumor cells were slowly injected under the subserosa, and translucent vesicles were seen to bulge (Greco et al. 2015). For GC model, a single cell suspension containing 2 × 10⁶ MFC cells in a volume of 200 µL was carefully subcutaneously injected into right flank of the 615‐line mice. (Cui et al. 2019) For the colon cancer model, we administered a subcutaneous injection of 200 µL of MC38 cells (1.0 × 10⁶) into the right flank of male C57BL/6 mice. The sham group was injected with an equal volume of PBS (Thibaut et al. 2025).

3.7. Detection of E. rectale Abundance

The qPCR was used to detect the abundance of E.rectale. The primers for the E.rectale used in this procedure are “Forward primer: AAGGGAAGCAAAGCTGTGAA; Reverse primer: TCGGTTAGGTCACTGGCTTC”. The results of E. rectale abundance were expressed as the average gene copy number / gram of feces (copy/g).

3.8. Grip Strength Measurement

To assess muscle strength, the maximum grip strength of the mice was measured thrice using a force meter (KW‐ZL, NJKEWBIO, Nanjing, China), and the mean maximum force was used for the analysis. All tests were performed in a blinded manner.

3.9. Fecal Microbiota Transplantation (FMT)

Before FMT, all mice received a one‐week antibiotic cocktail (ABX) in their drinking water to eliminate gut microbiota. The ABX consisted of vancomycin (0.5 g/L), metronidazole (1 g/L), ampicillin (1 g/L), and neomycin (1 g/L). Solutions and bottles were changed 2 times a week. Fresh stool samples were collected followed the procedure described above. The fecal suspension with an equivalent of 20 mg of human feces was transplanted into mice by oral gavage for 5 weeks. Stool samples were fully suspended in PBS in a ratio of 1:10 (w/v). Before gavage, the solid particulates were removed by centrifugation; the supernatant should be a homogenized solution containing fecal microbiota. All mice were fed a standard diet and were kept in sterilized cages and handled in sterilized hoods using sterile single‐use sterile protective gear and coating. Each recipient mouse received fecal microbiota from a single donor. In order to prevent spread of microbes between mice from different donors, we employed ventilated racks and appropriate barrier husbandry practices for cage changing, sanitation, and animal handling.

3.10. Bacterial Strains and Growth Conditions

Eubacterium rectale (ATCC 33656) was purchased from ATCC and cultivated anaerobically in modified GAM medium enriched with vitamin K1 and L‐cysteine. E. coli ATCC25922 was maintained in LB medium at 37°C. Cultures were incubated with shaking at 170 rpm under anaerobic conditions. Bacterial density was determined by measuring the optical density at 600 nm or by colony‐forming unit (CFU) enumeration after plating on agar supplemented accordingly.

3.11. LIVE/DEAD Bacterial Staining With DMAO & PI

Culture bacteria until reaching the logarithmic growth phase. Then take bacterial suspension and centrifuge at 10,000×g for 5 min at room temperature, discard the supernatant, wash once with the Stroke‐physiological Saline Solution, and then resuspend the bacteria to approximately 10^8 bacteria/mL (OD670≈0.3) with the Stroke‐physiological Saline Solution. Staining: Add 1 µL of Staining Solution (100X) per 100 µL of bacterial resuspension. After mixing well, incubate at 37°C in the dark for 15 min. After incubation, place 10 µL of bacterial suspension on a Microscope Slide, cover it with a 24 mm square Coverslip, and observe under a fluorescence microscope. DMAO emits green fluorescence (Ex/Em = 503/530 nm) and PI emits red fluorescence (Ex/Em = 535/617 nm).

3.12. Microbial Cell Viability Assay

Equilibrate the multi‐well plate containing microbial samples to room temperature. Add 100 µL of BacTiter‐Lumi Assay Reagent (Beyotime, No. C0052S) to each well of 96‐well plates. Shake the plate at room temperature for 5–25 min to stabilize the luminescence signal. Measure the luminescence in a compatible luminometer with appropriate parameters. Use an integration time of 0.25‐1 s per well as a guideline, which can be adjusted according to the sensitivity of the instrument. Calculate the relative microbial viability based on the luminescence readings or the ATP content derived from an ATP standard curve.

3.13. Medication Administration

Before establishing of PC model, mice were orally administered with 1.0 × 10⁹ CFU of bacteria three times per week for one week, followed by a total duration of five weeks. The suspension was sealed with a rubber stopper and aliquoted into anaerobic culture bags, along with an anaerobic gas‐generating pack and an indicator, to ensure the maintenance of anaerobic conditions. To investigate whether EV secretion is required for the E.rectale‐induced regulation of lipolysis, the PC mice were orally administered with GW4869‐pretreated E.rectale or DMSO‐pretreated E.rectale. To evaluate the effects of EVs on cachexia induced lipolysis, the PC mice were orally administered with 1.2 × 10¹⁰ vesicles of E.rectale EVs.

3.14. 16S rRNA Sequencing

Microbiome DNA isolation and 16S rDNA gene sequencing were conducted with the assistance of KAITAI‐BIO Co., Ltd. Microbial DNA was extracted from mouse feces using a Magnetic Soil and Stool DNA Kit (TianGen, China) and purified using electrophoresis and NanoDrop 2000. Briefly, PCR amplification targeted the entire 16S rRNA gene using specific primers 27F (5’‐AGRGTTYGATYMTGGCTCAG‐3’) and 1492R (5’‐RGY TACCTTGTTACGACTT‐3’) primers, followed by library construction and sequencing. The raw subreads were first corrected to obtain Circular Consensus Sequencing (CCS) reads using SMRT Link (version 8.0). Subsequently, CCS reads from different samples were identified and demultiplexed based on barcode sequences using Lima (v1.7.0), and chimeric sequences were removed to yield high‐quality CCS reads. The specific parameters used were as follows: (A) CCS Read Generation: In full‐length isoform sequencing, the size of the target DNA fragments is much smaller than the read length. As a result, the same fragment is sequenced multiple times (passes). Circular Consensus Sequencing (CCS) reads are generated by self‐correction of subreads derived from a single zero‐mode waveguide (ZMW). When the number of passes exceeds 4, the accuracy of the CCS reads typically exceeds 99%. CCS reads were identified using SMRT Link v8.0 with the following parameters: minPasses ≥ 5 and minPredictedAccuracy ≥ 0.9. (B) Barcode and Primer Identification: CCS reads were demultiplexed using lima v1.7.0 with default parameters to distinguish different samples based on barcode sequences. Subsequently, cutadapt v2.7 was used to identify both forward and reverse primers, allowing a maximum error rate of 20%. CCS reads lacking both primers were discarded. Finally, reads were filtered based on length thresholds: sequences not meeting the following size criteria were removed. (C) Chimera Removal: Chimeric sequences were identified and removed using UCHIME v4.2, resulting in high‐quality Effective‐CCS reads. Sequences were clustered into operational taxonomic units (OTUs) at a 97% similarity threshold using USEARCH (version 10.0). A default threshold of 0.005% of the total number of sequences was used to filter out low‐abundance OTUs, followed by species annotation and abundance analysis using the SILVA138 database. Alpha diversity was assessed using Chao1 and Shannon indices, while beta diversity was analyzed using principal coordinate analysis (PCoA) based on the Bray‐Curtis distance matrix. The evolutionary branching diagram was constructed using the LEfSe software.

3.15. Proteomic Analysis of EVs by 4D Label‐Free Mass Spectrometry

To identify the protein components of purified EV, the protein samples of EV were prepared for analysis performed mindlessly by Katimesbio (Hangzhou, China). (A) To extract and digest proteins, samples were lysed in SDT buffer, which contained 4% SDS and 100 mM Tris‐HCl at pH 7.6. Protein concentration was assessed using the BCA Protein Assay Kit sourced from Bio‐Rad (USA). Each sample, weighing 20 µg, was mixed with a 5X loading buffer, heated for 5 min, and then subjected to separation through a 4%–20% SDS‐PAGE gel at 180 V for 45 min, followed by staining with Coomassie Blue. (B) For LC‐MS/MS analysis, peptides were assessed using a timsTOF Pro mass spectrometer from Bruker, integrated with a Nano Elute system. Samples were introduced into a C18‐reversed phase column (Thermo) and eluted through a gradient of buffer B, consisting of 99.9% acetonitrile and 0.1% formic acid, at a flow rate of 300 nL/min. The mass spectrometer was operated in positive ion mode with an electrospray voltage of 1.5 kV. During analysis, the timsTOF Pro was utilised in parallel accumulation serial fragmentation (PASEF) mode, collecting data under specific conditions: the ion mobility coefficient (1/K0) was adjusted to a range of 0.6 to 1.6 versus cm², with one MS scan and ten MS/MS PASEF scans. Active exclusion was activated, allowing for a release time of 24 s. (C) The raw MS data were processed using MaxQuant 1.6.14 to facilitate the identification and quantification of proteins.

3.16. RNA Sequencing

Total RNA was extracted from each group (n = 3) using TRIzol reagent (Invitrogen, USA). Following library preparation, samples were sequenced on an Illumina platform (Beijing Allwegene Technology Company Limited) to produce 150 bp paired‐end reads. After read cleanup, the reads were aligned to the mouse genome utilizing STAR. To quantify gene expression, we calculated fragments per kilobase of transcript per million mapped fragments (FPKM). PCA was used to assess correlations between samples within each group. Differentially expressed gene (DEG) analysis was then carried out using the DEGSeq R package. DEGs were identified based on the following thresholds derived from FPKM values: an absolute log2 fold change greater than 1 (|log2 fold change| > 1), or less than negative one (|log2 fold change| < ‐1) with a P‐value and adjusted P‐value less than 0.05 (P and P adjust < 0.05). Finally, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed to pinpoint the pathways linked to these DEGs.

3.17. ORO and BODIPY 493/503 Staining

Following three washes with PBS, the cells were fixed in 4% formaldehyde for half an hour. Next, we stained the cells using an Oil Red O solution (C0158S, Beyotime) for 30 min at room temperature, making sure to wash them three times with water afterward. We then took a look at the staining under a bright‐field microscope. To quantify the dye retention, we removed the staining solution, eluted the retained dye with isopropanol, and measured the optical density (OD) at 510 nm. For BODIPY 493/503 staining (D3922, Thermo Fisher Scientific), fixed cells were incubated with the dye at a concentration of 0.1 mg/mL for 10 min before counterstaining with DAPI (D9542, Sigma Aldrich) and proceeding with imaging.

3.18. Histochemistry and Immunohistochemistry

Tissue was preserved through fixation with paraformaldehyde, subsequently dehydrated using ethanol, and embedded in paraffin. 5 µm thick sections were stained with hematoxylin and eosin to observe tissue structural changes.

For the immunohistochemistry analysis on the paraffin‐embedded WAT sections, antigen retrieval was achieved by heating the samples in a pressure cooker for 20 min using a pH 9.0 EDTA buffer. Once cooled, they underwent treatment with 3% hydrogen peroxide for 20 min to inhibit endogenous peroxidase activity. Following this step, the slides were rinsed with PBS and blocked with 10% BSA for 10 min. The sections were then incubated overnight at 4°C with the primary antibody F4/80. After three washes of 3 min each with PBS, the sections were treated with secondary antibody (PV‐9004, ZSGB‐Bio, China) for one h at room temperature.

3.19. Immunofluorescent Staining

Tissues were fixed in 4% paraformaldehyde, then sectioned into serial slices. The slides were deparaffinized with xylene and rehydrated using an ascending ethanol series. Antigen retrieval was performed by incubating the sections in citrate buffer (0.01 M, pH 6.0, 0.05% Tween‐20) in a steamer at 95°C for 20 min. After three washes with TBS, the slides were permeabilized and blocked with 10% normal goat serum in 0.3% Triton X‐100 PBST (PBS with 0.05% Tween‐20) for 1 h at room temperature. For immunofluorescence staining, tissue sections were incubated with primary antibodies, followed by Alexa Fluor‐conjugated secondary antibodies. Coverslips were mounted using an antifade reagent containing DAPI. The primary antibodies used were: anti‐F4/80, anti‐CD86, anti‐CD206

3.20. Flow Cytometry

To reduce non‐specific staining, anti‐mouse CD16/32 antibodies (BioLegend,101319) were incubated with macrophage samples for 10 min to block Fc receptors. Next, the anti‐mouse F4/80 (Biolegend, 123121), CD86 antibody (Thermo, 11‐0862‐82) and CD206 (BioLegend, 141705) were added and the cells were incubated for 30 min. Samples were processed using a BD LSRFortessa flow cytometer and analyzed by FlowJo software.

3.21. Macrophage Depletion Assay in Vivo

In the in vivo experiment involving macrophage depletion, PC mice were divided into two groups following tumor inoculation. Over the course of a week, each group received intraperitoneal injections of 100 µL of either clodronate‐liposomes (F70101C, FormuMax) or control liposomes (F70101‐N, FormuMax) every 2 days.

3.22. Free Fatty Acid Assay

The serum free fatty acid levels in serum were determined through the Free Fatty Acid Assay Kit (A042‐2‐1, Nanjing Jiancheng Bioengineering Institute), with all steps strictly adhering to the manufacturer's protocol.

3.23. Tissue TG Measurement

Adipose tissue TG was assayed using a commercial kit (A110‐1‐1, Nanjing Jiancheng Bioengineering Institute) in accordance with the manufacturer's protocol. Adipose tissue TG levels are expressed as mmol/mg protein.

3.24. Purification and Characterization of Bacterial EVs

After 4 days of anaerobic cultivation of E. rectale in GAM medium, the culture supernatant was harvested, centrifuged at 6000 × g for 30 min at 4°C, and passed through a 0.22 µm filter to remove the residual bacteria. The bacteria‐free supernatant was transferred to Amicon Ultra‐15 Centrifugal Filter Units and then concentrated by centrifugation at 4000 x g at 4°C. EVs were subsequently isolated from ultrafiltrate through Optiprep density gradient centrifugation. In brief, EVs were combined with Optiprep solution to generate a 50% density layer at the bottom in a polyallomer Beckman Coulter tube. Sequential layers consisting of 8 mL of 40%, 8 mL of 20%, 7 mL of 10% Optiprep, and finally 2 mL of PBS were layered on top of that 50% cushion. Next, the gradient fractions were then centrifuged at 100,000 x g for 18 h at 4°C using a Beckman Optima XPN ultracentrifuge. Fractions of 2 mL with the bottom fraction were collected and the EV‐enriched layers were pooled, diluted to 30 mL with PBS, and then centrifuged again at 100,000 x g for 3 h at 4°C. The resulting EVs pellet was either used immediately or stored at −80°C. The protein concentration of EVs was determined by BCA protein assay kit. The number of EVs were quantified by NTA. The morphologies of EVs were detected by TEM. (Liu et al. 2021)

3.25. Inhibition of EVs Secretion

To confirm whether GW4869 (52321ES08, YEASEN) is capable of inhibiting EVs secretion by E.rectale, E.rectale (10⁷ CFUs per 100 mL medium) was cultured in complete medium containing GW4869 (10×10⁻⁶ M) or an equivalent volume of vehicle control (DMSO). Following four days, viability of E. rectale was estimated by measuring the light absorption at a wavelength of OD600 to assess the bacterial density and evaluate the relative growth status of the bacteria. By Optiprep density gradient centrifugation, E. rectale EVs were isolated from conditioned medium. The E.rectale pellets were resuspended in fresh complete medium and continued to grow for another 4 days. The conditioned media were obtained for the isolation of E.rectale EVs, in order to evaluate the lasting inhibitory effects of GW4869 on EVs secretion. To determine if GW4869 inhibited the secretion of EVs, the protein content and particle number of E. rectale EVs were detected by BCA protein assay and NTA respectively.

3.26. Transmission Electron Microscopy (TEM) and Nanoparticle Tracking Analysis (NTA)

TEM negative staining was first performed by adding the EVs samples onto a carbon‐coated copper grid for a min. Next, we stained them using a 2% uranyl acetate solution, again for one min. EVs were imaged later using 120 kV FEI Tecnai G2 spirit electron microscope. To analyse the size distribution of EVs, NTA was performed using a NanoSight NS300 instrument (Malvern, UK).

3.27. In Vivo Tissue Distribution of EVs

The EVs of E. rectale were extracted from the culture medium and labelled using Dil (YEASEN, 41085‐99‐8) adhering to the manufacturer's protocol. Optiprep gradient centrifugation was used to remove redundant fluorescent dyes. The EVs were diluted using PBS and centrifuged at 100 000 ×g and 4°C for 3 h. Labeled EVs were resuspended in PBS and administered by gavage. Mice were given 100 uL PBS with or without 1.2 × 10¹⁰ vesicles of EVs. A control group received the same amount of the vehicle alone. At various points in time, the mice were euthanized, and major organs, including the heart, liver, spleen, lungs, kidneys, and WAT, were collected. We then used an IVIS spectrum imaging system (PerkinElmer) to take pictures of the organs. The intensity of the Dil dye was quantified using Living Image 3.1 Software.

3.28. Cellular Uptake of EVs Assay

The EVs were tagged with lipophilic membrane dye PKH26. After removing the redundant dye, the labelled EVs (6 × 10⁸ vesicles mL⁻¹) were incubated with RAW264.7 cells at 37°C. After washing with PBS, the cells were fixed with 4% paraformaldehyde, and the nuclei were stained with DAPI. The samples were subsequently analysed using a confocal fluorescence microscope (Olympus).

To investigate the pathway of EVs uptake, cells were pretreated with various pharmacological or chemical inhibitors prior to the addition. Genistein (G106673, Aladdin, Shanghai, China) and Chlorpromazine hydrochloride (CPZ, C407970, Aladdin, Shanghai, China) are major inhibitors of the macrophage phagocytic pathways. At 30 min before EVs treatment, 200 µM genistein or 50 µM CPZ was added to medium at 37°C and was present throughout the experiment.

3.29. Elisa

Inflammatory cytokines IL‐6 (E‐EL‐M0044, Elabscience) and TNF‐α (E‐EL‐M3063, Elabscience) in mice were quantified with ELISA kits.

3.30. Quantitative Real‐Time PCR Analysis

Total RNA was extracted in accordance with the manufacturer's guidelines using TRIzol Reagent (Qiagen, Hannover). The concentration of the RNA was quantified using a NanoDrop instrument provided by Agilent Technologies. To synthesize single‐stranded cDNA, HiScript III RT SuperMix for qPCR (R323‐01, Vazyme, China) was utilized. For relative quantification, the process involved SYBR Green Master Mix (4309155, Thermo Fisher Scientific, USA) along with the ABI SteponePlus PCR system (Thermo Fisher Scientific, USA). The changes in expression levels of the target genes were computed using the 2^{− ΔΔCT} method. The primers that were used for this study were produced by Generay Biotech and can be found in Table S2.

3.31. Western Blotting

Total proteins were extracted from target tissues and cells utilizing RIPA lysis buffer supplemented with phosphatase and protease inhibitors. Their concentrations were assessed using a BCA assay kit. Subsequently, the proteins underwent separation via SDS‐PAGE and were transferred onto PVDF membranes. To prevent non‐specific binding, the membranes were blocked with a 5% skim milk solution for 1.5 h at room temperature, followed by an overnight incubation with specific antibodies at 4°C. After washing, the incubation was continued for 1 h using the corresponding secondary antibody and then the bands were visualized using the ECL detection kit and Molecular Imager System (BIO‐RAD, Hercules, USA). The primary antibodies used in the study were as below: HSP90 (1:1000, Cell Signalling Technology, No. 4874S), HSL (1:1000, Abclonal, No. A15686), pHSL‐S660 (1:2000, Abclonal, No. AP1242) and ATGL (1:2000, Abclonal, No. A5126), GAPDH (1:1000, Cell Signalling Technology No. 5174S), Lamin B1 (1:5000, Abcam, No. ab133741), NF‐kB p65 (1:1000, Cell Signalling Technology, No. 8242), p‐p65 (1:1000, Cell Signalling Technology, No. 3033), p‐IkBα (S32/S36) (1:1000, Cell Signalling Technology, No.9246), and IkBα (1:1000, Cell Signalling Technology, No. 4814), ZO‐1 (1:5000, Abcam, No. ab59720), Occludin (1:5000, Abcam, No. ab216327), MUC2 (1:2000, Immunoway, No. YM8469), UCP‐1 (1:5000, Abcam, No. ab10983), PRDM16 (1:5000, Abcam, No. ab106410) and CIDEA (1:5000, Abcam, No. ab8402). Quantification of band density was performed by ImageJ.

3.32. Statistical Analysis

Data were expressed as means ± standard deviation (SD). For data that passed normality tests, the Student t test was used when differences between 2 groups were analysed, and analysis of variance was used when differences between 3 or more groups were compared. For comparisons among multiple groups, a one‐way ANOVA was utilized, followed by the Bonferroni post hoc test. Grouped analysis was conducted using two‐way ANOVA followed by Tukey's multiple comparisons test. Regular analysis, Pearson correlation test and linear regression were analysed with GraphPad Prism 9.5. Statistical significance was deemed acceptable at P < 0.05, with * denoting P < 0.05, ** indicating P < 0.01, and *** for further significance levels.

4. Discussion

CC is a multifaceted metabolic disorder characterized by significant muscle wasting, and marked reduction in adipose tissue, which impairs prognosis and limits treatment efficacy (Fearon et al. 2011). Adipose tissue depletion is mediated by enhanced lipolysis and adipose tissue browning, which collectively promote a systemic and local catabolic state (Rupert et al. 2021; Hu et al. 2023; Petruzzelli et al. 2014). Therefore, there is a pressing demand for innovative therapeutic interventions to alleviate symptoms and improve outcomes in patients with CC. In this study, we found that pancreatic CC patients exhibit more pronounced fat depletion accompanied by inflammatory cell infiltration and adipose tissue browning. Furthermore, studies have shown that patients with lower FMI have higher mortality and worse prognosis(Dunne et al. 2024; Yin et al., 2022; Khan et al. 2023). We also demonstrated that patients with a lower FMI have reduced survival rates, underscoring the pivotal role of adipose tissue in CC progression.

The accumulation of inflammatory cells and the release of pro‐inflammatory cytokines are central to the lipolytic processes in CC. Single‐cell sequencing studies have identified cachexia‐specific immune cell populations, such as CD8⁺ T cells and macrophages, in both subcutaneous and visceral adipose tissue of cachectic patients (Han et al. 2022). These immune cells secrete pro‐inflammatory cytokines that activate lipolysis, highlighting a complex interplay between immune cells and adipocytes during cachexia development. Studies have shown that, in the adipose tissue of gastrointestinal CC patients, adipocytes exhibit reduced cross‐sectional area, increased fibrous deposition, and macrophage infiltration (Molfino et al. 2022). In our study, we also found significant accumulation of pro‐inflammatory immune cells around adipose tissue in several CC models, reinforcing the critical role of immune cells in adipose tissue breakdown during CC.

Among the pro‐inflammatory cytokines implicated in CC, IL‐6 and TNF‐α have been extensively studied for their roles in muscle atrophy and lipolysis. IL‐6 activates lipolysis through the IL‐6/STAT3 signalling pathway in adipocytes, promoting the release of free fatty acids (FFA), which in turn induce mitochondrial dysfunction, oxidative damage, and muscle atrophy. (Rupert et al. 2021, Pototschnig et al. 2023) IL‐6 can also directly act on muscle, activating the JAK/STAT pathway and impairing muscle function. Additionally, elevated IL‐6 levels in the blood are linked to poor prognosis and increased mortality in PDAC (Yang et al. 2024; Holmer et al. 2014). Similarly, TNF‐α disrupts insulin signalling, suppresses adiponectin production, and promotes lipolysis through cAMP‐mediated activation of protein kinase A (PKA) (Yang et al. 2011; Zhang et al. 2002). At the same time, IL‐6 and TNF‐α are also the main cytokines secreted by M1 macrophages (Mantovani et al. 2004). In our study, we observed significantly increased levels of these cytokines in both pancreatic CC patients and mouse models, further supporting their role in adipose tissue wasting. These findings suggest that targeting pro‐inflammatory macrophages, inhibiting M1 macrophage polarization, and reducing the production of IL‐6 and TNF‐α may be effective strategies to ameliorate the progression of CC.

In addition to immune cells and cytokines, increasing evidence indicates that gut microbiota significantly influences cancer and related conditions. Some pathogenic bacteria such as Fusobacterium nucleatum have been implicated in cancer progression (Zepeda‐Rivera et al. 2024; Chen, et al. 2022), while some probiotics like Lactobacillus reuteri and Akkermansia muciniphila exhibit protective effects (Bender et al. 2023; Wu et al. 2025). However, the relationship between gut microbiota dysbiosis and lipolysis in CC has not been fully explored. Our study provides the first systematic investigation into the role of gut microbiota dysbiosis in lipolysis during pancreatic CC, focusing on the potential therapeutic effects of probiotics. Through correlation analysis with clinical indicators, we found that Eubacterium rectale abundance was negatively correlated with pro‐inflammatory cytokines levels and positively correlated with FMI, suggesting that E. rectale may play a protective role in the prevention of lipolysis in CC. These findings were further validated through single‐strain transplantation experiments that supplementation with E. rectale attenuated macrophage infiltration, cytokine production and improved CC symptoms.

While the relationship between E. rectale and other diseases has been explored, such as its reduced abundance in melanoma and its beneficial effects in lymphoma and Behçet's disease (Liu et al. 2023; Islam et al. 2021; Lu et al. 2022), the link between E. rectale and cachexia has not been previously studied. In this study, we developed a novel strategy using bacterial EVs to explore how E. rectale interacts with the host. These nano‐sized vesicles are produced by bacteria and contain various biological components of the parent bacteria. Due to their structure and composition, bacterial EVs have been developed for biomedical applications, including drug delivery and vaccines, and as mediators of microbiota‐host interactions (Hendrix and De Wever 2022). Here, we found for the first time that E. rectale can secrete EVs, which can translocate from the gastrointestinal tract to adipose tissue. Previous studies indirectly support this phenomenon (Luo et al. 2021), and our results confirmed that E. rectale EVs can be taken up by macrophages. This process was inhibited by endocytic pathway inhibitors of macrophages such as Genistein and CPZ. In addition, we performed a preventive intervention study to assess whether initiating EV treatment prior to tumor cell inoculation could delay or mitigate the progression of CC. The results showed that preventive administration markedly attenuated cachexia progression and systemic inflammation in a mouse model, suggesting that E. rectale EVs retain their functional activity after tissue uptake. These findings showed that E. rectale EVs can mediate host responses by interacting with macrophages, ultimately influencing the progression of CC.

An additional important discovery of this study is the mechanism by which E. rectale EVs affect macrophages. While the effects of E. rectale on NK cells and B lymphocytes have been reported (Liu et al. 2023; Lu et al., 2022) and it suppresses lymphomagenesis, its interaction with macrophages was previously unknown. Our study found that E. rectale EVs can alter the transcriptional profile of pro‐inflammatory macrophages. Transcriptomic analysis revealed that the NF‐kB signalling pathway was the main target of E. rectale EVs. The NF‐kB pathway is central to immune and inflammatory responses, with important roles in macrophage polarization and cytokine secretion (Guo et al. 2024). In this study, we demonstrated that E. rectale EVs can inhibit macrophage polarization by suppressing the nuclear translocation of p65 and its DNA binding activity, thus inhibiting the NF‐kB pathway. This inhibition leads to a reduction in pro‐inflammatory cytokine production such as IL‐6 and TNF‐α, improving CC in mouse models.

In conclusion, our study offers compelling evidence for the role of gut microbiota dysbiosis in lipolysis process of CC. We identified for the first time that E. rectale and its EVs are key mediators of host‐microbiota interactions, offering a novel therapeutic avenue for CC. By inhibiting macrophage polarization and pro‐inflammatory cytokine production via the NF‐kB pathway, E. rectale EVs ameliorate adipose tissue wasting in CC models. Future studies should aim to uncover the detailed mechanisms of gut microbiota‐host interactions and evaluate the preventive and therapeutic efficacy of E. rectale EVs in larger clinical cohorts. Overall, our findings offer new insights into the interaction between the gut microbiota and the host and further highlight that E. rectale and its EVs are promising targets for clinical intervention for CC.

Author Contributions

Jiaqi Wang and Xinying Wang spearheaded the conceptualization and design of the study, while Jiaqi Wang, Gulisudumu Maitiabula, Xinying Wang, and Shijie Wang jointly developed the methodology. Jiaqi Wang, Gulisudumu Maitiabula, Sirui Liu, Ruowen Li, Shijie Wang and Pinwen Zhou conducted the in vivo and in vitro experiments, with Jiaqi Wang and Shijie Wang handling the microbiologic aspects. Patient recruitment was managed by Jiaqi Wang, Shijie Wang, Yufei Xia, and Gulisudumu Maitiabula, while biospecimen collection was carried out by Jiaqi Wang, Xin Chen, Longchang Huang, and Yufei Xia. Pathological examinations were performed by Pinwen Zhou, Yufei Xia, and Gulisudumu Maitiabula. The bioinformatic analysis was a collaborative effort involving Jiaqi Wang, Shijie Wang, Longchang Huang, Xuejin Gao, and Li Zhang. Xinying Wang secured the funding, and Xinying Wang oversaw project administration and supervision. The writing of the manuscript was primarily the responsibility of Jiaqi Wang and Xinying Wang. All authors discussed the results and commented on the manuscript.

Funding

This study was supported by research funding from National Natural Science Foundation of China (No. 82370900, No. 82170575), Jiangsu Medical Innovation Center (CXZX202217) and Jiangsu Natural Science Fund ‐ Youth Fund (BK20231092).

Ethics Statement

All experiment animal and protocols were approved by the Institutional Animal Care and Use Committee of Nanjing University.

This study adhered to the ethical guidelines of the 1975 Declaration of Helsinki and received approval from the Ethics Committee of Nanjing Jinling Hospital (2023DZKY‐049‐03).

Consent

The written informed consent was obtained from the study participants. All authors have approved the publication of this manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information: Figure S1. Lipolysis and gut dysbiosis with a significant reduction of E. rectale in PC patients, related to Figure 1, Figure S2. Gut microbiota is closely associated with lipolysis, related to Figure 2, Figure S3. Gut microbiota is closely associated with lipolysis, related to Figure 2, Figure S4. E. rectale alleviates lipolysis induced by PC, related to Figure 3, Figure S5. E. rectale alleviates CC‐induced lipolysis via extracellular vesicles, related to Figure 4, Figure S6. The preventive study of prophylactic potential of E. rectale EVs, related to Figure 4, Figure S7. E. rectale EVs alleviated CC‐induced lipolysis by inhibiting M1 macrophage polarization, related to Figure 5, Figure S8. E. rectale EVs suppress NF‐κB signalling and promote anti‐inflammatory macrophage polarization independently of efferocytosis, related to Figure 7, Figure S9. E. rectale EVs Alleviate Lipolysis Induced by Colorectal and Gastric CC, Table S1. The baseline characteristics and clinical parameters of the two groups, related to Figure 1, Table S2. Primers used for qRT‐PCR, related to Methods.

Wang, J. , Maitiabula G., Liu S., et al. 2026. “The Eubacterium Rectale Derived Extracellular Vesicles Alleviate Cancer Cachexia Induced Lipolysis by Inhibiting Macrophage Polarization.” Journal of Extracellular Vesicles 15, no. 5: e70293. 10.1002/jev2.70293

Data Availability Statement

The data that support the findings of this study are openly available in Figshare at https://doi.org/10.6084/m9.figshare.29975377.v2. All other data files supporting this study are available on reasonable request from the corresponding author.