Early-onset Palmijihwang-hwan treatment modulates phospholipid metabolism and gut microbiota for healthy aging: reducing adipose inflammation and oxidative stress

Abstract

Aging is characterized by progressive physiological decline and increased vulnerability to metabolic and inflammatory disturbances. Palmijihwang-hwan (PM), a traditional East Asian herbal formula, has been used empirically for age-related complaints, but its mechanistic basis remains unclear. Here, we evaluated the effects of early-onset PM administration (starting at 2 months of age) on longevity-related phenotypes and metabolic regulation. PM significantly prolonged lifespan in Caenorhabditis elegans and improved survival in ICR mice without evident toxicity. Preventive PM administration reduced epididymal white adipose tissue (eWAT) mass and circulating insulin/adipokine levels. Lipidomic analysis showed a shift from lysophospholipids toward phospholipids, accompanied by downregulation of PLA2G7, indicating attenuation of adipose inflammation. PM also reshaped the gut microbiota, decreasing inflammation-associated taxa such as Oscillibacter valericigenes, and lowered adipose IL-6 and TNF-α expression. Collectively, these findings indicate that preventive early-onset PM administration modulates the gut microbial composition, counteracting the age-related enrichment of inflammation-related bacteria.

Introduction

Aging is characterized by a progressive loss of physiological integrity, leading to impaired metabolic homeostasis, low-grade inflammation, and increased vulnerability to chronic diseases. Recent research has extensively investigated the genetic, environmental, and metabolic determinants of aging^1,2. Among these factors, lipidomic remodeling and age-related shifts in the gut microbiota have emerged as central modulators of the aging trajectory. Alterations in lipid species such as ceramides and phospholipids accelerate aging and contribute to metabolic and inflammatory disorders³. Likewise, dysbiosis of the gut microbiota is linked to metabolic imbalance and chronic low-grade inflammation that exacerbate age-related diseases and shorten lifespan^4,5. These findings suggest that coordinated modulation of lipid metabolism and the gut microbiota may help slow age-related functional and metabolic decline, even if aging itself cannot be halted.

Traditional herbal medicines have long been recognized for their potential to influence longevity and mitigate age-related disorders. In traditional Chinese medicine, Palmijihwang-hwan (PM; also known as Ba-Wei-Di-Huang-Wan in China and Hachimijiogan in Japan) is a classical formula composed of eight medicinal herbs: Moutan Cortex, Poria, Corni Fructus, Dioscoreae Rhizoma, Rehmannia glutinosa, Alisma Rhizome, Cinnamomi Cortex, Aconiti Lateralis Radix Preparata, and contains well-characterized bioactive constituents such as morroniside, loganin, paeonol, alisol derivatives, and cinnamaldehyde. Consistent with these components, previous studies have reported that PM exerts therapeutic effects on osteoporosis, cognitive decline, and renal dysfunction^6–9. Clinical studies have also reported benefits in aging-related conditions, including mild cognitive impairment, dementia, Alzheimer’s disease, and frailty in older adults^10,11. These clinical observations support PM’s long-standing use in managing age-related functional decline.

In comparison, several modern pharmacological interventions such as metformin, rapamycin, and senolytics (e.g., dasatinib + quercetin) have shown promise in targeting fundamental aging pathways. However, their long-term or prophylactic use is often constrained by adverse effects including gastrointestinal intolerance, vitamin B12 deficiency, immunosuppression, and metabolic disturbances, as well as concerns related to polypharmacy in older adults^12,13. In contrast, PM has historically demonstrated favorable tolerability and multi-target regulatory actions, suggesting potential complementary advantages for addressing the multifactorial nature of age-related metabolic decline^14,15.

Despite these promising observations, the preventive, system-level effects of complex herbal formulas such as PM on both lipid metabolism and gut microbiota in the context of aging remain largely unexplored.

Therefore, the present study investigated the prophylactic administration of PM in Caenorhabditis elegans and ICR mice to determine whether it could delay age-related metabolic and inflammatory decline. Using integrated lipidomic profiling and 16S rRNA microbiota analysis, we sought to elucidate the mechanisms through which PM modulates lipid metabolism, gut microbial composition, and inflammatory signaling to confer geroprotective effects.

Results

Chemical components of PM

The chemical components of the PM extract were identified based on relative retention time, exact MS, and tandem mass spectrometry (MS/MS) fragmentation. In total, twenty compounds were identified, including major components such as trans-cinnamic acid, aconine, morroniside, albiflorin, paeonol, and alisol B acetate (Fig. 1B), along with derivatives originating from each constituent herb (Fig. 1C). In addition to qualitative profiling, quantitative LC-MS analysis was performed for the constituents amenable to quantification. The major compounds detected during the profiling step were present at measurable levels, ranging from 0.041 to 4.593 mg/g. Detailed quantitative results for constituents are provided in Supplementary Table S1.

Fig. 1. Chemical profiling of PM using UHPLC Q-TOF MS.

A Base peak chromatograms (BPC). B Extracted ion chromatograms (EIC) of major constituents in PM. C Table listing 20 identified chemical constituents in PM.

Early-onset preventive PM administration extends lifespan without toxicity

To evaluate the effects of PM on longevity, we first performed a lifespan assay using Caenorhabditis elegans (C.elegans) under standard conditions. PM supplementation significantly prolonged lifespan in a dose-dependent manner (PM 50 µg/mL, p = 0.0091; PM 100–400 µg/mL, p < 0.0001), increasing the mean lifespan by 16.7%, 30.7%, and 42.0% in the PM 50, 100, and 200 µg/mL groups, respectively (Fig. 2A).

Fig. 2. The preventive administration of PM extended the lifespans of worms and improved the survival rates of mice.

A Lifespan assay in C. elegans (Left: study design of C. elegans model, Right: lifespan assay). B Survival analysis in ICR mouse (Upper: study design of mouse model, Lower: survival analysis). *p < 0.05 for CNT vs. PM groups.

To confirm these effects in mammals, we conducted a long-term study using male ICR mice, in which early-onset PM administration began at 2 months of age and continued daily for 16 months. Mice received PM at doses of 50, 100, or 200 mg/kg via drinking water. At 16 months, survival rates were 56%, 61%, and 78% in the respective PM groups, compared with 42% in CNT group (p < 0.05; Fig. 2B). Body weight remained unchanged among groups (Fig. 3A), and no toxicity was observed based on organ-to-body-weight ratios (heart, kidney, liver, lung, spleen; Fig. 3B) or macroscopic inspection (Fig. 3C).

Fig. 3. Early-onset PM administration had no adverse effects on body weight or major organ morphology.

A Changes in body weight over the study period. B Organ indices of the heart, kidney, liver, spleen, and lungs in CNT, PM50, PM100, and PM200 mice, with no significant differences observed across groups. C Macroscopic examination of the heart, kidney, liver, spleen, and lungs revealed no visible abnormalities across CNT, PM50, PM100, and PM200 mice. D Representative hepatic SA-β-galactosidase staining, a marker of senescent cell accumulation, and quantification of SA-β-gal–positive area (n = 4 per group). Significance was determined by one-way ANOVA followed by Tukey’s post-hoc test; ***p < 0.001 was considered significant.

To further assess potential toxicity particularly given that PM contains Aconitum carmichaelii, a source of alkaloids with dose-dependent toxicity we examined serum biochemical markers in a preliminary mouse study. Levels of hepatic and renal injury indicators, including AST, ALT, and BUN, showed no significant differences between PM group and CNT group (Supplementary Fig. S2), indicating that long-term PM administration did not induce hepatic or renal toxicity. In addition, the estimated equivalent concentration of Aconitum carmichaelii within the PM200 condition based on the C. elegans assay ( ~ 7.4 μg/mL) was far below toxic thresholds and even extended lifespan in isolated-compound testing. Taken together, these biochemical and functional data support that PM is well tolerated at the doses used and does not exhibit measurable systemic toxicity in vivo.

Furthermore, to evaluate whether PM influences cellular senescence, hepatic SA-β-galactosidase activity a marker of senescent cell accumulation was examined using immunohistochemical staining. The results showed a marked reduction in SA-β-galactosidase activity in the PM-treated groups (PM 200, p < 0.001; Fig. 3D). To provide molecular support for this anti-senescence effect, we further assessed the protein expression of canonical cellular senescence markers in the intestinal tissue of aging mice. Western blot analysis revealed that the expression levels of p16 and p21 were significantly elevated in the aged control (CNT) group. However, early-onset PM administration suppressed the upregulation of both p16 and p21 in a dose-dependent manner (Supplementary Fig. S2A). These results corroborate the SA-β-gal findings by demonstrating PM’s ability to mitigate cellular senescence at the molecular level in mammals. Additionally, PM’s anti-aging efficacy was supported by independent functional and cellular markers in the C. elegans model, which showed that PM reduced lipofuscin accumulation and preserved locomotor activity (Supplementary Fig. S2B and S2C). Collectively, these findings demonstrate that PM safely improves survival outcomes and delays the progression of aging-associated changes via the mitigation of cellular senescence across species.

PM attenuates adipose and metabolic alterations associated with aging

PM treatment mitigated aging-associated adipose alterations in mice. Epididymal white adipose tissue (eWAT) mass decreased dose-dependently, reaching significance in both the PM 100 and PM 200 groups (p < 0.05 and p < 0.01, respectively; Fig. 4A). Elevated adipokines in aged control (CNT) mice leptin, resistin, and PAI-1 were significantly reduced by PM (leptin: PM 50 and PM 200, p < 0.01; PM 100, p < 0.001; resistin: PM 200, p < 0.05), while PAI-1 showed only a downward trend without statistical significance (Fig. 4B). Plasma insulin levels, markedly elevated in CNT mice relative to YCNT (p < 0.001), were significantly suppressed across all PM-treated groups (p < 0.001; Fig. 4C). Collectively, these results demonstrate that PM alleviates adipose tissue inflammation and metabolic dysregulation associated with aging.

Fig. 4. Early-onset PM treatment reduced adipose inflammation and insulin dysregulation associated with aging.

PM prevented adipose tissue aging (A, B), increase in insulin levels (C) A eWAT weights; B levels of adipokines (leptin, resistin, and PAI-1); C Changes in insulin levels; and Significance was determined by one-way ANOVA followed by Tukey’s post-hoc test; *p < 0.05, **p < 0.01, ***p < 0.001 was considered significant. The number of mice used was n = 7 for panels (A–C).

Because reductions in leptin, insulin, and adiposity can resemble metabolic features of calorie restriction (CR), we further evaluated whether PM affected feeding-related behavior or metabolic flexibility. In C. elegans, PM reduced pharyngeal pumping by 13.4% (Supplementary Fig. S3A), consistent with a CR-associated feeding phenotype. However, in mice, body weight trajectories did not differ among groups (Fig. 3A), and PM-treated mice showed a downward trend in glucose excursion during the OGTT compared with CNT mice, though the difference did not achieve statistical significance (Supplementary Fig. S3B and S3C). Although food intake was not directly measured, the absence of weight reduction and the improved OGTT response indicate that the metabolic benefits of PM occur independently of reduced caloric consumption. These findings support the interpretation that PM modulates specific metabolic pathways rather than inducing a CR-like suppression of feeding.

PM treatment mitigates phospholipid remodeling through regulation of PLA2G7

Comprehensive lipidomic profiling revealed distinct alterations in plasma lipid composition between YCNT group and CNT groups. Representative total ion chromatograms acquired in positive and negative ion modes are shown in Fig. 5A, B. The PLS-DA score plots demonstrated a clear separation between CNT and YCNT groups in both modes (Fig. 5C, D), and model validity was confirmed by 100-fold permutation testing (Fig. 5E, F). Notably, the lipid profiles of PM-treated mice clustered progressively closer to the YCNT group in a dose-dependent manner, indicating partial restoration of the youthful lipid signature.

Fig. 5. Early-onset PM treatment attenuated age-related alterations in circulating lipid profiles of aged mice.

Representative total ion chromatography obtained from mouse plasma from positive (A) and negative ion modes (B). Partial least squares-discriminant analyses (PLS-DA) score plots using unbiased lipid profiling of each treatment group obtained from positive (C R2X = 45.1%, R2Y = 47.2%, and Q2 = 26.5%, p = 0.00048) and negative (D R2X = 41.3%, R2Y = 53.0%, and Q2 = 24.3%, p = 0.00019) ion modes. Permutation plots of PLS-DA of plasma for positive (E) and negative (F) ion modes. The number of mice were YCNT = 15, CNT = 10, PM50 = 10, PM100 = 11, and PM200 = 13 for the positive mode, and YCNT = 15, CNT = 10, PM50 = 9, PM100 = 9, and PM200 = 13 for the negative mode.

A total of 57 lipid metabolites exhibiting high variable importance in projection (VIP) scores and significant differences (p < 0.05) between CNT and YCNT mice were visualized in a heatmap (Fig. 6A).

Fig. 6. Early-onset PM treatment modulated age-related lipidomic alterations through regulation of PLA2G7.

A Heatmap showing the significantly different lipids between CNT and YCNT mice. Red and underlined lipids indicate those with alterations in CNT that are by PM administration. B Lipids with alterations in CNT that are prevented by PM treatment. C Lipid enrichment analysis. D Change in PLA2G7 mRNA expression in eWAT (n = 5 per group). Significance was determined by one-way ANOVA followed by Tukey’s post hoc test; *p < 0.05, **p < 0.01, ***p < 0.001 was considered significant.

Early-onset PM treatment significantly modulated these age-related changes, decreasing 7 and increasing 12 lipid that were altered in CNT mice (p < 0.05; Fig. 6B; Supplementary Table S3). Pathway enrichment analysis identified glycerophospholipid metabolism as the most significantly affected pathway (Fig. 6C).

Among the lipids significantly altered by PM, the majority were phospholipids (PLs) or lysophospholipids (LPLs), comprising decreasing 7 and increasing 12 lipid relative to CNT mice. These lipid changes suggest that PM modulates phospholipase-related remodeling and attenuates oxidative stress–associated membrane alterations^16,17. Accordingly, changes in PLA2G7 in adipose tissue were investigated, as this newly identified regulator of immune-metabolic effects and lifespan is associated with longevity effect of calorie restriction^18,19. PLA2G7 mRNA levels were significantly reduced in all PM-treated groups (PM 50–200; p < 0.001) as well as in YCNT mice (p < 0.001) compared with CNT, consistent with suppressed phospholipase activity and restoration of phospholipid homeostasis (Fig. 6D). While these results strongly suggest that PM functionally suppresses PLA2G7 activity to restore phospholipid homeostasis, the direct validation of PLA2G7 protein expression and enzyme activity remains necessary for conclusive confirmation of its role as a direct target.

Preventive administration of PM reduced the inflammation-inducing microbes during aging

Gut microbiota also play a significant role in aging by altering homeostatic metabolism and influencing the host immune system²⁰. In the present study, the PM administration did not markedly alter alpha diversity, as reflected by the Chao1 and Shannon indices (Fig. 7A). Although no significant overall increase was detected, the Shannon index was significantly reduced in the PM 100 group (p < 0.01), indicating that PM treatment did not uniformly enhance microbial diversity.

Fig. 7. Early-onset PM treatment suppressed pro-inflammatory gut microbes associated with aging.

A Changes in alpha diversity. B Principal coordinates analysis of the gut microbiota composition. C Taxonomic composition of bacterial communities (Left) and the gut microbiota significantly altered by preventive administration of PM at the phylum level (Right). D Relative abundance of gut microbiota significantly altered by preventive administration of PM at the species level. Bold letters indicate inflammation-inducing microbes. Significance was determined by one-way ANOVA followed by Tukey’s post-hoc test; *p < 0.05, **p < 0.01, ***p < 0.001 was considered significant. The number of mice used for the gut microbiota analysis was YCNT = 15, CNT = 9, PM50 = 10, PM100 = 11, and PM200 = 14.

Principal coordinate analysis (PCoA) based on beta diversity revealed a clear separation between CNT and YCNT groups, consistent with age-related microbial shifts. Importantly, the microbial community structures of PM-treated mice were significantly distinct from those of CNT mice (p < 0.05, permutational multivariate analysis of variance [PERMANOVA]; Fig. 7B), suggesting that PM partially reshaped the gut microbiota composition toward a more youthful pattern.

At the phylum level, PM treatment markedly altered microbial composition. Firmicutes abundance was significantly increased in all PM groups compared with CNT mice (PM 50, p < 0.05; PM 100, p < 0.001; PM 200, p < 0.01), whereas no significant differences were detected in Bacteroidetes among the groups (Fig. 7C). In contrast, Proteobacteria a phylum commonly associated with gut inflammation and aging²¹ and Melainabacteria were more abundant in CNT mice than in both YCNT and PM groups (Proteobacteria: CNT > YCNT, p < 0.05; CNT > PM 50, PM 100, PM 200, p < 0.001; Melainabacteria: CNT > YCNT and PM 50, PM 100, PM 200, p < 0.001; Fig. 7C). These findings indicate that PM modulated the microbial profile of aged mice, suppressing the expansion of taxa that are elevated in the aged control (CNT) group and linked to inflammation.

At the species level, Helicobacter ganmani (phylum Proteobacteria) and Vampirovibrio chlorellavorus (phylum Melainabacteria) were significantly reduced by preventive PM administration (H. ganmani: PM 50, PM 100, PM 200, p < 0.01; V. chlorellavorus: PM 50, PM 100, PM 200, p < 0.001; Fig. 7D). Among Firmicutes genus, Oscillibacter valericigenes, Roseburia hominis were significantly decreased in the PM groups (O. valericigenes: PM 200, p < 0.05; R. hominis: PM 50, p < 0.05; PM 200, p < 0.01) and YNCT (O. valericigenes: p < 0.001; R. hominis: p < 0.05) compared with CNT (Fig. 7D). In contrast, Lactobacillus johnsonii abundance was significantly increased in the PM 50 and PM 100 groups (p < 0.05), while no change was observed in YCNT (p = 0.988). Within the Bacteroidetes phylum, Tidjanibacter massiliensis abundance was significantly decreased in both the PM 200 (p < 0.05) and YCNT (p < 0.05) groups compared with CNT (Fig. 7D).

H. ganmani and V. chlorellavorus are known for their pathogenic potential^22,23. Collectively, these findings indicate that preventive PM administration modulates the gut microbial composition, counteracting the age-related enrichment of inflammation-related bacteria. These parallel alterations suggest a potential interaction between microbial composition and age-related inflammatory changes; however, the present findings demonstrate association rather than causation.

Integration of phospholipid metabolism and gut microbiota modulates immune–metabolic balance during aging

The results of the lipidomic and microbiome analyses showed that preventive administration of PM inhibited PLA2G7 mRNA expression in eWAT and reduced the levels of pro-inflammatory microbes. Thus, the alteration of proinflammatory cytokines such as IL-6 and TNF-α levels in intestine and eWAT were investigated.

In the intestine, IL-6 and TNF-α mRNA expression levels were higher in the CNT group than in YCNT and PM-treated groups (Fig. 8A). IL-6 expression showed a non-significant trend toward increase in CNT mice compared with YCNT, but was significantly reduced by PM treatment (PM 50 and PM 200, p < 0.05; PM 100, p < 0.01). TNF-α expression was significantly elevated in CNT mice relative to YCNT (p < 0.05) and was also suppressed by PM treatment (PM 50 and PM 100, p < 0.05).

Fig. 8. Early-onset PM treatment modulated lipid metabolism and gut microbiota, alleviating immune–metabolic disturbances associated with aging.

A Change in IL-6 and TNF-α mRNA expressions in the intestine (n = 5 per group). B Change in IL-6 and TNF-α mRNA expression of in eWAT (n = 5 per group). Circulating lipids and microbiota altered by PM correlated with changes in insulin levels (C, D). C Bubble plot showing Spearman correlation analysis results of insulin levels and PM-regulated lipids. D Correlation between insulin levels and O. valericigenes. Significance was determined by one-way ANOVA followed by Tukey’s post hoc test; *p < 0.05, **p < 0.01 was considered significant.

In the eWAT, IL-6 mRNA expression in the CNT group was significantly higher than that in the YCNT (p < 0.05) and PM groups (PM100, p < 0.05; PM200, p < 0.01; Fig. 8B). TNF-α mRNA expression in the CNT group was significantly higher than that in the PM 100 group (p < 0.05; Fig. 8B). To complement these transcriptional findings, a preliminary Western blot analysis of eWAT cytokines was performed (Supplementary Fig. S4A). The protein levels of both IL-6 and TNF-α showed a downward trend in PM-treated mice compared with the aged control, supporting the interpretation that PM suppresses adipose inflammatory signaling. By contrast, the anti-inflammatory cytokine IL-10 showed no significant change in either intestinal or eWAT across groups (Supplementary Fig. S4B). These results suggest that PM alleviates adipose inflammation, consistent with the downregulation of PLA2G7, a key mediator of immune metabolic regulation during aging^18,19.

A total of 12 of the 9 altered lipid metabolites were significantly correlated with plasma insulin levels (p < 0.05; Fig. 8C). Specifically, four lysophospholipids (LPLs; LPC 17:0, LPC 18:0, LPE 20:1, and LPE 22:0) and phosphatidylinositol (PI) 36:4 showed significant positive correlations with insulin, whereas four phosphatidylcholines (PCs) and phosphatidylethanolamines (PEs) exhibited negative correlations, indicating an association between phospholipase-related lipid remodeling and insulin regulation.

In the correlation analysis between these lipids and microbes, LPLs showed a significant positive correlation with three inflammatory microbes (Supplementary Fig. S5). In contrast, L. johnsonii, a probiotic, positively correlated with the four PCs and negatively correlated with LPC 17:0²⁴ (Supplementary Fig. S5). Furthermore, the regularized canonical correlation analysis results indicated that V. chlorellavorus and O. valericigenes were closely related to LPLs and insulin (cut-off: 0.5; Supplementary Fig. S5). Despite this, O. valericigenes, a microbiota known to induce adipose tissue inflammation, was the only species that was positively correlated with insulin levels (p = 0.0490, r = 0.3173; Fig. 8D).

These findings suggest that changes in lipid metabolism and gut microbiota induced by PM administration contribute to the modulation of adipose inflammation, thereby mitigating age-related metabolic disturbances. Among the gut microbiota, O. valericigenes which was significantly reduced by PM and has been linked to adipose inflammation may act as a potential microbial mediator influencing insulin regulation during aging.

Discussion

In this study, we evaluated the long-term effects of early-onset preventive PM administration, which began at a young age before the onset of aging-associated decline. The present study demonstrated that PM administration enhanced survival and improved aging-associated phenotypes without evident toxicity. Although PM treatment did not involve dietary restriction, several of the observed changes including reductions in leptin, resistin, and insulin levels closely resembled the metabolic adaptations typically induced by CR, a well-established intervention known to enhance metabolic efficiency and extend lifespan¹⁹. These findings suggest that PM may influence CR-related metabolic pathways. However, the metabolic improvements observed in mice were not accompanied by reductions in food intake or body weight, and OGTT results showed only a modest downward trend without statistical significance. These findings indicate that PM influences selective metabolic pathways rather than acting as a classical calorie-restriction mimetic.

Lipidomic analysis further revealed pronounced alterations in plasma lipid profiles, characterized by an increase in PLs and a decrease in LPLs. This shift suggests that PM modulates phospholipase-related lipid remodeling, a process closely linked to inflammation and oxidative stress. In this context, Spadaro et al. reported that suppression of PLA2G7 in adipose tissue is a key mechanism by which calorie restriction reduces lipid-driven inflammation and promotes longevity¹⁹. Mechanistically, PLA2G7 hydrolyzes oxidized phospholipids into pro-inflammatory lysophospholipids and oxidized fatty acids, thereby promoting macrophage recruitment, cytokine production, and adipose tissue dysfunction²⁵. Elevated PLA2G7 activity is a known driver of chronic low-grade inflammation (“inflammaging”), impaired insulin sensitivity, and increased oxidative vulnerability. Conversely, genetic or pharmacological inhibition of PLA2G7 has been shown to reduce inflammatory lipid mediators, improve metabolic homeostasis, and attenuate age-related physiological decline^19,26. Consistent with this finding, the early-onset PM administration significantly reduced PLA2G7 mRNA expression, along with decreases in adipose mass, pro-inflammatory cytokines (IL-6 and TNF-α), insulin, and adipokines. Correlation analyses further demonstrated divergent associations of PLs and LPLs with insulin levels, supporting the idea that PLA2G7-linked lipid remodeling contributes to insulin regulation during aging.

Building on these findings, PM was also found to mitigate age-related alterations in the gut microbiota, notably suppressing the expansion of proteobacteria and inflammation-associated taxa such as H. ganmani and V. chlorellavorus, which have been associated with inflammatory states in previous studies, while showing increased levels of beneficial species including L. johnsonii. Of note, O. valericigenes a taxon previously linked to adipose macrophage–driven inflammation displayed a unique positive association with insulin levels in our dataset, suggesting a potential microbe–lipid–insulin relationship modulated by PM.

In addition to lipid remodeling, PM treatment alleviated gut microbiota alterations observed in aged (CNT) mice, suppressing the expansion of Proteobacteria and pro-inflammatory species such as H. ganmani and V. chlorellavorus, while enriching beneficial taxa including L. johnsonii. Among these microbes, O. valericigenes previously implicated in adipose macrophage-driven inflammation showed a significant positive correlation with insulin levels, whereas other taxa did not display such direct associations.

Increasing evidence indicates that aging, gut microbiota, and lipid metabolism are interconnected through bidirectional regulatory loops that influence systemic metabolic homeostasis. Aging is known to reduce microbial diversity and promote the expansion of inflammation-associated taxa²⁷. Moreover, age-related gut microbiota dysbiosis can impair intestinal barrier integrity and promote endotoxin leakage, thereby inducing inflammatory responses in peripheral metabolic organs such as the liver and adipose tissue. Recent studies further support this gut–organ inflammatory axis: for example, Oroxin B has been shown to alleviate metabolic-associated fatty liver disease by correcting gut microbiota dysbiosis and suppressing hepatic inflammation in high-fat diet–induced models²⁸. In addition, a recent review demonstrated that plant-derived natural products broadly ameliorate metabolic-associated fatty liver disease by modulating gut microbial composition and reducing downstream inflammatory signaling in metabolic tissues²⁹. Conversely, alterations in host lipid metabolism can reshape microbial communities through changes in bile acids, phospholipid substrates, and inflammatory mediators. Supporting this concept, interventions derived from Cornus officinalis have been shown to remodel gut microbiota and regulate host lipid droplet dynamics in metabolic disease models³⁰. In this context, our findings suggest that PM modulates this integrated aging–microbiota–lipid metabolism axis, evidenced by the simultaneous occurrence of altering PLA2G7-linked phospholipid remodeling and suppressing inflammation-associated microbes, thereby contributing to improved metabolic stability during aging.

In addition to these system-level changes, the data imply that PM may exert its anti-inflammatory and metabolic effects through both host-intrinsic pathways and microbiota-mediated mechanisms. The reduction of PLA2G7 expression in adipose tissue points to a direct modulation of lipid-inflammatory signaling within metabolic organs. Concurrently, PM suppressed inflammation-associated taxa including O. valericigenes, which uniquely correlated with insulin levels suggesting that microbiota remodeling may indirectly contribute to the observed metabolic improvements. While these findings support a coordinated lipid–microbiota inflammatory axis, the present study cannot determine whether these effects arise independently or sequentially. Future mechanistic work using germ-free mice, microbiota transfer, or adipose-specific gene manipulation will be required to establish their causal hierarchy.

Adipose tissue is increasingly recognized as a central regulator of organismal aging, with many lifespan-extending drugs targeting adipose tissue inflammation^31–33. Although the present data indicate that PM reduces inflammatory signaling in adipose tissue, these findings should be interpreted as supportive rather than definitive evidence due to the limited protein-level validation available. The downward trends observed in eWAT cytokine protein expression, together with reductions in adipose mass, PLA2G7 suppression, and improved circulating metabolic markers, collectively suggest that PM may attenuate adipose inflammation, but further studies with expanded sample sizes will be required to confirm this mechanism.

It is also worth noting that resistin showed a non-linear dose response. Resistin is inherently variable and primarily regulated by immune-cell activity in rodents, which often results in non-monotonic fluctuations independent of other adipokines^34,35. Given the consistent reductions in TNF-α, IL-6, and adipose mass across PM-treated groups, this isolated variation is unlikely to affect the overall interpretation that PM reduces inflammatory tone. In addition, although PAI-1 levels remained higher in all aged groups than in YCNT mice which reflects the well-established age-related increase in PAI-1 and its strong association with visceral adiposity they nevertheless showed a dose-dependent reduction in PM-treated mice compared with aged controls. This pattern suggests that PM partially attenuates, but does not fully reverse, the age-related elevation in PAI-1^36,37, and is therefore consistent with an overall anti-inflammatory effect rather than indicative of a contradictory pro-inflammatory response.

Although the transcriptional suppression of PLA2G7, together with restoration of phospholipid homeostasis, strongly suggests that PLA2G7 is a functional downstream target of PM, the present findings are based only on mRNA and lipid remodeling signatures. Definitive validation will require protein-level quantification and enzymatic activity assays to determine whether reduced gene expression corresponds to decreased PLA2G7 function. Furthermore, while PM lowered PLA2G7 expression and pro-inflammatory cytokines in adipose tissue, the current data cannot distinguish whether these effects originate from direct adipose modulation, microbiota-mediated regulation^18,19, or both. Germ-free mouse studies, microbiota transplantation, or tissue-specific genetic approaches will be needed to clarify these interactions. Likewise, although O. valericigenes correlated with insulin levels²⁴, its causal contribution to metabolic regulation remains unverified. Identifying the active PM constituents responsible for lipidomic and microbiome remodeling will also strengthen mechanistic insight.

This study has several additional limitations. First, although both C. elegans and mice showed survival benefits from PM, nematode data were used only to identify conserved aging signatures, as mechanistic conclusions rely solely on mouse experiments. Second, our microbiome findings are descriptive 16S profiles and cannot infer causality; functional assays such as targeted metabolomics or microbiota-transfer studies will be required. Third, the exclusive use of male mice limits generalizability, given known sex-specific differences in aging and metabolic responses. Future studies incorporating both sexes and functional microbiome analyses will be essential to enhance translational relevance.

Despite these limitations, our findings demonstrate that PM is associated with favorable shifts in phospholipid metabolism and gut microbiota composition, suggesting a coordinated action to mitigate adipose inflammation and metabolic dysregulation during aging. By integrating transcriptomic, lipidomic, and microbiome analyses, this study provides system-level evidence supporting PM as a promising multi-target intervention for delaying age-related metabolic decline.

Methods

Preparation and analysis of PM

PM was prepared using the following dried medicinal herbs: Rehmannia glutinosa (Gaertn.) DC., Dioscorea japonica Thunb., Cornus officinalis Siebold & Zucc., Paeonia × suffruticosa Andrews, Wolfiporia cocos (F.A.Wolf) Ryvarden & Gilb. Alisma orientale (Sam.) Juz., Cinnamomum verum J.Presl, and Aconitum carmichaelii Debeaux, mixed in a ratio of 8:4:4:3:3:3:1:1 (Heo, 2011). All herbs were sourced from Geumo-dang (Jeonju, Jeonbuk State, Korea). Dried herbs (405 g) were mixed in 3000 mL of distilled water and dried using a freeze dryer. This process resulted in a 26.03% yield and (105.4 g of the final product). A herbarium voucher specimen (HBM212-08) was deposited at Wonkwang University (Iksan, Jeonbuk State, Korea). Chemical component analysis was conducted (Fig. 1).

Chemical components analysis of PM

The extract of PM was dissolved in a 50% methanol solution and subjected to analysis utilizing ultra-high-performance liquid chromatography (UHPLC, 1290 Infinity II LC System, Agilent Technologies, Santa Clara, CA, USA) coupled with a quadrupole time-of-flight mass spectrometer (Q-TOF MS, 6546 Q-TOF, Agilent Technologies). This analysis was conducted using a Zorbax Extend-C18 column (80 Å, 2.1 × 50 mm, 1.8 μm) from Agilent Technologies, with the column temperature maintained at 40 °C. solvent A; water containing 0.1% formic acid, solvent B; acetonitrile contacting 0.1% formic acid (Flow rate: 0.1 min/mL). Solvent gradient; 0–3 min 5.0% B, 3–5 min 50% B, 5–20 min 85% B, 20–23 min 85% B, 23–24 min 5% B, 24–30 min 5% B. Acquired UHPLC-QTOF-MS data were processed using MS-Dial (Ver 4.80, http://prime.psc.riken.jp/) to detect features, perform alignments, and generate peak tables of m/z and retention time in the samples. Moreover, constituents including albiflorin (Sigma-aldrich), benzoylmesaconine (Chemfaces), coumarin (Chemfaces), and loganin (Sigma-aldrich), among all identified constituents were qualitatively analyzed by comparing their retention times and mass fragmentation patterns with those of authentic chemical standards.

C. elegans lifespan and aging-associated phenotype analysis

C. elegans was used as an in vivo model to evaluate the effects of PM on longevity and aging-associated phenotypes. Wild-type Bristol N2 worms were obtained from the Caenorhabditis Genetics Center (CGC) and maintained at 20 °C on nematode growth medium (NGM) agar plates seeded with Escherichia coli OP50. Lyophilized PM powder was dissolved in distilled water and added to autoclaved NGM agar cooled to approximately 50 °C to achieve the indicated concentrations.

For lifespan analysis, age-synchronized worms at the L4 larval stage were transferred to NGM plates containing floxuridine (5′-fluorodeoxyuridine) to prevent progeny production. To avoid dauer formation, worms were transferred to fresh plates on adult day 10. Survival was monitored daily using a dissecting microscope (SMZ1500; Nikon, Japan), and worms were scored as dead when they failed to respond to gentle stimulation with a platinum wire (Fig. 2A). Each independent experiment included 37–39 worms per group, and lifespan data were obtained from three independent biological replicates.

Aging-associated phenotypes were evaluated in parallel using the same synchronized worm populations. Intestinal lipofuscin accumulation, a hallmark of aging, was assessed on adult day 10. Worms were anesthetized with 2% sodium azide, and autofluorescence images were acquired using an upright fluorescence microscope (Eclipse Ci; Nikon, Japan). Lipofuscin levels were quantified using ImageJ software by measuring the average fluorescence intensity per worm.

Locomotory function was assessed as an indicator of functional aging at adult days 1, 3, 5, 7, and 9. At each time point, worms were transferred to fresh NGM plates, and locomotion was recorded for 20 s under a dissecting microscope. Locomotor velocity (μm/s) was calculated based on the distance traveled using Nikon NIS-Elements imaging software.

Animal study

Seven-week-old ICR male mice were obtained from KOATECH (Pyeongtaek-si, Gyeonggi-do, Korea) and acclimatized for one week before the start of the experiment. Animals were randomly assigned to five groups: natural-aging control (CNT, n = 24), PM50 (n = 18), PM100 (n = 18), PM200 (n = 18), and young control (YCNT, n = 16) (Fig. 2B). The sample size (CNT: n = 24; PM-treated groups: n = 18 each) falls within the range commonly used in murine aging and longevity studies^38,39 and is appropriate for Kaplan–Meier survival analysis with log-rank testing.

PM was administered prophylactically at doses of 50, 100, or 200 mg/kg via drinking water. Because PM treatment was initiated at 2 months of age well before the onset of natural aging this study represents an early-onset preventive intervention rather than a therapeutic approach in already aged animals. PM administration began at 2 months of age and continued for 16 months, corresponding to the time point at which survival in the CNT group declined below 50%. PM was dissolved in sterile drinking water at concentrations calculated to deliver the intended daily dose based on the average water intake of ICR mice (4–6 mL/day) and mean body weight at the start of each weekly cycle. Mice consumed PM-containing water ad libitum, and water bottles were replaced daily to ensure compound stability and freshness. The selected dose range was supported by preliminary experiments demonstrating improved survival at the mid-dose (100 mg/kg) and by previous studies using comparable PM-related prescriptions^40,41. To minimize handling-related stress and circadian-associated metabolic variability, all routine procedures, including water replacement and body weight measurements, were consistently performed between 9:00 and 10:00 AM. This voluntary oral administration approach is commonly used in long-term intervention studies, as it minimizes procedural stress while providing stable bioavailability through repeated small-volume ingestion across the 24-h cycle⁴². Results of the preliminary survival study are presented in Supplementary Fig. S1.

Mice were housed under controlled conditions (23 ± 2 °C, 50–60% relative humidity) with a 12-h light/dark cycle (lights on at 7:00 AM). Cages were changed three times per week in accordance with standard SPF husbandry guidelines, and all animals had ad libitum access to standard rodent chow (Purina 38507). For microbiome analysis, mice were individually placed in metabolic cages for 24 h to collect freshly excreted fecal pellets while minimizing coprophagy-related cross-contamination.

To reduce variability in metabolic hormone levels, mice were subjected to an overnight fasting period of approximately 10 h prior to blood collection. Food was withdrawn at approximately 23:00 on the day before sacrifice, and blood sampling was conducted between 09:00 and 10:00 the following morning, while water remained available ad libitum. At the study endpoint, mice were deeply anesthetized with isoflurane, and blood was collected via cardiac puncture, followed immediately by euthanasia through cervical dislocation. To control for circadian influences on metabolic and microbiome parameters, all tissue collections were performed at a consistent Zeitgeber Time (ZT), with autopsies conducted between ZT1 and ZT2. Whole blood was centrifuged at 3000 × g for 15 min at 4 °C, and plasma was stored at –80 °C for subsequent biochemical and lipidomic analyses. Liver, intestine, cecum, and epididymal white adipose tissue (eWAT) were also collected, snap-frozen or fixed as appropriate, and processed according to the requirements of each assay.

All experimental procedures were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) of Wonkwang University (approval number: WKU22-54), and animal care and handling were performed by trained personnel. Random sampling was applied prior to all analyses.

Immunohistochemistry

The expression of senescence-associated (SA) β-galactosidase in the liver was evaluated using Immunohistochemistry (IHC). 6 µm thick tissue sections were stained with β-galactosidase (Thermo Fisher Scientific Inc., A-11132) were incubated overnight at 4 °C in humidified chambers. Subsequently, all slides received a biotinylated secondary antibody, ImmPRESS® HRP Horse Anti-Rabbit IgG PLUS Polymer Kit (Vector Laboratories MP-7801-15) for 30 min at room temperature. Staining followed the manufacturer’s protocol, and the slides were visualized using a Zeiss Axioscan 7 slide scanner running Zeiss Zen Software (Carl Zeiss MicroImaging, Jena, Germany). Additionally, SA-β-galactosidase expression levels were analyzed using the ImageJ program.

Lipidomic profiling

Unbiased lipidomic profiling of plasma samples was performed using an ultra-high-performance liquid chromatography system (UHPLC, 1290 Infinity II LC system; Agilent Technologies, Santa Clara, CA, USA) coupled to a quadrupole time-of-flight mass spectrometer (Q-TOF MS, 6545XT Q-TOF system; Agilent Technologies). For sample preparation, 50 μL of plasma was mixed with 550 μL of chloroform:methanol (2:1, v/v) and vortexed for 1 min. Subsequently, 200 μL of water was added, followed by vortexing and incubation on ice for 10 min. After centrifugation at 13,000 rpm for 20 min at 4 °C, the lower organic phase (300 μL) was collected, dried, and reconstituted in 200 μL of isopropanol:acetonitrile (4:1, v/v) containing 10 μL of SPLASH® LIPIDOMIX® Mass Spec Standard (Avanti Polar Lipids).

Lipid separation was achieved using a CSH C18 column (100 × 2.1 mm, 1.7 μm; Waters Corp.) maintained at 35 °C. The mobile phase consisted of 10 mM ammonium formate (or ammonium acetate in negative ion mode) in acetonitrile:water (4:6, v/v; solvent A) and 10 mM ammonium formate (or ammonium acetate in negative ion mode) in acetonitrile:isopropanol (1:9, v/v; solvent B). The gradient program was as follows: 40–65% B (0–5 min), 65–70% B (5–12 min), 70–99% B (12–15 min), 99% B (15–17 min), and re-equilibration from 99 to 40% B (17–20 min). The flow rate was set to 0.25 mL/min, and the injection volume was 1.5 μL.

Quality control (QC) samples, prepared by pooling equal aliquots of all samples, were injected every 10 analytical runs. In addition, QC samples diluted 2-, 4-, and 8-fold were analyzed at the beginning and end of the sequence to assess analytical linearity and stability. Sample injection order was randomized to minimize batch effects.

Raw UHPLC-QTOF-MS data files (.d) were converted to ABF format and processed using MS-DIAL software (version 3.96; RIKEN, Japan) for peak detection, alignment, and lipid identification based on m/z values, retention times, and MS/MS fragmentation patterns. LOWESS normalization was applied using QC samples to correct for signal drift. Features exhibiting a coefficient of variation greater than 20% in QC samples, intensities below threefold the average of blank samples, or poor linearity in diluted QC samples were excluded from further analysis.

Lipid species were identified using built-in and public MS-DIAL libraries with a similarity cutoff of 80%, followed by manual verification of fragmentation patterns, head groups, and fatty acyl chain composition using MS-Finder and LipidCreator. Relative lipid quantification was performed using stable isotope-labeled internal standards (SPLASH® LIPIDOMIX®). Calibration curves generated from pooled QC samples spiked with known concentrations of isotope standards were used for semi-quantitative analysis. LPC 18:1(d7), LPE 18:1(d7), and PE 33:0 | PE 15:0_18:1(d7) were used as representative standards for LPC, LPE, and PE species, respectively.

Multivariate analysis of lipid profiling and lipid pathway enrichment analysis

Multivariate analysis of unbiased lipidomic data was performed using the SIMCA-P+ software (version 14.0; Umetrics, Umeå, Sweden). Partial least squares discriminant analysis (PLS-DA) was conducted and validated using permutation and cross-validation analyses of variance (CV-ANOVA) tests. To select the important lipids contributing to clustering or trends in the PLS-DA, variable importance on projection (VIP) analysis was performed. Statistical significance was evaluated using univariate statistical analysis. Lastly, enrichment analysis of significantly altered lipids was executed using the Lipid Pathway Enrichment Analysis tool available at https://lipea.biotec.tu-dresden.de/home.

Gut microbiome sequencing and analysis

Microbial genomic DNA was extracted from fecal samples using the FastDNA® SPIN Kit for Feces (MP Biomedicals, USA) according to the manufacturer’s instructions. DNA concentration and purity were assessed prior to downstream applications. For library preparation, 5 ng of genomic DNA was used as input for amplification of the V3–V4 hypervariable regions of the 16S rRNA gene using the Illumina 16S Metagenomic Sequencing Library Preparation protocol. PCR amplification was performed with universal primers containing Illumina adapter overhang sequences (forward primer: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG; reverse primer: 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC) and Herculase II Fusion DNA Polymerase (Agilent Technologies, Santa Clara, CA, USA).

The first PCR reaction was carried out under the following conditions: initial denaturation at 95 °C for 3 min, followed by 25 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 5 min. PCR products were purified using AMPure XP beads (Agencourt Bioscience, Beverly, MA, USA). Indexing was performed using Nextera XT indexed primers in a second PCR reaction (10 cycles), followed by an additional AMPure XP bead purification. Final libraries were quantified using the KAPA Library Quantification Kit for Illumina platforms and assessed for quality using the TapeStation D1000 ScreenTape system (Agilent Technologies). Paired-end sequencing was performed on an Illumina MiSeq™ platform (Illumina, San Diego, CA, USA) by Macrogen (Seoul, Korea).

Raw sequencing reads were demultiplexed into sample-specific paired-end FASTQ files and processed for amplicon sequence variant (ASV) analysis. Raw sequencing reads were demultiplexed into sample-specific paired-end FASTQ files and processed for amplicon sequence variant (ASV) analysis. Taxonomic assignment was performed using QIIME software (version 1.9) based on reference databases. Downstream microbiome analyses were conducted using the MicrobiomeAnalyst platform (https://www.microbiomeanalyst.ca). Alpha diversity was evaluated using Chao1 richness and Shannon diversity indices, while beta diversity was assessed using weighted and unweighted UniFrac distances. Principal coordinates analysis (PCoA) was applied to visualize differences in microbial community structure among experimental groups.

To explore the relationships between gut microbiota composition, metabolic parameters, and lipidomic features, regularized canonical correlation analysis was performed using the mixOmics package in R (version 3.3.1). This analysis was used to identify coordinated associations among microbial taxa, circulating insulin levels, and insulin-related lipid metabolites.

Western blot analysis

Adipose tissues were homogenized in ice-cold RIPA lysis buffer (Thermo Fisher Scientific) supplemented with protease and phosphatase inhibitor cocktails (Sigma-Aldrich). The homogenates were incubated on ice for 30 min and centrifuged at 15,000 × g for 10 min at 4 °C. The supernatants were collected, and protein concentrations were determined using the BCA Protein Assay Kit (Pierce, Thermo Fisher Scientific).

For samples requiring protein cleaning, four volumes of ice-cold 100% acetone were added to 60 µg of protein (per lane), mixed by gentle inversion, and incubated at −20 °C for at least 30 min or overnight. The mixtures were centrifuged at 15,000 × g for 10 min at 4 °C, the supernatants were discarded, and the resulting pellets were washed once with ice-cold 80% acetone. Equal amounts of protein were mixed with 2× Laemmli sample buffer (Bio-Rad), heated at 95 °C for 5 min, and separated on 10–12% SDS–polyacrylamide gels. Proteins were transferred onto nitrocellulose membranes (Bio-Rad) using a semi-dry transfer system. Membranes were blocked with EveryBlot Blocking Buffer (Bio-Rad) for 30 min at room temperature and incubated overnight at 4 °C with primary antibodies against IL-6(E-4), TNF-α(52B83), p16(C-3), p21(F-5), and β-actin (C4) (all from Santa Cruz Biotechnology; 1:1000).

Following three washes with TBS-T, membranes were incubated with HRP-conjugated secondary antibodies (anti-mouse IgG, 1:5000; Cell Signaling Technology) for 1.5 h at room temperature. Protein signals were detected using an enhanced chemiluminescence (ECL) reagent (Thermo Fisher Scientific) and imaged with the ImageQuant™ LAS 4000 mini system (Cytiva, USA).

Biochemical analysis

Insulin, leptin, resistin, and PAI-1 levels were measured using a BioPlex Pro Mouse 8-Plex Diabetes Kit (#171F7001M; Bio-Rad, Hercules, USA) on a Bio-Plex 200 system (Bio-Rad).

Real-time PCR analysis

Tissue samples, including eWAT, and intestine, were collected and homogenized for RNA extraction. Total RNA was extracted using the TRIzol reagent (Invitrogen, USA) following the manufacturer’s protocol. The RNA concentration and purity were determined using a NanoDrop spectrophotometer. Complementary DNA (cDNA) synthesis was carried out using a reverse transcription kit (Thermo Fisher Scientific, USA). For quantitative real-time PCR (qRT-PCR), SYBR Green Master Mix (Applied Biosystems, USA) was used with gene-specific primers. The primers used in the PCR amplification are listed in Supplementary Table S2. The reaction conditions were as follows: initial denaturation at 95 °C for 10 min, followed by 40 cycles of denaturation at 95 °C for 15 s and annealing/extension at 60 °C for 60 s. GAPDH was used as an internal control to normalize gene expression.

Statistical analysis

Kaplan-Meier survival curves were used to visualize data from the lifespan experiments, and survival differences between groups were evaluated using the log-rank test. Data are expressed as the mean ± standard error of the mean (SEM). Statistical differences among groups were assessed using one-way analysis of variance followed by Tukey’s post hoc test. Spearman’s correlation analysis was also performed. All statistical analyses were conducted using GraphPad Prism version 10 (GraphPad Software, La Jolla, CA, USA), and p < 0.05 was considered statistically significant.

Author contributions

J.J. conceived and designed the study. J.J.Y. and H.Y.K. performed the animal experiments. J.J., S.M.L., J.J.Y., H.Y.K., Y.M.A., S.S., and D.S.C. analyzed and interpreted the data. J.J., S.M.L., and H.Y.K. drafted the manuscript. S.M.L., D.S.C., J.J.Y., and Y.M.A. revised and edited the manuscript. J.J. and H.S.L. supervised the study. All authors read and approved the final manuscript.

Data availability

Raw data from this study are accessible in the Sequence Read Archive (SRA) under BioProject ID PRJNA1152724. These data are publicly available for further analysis and research.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41514-026-00334-4.